Untitled

in this issue 

© 2010 Nature America, Inc. All rights reserved. 

Sterile bollworms bolster Bt control 

Evolution of resistance 

by insect pests 

threatens to cut short 

the benefits of crops 

engineered to produce 

insecticidal proteins 

from the bacterium 

Bacillus thuringiensis 

(Bt). Tabashnik and 

colleagues use computer 

simulations and data 

from an extensive 4-year field trial to demonstrate that a multitactical 

program, including releases of sterile pink bollworm 

moths, can thwart the emergence of the pest’s resistance to 

Bt cotton. Although the sterile insect technique is not new, 

this is the first time it has been used jointly with a transgenic 

crop. From 2006 to 2009, 8 billion pink bollworm caterpillars 

were reared to maturity (pictured), and moths sterilized by 

irradiation were released over the cotton fields of Arizona. This 

substantially increased the likelihood that any rare resistant 

moths emerging from Bt cotton would mate with sterile moths, 

blocking their reproduction. The results were no increase in 

pink bollworm resistance to Bt cotton, a 99.9% decrease in 

the size of Arizona’s pink bollworm population and a sharp 

reduction in the use of insecticide sprays. It remains to be seen 

whether the sterile insect technique will be as effective for 

other combinations of Bt crops and the insect pests that they 

target. [Letters, p. 1304; News and Views, p. 1273] PH 

Synthesizing DNA aplenty 

Long synthetic DNA 

molecules have been 

assembled from inexpensive 

short oligonucleotides 

created on 

a microarray, but this 

approach has not been 

widely adopted owing 

to high error rates 

and limited scalability. 

Studies by two groups working with George Church address these 

problems. They describe clever approaches to selecting subsets of 

oligos from the complex pool released from the microarray. Without 

performing this selection step, the oligos from a microarray are not 

abundant enough and may cross-hybridize with each other, interfering 

with the assembly process. Kosuri et al. select subpools of oligos 

using PCR, and they lower error rates using high-fidelity microarrays 

Written by Kathy Aschheim, Michael Francisco, Peter Hare, 

Brady Huggett, Craig Mak & Lisa Melton 

and optimized assembly protocols. Matzas et al. describe a novel 

use of next-generation sequencing, reducing error rates and selecting 

oligos simultaneously by reading oligos with a high-throughput 

sequencer to identify the ones that do not have errors and then manually 

picking these off the sequencing instrument with a micropipette. 

The authors use their methods to synthesize a variety of functioning 

proteins, including fluorescent proteins and therapeutic antibody 

fragments. The two approaches each reduce DNA synthesis 

costs to


in this issue 

urine. The results of this study could help in designing nanoparticles 

with desired biodistribution properties after inhalation. 

[Letters, p. 1300; News and Views, p. 1275] 

KA 

Patent roundup 

Identifying and isolating DNA without further manipulation does 

not constitute an invention and is not patentable, a US district 

court said in October. The court stated its position in response to a 

lawsuit involving Myriad Genetics’ (Salt Lake City, UT, USA) breast 

cancer genes BRCA1 and BRCA2. In a Commentary, Kenneth 

Chahine discusses a new framework for assessing the eligibility of 

DNA patents [News, p. 1226; Commentary, p. 1251] LM 

Patent owners who can claim a humanitarian use for their 

patents will be offered a fast-track re-examination voucher 

under a system proposed by the US Patent and Trademark 

Office. [News, p. 1226] 

LM 

‘Blocking patents’ have killed many a startup. Here is advice on 

how to deal with them. [Building a Business, p. 1239] BH 

A whole host of approaches fall under the rubric ‘therapeutic 

nanoparticles’. Burgess et al. outline the reasons why innovative 

pharmaceutical products based on therapeutic nanoparticles offer 

particular advantages in terms of intellectual property protection 

from generic competition. [Patent Article, p. 1267] 

MF 

Recent patent applications in pharmacogenomics. 

[New patents, p. 1271] 

MF 

Minicircles for all 

Transgenes are often ferried into cells on plasmids, but an efficient way 

of producing minicircle DNA could make minicircles the method of 

choice for delivering exogenous genes to mammalian cells. Minicircles 

are circular expression vectors that lack the plasmid backbone. Longterm 

transgene expression from minicircles is much higher than from 

minicircles, but broad adoption of this technology has been hindered 

by a cumbersome production process. Kay and colleagues present 

an optimized protocol that yields purified minicircles in about the 

same time frame as that required for routine plasmid preparation. 

[Brief Communications, p. 1287] 

KA 

Next month in 

• Safe harbors in human iPS cells 

• Genome-wide profiling of cytosine hydroxymethylation 

• Experimental halotype phasing 

• Improving in vivo shRNA screens with fluorescent 

reporters 

• Genetic modification of rodents with zinc finger 

nucleases 

viii 

volume 28 number 12 December 2010 nature biotechnology


www.nature.com/naturebiotechnology 

EDITORIAL OFFICE 

biotech@us.nature.com 

75 Varick Street, Fl 9, New York, NY 10013-1917 

Tel: (212) 726 9200, Fax: (212) 696 9635 

Chief Editor: Andrew Marshall 

Senior Editors: Laura DeFrancesco (News & Features), Kathy Aschheim (Research), 

Peter Hare (Research), Michael Francisco (Resources and Special Projects) 

Business Editor: Brady Huggett 

Associate Business Editor: Victor Bethencourt 

News Editor: Lisa Melton 

Associate Editors: Markus Elsner (Research), Craig Mak (Research) 

Editor-at-Large: John Hodgson 

Contributing Editors: Mark Ratner, Chris Scott 

Contributing Writer: Jeffrey L. Fox 

Senior Copy Editor: Teresa Moogan 

Managing Production Editor: Ingrid McNamara 

Production Editor: Amanda Crawford 

Senior Illustrator: Katie Vicari 

Illustrator: Marina Corral 

Cover design: Erin DeWalt 

Senior Editorial Assistant: Ania Levinson 

MANAGEMENT OFFICES 

NPG New York 

75 Varick Street, Fl 9, New York, NY 10013-1917 

Tel: (212) 726 9200, Fax: (212) 696 9006 

Publisher: Melanie Brazil 

Exectutive Editor: Veronique Kiermer 

Chief Technology Officer: Howard Ratner 

Head of Nature Research & Reviews Marketing: Sara Girard 

Circulation Manager: Stacey Nelson 

Production Coordinator: Diane Temprano 

Head of Web Services: Anthony Barrera 

Senior Web Production Editor: Laura Goggin 

NPG London 

The Macmillan Building, 4 Crinan Street, London N1 9XW 

Tel: 44 207 833 4000, Fax: 44 207 843 4996 

Managing Director: Steven Inchcoombe 

Publishing Director: Peter Collins 

Editor-in-Chief, Nature Publications: Philip Campbell 

Marketing Director: Della Sar 

Director of Web Publishing: Timo Hannay 

NPG Nature Asia-Pacific 

Chiyoda Building, 2-37 Ichigayatamachi, Shinjuku-ku, Tokyo 162-0843 

Tel: 81 3 3267 8751, Fax: 81 3 3267 8746 

Publishing Director — Asia-Pacific: David Swinbanks 

Associate Director: Antoine E. Bocquet 

Manager: Koichi Nakamura 

Operations Director: Hiroshi Minemura 

Marketing Manager: Masahiro Yamashita 

Asia-Pacific Sales Director: Kate Yoneyama 

Asia-Pacific Sales Manager: Ken Mikami 

DISPLAY ADVERTISING 

display@us.nature.com (US/Canada) 

display@nature.com (Europe) 

nature@natureasia.com (Asia) 

Global Head of Advertising and Sponsorship: Dean Sanderson, Tel: (212) 726 9350, 

Fax: (212) 696 9482 

Global Head of Display Advertising and Sponsorship: Andrew Douglas, Tel: 44 207 843 4975, 

Fax: 44 207 843 4996 

Asia-Pacific Sales Director: Kate Yoneyama, Tel: 81 3 3267 8765, Fax: 81 3 3267 8746 

Display Account Managers: 

New England: Sheila Reardon, Tel: (617) 399 4098, Fax: (617) 426 3717 

New York/Mid-Atlantic/Southeast: Jim Breault, Tel: (212) 726 9334, Fax: (212) 696 9481 

Midwest: Mike Rossi, Tel: (212) 726 9255, Fax: (212) 696 9481 

West Coast: George Lui, Tel: (415) 781 3804, Fax: (415) 781 3805 

Germany/Switzerland/Austria: Sabine Hugi-Fürst, Tel: 41 52761 3386, Fax: 41 52761 3419 

UK/Ireland/Scandinavia/Spain/Portugal: Evelina Rubio-Hakansson, Tel: 44 207 014 4079, 

Fax: 44 207 843 4749 

UK/Germany/Switzerland/Austria: Nancy Luksch, Tel: 44 207 843 4968, Fax: 44 207 843 4749 

France/Belgium/The Netherlands/Luxembourg/Italy/Israel/Other Europe: Nicola Wright, 

Tel: 44 207 843 4959, Fax: 44 207 843 4749 

Asia-Pacific Sales Manager: Ken Mikami, Tel: 81 3 3267 8765, Fax: 81 3 3267 8746 

Greater China/Singapore: Gloria To, Tel: 852 2811 7191, Fax: 852 2811 0743 

NATUREJOBS 

naturejobs@us.nature.com (US/Canada) 

naturejobs@nature.com (Europe) 

nature@natureasia.com (Asia) 

US Sales Manager: Ken Finnegan, Tel: (212) 726 9248, Fax: (212) 696 9482 

European Sales Manager: Dan Churchward, Tel: 44 207 843 4966, Fax: 44 207 843 4596 

Asia-Pacific Sales & Business Development Manager: Yuki Fujiwara, Tel: 81 3 3267 8765, 

Fax: 81 3 3267 8752 

SPONSORSHIP 

g.preston@nature.com 

Global Head of Sponsorship: Gerard Preston, Tel: 44 207 843 4965, Fax: 44 207 843 4749 

Business Development Executive: David Bagshaw, Tel: (212) 726 9215, Fax: (212) 696 9591 

Business Development Executive: Graham Combe, Tel: 44 207 843 4914, Fax: 44 207 843 4749 

Business Development Executive: Reya Silao, Tel: 44 207 843 4977, Fax: 44 207 843 4996 

SITE LICENSE BUSINESS UNIT 

Americas: Tel: (888) 331 6288 

institutions@us.nature.com 

Asia/Pacific: Tel: 81 3 3267 8751 

institutions@natureasia.com 

Australia/New Zealand: Tel: 61 3 9825 1160 

nature@macmillan.com.au 

India: Tel: 91 124 2881054/55 

npgindia@nature.com 

ROW: Tel: 44 207 843 4759 

institutions@nature.com 

CUSTOMER SERVICE 

www.nature.com/help 

Senior Global Customer Service Manager: Gerald Coppin 

For all print and online assistance, please visit www.nature.com/help 

Purchase subscriptions: 

Americas: Nature Biotechnology, Subscription Dept., 342 Broadway, PMB 301, New York, NY 10013- 

3910, USA. Tel: (866) 363 7860, Fax: (212) 334 0879 

Europe/ROW: Nature Biotechnology, Subscription Dept., Macmillan Magazines Ltd., Brunel Road, 

Houndmills, Basingstoke RG21 6XS, United Kingdom. Tel: 44 1256 329 242, Fax: 44 1256 812 358 

Asia-Pacific: Nature Biotechnology, NPG Nature Asia-Pacific, Chiyoda Building, 

2-37 Ichigayatamachi, Shinjuku-ku, Tokyo 162-0843. Tel: 81 3 3267 8751, Fax: 81 3 3267 8746 

India: Nature Biotechnology, NPG India, 3A, 4th Floor, DLF Corporate Park, Gurgaon 122002, India. 

Tel: 91 124 2881054/55, Tel/Fax: 91 124 2881052 

REPRINTS 

reprints@us.nature.com 

Nature Biotechnology, Reprint Department, Nature Publishing Group, 75 Varick Street, Fl 9, 

New York, NY 10013-1917, USA. 

For commercial reprint orders of 600 or more, please contact: 

UK Reprints: Tel: 44 1256 302 923, Fax: 44 1256 321 531 

US Reprints: Tel: (617) 494 4900, Fax: (617) 494 4960

volume 28 number 12 DECEMBER 2010 

editorial 

1221 Vive la différence? 


Oligonucleotides generated on a 

microarray provide an abundant source 

of raw material for assembling long 

synthetic DNA molecules. Kosuri et al. 

and Matzas et al. reduce the error 

rate and improve the scalability 

of approaches that use microarray 

oligonucleotides to synthesize genes 

and other DNA elements (pp 1291 and 

1295). Credit: Marina Corral, based 

on idea courtesy of Sriram Kosuri and 

Mark Matzas. 

100 

80 

60 

40 

20 

0 

Weak 

Moderate 

Strong 

0 500 

1,000 

1,500 

Midpoint for antibody-based 

proteomics initiative, p 1248 

news 

1223 Will J&J turbocharge Crucell’s vaccine portfolio? 

1224 Amylin’s diabetes shock 

1225 Synthetic DNA firms embrace hazardous agents guidance but remain wary of 

automated ‘best-match’ 

1226 DNA not patentable 

1226 USPTO’s humanitarian vouchers 

1227 BARDA funds vaccine makers aiming to phase out eggs 

1229 Ethanol blend hike promises jump start for cellulosic investment 

1229 Singapore injects $12.5 billion 

1230 Good ideas across borders 

1230 Airlines ahead on algae 

1232 News feature: Crossing the line 

1236 News feature: New twists on proteasome inhibitors 

Bioentrepreneur 

Building a business 

1239 Around the block 

Y Philip Zhang 

opinion and comment 

CORRESPONDENCE 

1242 Wrong fixes for gene patents 

1243 Stem cell clinics in the news 

1246 Tracking and assessing the rise of state-funded stem cell research 

1248 Towards a knowledge-based Human Protein Atlas 

Nature Biotechnology (ISSN 1087-0156) is published monthly by Nature Publishing Group, a trading name of Nature America Inc. located at 75 Varick Street, 

Fl 9, New York, NY 10013-1917. Periodicals postage paid at New York, NY and additional mailing post offices. Editorial Office: 75 Varick Street, Fl 9, New York, 

NY 10013-1917. Tel: (212) 726 9335, Fax: (212) 696 9753. Annual subscription rates: USA/Canada: US$250 (personal), US$3,520 (institution), US$4,050 

(corporate institution). Canada add 5% GST #104911595RT001; Euro-zone: €202 (personal), €2,795 (institution), €3,488 (corporate institution); Rest of world 

(excluding China, Japan, Korea): £130 (personal), £1,806 (institution), £2,250 (corporate institution); Japan: Contact NPG Nature Asia-Pacific, Chiyoda Building, 

2-37 Ichigayatamachi, Shinjuku-ku, Tokyo 162-0843. Tel: 81 (03) 3267 8751, Fax: 81 (03) 3267 8746. POSTMASTER: Send address changes to Nature 

Biotechnology, Subscriptions Department, 342 Broadway, PMB 301, New York, NY 10013-3910. Authorization to photocopy material for internal or personal 

use, or internal or personal use of specific clients, is granted by Nature Publishing Group to libraries and others registered with the Copyright Clearance Center 

(CCC) Transactional Reporting Service, provided the relevant copyright fee is paid direct to CCC, 222 Rosewood Drive, Danvers, MA 01923, USA. Identification 

code for Nature Biotechnology: 1087-0156/04. Back issues: US$45, Canada add 7% for GST. CPC PUB AGREEMENT #40032744. Printed by Publishers 

Press, Inc., Lebanon Junction, KY, USA. Copyright © 2010 Nature America, Inc. All rights reserved. Printed in USA. 

i


COMMENTARY 

1251 Anchoring gene patent eligibility to its constitutional mooring 

Kenneth G Chahine 

1256 The environmental impact subterfuge 

Gregory Conko & Henry I Miller 


Curbing pink bollworm resistance 

to Bt, p 1273 

Simulating metabolic interactions 

between cells, p 1279 

feature 

1259 To selectivity and beyond 

George S Mack 

patents 

1267 On firm ground: IP protection of therapeutic nanoparticles 

Paul Burgess, Peter Barton Hutt, Omid C Farokhzad, Robert Langer, Scott Minick & 

Stephen Zale 

1271 Recent patent applications in pharmacogenomics 

NEWS AND VIEWS 

1272 Megabases for kilodollars 

Mikkel Algire, Radha Krishnakumar & Chuck Merryman 

see also pp 1291 and 1295 

1273 No refuge for insect pests 

Kongming Wu see also p 1304 

1275 Nanoparticles in the lung 

Wolfgang G Kreyling, Stephanie Hirn & Carsten Schleh see also p 1300 

1277 Research highlights 

computational biology 

Analysis 

1279 Large-scale in silico modeling of metabolic interactions between cell types in the 

human brain 

Nathan E Lewis, Gunnar Schramm, Aarash Bordbar, Jan Schellenberger, 

Michael P Andersen, Jeffrey K Cheng, Nilam Patel, Alex Yee, Randall A Lewis, 

Roland Eils, Rainer König & Bernhard Ø Palsson 

research 

BRIEF COMMUNICATIONS 

1287 A robust system for production of minicircle DNA vectors 

Mark A Kay, Cheng-Yi He & Zhi-Ying Chen 

Combining DNA reading and writing, 

p 1291 

letters 

1291 High-fidelity gene synthesis by retrieval of sequence-verified DNA identified using 

high-throughput pyrosequencing 

Mark Matzas, Peer F Stähler, Nathalie Kefer, Nicole Siebelt, Valesca Boisguérin, 

Jack T Leonard, Andreas Keller, Cord F Stähler, Pamela Häberle, 

Baback Gharizadeh, Farbod Babrzadeh & George M Church see also p 1272 

nature biotechnology 

iii


1295 Scalable gene synthesis by selective amplification of DNA pools from high-fidelity 

microchips 

Sriram Kosuri, Nikolai Eroshenko, Emily M LeProust, Michael Super, Jeffrey Way, 

Jin Billy Li & George M Church see also p 1272 

1300 Rapid translocation of nanoparticles from the lung airspaces to the body 

Hak Soo Choi, Yoshitomo Ashitate, Jeong Heon Lee, Soon Hee Kim, Aya Matsui, 

Numpon Insin, Moungi G Bawendi, Manuela Semmler-Behnke, John V Frangioni & 

Akira Tsuda see also p 1275 

Scalable DNA synthesis, p 1295 

1304 Suppressing resistance to Bt cotton with sterile insect releases 

Bruce E Tabashnik, Mark S Sisterson, Peter C Ellsworth, Timothy J Dennehy, 

Larry Antilla, Leighton Liesner, Mike Whitlow, Robert T Staten, Jeffrey A Fabrick, 

Gopalan C Unnithan, Alex J Yelich, Christa Ellers-Kirk, Virginia S Harpold, 

Xianchun Li & Yves Carrière see also p 1273 

1308 errata and corrigenda 


Nanoparticle behavior in the lungs, 

p 1300 

careers and recruitment 

1309 Catching the wave in China 

Connie Johnson Hambley 

1312 people 

ADVERTISEMENT 

Modern evolution of the ideal science park 

Since their inception, science parks have become much more than 

shelter for technical industries. Science parks are now a feature of the 

global landscape and their numbers are still growing in the 21 st century. 

The report takes a look at the policies and dynamics that help to shape 

the modern science parks, especially those policies that facilitate growth 

of the biotech industry. The report follows Errata and Corrigenda after 

page 1308. 


v

Editorial 

Vive la différence? 

The Obama administration’s $1 billion tax program at the very least signals a continued commitment to innovative 

biotech. The same cannot be said of plans afoot by the French government. 


At the beginning of November, the US Therapeutic Discovery 

Project Program (TDPP) handed out a billion dollars in tax credits/grants 

to small- to medium-sized (SME) companies in the life 

sciences. The program was overseen by the federal tax authorities 

and was a direct result of input by BIO, the Biotechnology Industry 

Organization. Although the amounts handed out to each company 

were rather small, the Obama administration should be congratulated 

on providing decisive support to the US biotech sector. But perhaps 

most important, this latest initiative continues decades of fiscal policy 

under different US administrations that has consistently supported the 

entrepreneurial life science sector. 

There is little doubt that the US biotech sector (and that in most 

other countries) is in need of a shot in the arm. BIO estimates that 

since 2007, more than 100 public biotech companies have closed their 

doors and countless more private companies have ceased operations. 

The percentage of companies with less than one year’s cash remains 

25% (down only slightly from 29% in 2007). 

TDPP was a one-off program, approved as part of the 2010 Patient 

Protection and Affordable Care Act, designed specifically for companies 

with fewer than 250 employees. The credits cover up to 50% of 

eligible R&D expenses in tax years 2009 or 2010. Applications were 

accepted from companies for one month from June 21 and the submissions 

reviewed by the US National Institutes of Health (NIH). 

The program was a roaring success—almost too successful. The US 

Internal Revenue Service had expected ~1,500 applications; it was deluged 

with over 5,600. To cope with this onslaught, a decision was made 

to turn the $1 billion cake into 4,606 crumbs, each worth a maximum 

of $244,479. Although some companies applied for and were granted 

multiple project awards (28% of total), 2,923 organizations in 47 states 

got a piece of the action. 

Project reviewers for TDPP were asked to assess, among a very few 

other criteria, whether a project was ‘likely’ to result in new therapies, 

or reduce long-term healthcare costs or contribute to the goal of curing 

cancer. The fact that 4,000 projects got the thumbs-up on this 

criterion at the very least severely stretches the meaning of ‘likely’. The 

best that can realistically be claimed is that it is likely that a handful 

of the projects will succeed, and that the chances of success in every 

project will be marginally enhanced by an extra couple of hundred 

thousand dollars. 

That said, by casting a wide net, the US authorities have ensured that 

their biotech sector retains more shots on the goal—a wise investment 

in the future. 

The same cannot be said for policy proposals under discussion in 

France. 

Back in September, the government of Nicolas Sarkozy introduced 

its 2011 finance bill in the French parliament. Amongst the cost-cutting 

measures and edicts designed to cut back on government spending 

was a small section, Article 78, which will result in an estimated annual 

saving to the French exchequer of $57 million. 

Article 78 alters radically the conditions attached to the Jeune 

Enterprise Innovante (JEI; Young Innovative Company) status in 

France. Currently, any research-based independent SME under 8 

years old is, in effect, excused from paying social taxes for any of its 

researcher employees. Because the social taxes represent 30–40% of 

salary costs, this represents a significant saving and an incentive to 

life science venture investment in the country (Nat. Biotechnol. 23, 

1187, 2005). Article 78 proposes to cap the savings at €103,680 per 

year per company and to tail off the benefit between years 4 and 8 of 

a company’s lifetime. 

A vote in the French Senate on the 2011 Finance Bill is expected 

early this month, with implementation in January. Meanwhile, the 

French industry association, France Biotech is furiously lobbying 

ministers, senators and senior figures in the civil administration in 

a desperate bid to axe Article 78. They fear, with justification, that it 

will discourage entrepreneurs and investors, and worse, that it will be 

a disincentive to innovate. 

What is most disingenuous about the French policy shift, though, 

is that the government is saying the benefit will actually provide 

SMEs with a ‘softer landing’ before returning to full taxation. In other 

words, they are claiming to understand the long-term effects of this 

volte face in policy. And when it comes to stargazing, the French government 

doesn’t have a good track record in biotech. Twenty years 

ago, the ‘BioAvenir’ initiative invested a billion francs in academic 

life science projects, but elected to make one (French) company— 

Rhone-Poulenc—the sole beneficiary of any commercialization. We 

don’t know what happened to the hundreds of BioAvenir projects, but 

Rhone Poulenc didn’t turn into the biotech driver the government 

envisaged—it is now known as Sanofi-aventis. 

As the BioAvenir experience illustrates, if there is one thing that 

can be guaranteed about the future, it is that it will not look like the 

present. That is something the US government seems to have grasped 

with TDPP, arguably to a fault: its $1 billion has been flung into any 

company that is both innovative and not established, into any company 

that might be the future simply by virtue of its not being a big part of 

the present. 

The French government, conversely, still clings to the idea that it can 

extrapolate the future from the present: each year, it gives an estimated 

€1 billion in R&D tax credits to Sanofi-aventis. There are worse places 

to spend a €1 billion, no doubt, but providing such largesse to one 

of the world’s biggest pharmaceutical companies while begrudging 

a fraction ($57 million) of it to innovative biotech doesn’t look like a 

government investing in the future. 

nature biotechnology volume 28 number 12 december 2010 1221

in this section 

Screening for 

malicious use of 

synthetic DNA 

p1225 

BARDA funds 

boost creative 

vaccine makers 

p1227 

news 

Ethanol blend 

for cars rises after 

35 years p1229 

Will J&J turbocharge Crucell’s vaccine portfolio? 


The emergence of contamination at Crucell’s 

production facility in Shingal, Yongin City, 

Korea, came at an awkward time. The vaccine 

maker, based in Leiden, The Netherlands, 

disclosed the news in late October as Johnson 

& Johnson (J&J) prepared to issue an update 

on progress of its $2.3 billion bid for outright 

ownership of the company. Nevertheless, J&J, 

of New Brunswick, New Jersey, intends to 

proceed according to plan. The large pharma’s 

tender offer for Crucell’s shares is due to 

be considered at a meeting of Crucell shareholders 

in Amsterdam on 10 December. The 

implications—if any—that the manufacturing 

disruption may have for the transaction are 

not clear at this point. “I don’t expect this to be 

a deal breaker, but there’s no guarantee it won’t 

be a deal adjuster,” says Jan Van den Bossche, 

analyst at Petercam in Brussels. “It will all 

depend on what’s going on out there.” 

Although shipments of the two products 

affected by the quality failures—the pediatric 

liquid pentavalent combination vaccine 

Quinvaxem (diphtheria (Corynebacterium 

diphtheriae), toxoid/tetanus (Clostridium 

tetani), toxoid/whole cell Bordetella pertussis 

(DTwP); Haemophilus influenzae type B (Hib) 

conjugate vaccine; and recombinant hepatitis 

B virus surface antigen (HBsAg) vaccine; 

trade name, Hepavax-Gene)—were, as Nature 

Biotechnology went to press, expected to 

resume by late November, full production will 

not be restored until February. Quinvaxem, 

which protects against five pediatric infections, 

is the company’s most important product, having 

secured tenders worth $910 million since 

the end of 2006, including a $110 million order 

from the New York–based United Nations 

Children’s Fund in May. Crucell has made 

provision for a €22.8 million ($31.8 million) 

charge on the inventory lost through the disruption. 

Even before the problems emerged, 

the company had been in the process of transferring 

production of the two vaccines to a new 

€50 million ($69.8 million) facility in Incheon, 

South Korea, which is scheduled to become 

fully operational next year. 

Crucell’s problems echo those of Shantha 

Biotechnics, of Hyderabad, India. Not 

long after Lyon-based Sanofi Pasteur 

The Dutch vaccine manufacturer Crucell (headquarters in Leiden pictured above) will run as an 

autonomous unit following J&J’s expected acquisition. 

gained an 80% stake in Shantha, the latter 

firm’s Shan5 (DTwP-HBsAg-Hib) vaccine, 

a direct competitor to Quinvaxem, 

lost its accreditation on the World Health 

Organization’s (Geneva) list of recommended 

vaccines because of the appearance of white 

sediment on vaccine vials. “They’ve had 

massive manufacturing setbacks,” says Peter 

Welford, analyst at Jefferies International, in 

London. “Crucell [has] had the least problems.” 

Others competing in this product 

segment include Panacea Biotech, of New 

Delhi, India, and the GSK Biologicals arm 

of London-based GlaxoSmithKline. 

J&J’s interest in the Dutch company came 

to the fore in September last year, when it paid 

around €302 ($407) million, or €20.6 ($27.8) 

per share, to amass an 18% stake in Crucell. 

The healthcare giant’s current $2.3 billion bid 

for the rest of the company it does not already 

own is well in excess of a reported $1.35 billion 

bid that Wyeth of Madison, New Jersey, 

was preparing early last year. That deal was 

derailed by New York–based Pfizer’s $68 billion 

acquisition of Wyeth. Recent contamination 

problems are unlikely to affect J&J’s offer, 

although Welford says, this will eliminate the 

possibility that dissenting shareholders may 

hold out for a higher price than the €24.75 

($35.18) per share that J&J has tabled. 

Crucell’s history can be traced back to Leidenbased 

IntroGene, which was formed in 1993 to 

commercialize the PER.C6 human cell culture 

technology, which can produce as much as 27 g/l 

of IgGs in bioreactor culture. The Crucell 

name came into being in 2000, after a merger 

with Utrecht Biotechnology Systems in The 

Netherlands, an antibody discovery firm that 

had developed a ‘subtractive’ phage display 

technology called MAbstract. In that platform, 

a phage antibody library is admixed with a 

heterogeneous cell mixture, where the particular 

cell of interest is labeled with a fluorescent 

tag and sorted by flow cytometry to identify 

binding phage antibodies, dispensing with the 

need for repeatedly constructing new libraries. 

These technologies, as well as an AdVac adenoviral 

vector platform, underpin the company’s 

development pipeline. However, its commercial 

product portfolio is based on recent acquisitions 

of Stockholm-based SBL Vaccin, and, especially, 

of Bern, Switzerland–based Berna Biotech. 

Berna had previously acquired Rhein Biotech, 

of Maastricht, The Netherlands, but never 

managed to achieve its stated growth ambitions 

(Table 1). Crucell, in contrast, has proven adept 

at winning sales, particularly in emerging markets. 

“They’ve done a marvelous job of turning 

around Berna Biotech,” Van den Bossche says. 

J&J plans to run Crucell as an autonomous 

Crucell 


NEWS 


in brief 

Amylin’s 

diabetes 

shock 

In a major 

setback for 

Amylin, the US 

Food and Drug 

Fears over long QT. Administration 

called for 

further tests on Amylin’s Bydureon, a onceweekly 

drug for type 2 diabetes. On October 

19, the San Diego-based biotech Amylin, and 

partner Eli Lilly of Indianapolis, announced that 

the agency had issued a Complete Response 

Letter calling for a study known as thorough 

QT to further investigate the potential effect of 

the drug’s cardiac effects. Bydureon is a longacting 

formulation of Amylin’s approved drug 

Byetta (exenatide), an analog of the insulinboosting 

glucagon-like peptide 1 (GLP-1) 

(Nat. Biotechnol. 28, 109, 2010). Both 

contain exenatide, but Bydureon contains 

a higher concentration of the active agent, 

delivered into the blood by controlled release. 

The first-in-class synthetic gut hormone drug 

Byetta generated $677 million in 2009 in the 

US alone. But Bydureon’s formulation—once 

weekly injections compared with twice a day—is 

predicted to gain competitive edge over its 

predecessor if approved. Simos Simeonidis, 

managing director and senior biotech analyst 

at Rodman & Renshaw in New York, says that 

Bydureon had “blockbuster” potential and could 

have brought in at least $1 billion per year 

for Amylin after the first few years. The FDA’s 

concerns over supratherapeutic concentrations 

of exenatide may have been triggered by 

observations in a Byetta study, in which a few 

patients with blood levels of 500 pg/ml had 

lengthened QT intervals. In patients with kidney 

problems, once-weekly Bydureon can reach 

five times the normal 200 pg/ml seen with 

Byetta. Given the delay in approving Bydureon, 

alternate treatments for type 2 diabetes like 

Victoza (liraglutide) from Novo Nordisk, of 

Princeton, New Jersey, and Amylin’s own Byetta, 

are now expected to flourish. London-based 

GlaxoSmithKline’s investigational Syncria(R) 

(albiglutide), also a GLP-1 analog, is currently 

undergoing phase 3 clinical trials. “They were 

considerably behind,” says Yaron Werber, senior 

biotech analyst at Citigroup in New York. “But 

now I think they have a chance to close a big 

part of the gap.” 

Nidhi Subbaraman 

iStockphoto/Andrius Gruzdaitis 

in their words 

“At this juncture, I don’t know that individuals 

will have all that much to gain from publishing 

their data, but I think that Genomes Unzipped 

will help to prove that there’s not all that much 

to lose, either.” Linda Avey, cofounder of 

23andMe, comments on a move by 11 Britishbased 

scientists and a US lawyer to make their 

own genetic tests publicly available in the hope 

of encouraging others to share their genome 

information. (The Times, 11 October 2010) 

Table 1 Vaccine mergers & acquisitions 

Acquirer Target (location) Value Year 

Berna Biotech Rhein Biotech $257 million 2002 

Crucell Berna Biotech $449 million 2005 

Crucell SBL Vaccin $52 million 2006 

GlaxoSmithKline ID Biomedical (Vancouver, Canada) $1.4 billion 2005 

GlaxoSmithKline Corixa (Seattle) $300 million 2005 

Intercell Iomai (Gaithersburg, Maryland) $189 million 2008 

Novartis Chiron (Emeryville, California) $5.4 billion a 2005 

Sanofi Pasteur Acambis $549 million 2008 

Sanofi Pasteur Shantha Biotechnics $781 million b 2009 

Sanofi Pasteur VaxDesign (Orlando, Florida) $55 million c 2010 

a The transaction involved the remaining 58% of Chiron’s stock not already held by Novartis. b Total valuation of the company implied 

by the terms of the deal. Sanofi Pasteur acquired an 80% stake. c The deal includes $5 million in potential milestone payments. 

unit, just as it has with several earlier acquisitions, 

such as antiviral drug developer Tibotec 

of Antwerp, Belgium, and antibody developer 

Centocor located in Horsham, Pennsylvania. 

J&J’s imminent acquisition of Crucell is further 

evidence that the vaccines area, although 

small in the context of overall pharma sales, is 

firmly back on the industry’s agenda, after a 

couple of decades during which it had fallen out 

of favor. Other acquisitions where vaccines have 

featured prominently (though not exclusively) 

are Pfizer’s takeover of Wyeth and Londonbased 

AstraZeneca’s $15.2 billion acquisition of 

Gaithersburg, Maryland–based MedImmune. 

The vaccines market’s steady growth equilibrium 

is punctuated by irregular spurts 

caused by significant new product introductions. 

For instance, Wyeth’s Prevnar (pneumococcal 

7-valent conjugate) or Gardasil (human 

papillomavirus quadrivalent (types 6, 11, 16 

and 18) recombinant vaccine) from Merck 

of Whitehouse Station, New Jersey, quickly 

added substantial chunks of revenue to the 

industry total. 

The complexities of manufacturing, combined 

with the economies of large-scale production, 

have conspired to make the sector 

strongly oligopolistic, with the market dominated 

by a small number of firms. The top 

five players, Sanofi Pasteur, GSK Biologicals, 

Merck, Pfizer and Novartis, control around 

85% of the market, with around $18 billion 

of sales in 2009, according to market analysts 

VacZine Analytics, of Bishop’s Stortford, 

UK. “The industry last year grew by about 

9, 10% over the previous year,” says VacZine 

Analytics director John Savopoulos. By his 

reckoning Crucell’s 2009 vaccine sales of €304 

million ($424.3 million) would rank the company 

in ninth place, behind CSL, of Parkville, 

Australia, and London-based AstraZeneca’s 

MedImmune unit. The H1N1 influenza pandemic 

had the biggest impact on sales in 2009, 

bringing in around $3.5 billion in additional 

revenue. In the first three-quarters of 2010, 

the top five companies posted combined sales 

of $16.9 billion, excluding the contribution of 

Sanofi Pasteur MSD, a European joint venture 

between Sanofi Pasteur and Merck. 

Nevertheless, the industry’s pipeline remains 

widely distributed. “Seventy companies are now 

targeting 40 plus pathogens,” Savopoulos says. 

Some 160 vaccines are in clinical development, 

around 100 of which are in phase 1 trials. Crucell 

has the industry’s fifth largest pipeline, he notes. 

It has clinical-stage programs in yellow fever, 

tuberculosis (TB), malaria, HIV, Ebola virus 

and Marburg virus as well as preclinical efforts 

in seasonal influenza, respiratory syncytial 

virus and in the development of a universal flu 

vaccine. Only GSK Biologicals, Sanofi Pasteur, 

Novartis and Merck have more vaccine programs 

in the clinic. “They look reasonably well 

positioned. The problem is when you probability-adjust 

their pipeline—malaria, TB—they’re 

tough areas to be in,” Savopoulos says. A preclinical 

antibody development program, which 

targets both seasonal and pandemic flu strains, 

and has attracted $40.7 million in US National 

Institutes’ of Health funding (with potentially 

$28.4 million more to come), was the main 

focus of J&J’s 2009 alliance with Crucell. “That 

technology, if it can be harnessed for an active 

vaccine, that’s a quantum leap above the existing 

flu market,” Savopoulos points out. Jan Van den 

Bossche identifies a combination monoclonal 

antibody development program for treating 

individuals exposed to rabies infection as particularly 

promising. “It’s a clear, understandable 

program,” he says. The program, in phase 2 trials, 

aims to replace an existing equine immunoglobulin 

therapy. 

Although J&J will seek further growth from 

Crucell’s existing product portfolio, J&J’s real 

measure of success will be in the extent to 

which it can convert pipeline promise into 

commercial reality. Sanofi Pasteur ‘turbocharged’ 

Acambis’s ChimeriVax platform 

when it acquired the Cambridge, UK–based 

firm, says Savopoulos. “The question is, can 

J&J do the same thing?” 

Cormac Sheridan, Dublin 

1224 volume 28 number 12 december 2010 nature biotechnology

NEWS 


in brief 

Amylin’s 

diabetes 

shock 

In a major 

setback for 

Amylin, the US 

Food and Drug 

Fears over long QT. Administration 

called for 

further tests on Amylin’s Bydureon, a onceweekly 

drug for type 2 diabetes. On October 

19, the San Diego-based biotech Amylin, and 

partner Eli Lilly of Indianapolis, announced that 

the agency had issued a Complete Response 

Letter calling for a study known as thorough 

QT to further investigate the potential effect of 

the drug’s cardiac effects. Bydureon is a longacting 

formulation of Amylin’s approved drug 

Byetta (exenatide), an analog of the insulinboosting 

glucagon-like peptide 1 (GLP-1) 

(Nat. Biotechnol. 28, 109, 2010). Both 

contain exenatide, but Bydureon contains 

a higher concentration of the active agent, 

delivered into the blood by controlled release. 

The first-in-class synthetic gut hormone drug 

Byetta generated $677 million in 2009 in the 

US alone. But Bydureon’s formulation—once 

weekly injections compared with twice a day—is 

predicted to gain competitive edge over its 

predecessor if approved. Simos Simeonidis, 

managing director and senior biotech analyst 

at Rodman & Renshaw in New York, says that 

Bydureon had “blockbuster” potential and could 

have brought in at least $1 billion per year 

for Amylin after the first few years. The FDA’s 

concerns over supratherapeutic concentrations 

of exenatide may have been triggered by 

observations in a Byetta study, in which a few 

patients with blood levels of 500 pg/ml had 

lengthened QT intervals. In patients with kidney 

problems, once-weekly Bydureon can reach 

five times the normal 200 pg/ml seen with 

Byetta. Given the delay in approving Bydureon, 

alternate treatments for type 2 diabetes like 

Victoza (liraglutide) from Novo Nordisk, of 

Princeton, New Jersey, and Amylin’s own Byetta, 

are now expected to flourish. London-based 

GlaxoSmithKline’s investigational Syncria(R) 

(albiglutide), also a GLP-1 analog, is currently 

undergoing phase 3 clinical trials. “They were 

considerably behind,” says Yaron Werber, senior 

biotech analyst at Citigroup in New York. “But 

now I think they have a chance to close a big 

part of the gap.” 


iStockphoto/Andrius Gruzdaitis 


“At this juncture, I don’t know that individuals 

will have all that much to gain from publishing 

their data, but I think that Genomes Unzipped 

will help to prove that there’s not all that much 

to lose, either.” Linda Avey, cofounder of 

23andMe, comments on a move by 11 Britishbased 

scientists and a US lawyer to make their 

own genetic tests publicly available in the hope 

of encouraging others to share their genome 

information. (The Times, 11 October 2010) 

Table 1 Vaccine mergers & acquisitions 

Acquirer Target (location) Value Year 

Berna Biotech Rhein Biotech $257 million 2002 

Crucell Berna Biotech $449 million 2005 

Crucell SBL Vaccin $52 million 2006 

GlaxoSmithKline ID Biomedical (Vancouver, Canada) $1.4 billion 2005 

GlaxoSmithKline Corixa (Seattle) $300 million 2005 

Intercell Iomai (Gaithersburg, Maryland) $189 million 2008 

Novartis Chiron (Emeryville, California) $5.4 billion a 2005 

Sanofi Pasteur Acambis $549 million 2008 

Sanofi Pasteur Shantha Biotechnics $781 million b 2009 

Sanofi Pasteur VaxDesign (Orlando, Florida) $55 million c 2010 

a The transaction involved the remaining 58% of Chiron’s stock not already held by Novartis. b Total valuation of the company implied 

by the terms of the deal. Sanofi Pasteur acquired an 80% stake. c The deal includes $5 million in potential milestone payments. 

unit, just as it has with several earlier acquisitions, 

such as antiviral drug developer Tibotec 

of Antwerp, Belgium, and antibody developer 

Centocor located in Horsham, Pennsylvania. 

J&J’s imminent acquisition of Crucell is further 

evidence that the vaccines area, although 

small in the context of overall pharma sales, is 

firmly back on the industry’s agenda, after a 

couple of decades during which it had fallen out 

of favor. Other acquisitions where vaccines have 

featured prominently (though not exclusively) 

are Pfizer’s takeover of Wyeth and Londonbased 

AstraZeneca’s $15.2 billion acquisition of 

Gaithersburg, Maryland–based MedImmune. 

The vaccines market’s steady growth equilibrium 

is punctuated by irregular spurts 

caused by significant new product introductions. 

For instance, Wyeth’s Prevnar (pneumococcal 

7-valent conjugate) or Gardasil (human 

papillomavirus quadrivalent (types 6, 11, 16 

and 18) recombinant vaccine) from Merck 

of Whitehouse Station, New Jersey, quickly 

added substantial chunks of revenue to the 

industry total. 

The complexities of manufacturing, combined 

with the economies of large-scale production, 

have conspired to make the sector 

strongly oligopolistic, with the market dominated 

by a small number of firms. The top 

five players, Sanofi Pasteur, GSK Biologicals, 

Merck, Pfizer and Novartis, control around 

85% of the market, with around $18 billion 

of sales in 2009, according to market analysts 

VacZine Analytics, of Bishop’s Stortford, 

UK. “The industry last year grew by about 

9, 10% over the previous year,” says VacZine 

Analytics director John Savopoulos. By his 

reckoning Crucell’s 2009 vaccine sales of €304 

million ($424.3 million) would rank the company 

in ninth place, behind CSL, of Parkville, 

Australia, and London-based AstraZeneca’s 

MedImmune unit. The H1N1 influenza pandemic 

had the biggest impact on sales in 2009, 

bringing in around $3.5 billion in additional 

revenue. In the first three-quarters of 2010, 

the top five companies posted combined sales 

of $16.9 billion, excluding the contribution of 

Sanofi Pasteur MSD, a European joint venture 

between Sanofi Pasteur and Merck. 

Nevertheless, the industry’s pipeline remains 

widely distributed. “Seventy companies are now 

targeting 40 plus pathogens,” Savopoulos says. 

Some 160 vaccines are in clinical development, 

around 100 of which are in phase 1 trials. Crucell 

has the industry’s fifth largest pipeline, he notes. 

It has clinical-stage programs in yellow fever, 

tuberculosis (TB), malaria, HIV, Ebola virus 

and Marburg virus as well as preclinical efforts 

in seasonal influenza, respiratory syncytial 

virus and in the development of a universal flu 

vaccine. Only GSK Biologicals, Sanofi Pasteur, 

Novartis and Merck have more vaccine programs 

in the clinic. “They look reasonably well 

positioned. The problem is when you probability-adjust 

their pipeline—malaria, TB—they’re 

tough areas to be in,” Savopoulos says. A preclinical 

antibody development program, which 

targets both seasonal and pandemic flu strains, 

and has attracted $40.7 million in US National 

Institutes’ of Health funding (with potentially 

$28.4 million more to come), was the main 

focus of J&J’s 2009 alliance with Crucell. “That 

technology, if it can be harnessed for an active 

vaccine, that’s a quantum leap above the existing 

flu market,” Savopoulos points out. Jan Van den 

Bossche identifies a combination monoclonal 

antibody development program for treating 

individuals exposed to rabies infection as particularly 

promising. “It’s a clear, understandable 

program,” he says. The program, in phase 2 trials, 

aims to replace an existing equine immunoglobulin 

therapy. 

Although J&J will seek further growth from 

Crucell’s existing product portfolio, J&J’s real 

measure of success will be in the extent to 

which it can convert pipeline promise into 

commercial reality. Sanofi Pasteur ‘turbocharged’ 

Acambis’s ChimeriVax platform 

when it acquired the Cambridge, UK–based 

firm, says Savopoulos. “The question is, can 

J&J do the same thing?” 

Cormac Sheridan, Dublin 


Synthetic DNA firms embrace hazardous 

agents guidance but remain wary of 

automated ‘best-match’ 

news 


After more than four years of public and 

private discussion and review, the US government 

has officially issued its Screening 

Framework Guidance for Providers of Synthetic 

Double-Stranded DNA (http://www.phe. 

gov/Preparedness/legal/guidance/syndna/ 

Documents/syndna-guidance.pdf), a set of 

voluntary guidelines intended to help gene 

synthesis companies intercept unauthorized 

purchases of genetic components from human 

or agricultural pathogens. Many are relieved that 

standardized guidelines have finally been established, 

enabling the industry to harmonize practices 

and provide reassurance to large corporate 

clients that represent their bread and butter. 

“Overall it’s not a bad framework, and I 

think it’s been designed with a lot of expertise,” 

says Markus Fischer, director and cofounder 

of Entelechon in Regensburg, Germany, a gene 

synthesis provider, also part of the International 

Association Synthetic Biology (IASB), one of 

two major industry groups representing gene 

synthesis companies. In the absence of clear government 

guidance, the IASB and its counterpart, 

the International Gene Synthesis Consortium 

(IGSC), each developed their own ‘best practices’ 

for screening both DNA orders and the customers 

that place them. To draw up its official guideline, 

the US Department of Health and Human 

Services (HHS), in Washington, DC, received 

comments from 22 organizations and individuals 

since publishing its draft Guidance in November 

2009 (Table 1). The final version—released in 

October—includes only a few notable changes, 

such as the elimination of a size cut-off for screening 

decisions on double-stranded segments. 

Although the Guidance structurally resembles 

preexisting protocols, critics such as Stephen 

Maurer of the University of California at Berkeley 

are concerned that its recommendations are 

weaker than what is needed and may encourage 

companies to cut corners in the future. “Industry 

had embraced a higher standard, and now the 

government is going to lead us to a lower standard,” 

he says. Chief among his concerns is the 

proposed mechanism for screening sequences. 

The government proposes a ‘best match’ 

strategy, in which orders are compared against 

GenBank in 200-bp segments, based on both 

nucleotide and all six possible peptide sequences. 

If the top ‘hit’ is from a pathogen on the government’s 

list of select agents and toxins (http://www. 

selectagents.gov/Select%20Agents%20and%20 

Toxins%20List.html) or, for international orders, 

the ‘Commerce Control List’ (http://www.gpo. 

The select agents list omits many known human 

pathogens, such as the SARS coronavirus. 

gov/bis/ear/pdf/ccl1.pdf), it should be considered 

a ‘sequence of concern’ for further expert 

analysis, in conjunction with a careful assessment 

of the ordering customer’s credentials. 

Maurer suggests that by not expressly calling 

for human review of database matches— 

regardless of whether or not they are on the 

select agents list—this strategy is inherently 

less effective than the ‘top homology’ method 

already in use at several companies, including 

Entelechon, in which all GenBank results are 

manually assessed. “We mandate that one of 

our employees reviews the complete list of hits, 

and not just the ones that have been automatically 

flagged,” says Fischer. “A fully automated 

screening system leaves significant biosecurity 

questions unanswered.” 

According to Theresa Lawrence, a senior 

science advisor with HHS, top homology was 

rejected in the interest of applying a consistent 

standard for distinguishing potential threats based 

on analysis of an established data source. “There 

was concern with the top homology approach that 

we would have to designate an arbitrary threshold,” 

she says, “and this approach needs human 

screeners, which can represent an inconsistent 

mechanism from provider to provider.” 

Although GenBank represents a rich resource 

for genetic data and is therefore a powerful 

foundation for such screens, it is nevertheless 

a product of community curation and potential 

‘sequences of concern’ may be inconsistently 

designated. “GenBank is just a repository,” says 

Sean Eddy, a computational biologist at the 

Howard Hughes Medical Institute’s Janelia Farm 

Research Campus in Loudon County, Virginia. 

CMSP 


NEWS 


in brief 

DNA not patentable 

In a spine-chilling announcement, the 

US Department of Justice in October said 

unmodified human DNA should not be eligible 

for patent. The department’s stance conflicts 

with the body of case law on the matter and a 

longstanding position held by the US Patent 

and Trademark Office, which has issued 

more than 10,000 of these patents. The 

Justice Department announced its position 

in response to a lawsuit involving patents on 

breast cancer genes BRCA1 and BRCA2. A US 

district court in March declared the patents 

invalid, saying that the genes are products of 

nature rather than human-made inventions. 

Patent holders University of Utah and Myriad 

Genetics, based in Salt Lake City, appealed 

in June. In an amicus brief filed with the 

Federal Circuit Court of Appeals the Justice 

Department agreed that identifying and 

isolating DNA without further manipulation 

is not an invention, or patent eligible. How 

the agency’s declaration will influence 

justices and the patent office worries biotech 

companies. But the patent office isn’t easily 

swayed, says Thomas Kowalski, an attorney 

with Vedder Price in New York. “The patent 

office is not going to change what it’s doing in 

view of what the Department of Justice says,” 

he says. Besides, international agreements 

between the American, European and 

Japanese patent offices, known as Trilateral 

Co-operation, have concluded that unmodified 

DNA is patentable. 

Emily Waltz 

USPTO’s do-good vouchers 

Patent owners may soon be able to cut the 

time needed to have a patent reexamined by 

up to two-thirds if they can demonstrate a 

humanitarian use. The US Patent & Trademark 

Office (USPTO) is proposing a system that would 

offer fast-track reexamination vouchers as an 

incentive to stimulate “creation or licensing 

that addresses humanitarian needs.” Because 

patents under reexamination are often the most 

valuable commercially, a fast-track procedure 

would let patent owners “more readily and less 

expensively affirm the validity of their patents,” 

according to the Federal Register notice. The 

USPTO currently takes 19 to 20 months for 

such reexaminations, whereas the expedited 

review promises a six-month turnaround. The 

system is modeled on the US Food & Drug 

Administration’s priority review vouchers given 

to entities that develop drugs to treat neglected 

tropical diseases. In this case, patent holders 

who receive the fast-track reexamination voucher 

could use it on any other patent they own or 

transfer it to the open market. Although the 

intent is worthy, there are too many unanswered 

questions, worries Thomas Kowalski of Vedder 

Price. “What will the USPTO do to ensure that 

those in the developing world as well as the 

poor in the developed world can gain access 

to the technology? Also, the voucher should 

be tied specifically to the technology with the 

humanitarian use instead of being independent 

and transferable.” Michael Francisco 

Table 1 Timeline of events leading up to the synthetic DNA guidance 

December 2006 The US government’s National Science Advisory Board for Biosecurity issues a report 

recommending the establishment of “uniform and standardized screening practices 

among providers of synthetic DNA.” 

June 2007 The US Government convenes an interagency working group on synthetic nucleic acid 

screening. 

November 3, 2009 Companies of the International Association Synthetic Biology (Entelechon, 

ATG:biosynthetics, Biomax, febit and Sloning Biotechnology) present their “Code of 

Conduct” for the screening of gene orders and customers. 

November 19, 2009 Companies of the International Gene Synthesis Consortium (Blue Heron Biotechnology, 

DNA2.0, GENEART, GenScript and Integrated DNA Technologies) release a similar set of 

‘best practices’ guidelines, the “Harmonized Screening Protocol.” 

November 29, 2009 The US Department of Health and Human Services releases a draft version of its 

Screening Framework Guidance for Providers of Synthetic Double-Stranded DNA for 

public comment. 

January 11, 2010 The American Association for the Advancement of Science (AAAS) hosts a meeting for policy 

specialists and scientists from academia and industry on “Minimizing the Risks of Synthetic 

DNA: Scientists’ Views on the U.S. Government’s Guidance on Synthetic Genomics.” 

January 26, 2010 Public comment period ends on the draft guidance. 

October 13, 2010 HHS publishes its final version of the Screening Framework Guidance. 

He adds, “The annotation is as provided by the 

person that deposited the sequence.” Screening 

effectiveness could also be constrained by 

biases in the database contents, according to 

James Diggans, a researcher at MITRE, a notfor-profit 

national technology resource that 

focuses on security issues, located in Bedford, 

Massachusetts and McLean, Virginia. “There are 

far more harmless sequences in these databases 

than there are sequences that could be used to 

harm human health.” 

For other scientists, the reliance on the select 

agents list is also problematic. Eighty-two items 

currently listed represent known risks to human, 

plant or animal health, and are unambiguously 

regulated by federal law. But many known 

human pathogens, such as severe acute respiratory 

syndrome virus, are omitted, and others 

worry about the problems that could be posed 

by the yet-unknown sequence variants. “If you 

synthesize a genome without creating the actual 

organism it encodes—and where now you aren’t 

even limited to the variability found in nature— 

how do you taxonomically classify that genome 

sequence?” says Eddy. 

Eddy and other scientists recently partnered 

with the US National Research Council in an 

effort to bring some clarity to the characterization 

of high-risk genes. The resulting report, 

Sequence-Based Classification of Select Agents: 

A Brighter Line (http://www.nap.edu/catalog/ 

12970.html), concludes that although it is presently 

impossible to reliably predict gene function 

based on sequence, it should nevertheless 

be within reach to develop mechanisms that 

can help categorize sequences as belonging to 

predefined ‘hazardous’ or ‘safe’ classes of genes, 

an effort that could greatly improve the future 

efficiency of synthetic gene order screening. 

Several parallel efforts are also underway to 

develop more sophisticated and comprehensive 

pathogen databases. Fischer and Maurer are 

collaborating on Virulence Factor Information 

Repository (VIREP), a repository for annotated 

information about known virulence genes, 

based at UC, Berkeley. The IGSC has also stated 

its intention to develop an extensive regulated 

pathogens database, which could offer a broadly 

useful community resource. However, both 

groups are waiting on government support to 

help move these projects forward. 

For now, the member companies of the IGSC, 

which are predominantly based in the US, are 

moving to adapt their standards to comply with 

the HHS recommendations. However, the guidance 

also invites companies to apply their own 

“equivalent or superior” screening standards 

and several companies indicate that they will 

continue to err on the side of caution in their 

screening procedures. “If we get a gene in, we 

screen it,” says Robert Dawson, director of bioinformatics 

at Coralville, Iowa–based Integrated 

DNA Technologies. “There’s never a case where 

we would have a gene go right into production 

without a human being having looked at both 

the sequence and the prospective customer.” 

HHS has also made it clear that these are minimum 

screening recommendations and not the 

final word, and discussions are ongoing. 

Given the early stage of the field, when the 

risk from synthetic biology is still seen as relatively 

low—to date, no IGSC member company 

reports having received an order for a ‘sequence 

of concern’ that also came from a dubious customer—some 

hope that there will be sufficient 

opportunity for these guidelines to grow into a 

more effective monitoring strategy. “It’s a line 

in the sand drawn by the US government that 

now serves as something to be improved over 

time,” says Diggans. “All of these things make 

a direct contribution to maintaining near-term 

biosecurity, but it will need to evolve quickly— 

the technology is moving ever faster.” 

Michael Eisenstein, Philadelphia 


NEWS 


in brief 

DNA not patentable 

In a spine-chilling announcement, the 

US Department of Justice in October said 

unmodified human DNA should not be eligible 

for patent. The department’s stance conflicts 

with the body of case law on the matter and a 

longstanding position held by the US Patent 

and Trademark Office, which has issued 

more than 10,000 of these patents. The 

Justice Department announced its position 

in response to a lawsuit involving patents on 

breast cancer genes BRCA1 and BRCA2. A US 

district court in March declared the patents 

invalid, saying that the genes are products of 

nature rather than human-made inventions. 

Patent holders University of Utah and Myriad 

Genetics, based in Salt Lake City, appealed 

in June. In an amicus brief filed with the 

Federal Circuit Court of Appeals the Justice 

Department agreed that identifying and 

isolating DNA without further manipulation 

is not an invention, or patent eligible. How 

the agency’s declaration will influence 

justices and the patent office worries biotech 

companies. But the patent office isn’t easily 

swayed, says Thomas Kowalski, an attorney 

with Vedder Price in New York. “The patent 

office is not going to change what it’s doing in 

view of what the Department of Justice says,” 

he says. Besides, international agreements 

between the American, European and 

Japanese patent offices, known as Trilateral 

Co-operation, have concluded that unmodified 

DNA is patentable. 

Emily Waltz 

USPTO’s do-good vouchers 

Patent owners may soon be able to cut the 

time needed to have a patent reexamined by 

up to two-thirds if they can demonstrate a 

humanitarian use. The US Patent & Trademark 

Office (USPTO) is proposing a system that would 

offer fast-track reexamination vouchers as an 

incentive to stimulate “creation or licensing 

that addresses humanitarian needs.” Because 

patents under reexamination are often the most 

valuable commercially, a fast-track procedure 

would let patent owners “more readily and less 

expensively affirm the validity of their patents,” 

according to the Federal Register notice. The 

USPTO currently takes 19 to 20 months for 

such reexaminations, whereas the expedited 

review promises a six-month turnaround. The 

system is modeled on the US Food & Drug 

Administration’s priority review vouchers given 

to entities that develop drugs to treat neglected 

tropical diseases. In this case, patent holders 

who receive the fast-track reexamination voucher 

could use it on any other patent they own or 

transfer it to the open market. Although the 

intent is worthy, there are too many unanswered 

questions, worries Thomas Kowalski of Vedder 

Price. “What will the USPTO do to ensure that 

those in the developing world as well as the 

poor in the developed world can gain access 

to the technology? Also, the voucher should 

be tied specifically to the technology with the 

humanitarian use instead of being independent 

and transferable.” Michael Francisco 

Table 1 Timeline of events leading up to the synthetic DNA guidance 

December 2006 The US government’s National Science Advisory Board for Biosecurity issues a report 

recommending the establishment of “uniform and standardized screening practices 

among providers of synthetic DNA.” 

June 2007 The US Government convenes an interagency working group on synthetic nucleic acid 

screening. 

November 3, 2009 Companies of the International Association Synthetic Biology (Entelechon, 

ATG:biosynthetics, Biomax, febit and Sloning Biotechnology) present their “Code of 

Conduct” for the screening of gene orders and customers. 

November 19, 2009 Companies of the International Gene Synthesis Consortium (Blue Heron Biotechnology, 

DNA2.0, GENEART, GenScript and Integrated DNA Technologies) release a similar set of 

‘best practices’ guidelines, the “Harmonized Screening Protocol.” 

November 29, 2009 The US Department of Health and Human Services releases a draft version of its 

Screening Framework Guidance for Providers of Synthetic Double-Stranded DNA for 

public comment. 

January 11, 2010 The American Association for the Advancement of Science (AAAS) hosts a meeting for policy 

specialists and scientists from academia and industry on “Minimizing the Risks of Synthetic 

DNA: Scientists’ Views on the U.S. Government’s Guidance on Synthetic Genomics.” 

January 26, 2010 Public comment period ends on the draft guidance. 

October 13, 2010 HHS publishes its final version of the Screening Framework Guidance. 

He adds, “The annotation is as provided by the 

person that deposited the sequence.” Screening 

effectiveness could also be constrained by 

biases in the database contents, according to 

James Diggans, a researcher at MITRE, a notfor-profit 

national technology resource that 

focuses on security issues, located in Bedford, 

Massachusetts and McLean, Virginia. “There are 

far more harmless sequences in these databases 

than there are sequences that could be used to 

harm human health.” 

For other scientists, the reliance on the select 

agents list is also problematic. Eighty-two items 

currently listed represent known risks to human, 

plant or animal health, and are unambiguously 

regulated by federal law. But many known 

human pathogens, such as severe acute respiratory 

syndrome virus, are omitted, and others 

worry about the problems that could be posed 

by the yet-unknown sequence variants. “If you 

synthesize a genome without creating the actual 

organism it encodes—and where now you aren’t 

even limited to the variability found in nature— 

how do you taxonomically classify that genome 

sequence?” says Eddy. 

Eddy and other scientists recently partnered 

with the US National Research Council in an 

effort to bring some clarity to the characterization 

of high-risk genes. The resulting report, 

Sequence-Based Classification of Select Agents: 

A Brighter Line (http://www.nap.edu/catalog/ 

12970.html), concludes that although it is presently 

impossible to reliably predict gene function 

based on sequence, it should nevertheless 

be within reach to develop mechanisms that 

can help categorize sequences as belonging to 

predefined ‘hazardous’ or ‘safe’ classes of genes, 

an effort that could greatly improve the future 

efficiency of synthetic gene order screening. 

Several parallel efforts are also underway to 

develop more sophisticated and comprehensive 

pathogen databases. Fischer and Maurer are 

collaborating on Virulence Factor Information 

Repository (VIREP), a repository for annotated 

information about known virulence genes, 

based at UC, Berkeley. The IGSC has also stated 

its intention to develop an extensive regulated 

pathogens database, which could offer a broadly 

useful community resource. However, both 

groups are waiting on government support to 

help move these projects forward. 

For now, the member companies of the IGSC, 

which are predominantly based in the US, are 

moving to adapt their standards to comply with 

the HHS recommendations. However, the guidance 

also invites companies to apply their own 

“equivalent or superior” screening standards 

and several companies indicate that they will 

continue to err on the side of caution in their 

screening procedures. “If we get a gene in, we 

screen it,” says Robert Dawson, director of bioinformatics 

at Coralville, Iowa–based Integrated 

DNA Technologies. “There’s never a case where 

we would have a gene go right into production 

without a human being having looked at both 

the sequence and the prospective customer.” 

HHS has also made it clear that these are minimum 

screening recommendations and not the 

final word, and discussions are ongoing. 

Given the early stage of the field, when the 

risk from synthetic biology is still seen as relatively 

low—to date, no IGSC member company 

reports having received an order for a ‘sequence 

of concern’ that also came from a dubious customer—some 

hope that there will be sufficient 

opportunity for these guidelines to grow into a 

more effective monitoring strategy. “It’s a line 

in the sand drawn by the US government that 

now serves as something to be improved over 

time,” says Diggans. “All of these things make 

a direct contribution to maintaining near-term 

biosecurity, but it will need to evolve quickly— 

the technology is moving ever faster.” 

Michael Eisenstein, Philadelphia 


news 

BARDA funds vaccine makers aiming to phase out eggs 


The US government handed out contracts worth 

potentially more than $100 million to boost 

innovative technologies that can deliver vaccines 

in large quantities—and fast—as part of a 

$1.9 billion initiative to protect Americans from 

biologic threats of the future. The Biomedical 

Advanced Research and Development Authority 

(BARDA) has awarded money to eight firms 

engaged in creative ways to make vaccines 

against naturally occurring diseases, such as 

the H1N1 influenza, which was pandemic last 

year, and others it views as potential bioterrorist 

weapons. The goal, Kathleen Sebelius, 

Department of Health and Human Services 

(HHS) secretary says, is to create “nimble, flexible 

capacity to produce medical countermeasures 

rapidly in the face of any attack or threat.” 

Contract winners announced in September 

(Table 1) include VaxDesign, Novartis partnered 

with Synthetic Genomics Vaccines, the 

nonprofits PATH (Program for Appropriate 

Technology in Health) and the Infectious 

Disease Research Institute, Pfenex, Rapid 

Micro Biosystems, Emergent Biosolutions, 3M 

and Northrop Grumman Security Systems. 

BARDA’s push to make the vaccine development 

and production process move beyond 

half-century-old systems that rely on fertilized 

eggs to culture most vaccines has been 

several months in gestation. In August, two 

reports looked at the US’s ability to produce 

vaccines quickly and efficiently and came 

away holding their noses. “The closer we 

looked at the countermeasure pipeline, the 

more leaks, choke points and dead ends we 

saw. In this age of new threats, we aren’t generating 

enough new products,” said Sebelius 

A half-century-old technology. Fertilized eggs were used to produce a vaccine for the H1N1 flu 

virus at the Sinovac plant in Beijing during the 2009 outbreak. 

at a press conference on August 19 during 

which both reports were presented. 

The first report, The Public Health Emergency 

Medical Countermeasures Enterprise Review 

(https://www.medicalcountermeasures.gov/ 

documents/MCMReviewFinalcover-508. 

pdf), commissioned by the HHS called for 

modernization and speed-up of the regulatory 

procedures, the development of more flexible 

vaccine manufacturing processes to allow 

more than one product to be produced at the 

same facility, and an upgrade and modernization 

of everything related to rapid influenza 

vaccine production. Nearly $2 billion was put 

aside to reach the report’s goals. 

More quantitative was the President’s 

Council of Advisors on Science and 

Technology’s (PCAST) report entitled 

Reengineering the Influenza Vaccine Production 

Enterprise to Meet the Challenges of Pandemic 

Influenza (http://www.whitehouse.gov/sites/ 

default/files/microsites/ostp/Influenza%20 

Vaccinology.pdf). It outlines every stumbling 

block to speeding up vaccine development but 

more significantly lists how much time could 

be saved by adapting improvements already 

in the pipeline. The Influenza Vaccinology 

Working group, chaired by Eric Lander, director 

of the Broad Institute of Harvard and MIT 

in Cambridge, Massachusetts, and Harold 

AP Photo/Greg Baker 

Table 1 Vaccine manufacturers awarded BARDA contracts 

Company/location Project aims Contract value 

Novartis Vaccines and Diagnostics/ 

Cambridge, Massachusetts 

Pfenex/ 

San Diego 

VaxDesign/ 

Orlando, Florida 

Northrop Grumman Security Systems/ 

Baltimore 

PATH/ 

Seattle 

Infectious Disease Research 

Institute (IDRI)/ 

Seattle 

Rapid Micro Biosystems/ 

Bedford, Massachusetts 

3M/ 

St. Paul, Minnesota 

Novartis Vaccines and Diagnostics will investigate techniques for the rapid development of optimized influenza 

seed virus. 

Pfenex will apply its Pfenex Expression Technology Platform to the development of optimized bioprocesses for 

high-yield production of a stable candidate anthrax vaccine. 

VaxDesign will further develop its MIMIC platform, an in vitro human immune system mimetic designed to accelerate 

evaluation of candidate and stockpiled vaccine safety and effectiveness by supplementing animal testing. 

Northup will develop integrated diagnostic capabilities for rapid, high-throughput surveillance and molecular 

diagnostics. 

~ $24 million 

over three years 

~$18.8 million 


~$17.1 million 


~$9.8 million 

over one year 

PATH will test multiple innovative formulation chemistries and strategies to increase the shelf life of influenza ~$9.4 million 

vaccines, which has implications for the national vaccine stockpile as well as cold-chain requirements domestically 

and in developing 


countries. 

IDRI will develop and evaluate innovative adjuvant formulations to enhance influenza vaccine immunogenicity 

and cross-protection to make them more effective against novel viral strains that may cause the next pandemic. 

Rapid Micro Biosystems will develop methods for accelerated sterility testing. Together, these improvements could 

shave weeks off the influenza vaccine manufacturing and product release schedule. 

3M will develop integrated diagnostic capabilities for rapid, high-throughput surveillance and molecular 

diagnostics. 

~$8.5 million 


~$6.8 million 


~$6 million 

over two years 


NEWS 


Vaccine makers’ immunity questioned in court 

The US Supreme Court has begun considering how much liability vaccine makers have if 

the side effects of their products are believed to have injured or killed someone. The case 

was brought against Wyeth (now merged with Pfizer of New York) by parents of Hannah 

Bruesewitz, who in 1992 began suffering seizures and developmental problems after 

being given the combined Corynebacterium diphtheriae toxoid/Clostridium tetani toxoid/ 

polio (DTP) vaccine against diphtheria, tetanus and pertussis (whooping cough). A few 

years later, DTP was removed from the market and replaced by a vaccine with fewer side 

effects. The Bruesewitzes believed their daughter’s injuries were avoidable because Wyeth 

should have put a product with fewer side effects on the market earlier. 

What is most notable about the Bruesewitz v. Wyeth case, which was argued on October 

12 in Washington, DC, is that many in the US drug industry had believed that the issue 

had been completely resolved with the adoption in 1986 of the National Childhood 

Vaccine Injury Act. The act set up a Vaccine Court to adjudicate claims of injury on a nofault 

basis and pay successful claimants with money generated from a tax on vaccines. 

The Vaccine Act was put into effect because of a fear at the time that lawsuits claiming 

‘design defects’ would force companies to stop making vaccines. Accordingly, the act 

says suits cannot be filed against manufacturers “if the injury or death resulted from side 

effects that were unavoidable, even though the vaccine was properly prepared and was 

accompanied by proper directions and warnings.” 

There is a back door to the law that allows families to go to a federal court if they lose 

in Vaccine Court or they don’t like the amount of its judgment. However, those suits are 

governed by the Vaccine Act, too. 

But neither the Vaccine Court nor a lower US federal court accepted the 

Bruesewitzes’ argument that their daughter’s injuries could have been avoided by 

the manufacturer. However, the justices found the wording in the Act, and especially 

its use of the word “unavoidable” quite confused. Justice Stephen Breyer remarked 

“it’s pretty hard to say the word unavoidable means avoidable.” A final judgment is 

expected in early spring of 2011. 

Stephen Strauss 

Varmus, then at the Memorial Sloan-Kettering 

Cancer Center, also estimates the time it would 

take before these changes could be instituted. 

From rapid sterility testing, to accelerated 

virus seed production and improved adjuvants, 

each advance could slice several weeks off the 

time for the first dose to reach the market. The 

PCAST committee members believed these 

changes individually could be put in place within 

1 to 3 years. To change egg-based vaccine production 

systems for alternative cell or recombinant 

DNA platforms would require longer—up 

to a decade—to reach market penetration. 

BARDA’s R&D money will help push a 

broad swath of potentially game-changing new 

technologies, but deputy assistant secretary 

of BARDA, Robin Robinson, admits that it 

doesn’t cover the gamut of vaccine innovations 

in development. In particular, Robinson points 

to efforts to grow vaccines in plants and insect 

cells. Some of these projects are being funded 

by other US government agencies, most notably 

the Defense Advanced Research Project 

Agency, which is supporting four tobaccobased 

vaccine production platforms. 

In plants, the process is quicker than in 

eggs. Andy Sheldon, president and CEO of 

Medicago of Quebec City, Canada, says “it 

takes five weeks to grow the tobacco, the 

plants start expressing the protein in five days, 

and then it takes two days to purify the VLPs 

[virus-like particles].” He compares this to 

the six months egg-based vaccine production 

takes. Medicago is entering phase 2 clinical trials 

with its plant-derived flu vaccine. 

Plant-based production is cheaper too. The 

manufacturing facilities Medicago plans for 

Raleigh, North Carolina, will cost $25 million 

to build, a far smaller investment than the $250 

million required for an egg-based production 

plant and the $1 billion that Novartis recently 

spent on a new Holly Springs, North Carolina 

facility. If approval is granted, it will become 

the first facility in the United States licensed to 

use mammalian cells to produce flu vaccines 

and is expected to be operational in 2013. 

A seasonal influenza vaccine, FluBlok, 

produced in insect cell culture, could be on 

the market next year. Protein Sciences of 

Meriden, Connecticut, received a BARDA 

contract in 2009 to use cells from fall armyworm 

(Spodoptera frugiperda) with a baculovirus 

system to generate influenza VLPs. 

Protein Science’s president and CEO, Manon 

Cox, says it takes about two months from virus 

discovery to vaccine production using insect 

cells for production. It is also cheaper than 

egg-based vaccine manufacturing. “Licensed 

vaccines cost approximately $1 a dose to make 

the active ingredients. Our estimates are that 

we can make three times more product for that 

price,” claims Cox. Protein Sciences is waiting 

for US Food and Drug Administration 

approval of FluBlok. 

Pfenex, too, says its technology is nearing 

the market. Last year, the Defense Threat 

Reduction Agency provided the company 

with a DNA sequence of an unknown antigen 

and challenged them to develop both a production 

strain and a high-speed, high-quality, 

low-cost, antigen-production process. In conjunction 

with partner organizations, Pfenex 

used its screening technology to do this within 

42 days. And there were cost savings. “If you 

scaled up to production levels, the antigen 

can be produced for ~50 cents per dose,” says 

Patrick Lucy, Pfenex’s vice president of business 

development. 

But the issue that looms largest in the push 

to modernize and speed up vaccine development 

relates to the business end of things. How 

are these innovative technologies going to fit 

into an existing vaccine marketplace that— 

influenza pandemics and potential terrorist 

bio-attacks aside—generally satisfies the 

world’s vaccine needs? 

For example, the PCAST report pointed 

out that although the Novartis cell culture 

facility was likely to generate annual profits 

of $30 million, it “would take over 30 years to 

recover the [$1 billion] investment in nominal 

dollars (leaving aside the need for a return 

on investment).” 

Protein Science’s Cox argues this naturally 

leads vaccine manufacturers, using egg-based 

technologies, to resist any change. “They are 

not going to easily let that [advantage] be 

taken away by a new technology in which 

their learning curve is going to be as steep 

as anybody’s else’s,” she says. Rafick-Pierre 

Sékaly, co-director and scientific director of 

the Vaccine and Gene Therapy Institute of Port 

St. Lucie, Florida, concurs. “The president and 

the committee can make all the recommendations 

they want but if the big vaccine makers 

say it is too costly or there is too much R&D, 

then changes are going to be treated not as a 

solution but as an added burden.” 

On this point, BARDA’s Robinson says, “we 

understand, and that is why we are pushing 

things that will definitely benefit all vaccines, 

including eggs.” Indeed, in October BARDA 

awarded Sanofi Pasteur of Lyon, France, a 

3-year, $57 million contract to make more fertilized 

eggs available for vaccine production on 

a year-round basis. 

Stephen Strauss, Toronto 


NEWS 


Vaccine makers’ immunity questioned in court 

The US Supreme Court has begun considering how much liability vaccine makers have if 

the side effects of their products are believed to have injured or killed someone. The case 

was brought against Wyeth (now merged with Pfizer of New York) by parents of Hannah 

Bruesewitz, who in 1992 began suffering seizures and developmental problems after 

being given the combined Corynebacterium diphtheriae toxoid/Clostridium tetani toxoid/ 

polio (DTP) vaccine against diphtheria, tetanus and pertussis (whooping cough). A few 

years later, DTP was removed from the market and replaced by a vaccine with fewer side 

effects. The Bruesewitzes believed their daughter’s injuries were avoidable because Wyeth 

should have put a product with fewer side effects on the market earlier. 

What is most notable about the Bruesewitz v. Wyeth case, which was argued on October 

12 in Washington, DC, is that many in the US drug industry had believed that the issue 

had been completely resolved with the adoption in 1986 of the National Childhood 

Vaccine Injury Act. The act set up a Vaccine Court to adjudicate claims of injury on a nofault 

basis and pay successful claimants with money generated from a tax on vaccines. 

The Vaccine Act was put into effect because of a fear at the time that lawsuits claiming 

‘design defects’ would force companies to stop making vaccines. Accordingly, the act 

says suits cannot be filed against manufacturers “if the injury or death resulted from side 

effects that were unavoidable, even though the vaccine was properly prepared and was 

accompanied by proper directions and warnings.” 

There is a back door to the law that allows families to go to a federal court if they lose 

in Vaccine Court or they don’t like the amount of its judgment. However, those suits are 

governed by the Vaccine Act, too. 

But neither the Vaccine Court nor a lower US federal court accepted the 

Bruesewitzes’ argument that their daughter’s injuries could have been avoided by 

the manufacturer. However, the justices found the wording in the Act, and especially 

its use of the word “unavoidable” quite confused. Justice Stephen Breyer remarked 

“it’s pretty hard to say the word unavoidable means avoidable.” A final judgment is 

expected in early spring of 2011. 

Stephen Strauss 

Varmus, then at the Memorial Sloan-Kettering 

Cancer Center, also estimates the time it would 

take before these changes could be instituted. 

From rapid sterility testing, to accelerated 

virus seed production and improved adjuvants, 

each advance could slice several weeks off the 

time for the first dose to reach the market. The 

PCAST committee members believed these 

changes individually could be put in place within 

1 to 3 years. To change egg-based vaccine production 

systems for alternative cell or recombinant 

DNA platforms would require longer—up 

to a decade—to reach market penetration. 

BARDA’s R&D money will help push a 

broad swath of potentially game-changing new 

technologies, but deputy assistant secretary 

of BARDA, Robin Robinson, admits that it 

doesn’t cover the gamut of vaccine innovations 

in development. In particular, Robinson points 

to efforts to grow vaccines in plants and insect 

cells. Some of these projects are being funded 

by other US government agencies, most notably 

the Defense Advanced Research Project 

Agency, which is supporting four tobaccobased 

vaccine production platforms. 

In plants, the process is quicker than in 

eggs. Andy Sheldon, president and CEO of 

Medicago of Quebec City, Canada, says “it 

takes five weeks to grow the tobacco, the 

plants start expressing the protein in five days, 

and then it takes two days to purify the VLPs 

[virus-like particles].” He compares this to 

the six months egg-based vaccine production 

takes. Medicago is entering phase 2 clinical trials 

with its plant-derived flu vaccine. 

Plant-based production is cheaper too. The 

manufacturing facilities Medicago plans for 

Raleigh, North Carolina, will cost $25 million 

to build, a far smaller investment than the $250 

million required for an egg-based production 

plant and the $1 billion that Novartis recently 

spent on a new Holly Springs, North Carolina 

facility. If approval is granted, it will become 

the first facility in the United States licensed to 

use mammalian cells to produce flu vaccines 

and is expected to be operational in 2013. 

A seasonal influenza vaccine, FluBlok, 

produced in insect cell culture, could be on 

the market next year. Protein Sciences of 

Meriden, Connecticut, received a BARDA 

contract in 2009 to use cells from fall armyworm 

(Spodoptera frugiperda) with a baculovirus 

system to generate influenza VLPs. 

Protein Science’s president and CEO, Manon 

Cox, says it takes about two months from virus 

discovery to vaccine production using insect 

cells for production. It is also cheaper than 

egg-based vaccine manufacturing. “Licensed 

vaccines cost approximately $1 a dose to make 

the active ingredients. Our estimates are that 

we can make three times more product for that 

price,” claims Cox. Protein Sciences is waiting 

for US Food and Drug Administration 

approval of FluBlok. 

Pfenex, too, says its technology is nearing 

the market. Last year, the Defense Threat 

Reduction Agency provided the company 

with a DNA sequence of an unknown antigen 

and challenged them to develop both a production 

strain and a high-speed, high-quality, 

low-cost, antigen-production process. In conjunction 

with partner organizations, Pfenex 

used its screening technology to do this within 

42 days. And there were cost savings. “If you 

scaled up to production levels, the antigen 

can be produced for ~50 cents per dose,” says 

Patrick Lucy, Pfenex’s vice president of business 

development. 

But the issue that looms largest in the push 

to modernize and speed up vaccine development 

relates to the business end of things. How 

are these innovative technologies going to fit 

into an existing vaccine marketplace that— 

influenza pandemics and potential terrorist 

bio-attacks aside—generally satisfies the 

world’s vaccine needs? 

For example, the PCAST report pointed 

out that although the Novartis cell culture 

facility was likely to generate annual profits 

of $30 million, it “would take over 30 years to 

recover the [$1 billion] investment in nominal 

dollars (leaving aside the need for a return 

on investment).” 

Protein Science’s Cox argues this naturally 

leads vaccine manufacturers, using egg-based 

technologies, to resist any change. “They are 

not going to easily let that [advantage] be 

taken away by a new technology in which 

their learning curve is going to be as steep 

as anybody’s else’s,” she says. Rafick-Pierre 

Sékaly, co-director and scientific director of 

the Vaccine and Gene Therapy Institute of Port 

St. Lucie, Florida, concurs. “The president and 

the committee can make all the recommendations 

they want but if the big vaccine makers 

say it is too costly or there is too much R&D, 

then changes are going to be treated not as a 

solution but as an added burden.” 

On this point, BARDA’s Robinson says, “we 

understand, and that is why we are pushing 

things that will definitely benefit all vaccines, 

including eggs.” Indeed, in October BARDA 

awarded Sanofi Pasteur of Lyon, France, a 

3-year, $57 million contract to make more fertilized 

eggs available for vaccine production on 

a year-round basis. 

Stephen Strauss, Toronto 



Ethanol blend hike to jump start cellulosic 

investment 

A move announced on October 13 by the US 

Environmental Protection Agency (EPA) to 

raise the permitted ethanol blend in gasoline 

for motor vehicles has been widely welcomed 

by ethanol manufacturers and biofuel 

advocate groups. The increase of ethanol in 

gasoline, from 10% to 15%, revises a mandate 

that has stood for 35 years and is thought to 

be a critical step toward bringing cellulosic 

ethanol to consumers. 

The EPA’s decision applies only to light 

motor vehicles made since 2007—a second 

decision on vehicles made from 2001 to 2006 

is expected by the year end—but it’s still 

enough to give hope to biotechs working 

on cellulosic ethanol (and other non-corn 

feedstock ‘second-generation’ biofuels). The 

problem for cellulosic ethanol producers 

was that the so-called E10 blend limit—10% 

ethanol—left the market saturated with 

corn ethanol, says Jim Sturdevant, director 

of Project Liberty, the cellulosic operation 

of Poet, of Sioux Falls, South Dakota. Raising 

that limit should open the door to companies 

like his—critical, Sturdevant says, because 

“without a demand, it’s not worth investing 

in cellulosic ethanol.” 

The E15 decision could indeed revive 

investor interest in non-grain-based ethanol 

companies. Investor interest—in both biofuels 

and bio-based materials—grew during the 

first few years of this decade, reaching a peak 

in 2007 (Fig. 1). That meant companies were 

able to quickly acquire seed money and take 

their technologies to the pilot plant level, 

but as financing became harder to acquire 

and interest petered out, progress stalled. 

That was all part of “a natural dropoff ” in 

the cycle of early technology development, 

says David Berry, a principal at Flagship 

Ventures in Cambridge, Massachusetts, but 

the blend wall decision could start money 

flowing again. 

Berry says the biggest impact would be 

on the later-stage companies in need of 

financing to “cross the chasm” from pilotscale 

facilities to large manufacturing plants. 

An example is Mascoma, of Lebanon, New 

Hampshire, which is in the process of securing 

funding for its first industrial-scale cellulosic 

plant, in Kinross, Michigan. After the 

EPA’s decision, it will be “considerably easier 

to raise equity for this sort of facility,” says 

Justin van Rooyen, Mascoma’s director of 

business development. 

But corn ethanol—for all the roads it 

has paved—can still be an impediment to 

Car drivers in the US can now fill their tanks with 

a higher blend of ethanol. But cellulosic biofuel 

producers want to see the limit rise further. 

second-generation biofuels. It’s part of the 

reason 54 ethanol manufacturers and Growth 

Energy, an ethanol supporters’ group based in 

Washington, DC, submitted the E15 Green Jobs 

Waiver to the EPA in the spring of 2009, petitioning 

the organization to amend the ethanol 

blend limit to 15%. The lobby group realized 

the 10% ethanol opportunity in gasoline was 

becoming saturated by existing corn ethanol 

manufacturers. 

Cynthia Bryant, marketing manager, global 

fuels at Novozymes North America, with 

headquarters in Bagsvaerd, Denmark calls 

the EPA’s decision “a validation of what we’ve 

been saying all along—that higher blends of 

ethanol can be used in our cars today.” A 15% 

blend puts ethanol into 43 million cars; if the 

EPA allows the blend into cars built in and 

after 2001, it would add another 83 million 

cars. That could, in effect, reach 54% of the 

cars in the US. 

That’s a huge step, though the government 

is also assisting the growth of biofuels in 

other ways. In October the US Department of 

Agriculture said its Biomass Crop Assistance 

Program will supply $525 million in subsidies 

to farmers growing second-generation biomass 

crops over a 15-year period. But the future of 

cellulosic ethanol in the US will eventually 

come down to the hypothetical consumer, 

and he or she is already being discussed. The 

EPA E15 decision came with a suggested 

label to identify pumps dispensing the new 

blend: bright orange, with the word ‘caution’ 

printed in bold at the top. It is not final, but 

has raised eyebrows anyway, as those in the 

biofuels industry say the label could scare away 

customers. “It’s a little ominous looking,” says 

Dreyer, particularly because “all of our testing 

has shown that E15 is perfectly safe.” 

Growth Energy is seeking federal legislation 

that will require country-of-origin 

Poet 

in brief 

Singapore injects $12.5 

billion 

news 

The Singaporean government will spend 

S$16.1 ($12.5) billion on research 

innovation over the next 4 years—a 20% 

increase over the previous budget. A quarter 

of the funding—S$3.7 ($2.9) billion —is 

allotted to biomedical science, according to 

the September announcement. The funding 

boost “reflects our steady commitment 

to transforming our economy,” says Beh 

Kian Teik, director of biomedical sciences 

of Singapore’s Economic Development 

Board. Since 2000, when the Singaporean 

government outlined its long-term plan 

to transform this tiny nation-state into a 

knowledge-based economy, the government 

ploughed roughly S$25 ($19.3) billion to 

shore up research infrastructure and create a 

talent pool to attract private investment. The 

figures suggest Singapore has succeeded. 

The country’s biomedical manufacturing 

output alone more than tripled from S$6.3 

billion ($4.9 billion) in 2000 to S$21 billion 

($16.2 billion) in 2009. The biomedical 

sector now represents 10% of the country’s 

manufacturing output, up from 4% in 2000. 

Most major pharma companies have a 

presence in Singapore, although in October, 

Eli Lilly of Indianapolis announced the 

closure of its Singapore Centre for Drug 

Discovery. The new funding budget will 

make “more money available for everyone,” 

says Beh, with an emphasis on initiatives 

that favor “economic outcomes.” In fact, 

some are questioning whether the shift in 

emphasis toward translational outcomes and 

away from blue sky research is too myopic. 

Some of the funding has been earmarked 

to support the Biomedical Science Industry 

Partnership Office (IPO), a fledgling agency 

that supports public-private research projects 

and helps companies pool expertise across 

Singaporean research institutes and medical 

centers. The IPO office has succeeded in 

setting up collaborations for Basel-based 

Roche and London-based GlaxoSmithKline 

with Singaporean researchers. Roche 

plans to spend S$130 ($100.6) million to 

support a center for translational medicine 

to develop drugs for the Asian market; 

whereas GlaxoSmithKline will participate in 

four public-private research collaborations 

focused on early-stage research in 

ophthalmology, regenerative medicine and 

neurodegeneration. “This should be good 

news for the biotech sector,” says Jan-Anders 

Karlsson, CEO of S*BIO, a drug discovery 

company set up as a joint venture between 

the Singapore Economic Development Board 

Investments and Chiron Corporation in 2000. 

Government estimates that the nation’s 

economy will have grown by 13–15% by 

year’s end—higher than estimated growth 

for China and India. In October, Basel-based 

agribusiness Syngenta opened an R&D 

facility in Singapore to support technology 

development in the Asia Pacific region. 

 

Gunjan Sinha 


NEWS 


in brief 

Good ideas across borders 

The European Commission (EC) has made a 

bold move to improve Europe’s competitiveness 

and create an environment conducive to 

innovation with the launch of Innovation Union. 

Part of the Europe 2020 initiative for recovery 

from the recession and economic growth over 

the next decade, the program, announced 

October 6, aims to “boost the whole innovation 

chain from ‘research to retail’,” says Mark 

English, spokesperson for the Commissioner for 

Research, Innovation and Science in Brussels. 

The EC is calling on member states to spend 

3% of the gross domestic product (GDP) on 

R&D by 2020, a hike from the current 1.9% of 

GDP. Europe is the largest market in the world, 

but its policies and support for R&D remain 

fragmented. The EC is proposing to boost 

cooperation across borders, among companies, 

and between private and public sectors. Plans 

to adopt a single EU patent system, and provide 

tax incentives and legislation that will allow 

venture capital funds to invest freely anywhere 

in the EU are also part of the Innovation Union’s 

commitments. Monika Wcislo, press officer for 

Research, Innovation and Science, notes that 

funding is not earmarked for particular sectors 

so they expect the impact to be widespread. 

“[Innovation Union] will generate more R&D, 

more startups and more major EU companies 

with the potential to be international players,” 

says English. 


Airlines ahead on algae 

Plans to grow algal biomass in airport grounds 

for aircraft fuel are moving forward supported 

by Airbus, British Airways and Cranfield 

University. On September 16, European aircraft 

manufacturer Airbus of Blagnac, France, 

London-based British Airways, and Gatwick 

airport in Surrey, announced a collaboration 

with Cranfield University, UK, through the 

Sustainable Use of Renewable Fuels (SURF) 

consortium. For the past three years, Cranfield 

University scientists have been researching 

algal-derived biofuel for use in aviation. Now, 

with their partners at SURF, they are working 

to scale up from their pilot plant (a thousand 

gallons per batch) for commercial output. Unlike 

the automobile industry, the aviation sector 

lacks the option to use electric alternatives for 

energy, says Feargal Brennan, head of Cranfield 

University’s Department of Offshore, Process and 

Energy Engineering. “For the foreseeable future 

the aviation industry will depend on biofuels, so 

they have to take a lead in commercial biofuel 

replacements,” Brennan says. Paul Nash, 

Airbus’ head of New Energies and Environmental 

Affairs notes that the firm is collaborating on 

biofuel projects with research groups from Brazil 

and Qatar. Later this year Brazilian airline TAM 

will test a fuel mixture of bio-kerosene derived 

from the native jatropha plant in an Airbus 

aircraft. “What we’re finding today is that the 

industry is moving faster [through partnerships] 

than R&D or governments would do,” says Nash. 

“As an industry, we can say: ‘our aircrafts are 

ready for this’.” 


Total transaction value (millions) 

$1,500 

$1,200 

$900 

$600 

$300 

$0 

88.0 112.8 

2004 

2005 

788.7 

2006 

Year 

1,378.3 

2007 

1,190.9 

2008 

876.7 

2009 

Figure 1 Investment into industrial biotech. After 

cresting in 2007, total investment into biofuels 

and bio-based materials has been in decline. 

labeling on all fuel and fuel sources to entice 

consumers by letting them know where their 

fuel is coming from. 

“You have labels on individual pieces of 

fruit to tell you where they came from,” says 

Glenn Nedwin of Genencor, in Rochester, 

New York, a division of Danish Danisco, and 

one of the biggest developers of industrial 

enzymes in the world. “But you don’t know 

where fuel is coming from. If [the pump] said 

it was coming from Venezuela, Saudi Arabia 

or Iowa, let the consumer make the choice.” 

In fact, much of the current biofuels market 

is being propped up by government 

decrees and subsidies. The EPA’s Renewable 

Fuel Standard (RFS) mandate of 2007 

requires that 36 billion gallons of biofuel 

be blended with gasoline tanks by the year 

2022. Cellulosic ethanol itself, the mandate 

says, will furnish 16% of that, with the rest 

supplied by corn ethanol and other secondgeneration 

biofuels such as biodiesel. 

That RFS mandate has helped drive 

demand, and experts like Nedwin believe an 

aggressive infrastructural overhaul is neces- 


“There’s no scientist out 

there that has been wary 

of me, and guys in the 

venture-capital world I 

dealt with before—it’s 

as if I wasn’t gone a 

day!” Ex-convict Sam 

Waksal, former ImClone 

Systems CEO, is out 

of jail and back in 

business, raising $50 

million to acquire hepatitis C treatment maker 

Three Rivers Pharmaceuticals. (Pharmalot, 26 

October 2010) 

“The pendulum has swung too far at the FDA if 

they influence Advisory Committees to vote no for 

a drug that poses very little risk to the public at 

large…and where there is a demonstrable clinical 

sary to eventually cement biofuels’ place in 

everyday life. Growth Energy has a ‘Fueling 

Freedom’ plan designed to do just that, calling 

for additional tax credits for retailers to 

build 200,000 blender pumps (which allow 

consumers to regulate the percentage of 

ethanol they put in their car using a mixture 

of gasoline and E85), a loan guarantee 

for an ethanol pipeline and flex-fuel vehicles 

to be released on the roads. Those moves 

would require more involvement by the government, 

but there is still another concern: 

price. Without subsidies, gasoline remains 

the cheapest fuel drivers can put into their 

cars. “We’re seeing technologies that are projecting 

the ability to be competitive [with 

gasoline],” says Flagship Ventures’ Berry, 

“But…people aren’t ready to be cost competitive 

without subsidy yet.” 

The unknowns with consumers and price 

explain why the move to a 15% blend falls 

short of making cellulosic producers feel the 

golden years are just ahead. “I think this is a 

positive first step and investors will look at 

this, see it as a good sign. [But i]t doesn’t add 

enough gallons to where we ultimately need 

to get to,” says Wes Bolsen, chief marketing 

officer and vice president, government affairs 

at Coskata, of Warrenville, Illinois. 

As for the next steps, Sturdevant would 

like to see the limit moved to 20%, 30% 

and higher. “I believe that the scientific 

data supports moving more quickly now to 

higher blends. I believe that the EPA and 

[Department of Energy] will see that, and 

they’ll see that there’s no harm to engines at 

higher blends, and I can only hope that they 

will move faster in the future.” 

Nidhi Subbaraman, New York 

benefit in terms of additional weight loss and 

reduction in cardiovascular risk factors.” A petition 

from ‘Citizens for Lorcaserin’ to FDA Commissioner 

Margaret Hamburg protesting the 16 September 

advisory committee decision to turn down approval 

of Arena Pharmaceuticals diet drug. 

“There are things I know something about, and 

things I know little about. The stock market is one 

of the latter.” Sangamo CEO Ed Lanphier admits, 

following a 52-week slide in the company’s stock 

value. (Xconomy, 18 October 2010) 

“We are left relying on 20th century approaches 

for the cures of the 21st century.” FDA 

Commissioner Margaret Hamburg argues for a 

move away from randomized controlled trials 

to testing drugs in combination and using 

biomarkers. (Financial Times, 15 October 2010) 


NEWS feature 


Crossing the line 

The list of drug companies forced into several hundred million 

dollar settlements for making fraudulent product claims continues 

to lengthen. And the signs are that the US government will 

continue to ramp up its efforts, using new theories of liability and 

handing out even stiffer penalties. Mark Ratner investigates. 

It has been more than a year since the US 

Department of Justice (DoJ) announced its 

$2.3 billion settlement with New York–based 

Pfizer for Pfizer’s illegal, off-label marketing 

of four drugs—the largest fraud suit to hit the 

pharmaceutical industry (Nat. Biotechnol. 27, 

961–962, 2009). And the beat goes on, with 

several other high-profile fraud settlements 

coming to light (Table 1). 

Over the past five years, there has been 

no letup in government investigations 

of the marketing and pricing practices 

of pharma, biotech and medical device 

companies. These activities are being 

heavily scrutinized on many fronts—by 

the US Food and Drug Administration, 

DoJ, US Congress and even investors. 

But whereas industry watchers consider 

these ‘plain vanilla’ off-label suits as ‘legacy 

conduct’—part of a bygone era—it is 

unclear whether the government’s and 

industry’s perspectives on this are truly 

aligned. Indeed, the government is now 

focusing on new theories of liability. 

“There is a very careful mining of 

operations at pharmaceutical companies 

to find cases,” says T. Reed Stephens, 

a former DoJ lawyer now with the 

Washington, DC, law firm McDermott 

Will & Emery. “The government now 

has the resources available to it to evaluate 

a higher number of new cases,” and 

to find cases that are consistent with the 

dual policy goals of deterring inappropriate 

conduct and maintaining patient 

safety, he says. By most accounts, there 

may be 150 or more cases still in the 

prosecutorial pipeline. 

Filling government coffers 

For every dollar the government puts into 

healthcare fraud enforcement, it gets back 

more than fifteen times that amount, notes 

Shelley Slade, a former DoJ attorney now 

representing corporate whistle-blowers 

through the Washington, DC, law firm of 

Vogel, Slade & Goldstein. Government press 

releases tout the amounts: according to DoJ, it 

secured more than $2.5 billion in settlements 

and judgments in False Claims Act (FCA) 

matters alleging healthcare fraud in the fiscal 

year ending September 30, 2010. This 

was more than ever before obtained in a 

single year and represents a 66% increase 

over fiscal year 2009, in which $1.68 billion 

was obtained. Just after the Pfizer settlement 

in September 2009, recoveries over the last 

15 years in the federal judicial district of 

Although Pfizer’s multi-billion dollar 2009 settlement 

remains the poster child for healthcare fraud, significant 

settlements continued to pile up this year including a 

half-billion dollars or more each by drug makers Novartis, 

AstraZeneca and Allergan. (© Novartis AG) 

Massachusetts alone, where that suit was 

prosecuted, had reached $6.8 billion. “It’s a 

no-brainer in terms of financial investment 

by the government,” says Slade. 

Whistle-blowers also have a vested interested 

in finding new theories they can file on, 

Stephens points out, as do the attorneys who 

specialize in whistle-blower cases. The private 

incentive alone—whistle-blowers may receive 

upwards of 15% of any amounts recovered in 

claims by the federal government—is enough 

to assure a steady flow of informants and creative 

new theories of potential liability. 

On the other hand, if whistle-blowing is effective 

at rooting out fraud scenarios, it is a relatively 

inefficient way to protect patient interests. 

“I think it’s specious for someone to go to the 

government because they think something is 

dangerous,” says Retta Riordan, an independent 

compliance consultant. “We know how long it 

takes for the government to develop a case— 

four to six years. If you really want the activity 

to stop you need to do that immediately and the 

best way to do that is in-house.” 

To that end, almost every fraud settlement 

is accompanied with a corporate integrity 

agreement (CIA) between the company 

and the government—specifically the US 

Department of Health & Human Services 

(HHS) Office of the Inspector General 

(OIG). CIAs mandate a set of procedures and 

reporting responsibilities to monitor compliance 

and prevent similar conduct from 

recurring. CIAs also help ensure that 

internal processes exist to bring future 

bad conduct to light (Box 1). 

Medical affairs or promotion? 

Recent cases and the resulting terms of the 

CIAs show that government officials have 

been focusing on fraud arising from medical 

affairs operations, both internally and 

externally. In many situations, the misconduct 

has centered on some form of kickback. 

“It’s a somewhat consistent theme 

in our investigations,” says Mary Riordan, 

senior counsel in the HHS OIG. But the 

form of those kickbacks has changed over 

time. “I think we see less blatant cash-tophysicians 

type[s] of kickbacks,” she says. 

“Now the type of activity we are looking 

at is more sophisticated—contractual 

arrangements between doctors and the 

drug companies who engage them.” 

The recent AstraZeneca settlement, 

for one, involved clinical trials activities, 

publications and grants in addition to 

inappropriate product promotion. The 

Pfizer case, which focused largely on the 

promotion of their pain medication Bextra 

(valdecoxib), also had a medical affairs– 

publications component. These areas had not 

traditionally been part of the government’s 

focus. Companies are also now being investigated 

more often for whether they may have 

retained clinicians to conduct ‘sham’ clinical trials—essentially 

another form of kickback. Such 

a claim was made against Boston Scientific’s 

Guidant subsidiary, in Indianapolis, as part of a 

$22 million December 2009 settlement in a case 

that included the implantation of pacemakers 

and defibrillators in clinical trials. 

1232 volume 28 number 12 DECEMBER 2010 nature biotechnology

news feature 


Table 1 Recent settlements of healthcare fraud cases 

Company (date announced) 

Novartis 

(Basel) (September 2010) 

Allergan 

(Irvine, California) (September 2010) 

Forest Laboratories 

(New York) (September 2010) 

Novartis 

(May 2010) 

Ortho-McNeil-Janssen 

(Titusville, New Jersey) (April 2010) 

Schwarz Pharma 

(UCB, Brussels) (April 2010) 

AstraZeneca 

(London) (April 2010) 

Alpharma 

(King Pharmaceuticals, 

Bristol, Tennessee) (March 2010) 

Boston Scientific 

(Guidant division, Indianapolis) 

(December 2009) 

Biovail 

(Valeant, Mississauga, Canada) 

(September 2009) 

Elan 

(Dublin) (settlement pending) 

Teva 

(Jerusalem) (settlement pending) 

There’s also been more focus on the intersection 

of medical affairs and commercial 

functions within companies—the grant evaluation 

process, how publications are developed 

and whether they are being used as disguised 

promotion. An abundance of ten-patient studies 

in off-label areas could be seen as geared 

merely to generate interest in providers. The 

government also is concerned that physicians 

are brought on to company boards so that these 

key opinion leaders can be influenced favorably 

toward in-house products and promote 

the message in the community, rather than 

using the physicians in their advisory capacity. 

Or if a speakers’ bureau brings 500 speakers 

to a nice hotel for training but only uses 100 

of them, it would in effect be a promotional 

activity. “I think our analysis of the situations 

is becoming more complex, as are the situations 

we are seeing,” says Mary Riordan. 

Within a company, even the seemingly 

simple act of responding to an information 

request can blur the line between medical and 

commercial activities and put a company in a 

difficult situation. The Frazer, Pennsylvania– 

based biotech Cephalon gets thousands of such 

requests every quarter, says chief compliance 

officer Valli Baldassano. “We get all kinds of 

requests, not just for off-label uses—about the 

Amount 

($ millions) Issues settled 

422.5 Criminal and civil charges relating to the off-label marketing of Trileptal (oxcarbazepine) and five other drugs in 

2000 and 2001. Whistle-blowers, all former employees of Novartis, receive over $25 million from the settlement. 

600.0 Criminal and civil charges relating to the off-label marketing of Botox (onabotulinumtoxinA) in 2003. 

Whistle-blowers receive $37.8 million. 

313.0 Criminal charges (including obstruction of justice) around the distribution of an unapproved drug, Levothroid 

(levothyroxine), civil liabilities relating to the off-label marketing of Celexa (citalopram) and fraud in the pricing 

of those drugs as well as Lexapro (escitalopram). Conduct occurred between 2001 and 2003. Whistle-blowers 

receive ~$14 million. 

72.5 Civil liabilities relating to the off-label marketing of inhaled tobramycin between 2001 and 2006. Whistleblowers, 

all former employees of Chiron (now part of Novartis), receive $7.8 million. 

81.0 Criminal and civil charges relating to the off-label marketing of Topamax (topiramate). Whistle-blowers receive 

more than $9 million. 

22.0 Civil charges relating to the billing of two unapproved drugs to the Centers for Medicare and Medicaid Services. 

Whistle-blowers receive more than $1.8 million. 

520.0 Civil charges relating to the off-label marketing of Seroquel (quetiapine) from 2001 to 2006. The whistle-blower 

receives more than $45 million. 

42.5 Civil charges relating to bribes paid to healthcare providers to induce them to promote or prescribe Kadian 

(morphine sulfate), and misrepresentations about the drug’s safety and efficacy, from 2000 to 2006. 

The whistle-blower receives $5.33 million. 

22.0 Civil charges that its subsidiary, Guidant, used post-market studies as vehicles to funnel kickbacks to physicians 

to induce them to use the company’s pacemakers and defibrillators. 

24.6 Guilty plea on criminal conspiracy and kickback charges relating to its drug Cardizen (diltiazem) beginning in 

2003. Amount includes a $2.4 million civil fine. 

203.5 Company has announced an agreement with the US Attorney’s Office in Massachusetts to settle claims around 

the marketing of Zongran (zonisamide). 

169.0 Texas Attorney General’s Office announced an agreement with the company to settle claims relating to improper 

price reporting for its drugs. Whistle-blower Ven-A-Care triggered the investigation. 

Source: Press releases from the US Department of Justice; US Attorney’s Office District of Massachusetts; and Attorney General of Texas. 

science of the drugs, their pharmacokinetics, 

their uses and what data are available for both 

on- and off-label uses. They can come in 

through a phone call, e-mail or because a doctor 

has asked a sales rep, which is how many of 

the sensational cases arise.” 

Follow the money 

By dint of its settlement over off-label promotion 

of its narcolepsy drug Provigil (modafinil), 

which included a CIA, Cephalon has helped 

set the industry standard for the processes for 

tracking spending on speakers, consultants 

and the like—a focus of DoJ’s inquiry into the 

company and also an element in the Physicians 

Payment Sunshine Act of 2009, now incorporated 

into the recent federal healthcare reform 

legislation, which takes effect for all companies 

in 2013. ProPublica now posts these payments 

by Cephalon on its website, along with those 

from the six large drug companies that have 

also made them public (http://projects.propublica.org/docdollars/). 

The mere disclosure of payments is an area 

where consumers, providers, competitors 

and the public at large can see if there are particular 

trends or payments. “With ‘Sunshine,’ 

presumably comes a desire to reduce potential 

conflicts of interest,” says Howard Young, an 

attorney with Morgan Lewis in Washington, 

DC. (Academic institutions, for their part, are 

also cracking down on this potential conflict 

of interest; Harvard University, for example, 

is implementing a policy starting January 1, 

2011, prohibiting faculty from giving promotional 

talks for drug and device makers or 

receiving gifts, travel stipends and meals.) 

Pharmaceutical companies pay doctors to 

do services for them under fee-for-service 

arrangements. These can include speaker 

programs, promotional talks or participation 

advisory boards, including commercially oriented 

boards that brainstorm marketing messages. 

There’s also a bucket of spending for fees 

around research participation, including conducting 

clinical trials. What’s more, there are 

incidental expenses for travel and meals. 

“The popular thought is that pharma companies 

can push a button and miraculously 

every dollar you paid to Dr. X shows up in 

a database,” says Baldassano. “That’s just not 

true. It’s a huge undertaking and different for 

every company.” When the government wanted 

Cephalon to calculate aggregate expenditure as 

part of its CIA, the company balked. “We said 

we’ve been trying to build it, have some components 

of it for state reporting, but we don’t have 

it the way you are envisioning it,” she says. “So 

nature biotechnology volume 28 number 12 DECEMBER 2010 1233

NEWS feature 


you need to give us time to build it.” As a result, 

the aggregate expenditure system in the CIA is 

staged, with various deadlines. 

For example, it’s easy to capture a consulting 

fee paid for services; there’s usually a contract 

and a direct payment. Cephalon is already 

posting those payments publicly and every 

quarter going forward. But other spending— 

for a pizza or sandwich, and for travel fees 

ancillary to research services—may come 

through an expense report system or a logistics 

vendor. “It doesn’t sound like it may be complicated,” 

says Baldassano. “But it is, to make 

sure you capture it all accurately.” 

Trend spotting 

A close examination of the provisions in 

CIAs can help other companies determine if 

their own processes represent the best practices 

within the industry, and whether they 

should be adopting other measures. “From 

an educational perspective, they’ve had that 

kind of broad-based effect on the industry,” 

says Young. 

Box 1 Getting the house in order 

Many pharma and biotech companies have responded to increased 

government scrutiny of their sales and marketing practices by 

adding substantially more training of staff with respect to off-label 

promotion. “We are seeing a real effect,” says Howard Young, an 

attorney with Morgan Lewis in Washington, DC, and former senior 

attorney and deputy branch chief with the Office of the Inspector 

General (OIG) of the Department of Health & Human Services. 

“The prosecutors themselves have stated that the industry appears 

to be moving in the right direction,” he notes. 

“I think the amount of money being spent and the number 

of FTEs [full-time employees] being devoted toward regulatory 

compliance over the past five to ten years reflects a real 

understanding of most of the industry that these are obligations 

that are here to stay and they are being institutionalized in a 

meaningful way,” adds T. Reed Stephens, a former DoJ lawyer now 

with the Washington, DC, law firm McDermott Will & Emery. 

OIG came out with a guidance for pharmaceutical companies on 

compliance practices in 2003, to prevent and reduce fraud and 

abuse in federal healthcare programs (http://oig.hhs.gov/authorities/ 

docs/03/050503FRCPGPharmac.pdf). Since then, most large 

pharma companies have established formal programs, as have most 

mid-sized biotech companies and even many smaller companies. 

For companies that have run afoul of the law, OIG requires—almost 

without exception—that those companies sign a CIA mandating 

the establishment of compliance procedures aimed at preventing 

the prior contentious conduct from recurring. This is not unique to 

the healthcare industry: Any company involved with civil or even 

criminal claims is going to be entering into a CIA whether in the 

insurance business, the defense business or pharma. 

“At that point, DoJ has taken care of your sins of the past 

and OIG is there to make sure you fly right in the future,” says 

Thomas Gunning, general counsel of EMD Serono in Rockland, 

Massachusetts, which settled a case involving off-label promotion 

of its human growth hormone, Serostim, with DoJ in 2005. 

“Every CIA is an opportunity to look at new 

provisions and see what the government is focusing 

on,” adds Retta Riordan. “It’s an opportunity 

to see what went wrong. Was it a lack of knowledge? 

A lack of processes in place—policies and 

procedures? A lack of training? And use that as 

a tool to reassess an existing program.” When 

Indianapolis-based Eli Lilly, for example, settled 

a case with DoJ in 2005 over allegations it had 

been promoting its osteoporosis drug Evista 

(raloxifene) for three indications where it was 

only approved for one, Riordan pondered how 

it could have happened, “especially in a company 

the size of Lilly that has a very good compliance 

program,” she says. 

There was evidence that Lilly had developed 

an internal best practices video emphasizing 

the off-label message. Yet although all companies, 

including Lilly, review material going 

out to the public, as it turned out, Lilly’s promotional 

materials review committee was not 

reviewing internal training materials. 

Experienced whistle-blower attorneys will 

focus on these kinds of situations: products 

that have been on the market for a long time 

and have a track record of sales and profits 

that will potentially constitute the measure of 

damages to gauge the expected payoff from 

the case, says McDermott Will & Emery’s 

Stephens. “You have to be looking at cases 

where there’s been a systematic approach to 

a claim backed by the organization and it is a 

strategy they have undertaken. The cases with 

the greatest exposure will be those where the 

company took a sustained approach and where 

the plaintiff ’s counsel can at least theorize a 

substantial amount—tens of millions or hundreds 

of millions of revenue and on top of that 

a proportionate amount of profit that involves 

federal dollars.” 

Warning letters to companies from FDA, 

many of them directed at communications to 

medical professionals about their products 

that do not adequately discuss the risks of 

therapies, may also tip off a whistle-blower 

attorney to possible fertile ground. They 

could even foreshadow a future DoJ investigation—or 

a lawsuit by competitors—as DoJ 

In virtually every case over the past six years, CIAs have 

included provisions forcing a separation of the compliance 

function from the legal department, in effect instituting another 

set of checks and balances. CIAs also call for the retention of an 

outside independent review organization, which performs annual 

audits of compliance processes and training, and reports to the 

company as a financial auditor would. In cases involving offlabel 

promotion, a CIA would likely have provisions that require 

the company to get ‘verbatims’ from doctors—an independent 

review of what was said in face-to-face sessions with sales 

reps. If a case is focused on kickbacks, the CIA might focus 

on how consultants and speakers are hired and the exact use 

of those monies. CIAs and investigations that focused on best 

pricing issues, rebates and whether Medicare and Medicaid 

bill the right price would get deep into the internal policies and 

practices for price reporting. 

Compliance may be “a further opportunity to embed, from a 

business standpoint, the entire area of process improvement,” 

notes Paul Silver, managing director of The Huron Group, a 

consultancy in Atlanta that often acts as an independent review 

organization. For example, after Cephalon’s settlement of claims 

that the company promoted its narcolepsy drug Provigil (modafinil) 

off-label to treat fatigue associated with multiple sclerosis, 

it essentially “redesigned how we did business,” says chief 

compliance officer Valli Baldassano. 

Compliance systems also help to identify areas of abuse 

proactively. “We’ve made it into a much broader risk management 

function that is much more engrained into the business,” 

Baldassano says. “If you can design a good way to slice and dice 

the data, you may be able to find the trend you are looking for to 

see if you’ve got a problem out there. Then you can either do more 

investigation, stronger training, some kind of intervention that can 

affect that behavior. You have to have the system to smoke out the 

behavior, but the system itself cannot be the panacea.” 


news feature 


closely examines marketing claims that are 

regulated by FDA to determine whether they 

also constitute off-label violations. 

Expanding on the claims beyond what are 

in FDA-approved statements is a little bit different 

from off-label promotion, Stephens 

says. It’s typically an FDA-focused issue but 

also raises the specter of making comparative 

claims about a product that run afoul of the 

federal civil FCA. The FCA is the principal 

mechanism for pursuing claims of off-label 

promotion and, in cases where the government 

is the customer, kickbacks and illicit pricing 

schemes, including rebates to Medicare. The 

so-called qui tam provisions of FCA also keep 

in place a structure for whistle-blowers to 

come forward and alert the government to 

some of those schemes. And once it becomes 

known that a company is under investigation, 

many states then start class actions. 

Because of the huge advertising potential of 

social media—Twitter, Facebook, Wikipedia 

and the like—their use is bound to be fertile 

ground for pharmaceutical companies to 

get out their message, and for investigation. 

“It’s going to create some real challenges for 

pharmaceutical companies to track where 

their content goes,” says Stephens. FDA is 

already working on guidelines for how the 

pharmaceutical industry can use social media 

(Nat. Biotechnol. 28, 641–643, 2010). But the 

mere fact that FDA is going to regulate those 

types of interactions means there are going 

to be opportunities for mistakes to be made 

and for companies to be overly aggressive, 

before guidelines come into place, according 

to Stephens. 

In addition, companies operating under 

CIAs are required to submit marketing and 

promotional plans to the government ahead of 

time. “That’s one of the most exacting aspects 

of these CIAs, giving the government the ability 

to preview what the company is planning 

on doing,” Stephens says. “DoJ could make 

the allegation, it’s not just an FDA violation 

but that you are promoting off-label by saying 

something lacking a scientific basis,” he 

says. “[In addition], they could solicit whistle-blowers 

from within that company—they 

would explain what the theory is and hope 

they could bring to them factual allegations 

they could attach to their theory and quickly 

file a case,” Stephens says. 

Criminal prosecutions 

According to Shelley Slade, government 

officials are coming to the conclusion that 

although the FCA “is somewhat of a deterrent, 

it’s not enough.” Increasingly, DoJ is turning 

to the Foreign Corrupt Practices Act (FCPA), 

which gives it some jurisdiction over conduct 

outside the US. For example, this August, 

Merck disclosed that it is being investigated 

for violations under the FCPA, which DoJ 

had used to penalize Novo Nordisk in 2009, 

for kickbacks paid to the former Iraqi government 

as an incentive to obtain insulin supply 

contracts. FCPA-related investigations are also 

ongoing for several other pharmas. 

But the real big stick will come in the form 

of criminal prosecutions of individuals spearheading 

these schemes, Slade says. Indeed, a 

trend over the past several years has been for 

DoJ and OIG to assess individual liability and 

responsibility for fraudulent schemes, along 

with requiring boards and senior management 

to be more active in compliance programs. 

As with any industry, compliance units 

within pharma and biotech companies come 

up against the fact that people are rewarded 

for improving a firm’s financial performance— 

right up the chain to the CEO, Slade says. 

“Their concerns are pleasing Wall Street and 

shareholders, and bonuses are driven by the 

financial performance. There’s a real conflict 

between what the compliance folks want them 

to do and what’s in their immediate financial 

self-interest. Sometimes the compliance people 

don’t get the full story from the people in 

operations.” Thus, despite the best of intentions, 

prosecutors know that when they speak 

at compliance conferences they are basically 

preaching to the choir. 

“Holding individuals accountable, and 

criminal settlements—the whole combination 

of steps—will lead to better compliance,” 

says OIG’s Mary Riordan. “We are increasingly 

looking at ways of holding individuals 

accountable,” she says, both through CIAs 

and also through so-called exclusion actions 

against individuals, which would bar them 

from participating in the industry. OIG currently 

has an initiative underway to target 

the responsible corporate executives through 

exclusion actions of individuals— individuals 

they believe were in positions to prevent 

alleged fraudulent conduct or should have 

been aware of it and stopped it. 

The most dramatic case in point has been 

the criminal conviction (now under appeal) 

of three individuals from Purdue Pharma 

in Stamford, Connecticut, who were found 

responsible for not preventing deceptive marketing 

practices for the narcotic Oxycontin 

(oxycodone). And on November 8, DoJ 

indicted a former attorney for London-based 

GlaxoSmithKline (GSK, Brentford, UK) charging 

her with obstruction of justice and making 

false statements to FDA in connection with 

an investigation into GSK’s promotion of the 

antidepressant Welbutrin (buproprion) early 

in the decade. That case is still pending, but in 

the fourth quarter of 2008, GSK set aside $400 

million, anticipating a settlement. 

To date, the Purdue case is the only situation 

where OIG has exercised its authority in 

this way. But there’s whispering among attorneys 

that more such cases are coming down 

the pike. Given an appropriate case with clear 

criminal violations and egregious circumstances 

where patient health is being compromised 

by the off-label use, and executives well 

aware of what the sales force has been doing 

and of the risk of harm to patients, “the only 

way the executives’ behavior is going to change 

is if the government treats these as criminal 

violations and goes after the executives,” says 

Slade. “If they see their peers going to jail, it 

will offset the financial calculus for them.” 

OIG could also exclude an entire company 

from the industry, but that’s a ‘nuclear bomb’ 

scenario that would raise serious issues of 

patients’ rights and access to drugs, as well as 

antitrust. 

Too early to know 

In the meantime, the number of fines and 

prosecutions continues to increase. And 

industry is devoting more and more resources 

to ensure that it does not fall foul of government 

officials. “We know that companies are 

absolutely continuing to ramp up their compliance 

efforts and staffing up their compliance 

groups internally,” says New York attorney 

Arnold Friede. “They are hiring more people 

and propagating their compliance policies 

more forcefully internally. But whether that’s 

accompanied by a cultural shift, particularly in 

an era of organizational uncertainty, is an open 

question. Process and culture are not necessarily 

coincident, and it probably takes an equal 

measure of both to make sure that you are 

working towards effective compliance.” 

Overall, “it’s too early to tell” whether the 

recent ramp-up of enforcement activities will 

have the desired deterrent effect, says Jerry 

Avorn, professor of medicine at Harvard 

Medical School and the author of Powerful 

Medicines, one of the first books to examine 

pharmaceutical detailing practices—the oneon-one 

conversations between sales reps and 

physicians—in detail. 

“We don’t have a good objective way to measure 

changes in corporate behavior, but we certainly 

hope the CIAs are an effective tool,” says 

HHS’s Mary Riordan. 

Mark Ratner, Cambridge, Massachusetts 


NEWS feature 


New twists on proteasome inhibitors 

Only one drug targeting the proteasome protein degradation 

pathway has been approved, but several second-generation 

inhibitors are making progress in trials. Jim Kling reports. 

In July, Onyx Pharmaceuticals of Emeryville, 

California, announced positive phase 2 results 

for its proteasome inhibitor carfilzomib, which 

is currently being tested in multiple myeloma. 

The drug achieved a 24% response rate in individuals 

with advanced cancer, who had failed 

on all prior therapies, more than double the 

expected response rate (11%). This was not 

only good news for the company—whose stock 

jumped 21% and which clinched a marketing 

deal worth >$300 million with Japanese firm 

Ono Pharmaceuticals of Osada—but also a 

boost for the field of proteasome inhibition, 

which thus far has only a single success story, 

Velcade (bortezomib), which was first approved 

in multiple myeloma in 2003. 

Carfilzomib, however, is one of several 

proteasome inhibitors currently progressing 

through the clinic. Many of the compounds 

causing the most buzz are those that target 

another part of the cell’s proteolytic machinery, 

the ubiquitin ligases, which number in the 

hundreds (Fig. 1) and could potentially provide 

greater clinical specificity in the clinic. 

The trailblazer 

First approved as a third-line therapy for 

refractory multiple myeloma, Velcade was 

extended to mantle cell lymphoma in 2006 

and became a frontline treatment for multiple 

myeloma in the US in 2008. An injectable 

tripeptide (modified with a boronic acid at 

one end, Pyz-Phe-boroLeu), it binds the catalytic 

site of the 26S proteasome (Box 1). The 

mechanism of action, with respect to multiple 

myeloma, is straightforward: myeloma 

cells are essentially protein factories, having 

descended from B lymphocytes, which create 

antibodies. “They fail to turn off their 

protein production machinery, causing proteotoxic 

stress. Cells deal with it by upregulating 

the proteasome. So the cell is gummed 

up with proteins anyway, and then we shut 

off the primary mechanism by which they’re 

getting rid of the stress. That’s why Velcade 

is effective,” says Joe Bolen, chief scientific 

officer of Millennium (now part of Takeda in 

Osaka), the company that brought the drug 

to market. 

Sensitivity of tumor cells to proteasome 

inhibitors has been documented 1 , and such 

observations have prompted companies to 

consider them for treating other cancers, 

particularly solid tumors. But Velcade turns 

out to be not all that precise in its targeting. In 

disrupting the proteasome, it affects the breakdown 

of a range of proteins and interferes with 

some proteases, which is thought to be associated 

with the observed neurotoxicity. 

The next generation 

Velcade is an unqualified success in the marketplace—the 

drug generates about $1.5 billion 

in revenues worldwide annually, according 

to David Moskowitz, who follows Onyx in 

his role as managing director with Madison 

Williams investment bank in New York. But 

its limitations, which include the development 

of resistance to it in addition to neurotoxicity, 

have inspired the development of a second 

generation of proteasome inhibitors, including 

compounds like carfilzomib. Carfilzomib is an 

irreversible epoxomicin-related peptide that 

binds specifically the chymotrypsin-like protease 

activity of the proteasome, hitting fewer 

nontarget proteases than Velcade, which may 

account for its reduced toxicity. 

Nereus Pharmaceuticals in San Diego is 

developing an entirely different class of compounds 

represented by NPI-0052 (salinosporamide 

A), a β-lactam isolated from the 

marine bacterium Salinispora tropica. NPI- 

0052 blocks all three catalytic activities— 

chymotrypsin-like hydrolysis, trypsin-like 

hydrolysis and peptidylglutamyl peptide 

hydrolysis—whereas Velcade inhibits just 

the chymotrypsin-like activity. NPI-0052 has 

Box 1 A protein degrading machine 

shown activity against multiple myeloma cells 

resistant to Velcade, and in animal models it 

inhibits colon, pancreatic and lung cancers. 

The drug is currently in phase 1 clinical trials 

for multiple myeloma and lymphoma, as a 

single agent and in combinations (Table 1). 

In addition to carfilzomib, which entered 

the Onyx pipeline with its 2009 acquisition of 

the biotech company Proteolix, Onyx is developing 

two other Proteolix compounds. ONX 

0912, an oral proteasome inhibitor, is currently 

in phase 1 clinical trials for advanced refractory 

or recurring solid tumors. ONX 0914 

inhibits the related immunoproteasome, a 

specialized type of proteasome mainly found 

in monocytes and lymphocytes. In preclinical 

models, it was shown to be 10- to 50-fold more 

selective for the immunoproteasome than for 

the general proteasome. Possible indications 

include rheumatoid arthritis, inflammatory 

bowel disease and lupus. 

Michael Kauffman, Onyx’s chief medical 

officer, thinks carfilzomib has the potential to 

be a frontline treatment for multiple myeloma 

because it shows low levels of neuropathy, neutropenia 

and other toxicities associated with other 

multiple myeloma treatments. “We’re seeing very 

minimal long-term side effects in patients, and 

none so far that have precluded indefinite dosing,” 

he adds. That’s important because the goal is 

to make multiple myeloma a chronic condition. 

Kauffman also believes that the ability to keep 

patients on the drug should help prevent tumors 

from developing resistance. 

Millennium is also working on a secondgeneration 

proteasome inhibitor, an orally 

active peptide boronic acid analog, MLN 9708. 

Unlike Velcade, MLN 9708 has activity against 

a range of solid tumors in preclinical studies 

and is currently in phase 1 trials for hematologic 

malignancies and solid tumors. “We’re 

really striving to bring this next generation of 

proteasome inhibitors beyond the hematologic 

malignancies,” says Bolen. 

The proteasome is a multicatalytic enzyme comprising a hollow 20S core particle complexed 

with two 19S regulatory particles to form a 26S structure where protein degradation takes 

place. The two regulatory caps are bound to the core particle through ubiquitin-binding sites. 

The 20S particle has three catalytic actions: chymotrypsin-like activity, trypsin-like activity 

and peptidylglutamyl peptide hydrolyzing activity. 

Proteasome activity occurs through a multistep process. Proteins are marked for 

destruction by ubiquitin tags. This process begins with activation of ubiquitin (E1 enzyme), 

followed by conjugation (E2 enzyme) and finally specific ligases (E3) that mediate 

attachment of ubiquitin chains to target proteins. The pathway is hierarchical, with a single 

(or small number of) E1 enzyme(s) interacting with a small number of E2 enzymes, and 

several hundred E3 ubiquitin ligases serving specific subsets of proteins (Fig. 1). Specific 

ubiquitin ligases tag target proteins with a string of ubiquitins, which act as a signal to the 

proteasome to destroy it. 


news feature 


Even so, achieving success in solid tumors 

rather than hematological cancers will be a challenge 

as blood cells have high concentrations of 

proteasomes; indeed, Millennium researchers 

suspect that they act as a sink for tight-binding 

proteasome inhibitors, such as Velcade, thereby 

conferring efficacy in blood cancers. “We reasoned 

that a proteasome inhibitor with a shorter 

half-life [on the proteasome] might have better 

tissue distribution. This is of particular interest 

for the potential utility of proteasome inhibitors 

for treating nonhematological malignancies,” 

says Bolen. From a drug discovery textbook 

point of view, however, it was completely counterintuitive. 

“The mantra of drug development is 

to be specific and to stick to the target with high 

affinity or irreversibly,” he adds. 

Two hits are better than one 

Preclinical data support using proteasome 

inhibitors in combination with other drugs. 

“They should combine favorably with pretty 

much any type of chemotherapy,” says 

Kauffman. Some combination strategies have 

a theoretical basis as well, according to Robert 

Orlowski, a professor of cancer medicine at the 

University of Texas M.D. Anderson Cancer 

Center (Houston) 2 . 

The website ClinicalTrials.gov lists almost 

300 clinical trials with Velcade in combination 

with other drugs, including other targeting 

agents, such as Bristol-Myers Squibb’s 

(Princeton, NJ, USA) elotuzumab (an antibody 

targeting the myeloma cell-surface protein 

CS-1), Centocor Ortho Biotech’s (Horsham, 

PA, USA) siltuximab (an anti-interleukin-6 

antibody) as well as Pfizer’s (New York) Akt 

inhibitors (Viracept, nelfinavir mesylate) and 

heat shock protein 90 inhibitors (tanespimycin, 

Bristol-Meyers Squibb, New York). Velcade 

has also been combined with DNA-damaging 

therapies such as radiation and anthracyclines, 

because proteasome inhibitors suppress DNA 

E1 

E2s 

E3s 

? 

Substrates 

repair proteins, says Ken Anderson, a professor 

of medicine at the Dana-Farber Cancer Institute 

and Harvard Medical School and collaborator 

with Millennium and Nereus. “We’re nowhere 

near yet to learning the best combination, nor 

are we near to maximizing the benefit of this 

drug for patients,” says Orlowski. 

Palladino and others are hopeful that these 

combinations will allow proteasome inhibitors 

to expand beyond multiple myeloma, but 

not everyone is convinced. “It feels like they’re 

just throwing stuff up against the wall, looking 

for a signal. The risk-reward ratio (for solid 

tumors) is viewed favorably [by investors], but 

expectations are very low,” says Moskowitz. 

Beyond the proteasome 

Farther on the horizon, some companies are 

looking to attack the other half of the proteasome-ubiquitin 

pathway. Ubiquitin ligases, of 

which there are hundreds in mammalian cells, 

tag proteins with ubiquitin, marking them 

? 

? 

? 

? ? 

Figure 1 The pyramidal ubiquitin cascade. The ubiquitin cascade is pyramidal in design. A single 

E1-activating enzyme transfers ubiquitin to roughly three dozen E2s, which function together with 

several hundred different E3 ubiquitin ligases to ubiquitinate thousands of substrates. (Reprinted with 

permission from Nalepa, G. et al. Nat. Rev. Drug Discov. 5, 596–613, 2006) 

for destruction by the proteasome (Box 1). 

If researchers can find ways to modulate 

specific ligases, or their counterparts, the deubiquitinases, 

they may obtain greater specificity. 

“Proteasome inhibitors block the whole 

protein degradation machinery. I think at the 

end of the day, people will realize that this 

is not where selective therapy is going,” says 

Tauseef Butt, president and CEO of Progenra 

(Malvern, PA, USA), which is dedicated to 

drug discovery in the ubiquitin area. 

Millennium has such an inhibitor in early 

trials: MLN 4924, currently in phase 1 trials 

for hematological and solid tumors, inhibits 

NEDD8-activing enzyme. NEDD8 is a 

ubiquitin-like protein that, when activated, 

controls the activity of cullin-RING ligases, 

the largest class of E3 ligases. The company 

also has early-stage programs in ubiquitin and 

ubiquitin-like protein pathways. 

To date, ubiquitin ligases have shown a 

high degree of substrate specificity, making 

? 

Table 1 Selected agents targeting protein degradation in trials 

Company Drug/drug target Phase of development 

Takeda/Millennium 

Onyx Pharmaceuticals 

Velcade/proteasome 

MLN 9708/proteasome 

MLN 4924/Nedd8-activating enzyme 

MLN 519 

Carfilzomib/proteasome 

Approved for multiple myeloma (5/03) 

Approved for NHL (12/06) 

Phase 2 for breast cancer, ovarian cancer, myelodysplastic syndrome or 

Waldenstrom macroglobulinemia 

Phase 1/2 for amyloidosis or multiple myeloma (relapsed, refractory) 

Phase 1 for small cell lung cancer or solid tumors 

Phase 1/2 for multiple myeloma 

Phase 3 for multiple myeloma 

Phase 1/2 for solid tumors 

Phase 1 for solid tumors 

ONX 0912/proteasome 

Cephalon (Fraser, PA, USA) CEP-1877/proteasome Phase 1/2 for multiple myeloma 

Nereus Pharmaceuticals NPI-0052/proteasome Phase 1 for solid tumor or multiple myeloma 

Johnson & Johnson JNJ-26854165/MDM2 Phase 1 for solid tumors 

Roche RG7112/MDM2 Phase 1 for hematological cancers or solid tumors 

NHL, non-Hodgkin lymphoma. 


NEWS feature 

Box 2 The other protein degrading system 

The proteasome and ubiquitin systems are not the sole systems for 

degrading protein in cells. Autophagy—whereby proteins are first 

packaged in vesicles that then fuse with lysozymes—is another 

process that has attracted attention. 

Tony Wyss-Coray, at the Stanford University School of 

Medicine, along with Beth Levine, at the University of Texas 

Southwestern Medical Center (Dallas), found that knocking 

out a key autophagy protein, Beclin-1, in mice could lead to 

greater neurodegeneration and beta-amyloid accumulation 

in the brain. Furthermore, there is evidence that Beclin-1 

levels are low in Alzheimer’s disease patients’ brains, whereas 

some other proteins involved in autophagy are unchanged. 

In mice, restoration of beclin-1 levels can reduce β-amyloid 

accumulation and reverse some of the degenerative changes. 

“I think there is a possibility that this may work in humans 

as well,” says Wyss-Coray. His team is now screening for 

compounds that can restore autophagy in cells that are 

deficient in Beclin-1. 

Autophagy also has a strong connection to cancer. The beclin- 

1–deficient mice are more prone to developing cancer at an 

advanced age than normal mice, and Beclin-1 is deleted in humans 

in about 40% of prostate cancers, 50% of breast cancers and 75% 

of ovarian cancers, according to Shengkan Jin, at Rutgers University 

in Piscataway, New Jersey. Autophagy also provides a means for cells 

to survive periods of stress, such as starvation, and some tumor cells 

may use it to survive the onslaught of chemotherapy. 

Like the proteasome, autophagy appears able to show some 

specificity in the proteins it degrades, but that process isn’t well 

understood. The two processes are somewhat complementary. 

“There are suggestions that if you inhibit one pathway, the other can 

take over. But I’m not aware of key proteins that play roles in both 

processes,” says Wyss-Coray. 


them attractive drug targets. “These [ubiquitin 

ligases] are the regulatory proteins of 

the ubiquitin system. It is the most selective 

intervention point,” says Avishai Levy, CEO 

of Proteologics (Orangeburg, NY, USA). His 

company is collaborating with Teva (Tel Aviv, 

Israel) to discover inhibitors of ring finger E3 

ligases, including drugs against inhibitors of 

the POSH polypeptide currently in preclinical 

testing for the potential treatment of HIV 

infection and cancer. 

Several groups are looking at key points in 

signal transduction pathways where ubiquitin 

ligases play a role. “There is [evidence] that G 

protein–coupled receptors are being managed 

by ubiquitination. Channels are being ubiquinated 

and recycled,” says Butt. For example, 

growth hormone receptors are internalized 

after binding growth hormone, followed by 

phosphorylation, ubiquitination and deposition 

into the lysosome. There, the ligand 

drops off the receptor, the terminal ubiquitin is 

cleaved by a de-ubiquitinase, and the receptor 

is returned to the cell surface, which can contribute 

to tumorigenesis. “If you inhibit the deubiquitinases 

involved, the ubiquitin remains 

in place and the receptor becomes a target for 

the proteasome. Now the cell doesn’t make a 

commitment to proliferation,” says Butt. 

The key is determining the ubiquitin status 

of disease-causing proteins. “It’s like the 

phosphorylation status of a kinase substrate,” 

says Donald Payan, chief scientific officer at 

Rigel Pharmaceuticals, which has a ubiquitin 

ligase program in cancer that focuses 

on the SCF/p27 complex. The p27 protein is 

a cyclin-dependent kinase inhibitor involved 

in the transition of a cell from G1 to S phase. 

SCF ubiquitinates p27. Inhibiting SCF should 

leave more p27 in place to act as a brake on 

cell cycle progression. 

Just as with kinases, which took 20 years to 

become established as drug targets, it could 

easily take as long for the ubiquitin ligase field 

to mature. One problem with designing U3 

ligase inhibitors is the difficulty of interfering 

with protein-protein interactions. “You have 

to look for sites or regions that are druggable,” 

says Payan. Even so, Orlicky et al. described 

a small-molecule inhibitor of the yeast F-box 

protein Cdc4, which is a central component 

of the cullin-RING ligases 3 . The inhibitor 

targets the WD40 propeller domain, and the 

structural changes induced by the inhibitor 

affect the substrate-binding pocket. If similar 

interactions occur in mammalian ligases, 

the strategy could open up a number of novel 

drug targets. 

Indeed, Payan claims an ongoing collaboration 

between Rigel and Daiichi Sankyo 

(Tokyo), initiated in 2002, is about to start 

human testing of a ubiquitin ligase inhibitor 

in an undisclosed cancer. Elsewhere, Apeiron 

Biologics (Vienna) is investigating short interfering 

RNAs that silence the E3 ubiquitin 

ligase Casitas B-cell lymphoma-b (Cbl-b) as a 

means of modulating T-cell activity for cancer 

treatment. “It’s a field where people have been 

wringing their hands about how the targets are 

not druggable, but there are four or five companies 

(nearing) the clinic,” he says 

Rigel’s own focus is not on cancer, but muscle 

atrophy. The ligases atrogin-1 and MuRF-1 

are upregulated in patients who experience 

muscle atrophy when being intubated 4 . “If 

you could block MuRF-1 and/or atrogin, you 

might have a big impact on muscle atrophy, 

which is a big opportunity,” says Payan. 

In many ways, the ubiquitin field is even 

more diverse than the kinase field, and significantly 

more difficult. “Key proteins all 

have their ligase hovering above them, ready 

to pounce. What has held the field back is that 

these are intracellular targets, so you’re stuck 

with small molecules. Companies must determine 

what’s the best way to go after these [protein 

interactions] chemically,” says Payan. 

The yin and the yang 

Not everyone is convinced of the potency of 

ligases. “There have been failures with ligases,” 

says Butt. He thinks the future may lie in deubiquitinases, 

which remove ubiquitin from 

proteins, about 100 of which have been identified 

in humans. 

“If the yin is the ligase, the yang is the deubiquitinase, 

and the yang is really taking off 

at the moment,” says Butt. Progenra has an 

inhibitor of the de-ubiquitinase USP7, which 

acts on the p53 tumor suppressor protein. The 

drug candidate has shown good anti-tumor 

activity in animal models and will be developed 

for multiple myeloma, Butt says. 

Ubiquitinases and de-ubiquitinases are 

just the latest targets in protein homeostasis, 

which will continue to evolve (Box 2). 

The ubiquitin field has significant challenges 

ahead. Researchers must achieve selectivity 

for the ligase of interest, and then build the 

infrastructure to study it. 

It’s a story Payan has heard before. “The 

guys who were involved in developing Velcade 

kept pushing and pushing, when people said it 

would never work. Kudos to the proteasome 

guys [for getting to the market]. That’s probably 

the least specific site to intervene.” 

Jim Kling, Bellingham, Washington 

1. Adams, J. Cancer Cell 5, 417–421 (2004). 

2. Rumpold, H. Biochem. Biophys. Res. Commun. 361, 

549–554 (2007). 

3. Orlicky, S. et al. Nat. Biotechnol. 28, 733–737 

(2010). 

4. Levine, S. et al. N. Engl. J. Med. 358, 1327–1335 

(2008). 


uilding a business 

Around the block 

Y Philip Zhang 

A blocking patent can kill your business before it’s off the ground. 


One of the most unpleasant surprises for 

any entrepreneur at a startup company is 

learning that a blocking patent threatens the 

commercial viability of their lead product. This 

could seriously jeopardize the company’s ability 

to raise financing, its attractiveness to strategic 

partners and ultimately its overall viability 

as a business. 

In this article, I provide a few pointers on 

how to avoid blocking patents in the first place 

and then go on to explain how to take evasive 

action if you are in the unfortunate position of 

sweating one out. 

First, avoid 

Officially, a blocking patent is one that is 

valid and enforceable and could prevent 

your potential product from coming to or 

perhaps staying on the market. In the United 

States and most industrialized countries, an 

actual or imminent patent infringement is 

grounds for government-sanctioned injunctions 

and monetary damages. Faced with 

hefty capital demands and long, perilous 

clinical trials, drug developers tend to avoid 

product candidates that present patent uncertainties 

and litigation risks, even if the drug 

satisfies unmet needs of patients. Concerns 

over patent infringement routinely interrupt 

and sometimes derail commercialization of 

drugs. Thus, freedom to operate (FTO)—the 

ability to develop and market a drug or device 

without infringing the valid and enforceable 

patent rights of others—should be high on 

your CEO agenda. 

The starting point for securing FTO is to 

clearly identify the business space for your 

company. A blocking patent is relevant only 

if it affects your business. A first-level inquiry 

asks: What are the current and future (including 

Y. Philip Zhang is cofounder and co–managing 

principal at Milstein Zhang & Wu, Newton, 

Massachusetts, USA. 

e-mail: philip.zhang@mzwiplaw.com 

potential) businesses that you are conducting 

and plan to conduct? A company in the drug 

discovery space faces different challenges than a 

company that is in the in vitro diagnostics space, 

for instance. A developer of new drugs typically 

must have a good understanding of discrete 

patent spaces around its drug targets and candidates 

in the pipeline. A company in the diagnostic 

or genetic testing business, on the other 

hand, may have to walk through the minefield 

of ever-increasingly patented biomarkers and 

immunoassays. A small-molecule drug developer 

generally is not too concerned with drug 

manufacturing. If you are developing recombinant 

proteins, however, manufacturing patents 

are plentiful and should be watched for. 

Next, you should identify the technologies 

that are vital to your company, and this may be 

tougher than determining your business space. 

It is important for you to have a mental list of the 

critical and enabling technologies for successful 

commercialization of your products. You need 

to know what technologies are necessary to 

build your products and clarify which are your 

own innovations and which are other people’s 

technology that you need access to. A developer 

of a single-molecule DNA-sequencing 

technology, for example, must evaluate FTO 

for its core sequencing platform but may also 

need to review reagents, optics, electronics and 

other enabling technologies. A bioethanol producer 

developing a new fermentation process 

may have to review its biomass pretreatment 

protocols and microorganism production steps 

as well as overall process integration. 

Once you understand your business and 

technology space, you’ve paved the way for 

the next step: understanding the patent environment 

in which your business operates. The 

patent landscape affects your patent needs for 

protecting your business and your FTO. A 

common misconception is that if you have a 

patent on something then you are free to practice 

the technology disclosed or claimed in 

your patent. This is not true. 

A patent grants its owner the right to exclude 

others from practicing the claimed invention 

without the owner’s authorization, but it does 

not grant the owner a right to practice the 

claimed invention or technology disclosed in 

the patent. So even if you patented a technology, 

it doesn’t mean that you necessarily can 

use it yourself because your patented technology 

could be dominated by another patent. 

Thus, it is important early on to have a 

solid grasp of the patent environment surrounding 

key product candidates through 

rigorous patent search and review. The patent 

search and review process is preferably conducted 

by experienced legal counsel working 

closely with your project team. Scientists 

should resist the temptation to play patent 

lawyer. Understanding the science in a patent 

is important, but it’s only the first step 

in understanding the impact of the patent, 

which is better done by, or with the guidance 

of, experienced counsel. A good patent lawyer 

is one who understands your business and 

the industry you operate in and can effectively 

communicate with your management and 

scientists. He or she understands the business 

ramifications of the patent issue at hand and 

helps you make the right judgment calls. 

You and your counsel should map out the 

existing patent landscape along the entire commercialization 

path, even if your business plan 

is to seek exit before market entrance. If you 

find a patent that presents FTO risks, do not 

panic. The good thing is that not everything is 

patented (even though it might seem like it), 

and more often than not you will be able to find 

a path forward. Ideally, patent searches should 

cover both granted patents and published patent 

applications and additionally encompass all 

intended international markets for the product. 

Knowing where a patent is granted or pending 

helps with assessing its overall impact and 

affords you an early appreciation for the scope 

of your need for licensing, design-around 

options or patent challenge. 




Box 1 Blocking patent? No 

Determining patent infringement involves a two-step analysis: first, construction of the 

claim to determine the coverage of a patent, and second, comparison of the alleged 

infringing composition, device or method with the construed claims. Thus, overcoming a 

patent means understanding its claim scope and designing the product outside such scope. 

Consider the patent fight between Novartis Pharmaceuticals and Eon Labs 

manufacturing. Novartis has a patent on a drug formulation that claims a hydrosol, 

which includes solid particles of the drug compound and a stabilizer that maintains 

the size distribution of the particles. The claim further specifies the water solubility 

of the drug compound as well as the ratios of drug compound to water and drug 

compound to the stabilizer in the hydrosol. 

Along comes Eon. It sells a drug product that includes the same drug compound 

but not in hydrosol form. Eon instead makes capsules that contain the drug compound 

dissolved in a small amount of ethanol. There is no water in these capsules, and the drug 

compound inside the capsule is completely dissolved in ethanol (not in particle form). 

Novartis sued Eon for infringement, asserting that when Eon’s capsule is ingested 

by a patient, an infringing hydrosol is formed when the capsule mixes with the 

aqueous environment of the patient’s stomach. This unusual infringement theory did 

not survive summary judgment at the trial court (and on appeal) as Novartis’ hydrosol 

claim was construed to be limited to a medicinal preparation consisting of a dispersion 

of solid particles in an aqueous colloidal solution prepared outside of the body, which 

Eon’s capsules clearly are not 1 . 

Another factor that may require some soul 

searching on your part is the risk profile of 

your company. Clearly articulating your risk 

tolerance can be difficult, but in many ways 

it shapes the appropriate approach to a patent 

problem. Your risk profile may change over 

time and is determined by a number of factors, 

including the business sector you operate 

in, the stage of your company, your financial 

situation, the liability exposure and the personalities 

of the management. The life science 

community in general and venture capital 

investors in particular are quite risk averse 

when it comes to patents and FTO. Because 

of the large capital need and lengthy and risky 

development process, investors do not like to 

deal with patent uncertainties and litigation 

threats. Technology and regulatory risks are 

already plentiful in biotech—you don’t need 

anything more. Heightened patent risks could 

drive your board and the investors nuts. 

A review of your FTO should not be a onetime 

occurrence, because a thorough search 

and review doesn’t necessarily uncover 

all relevant patents. For one thing, patent 

applications are published 18 months after 

filing, so there is always a black box in the 

patent space. For another thing, your product 

design could change and your need for 

enabling technologies may evolve over time. 

A good grasp of the patent landscape affords 

you the ability to make adjustments along 

the product development path. In the life 

science sector, the point of each significant 

go or no-go decision is often a good time 

to revisit the patent situation. Competent 

counsel can guide you through the process. 

Ignorance is rarely a viable approach— 

sophisticated investors and collaborators will 

conduct thorough due diligence on your patent 

position. It is better that you know your 

shortcomings before potential investors or 

partners point them out. 

Next, overcome 

If the worst happens and you do identify a 

potential blocking patent or patent application, 

it is critical to fully understand how it 

impacts your business. For example, does it 

affect the technology platform at the core 

level, a particular product composition, a 

manufacturing step or a specific use of a 

product? You and your patent counsel should 

understand the patent’s legal status and the 

claim scope. Information on the patent 

(whether the claims are allowed, issued, on 

appeal, in interference, under reexamination 

or challenged in litigation) can be obtained 

by your counsel. Often a review of the patent 

file history is needed to fully grasp the scope 

of a patent and its real or potential weaknesses. 

Claims in different countries quite 

often are not identical; therefore, you should 

consider obtaining legal advice from a lawyer 

in each of the countries of interest. 

In addition, you should gather background 

information about the patent owner as a prelude 

to a licensing effort or a preparation for 

challenge. Where the patent is licensed to a 

third party, information on the licensee could 

be valuable as well. They could become your 

competitor, collaborator, licensor, licensee 

or all of the above. Knowing their business 

situation, their financial clout, their technology 

and their patent needs can help you 

gain leverage and affect how you deal with 

the blocking patent and its owner. Patent 

ownership and licensing information may 

be obtained from appropriate databases, 

such as ones found at the US Patent and 

Trademark Office, the World Intellectual 

Property Organization and the US Securities 

Exchange Commission. 

You’ll then need to explore design-around 

options. Quite often, a blocking patent is relevant 

only because of one or two features of 

a candidate compound or device. When such 

features can be removed or modified without 

sacrificing the needed functionality, you 

might find a successful design-around solution. 

Working closely with your scientific 

team, an experienced lawyer can make significant 

contributions in identifying and finetuning 

successful design-around options. 

In a hypothetical drug development scenario, 

a drug is approved by the US Food and 

Drug Administration for treating disease A. 

The owner of the drug has obtained patents 

on the drug compound, formulation and use 

of the compound to treat disease A. The drug, 

however, is now suspected to have potency 

against disease B. The suspicion is further 

borne out by your preclinical studies. 

In this scenario, FTO issues are clearly 

present due to the patent estate around the 

approved drug. These patents, however, 

might be limited to the approved drug compound 

itself and certain close derivatives. 

Some chemical compound patents suffer 

from insufficient enablement regarding 

distant derivatives and analogs. Analysis of 

the patents might reveal that all valid patent 

claims are tied to a certain core chemical 

scaffold or more commonly to certain 

specific functionalities on the core chemical 

scaffold. Working together, your scientists 

and patent counsel can design exploratory 

compound libraries that investigate the surrounding 

chemical space uncovered by the 

patent. Knowledge of the structure-activity 

relationship and the patent landscape is critical 

in successfully navigating the hit-to-lead 

evolution process. A good practice from a 

patent counsel’s perspective is to fully grasp 

both technical and legal parameters while 

communicating effectively and working intimately 

with your project team. The objective 

should be to obtain a design-around solution 

that minimizes infringement risks and 

allows solid, fresh patent protection for the 

redesigned compound (Box 1). 

When faced with a blocking patent, you 

should also explore the availability and cost of a 




patent license. Too often the project team drops 

a compound after the discovery that it is covered 

by a third-party patent. Licensing opportunities 

often are not carefully investigated and 

followed up, resulting in premature loss of good 

drug candidates. It is more desirable from efficiency 

and productivity perspectives that biotech 

startups become more open to licensing 

both at the receiving and giving ends. 

You’ll need your counsel to closely guide 

the licensing process, as many pitfalls exist 

that could entrap the unwary. A misstep 

could expose you to various risks, ranging 

from prematurely exposing your licensing 

needs to the patent owner to a loss of your 

right to later challenge the patent. It is generally 

a good idea to enter a confidentiality 

agreement before exchanging sensitive information 

and starting negotiations. You can 

also sometimes enter a joint privilege agreement 

if sensitive opinions or legal documents 

are shared with the other side. 

Keep in mind, though, that the patent 

licensing process could be lengthy and 

exhaustive, and it can easily take months to 

complete. This means it’s important to set 

a timetable and move forward accordingly. 

Your knowledge of the weaknesses of the 

desired patent estate and its holder could help 

steer the process favorably for you. In this 

regard, due diligence on the desired patent 

estate and its owner is must-do homework 

that should be completed before approaching 

the owner. Understanding the value of 

the license to you is important because the 

deal inevitably involves setting licensing 

fees and royalty rates. Experienced counsel 

might know the prevailing licensing rates 

and industry terms. Your ultimate objective 

in a licensing discussion is to fully explore 

the potential of a license in view of all other 

options available to you so that you can make 

a cogent business decision. 

Though rarely enjoyable, sometimes the 

best approach is to simply challenge a blocking 

patent face on. It’s preferable to have proactive 

patent challenges at favorable times and 

venues rather than mere reactive defenses. 

Well-planned and executed strategies for 

patent challenges often help realize the full 

commercial potential of your product. 

Legal procedures vary from country to 

country, but some forms of patent validity 

challenge are available in all major jurisdictions. 

For example, Europe, Japan, China, 

Australia and Canada all have post-grant 

opposition or cancellation procedures. In 

the United States, the primary procedures 

for validity challenge at the US Patent and 

Trademark Office are ex parte and inter parte 

reexaminations. Currently, post-grant opposition 

is not available in the United States, 

but it is a part of most patent reform bills and 

could become available soon, as it’s believed 

that some iteration of patent reform is likely 

to pass into law. 

At any time after US patent issuance 

and before expiration, anyone may request 

reexamination on the grounds of a substantial 

new question of patentability based on 

printed prior art or patent references. A 

notable benefit is that the outcome of the ex 

parte reexamination does not have a binding 

effect on the requester, which means that you 

may raise the same basis of invalidity again in 

a pending or future patent litigation, essentially 

allowing you two bites at the apple. A 

potential downside, however, is that you do 

not have much direct participation in the proceedings, 

and the patent holder could steer 

that reexamination process to strengthen his 

or her patent. Inter parte reexamination, in 

contrast, allows more direct involvement, 

but the outcome has a binding effect on the 

requester. Heed your counsel’s advice when 

formulating and implementing effective 

patent challenge strategies. 

Conclusions 

FTO is critical to life science startups. Not having 

it from the beginning of a venture is like 

a building without a foundation. A biotech 

startup is wise to craft and implement an appropriate 

patent strategy early on that steers it clear 

of and minimizes the impact of hostile patents. 

But should one appear, there are approaches to 

try in order to maneuver around it. Still, nothing 

can substitute for the commitment of the 

company and the teamwork of management, 

scientists and counsel. 

1. Novartis Pharms. Corp. v. Eon Labs Mfg., Inc., 363 F. 

3d 1306 (Fed. Cir. 2004). 

To discuss the contents of this article, join the Bioentrepreneur forum on Nature Network: 

http://network.nature.com/groups/bioentrepreneur/forum/topics 


correspondence 

Wrong fixes for gene patents 


To the Editor: 

In your August issue, the Commentary 

entitled “DNA patents and diagnostics: not 

a pretty picture” by Robert Cook-Deegan 

and colleagues 1 draws conclusions and 

recommendations that are unwarranted by 

the full scope of available data and unfairly 

criticizes the activities of the Biotechnology 

Industry Organization (BIO; Washington, 

DC, USA) with respect to this issue. As 

general counsel and senior vice president for 

legal & intellectual property for BIO, I would 

like to respond to the criticisms in this article. 

The authors admit that there is “limited” 

evidence that patents are negatively impacting 

research on, or the accessibility of, genetic 

diagnostic tests, but then go on to assert that, 

despite this lack of evidence, “concerns about 

DNA patents persist.” They reference many 

studies for this proposition, yet the cumulative 

evidence contained in those studies does not 

support any systemic basis for such concerns. 

Anecdotal examples of access problems, 

such as those contained in the specific case 

studies heavily relied upon by the authors, 

may be helpful to consider but, as the authors 

admit, the evidence of problems from 

these particular examples is “not uniform, 

consistent or pervasive.” To draw from such 

anecdotal evidence the general conclusion 

that patents, or the exclusive licensing thereof, 

are the cause of such problems, or are not 

necessary to incentivize the development and 

marketing of diagnostic tests except in rare 

cases, is simply not justified. 

Yet the authors assert that these case 

studies—which, although well done, are 

truly case studies and are not purported to 

be representative of the entire field of genetic 

diagnostic testing—show that “Exclusive 

licensing practices consistently reduce 

availability” of genetic diagnostic tests. The 

weakness in this argument, beyond the 

unsupported generalization noted above, is 

found in the phrase immediately following 

this conclusion—“at least as measured by the 

number of available laboratories offering a 

test” (emphasis added). Of course, it should 

come as no surprise—and certainly we did 

not need a series of case studies to teach us— 

that exclusive licensing limits the number of 

laboratories offering a test and, as the authors 

further complain, “reduce[s] competition.” 

In fact, this is true of every product that is 

patented or exclusively licensed, regardless of 

the field. The key fact that should concern us, 

however, is not how many entities provide the 

test, but whether individuals can reasonably 

gain access to it. Just because several academic 

research laboratories may offer a particular 

test absent an exclusive license to another 

party, it does not automatically follow that 

people would have greater access to that test. 

Indeed, it is just as reasonable to project that, 

without a single entity having a financial 

incentive to develop not just the test but 

also the market for the test, many patients 

who should be screened (and their doctors) 

may never learn about the availability and 

usefulness of the test at all. Exclusive licensees 

also would seemingly have a much greater 

incentive to validate and improve the test, and 

to help obtain coverage for such tests from 

payors and insurers. As the studies do reveal, 

lack of insurance coverage can be a major 

obstacle to patient access to diagnostic tests, 

particularly for poor women. In addition, 

concern about reduced competition in and 

of itself is misplaced; the concern should be 

whether, in light of the reduced competition, 

there is public harm in the form of higher 

prices or inferior products. Again, the case 

studies do not reveal any significant evidence 

of either. 

The Commentary then criticizes BIO 

for having “chosen not to acknowledge the 

real problems” that exist with respect to 

the genetic diagnostics market, and for its 

“quick and vociferous” opposition to the 

recommendations of the US Secretary’s 

Advisory Committee on Genetics, Health 

& Society (SACGHS). To the contrary, 

BIO publicly testified before SACGHS on 

its original, more balanced draft report 

and praised the work of the committee. 

BIO cautioned the committee that some 

of the proposals under consideration 

could undermine the goal of enhancing 

patient access to genetic tests and urged the 

committee to partner with BIO in crafting 

sensible recommendations that would actually 

enhance patient access to such tests (e.g., 

expanded insurance coverage). Instead, the 

committee rewrote its original draft report 

in an apparent attempt to support a series of 

divisive, controversial and highly impractical 

recommendations to alter patent law and 

impose one-size-fits-all licensing strategies 

enforced by the federal government. Not only 

were these recommendations not supported 

by the committee’s own review and case 

studies, but they also chased red herrings such 

as patenting and exclusive licensing practices, 

while avoiding more practical approaches that 

would not only have garnered more consensus 

among external stakeholders and the 

committee’s own members (three of whom 

dissented in a highly unusual maneuver) 

but also would have been more effective in 

achieving our shared goals. 

The authors also chastise BIO for not 

proposing constructive solutions, such as the 

development of a ‘gene patent supermarket’ 

or similar licensing pools that could help 

facilitate broader licensing of such patents, 

while maintaining appropriate incentives. 

In actuality, BIO has been discussing such 

approaches with our members and with 

the US National Institutes of Health for 

several months now—not because there is 

evidence that DNA patents are impeding 

the development of more complex or ‘whole’ 

genomic testing products, but to help avoid 

any chilling effect that might arise in the 

future. BIO also has been working with 

other organizations in developing ideas 

to address the ‘second opinion’ quandary 

that may arise when there is a sole provider 

of an important medical test that drives a 

particular treatment regimen. The key point, 

however, is that proposed solutions must be 

targeted to deal with specific problems, while 

avoiding unintended and counterproductive 

consequences for innovation and patient care. 

Finally, the authors criticize BIO for 

“failing to enforce the established norms” 

of nonexclusive licensing and for not 

“criticiz[ing] deviations from them” by our 

members. Leaving aside the obvious antitrust 

and other legal and practical concerns about 




a trade association seeking to require or 

‘enforce’ specific licensing practices by its 

membership, the authors’ suggestion that 

nonparties to a specific licensing transaction 

would have the information necessary to 

reasonably judge (presumably, after the 

fact) whether an exclusive agreement is 

or is not appropriate under the particular 

circumstances involved is simply not realistic, 

as much of the relevant information would 

likely be proprietary. In addition, BIO’s 

membership is largely made up of therapeutic 

research and development companies, rather 

than the type of companies whose business 

models are the focus of the authors’ critique. 

In short, BIO never has denied that certain 

problems exist with respect to access to 

genetic diagnostic testing; we just disagree 

as to the causes of such problems and how 

best to fix them. BIO will continue its efforts 

to work with other organizations to help 

improve patient access to genetic testing, but 

will also continue to oppose—vigorously 

when necessary—any ill-considered and 

misguided proposals that would undermine 

the development of new diagnostics and 

therapies and do more harm than good for the 

patients of today and tomorrow. 

COMPETING FINANCIAL INTERESTS 

The author declares no competing financial interests. 

Tom DiLenge 

Biotechnology Industry Organization, 

Washington, DC, USA. 

e-mail: tdilenge@bio.org 

1. Carbone, J. et al. Nat. Biotechnol. 28, 784–791 

(2010). 

Robert Cook-Deegan, 

Subhashini Chandrasekharan, 

Misha Angrist, Bhaven Sampat, 

E Richard Gold, Julia Carbone & 

Lori Knowles reply: 

We thank Tom DiLenge of BIO for his 

thoughtful comments. We agree with many 

points, but focus here on remaining points of 

disagreement. 

First, although we agree there is no 

evidence of systematic and pervasive 

harm from patenting and licensing in 

DNA diagnostics, we reiterate that there 

is unequivocal evidence of problems in 

some cases. We agree there may well be a 

role for patent incentives in DNA testing; 

we do not believe, however, that this means 

carte blanche for patent holders. We are 

particularly wary of exclusive licensing 

to sole providers of genetic tests unless 

nonexclusive licensing will fail to bring a 

product to market. This is decidedly not the 

case in empirical studies to date. We say this 

for three main reasons. First, in instances 

where no test is available and yet patents are 

being enforced, as was the case with long-QT 

testing from 2002 to 2004, there are clearly 

access problems by any definition. These 

situations may be rare, and we hope they are, 

but denying a problem that has historically 

occurred is not a winning argument. The 

BIO letter is silent on such problems. 

Second, it is simply not true that exclusive 

licensing needs to lead to monopolies. If 

a particular laboratory does not offer a 

particular form of service (e.g., prenatal 

testing), does not have a payment agreement 

with an insurer or health plan, or has already 

gotten paid to do a test, and the patient (or 

doctor) wants verification, then prudent 

business practice would suggest sublicensing, 

a permissive testing policy or some other 

way to ensure testing can be done by others. 

Policies on sublicensing or testing by others 

are under control of the patent holder and 

could be remedied by them without breaking 

patents. It is thus puzzling that patentholders 

have not adopted such policies. 

Third, although we agree that reducing 

the number of laboratories offering a test 

does not necessarily reduce patient access, 

there is a very consistent pattern revealed 

in our case studies and the survey of 

laboratory directors that we cited by Cho 

et al. 1 : the holder of exclusive patent rights 

is consistently not first to market with a 

genetic test. The effect of patents has been 

solely to reduce competition, not to create 

new products that would not otherwise exist. 

Suppression of competitors who have beaten 

the holder of exclusive rights to market is 

not what is usually observed with patents. 

Pharmaceutical firms and instrument 

companies generally do not enforce patents 

against universities and research institutions, 

for example, and yet this is what we find in 

DNA diagnostics in several cases. In this 

respect, diagnostics are unusual compared 

with other domains where patent exclusivity 

has a role. We agree the evidence of harms 

from exclusive licensing is not systematic, 

but the evidence of benefit from patents 

in genetic diagnostics historically is even 

weaker. 

Finally, we appreciate there are indeed 

limits to BIO’s actions when questions 

of antitrust would arise in enforcing the 

existing norms on patenting and licensing 

genomic inventions. The licensing norms 

developed by the Organization for Economic 

Cooperation and Development 2 (Paris), 

the US National Institutes of Health 3 and 

the ‘Nine Points’ document on university 

technology licensing 4 are all pro-competitive 

however, not anti-competitive. If a company 

is deviating from those norms, therefore, 

antitrust concerns would not arise; quite 

the reverse. We don’t suggest BIO act 

when antitrust would loom as an issue, but 

commenting on policies—such as enforcing 

patents when no test is available to patients— 

would rarely confront antitrust policy. 

The main underlying point is that 

problems with patents and exclusive licensing 

distinctive to diagnostics can be identified 

and dealt with, but only if the problems are 

acknowledged and acted upon. If BIO is 

turning its attention to these issues, then we 

will all benefit. 


The authors declare no competing financial interests. 

1. Cho, M.K., Illangasekare, S., Weaver, M.A., Leonard, 

D.G.B. & Merz, J.F. J. Mol. Diagn. 5, 3–8 (2003). 

2. Organisation for Economic Co-operation and 

Development. Guidelines for the Licensing of Genetic 

Inventions (OECD, Paris, 2006). 

3. National Institutes of Health. “Best Practices for the 

Licensing of Genomic Inventions,” Federal Register 70 

(No. 68): 18412–18415. 

4. Association of University Technology Managers. In the 

Public Interest: Nine Points to Consider in Licensing 

University Technology (AUTM, Deerfield, Illinois, USA, 

2007). 

Stem cell clinics in the news 


As highlighted in a News Feature “Trading 

on hope” published in this journal 1 , stem 

cell tourism is a growing and increasingly 

contentious phenomenon. By ‘stem cell 

tourism’, we refer to the emerging practice 

that sees patients travel abroad to receive 

(largely) unproven stem cell treatments that 

are generally not approved or available in their 

home country 2 . Although precise numbers are 

unknown, current information suggests that 

potentially thousands of patients each year 

from various countries are travelling around 

the world to receive stem cell therapies for a 

wide range of conditions 3–5 . 

The stem cell tourism phenomenon 

is highly controversial. The therapeutic 

possibilities promised by the clinics involved 

engage the hopes of often desperate patients 

and their families, including those who feel 

there are no other options. These individuals 

are understandably anxious to have a 




a trade association seeking to require or 

‘enforce’ specific licensing practices by its 

membership, the authors’ suggestion that 

nonparties to a specific licensing transaction 

would have the information necessary to 

reasonably judge (presumably, after the 

fact) whether an exclusive agreement is 

or is not appropriate under the particular 

circumstances involved is simply not realistic, 

as much of the relevant information would 

likely be proprietary. In addition, BIO’s 

membership is largely made up of therapeutic 

research and development companies, rather 

than the type of companies whose business 

models are the focus of the authors’ critique. 

In short, BIO never has denied that certain 

problems exist with respect to access to 

genetic diagnostic testing; we just disagree 

as to the causes of such problems and how 

best to fix them. BIO will continue its efforts 

to work with other organizations to help 

improve patient access to genetic testing, but 

will also continue to oppose—vigorously 

when necessary—any ill-considered and 

misguided proposals that would undermine 

the development of new diagnostics and 

therapies and do more harm than good for the 

patients of today and tomorrow. 



Tom DiLenge 

Biotechnology Industry Organization, 

Washington, DC, USA. 

e-mail: tdilenge@bio.org 


(2010). 

Robert Cook-Deegan, 

Subhashini Chandrasekharan, 

Misha Angrist, Bhaven Sampat, 

E Richard Gold, Julia Carbone & 

Lori Knowles reply: 

We thank Tom DiLenge of BIO for his 

thoughtful comments. We agree with many 

points, but focus here on remaining points of 

disagreement. 

First, although we agree there is no 

evidence of systematic and pervasive 

harm from patenting and licensing in 

DNA diagnostics, we reiterate that there 

is unequivocal evidence of problems in 

some cases. We agree there may well be a 

role for patent incentives in DNA testing; 

we do not believe, however, that this means 

carte blanche for patent holders. We are 

particularly wary of exclusive licensing 

to sole providers of genetic tests unless 

nonexclusive licensing will fail to bring a 

product to market. This is decidedly not the 

case in empirical studies to date. We say this 

for three main reasons. First, in instances 

where no test is available and yet patents are 

being enforced, as was the case with long-QT 

testing from 2002 to 2004, there are clearly 

access problems by any definition. These 

situations may be rare, and we hope they are, 

but denying a problem that has historically 

occurred is not a winning argument. The 

BIO letter is silent on such problems. 

Second, it is simply not true that exclusive 

licensing needs to lead to monopolies. If 

a particular laboratory does not offer a 

particular form of service (e.g., prenatal 

testing), does not have a payment agreement 

with an insurer or health plan, or has already 

gotten paid to do a test, and the patient (or 

doctor) wants verification, then prudent 

business practice would suggest sublicensing, 

a permissive testing policy or some other 

way to ensure testing can be done by others. 

Policies on sublicensing or testing by others 

are under control of the patent holder and 

could be remedied by them without breaking 

patents. It is thus puzzling that patentholders 

have not adopted such policies. 

Third, although we agree that reducing 

the number of laboratories offering a test 

does not necessarily reduce patient access, 

there is a very consistent pattern revealed 

in our case studies and the survey of 

laboratory directors that we cited by Cho 

et al. 1 : the holder of exclusive patent rights 

is consistently not first to market with a 

genetic test. The effect of patents has been 

solely to reduce competition, not to create 

new products that would not otherwise exist. 

Suppression of competitors who have beaten 

the holder of exclusive rights to market is 

not what is usually observed with patents. 

Pharmaceutical firms and instrument 

companies generally do not enforce patents 

against universities and research institutions, 

for example, and yet this is what we find in 

DNA diagnostics in several cases. In this 

respect, diagnostics are unusual compared 

with other domains where patent exclusivity 

has a role. We agree the evidence of harms 

from exclusive licensing is not systematic, 

but the evidence of benefit from patents 

in genetic diagnostics historically is even 

weaker. 

Finally, we appreciate there are indeed 

limits to BIO’s actions when questions 

of antitrust would arise in enforcing the 

existing norms on patenting and licensing 

genomic inventions. The licensing norms 

developed by the Organization for Economic 

Cooperation and Development 2 (Paris), 

the US National Institutes of Health 3 and 

the ‘Nine Points’ document on university 

technology licensing 4 are all pro-competitive 

however, not anti-competitive. If a company 

is deviating from those norms, therefore, 

antitrust concerns would not arise; quite 

the reverse. We don’t suggest BIO act 

when antitrust would loom as an issue, but 

commenting on policies—such as enforcing 

patents when no test is available to patients— 

would rarely confront antitrust policy. 

The main underlying point is that 

problems with patents and exclusive licensing 

distinctive to diagnostics can be identified 

and dealt with, but only if the problems are 

acknowledged and acted upon. If BIO is 

turning its attention to these issues, then we 

will all benefit. 



1. Cho, M.K., Illangasekare, S., Weaver, M.A., Leonard, 

D.G.B. & Merz, J.F. J. Mol. Diagn. 5, 3–8 (2003). 

2. Organisation for Economic Co-operation and 

Development. Guidelines for the Licensing of Genetic 

Inventions (OECD, Paris, 2006). 

3. National Institutes of Health. “Best Practices for the 

Licensing of Genomic Inventions,” Federal Register 70 

(No. 68): 18412–18415. 

4. Association of University Technology Managers. In the 

Public Interest: Nine Points to Consider in Licensing 

University Technology (AUTM, Deerfield, Illinois, USA, 

2007). 

Stem cell clinics in the news 


As highlighted in a News Feature “Trading 

on hope” published in this journal 1 , stem 

cell tourism is a growing and increasingly 

contentious phenomenon. By ‘stem cell 

tourism’, we refer to the emerging practice 

that sees patients travel abroad to receive 

(largely) unproven stem cell treatments that 

are generally not approved or available in their 

home country 2 . Although precise numbers are 

unknown, current information suggests that 

potentially thousands of patients each year 

from various countries are travelling around 

the world to receive stem cell therapies for a 

wide range of conditions 3–5 . 

The stem cell tourism phenomenon 

is highly controversial. The therapeutic 

possibilities promised by the clinics involved 

engage the hopes of often desperate patients 

and their families, including those who feel 

there are no other options. These individuals 

are understandably anxious to have a 




Table 1 Demographics of patient undergoing stem cell treatment as reported in news 

media articles from October 2006 to September 2009 

Gender 

Male 

(134; 59.8%) 

Female 

(87; 38.8%) 

Unspecified 

(3; 1.3%) 

Age (adult 

versus minor) 

Adult 

(126; 56.3%) 

Minors 

(98; 43.8%) 

Country of origin (in 

descending order) 

US (84; 37.5%) 

UK (80; 35.7%) 

Australia (32; 14.3%) 

Canada (12; 5.4%) 

reason to feel optimistic about their medical 

prospects, and often frustrated with their 

native medical systems. Patients pay for these 

treatments personally, and research reveals 

costs ranging from $5,000 to $39,500 (ref. 4), 

or an average cost of $21,500 (excluding travel 

and accommodation) 3 . These treatments 

generally have not proceeded through the 

clinical trial process, and published research 

reveals little to no evidence of efficacy for the 

services advertised by these clinics 3 . There also 

appears to be an overall lack of transparency 

regarding the specifics of treatment protocols 

and an absence of comprehensive posttreatment 

follow-up. Accordingly, many 

experts are highly skeptical of the claims 

made by stem cell clinics and, particularly in 

light of emerging evidence regarding adverse 

effects from such treatments 6 , concerned 

about potential risks 7,8 . Other emerging 

issues include a lack of informed consent 

because of inadequate information provided 

to prospective patients 4,9 , the potential for 

vulnerable individuals (including children) to 

be taken advantage of or put at risk 10 , and the 

obligation to provide continuing care upon a 

patient’s return. 

Not surprisingly, treatment providers are 

often highly motivated to defend their work 

and to advocate for the use of novel treatments. 

This range of conflicting and competing 

perspectives regarding stem cell tourism serves 

to create a confusing picture of the field in the 

news media, especially for patients and their 

families. Previous research points to two key 

mechanisms by which news media portrayals 

are likely to influence audiences. First, through 

a process of agenda setting, the news media 

calls attention to specific issues, events or 

developments 11 , such as the claims of stem 

cell clinics or developments in the field. Given 

high levels of motivation, patients and their 

families are likely to pay closer attention to 

stem cell-related news coverage and to actively 

Five most common 

conditions 

Multiple sclerosis 

(22; 9.8%) 

Cerebral palsy 

(18; 8.0%) 

Septo-optic dysplasia 

(15; 6.7%) 

Optic nerve hypoplasia 

(13, 5.8%) 

New Zealand (12; 5.4%) Unspecified blindness 

(13; 5.8%) 

Israel (1; 0.5%) 

Brazil (1; 0.5%) 

Three most common 

treatment destinations 

China (100; 44.6%) 

Germany (17; 7.6%) 

Mexico (17; 7.6%) 

seek information about therapies online 

(Supplementary Discussion 1). 

Not only do the news media—and 

newspapers especially—call the public’s 

attention to medical claims and (unproven) 

therapies, but news coverage also selectively 

frames the nature of these claims. ‘Frames’ 

is the conceptual term for interpretative 

storylines that emphasize specific dimensions 

of a complex topic over others, often reducing 

complexity and uncertainty, and leading 

audiences to consider certain considerations 

over others in reaching judgments and making 

decisions (Supplementary Discussion 1) 12 . 

Given the likely importance of the news 

media—and newspapers in particular—in 

setting the agenda of the public and in shaping 

public interpretations, we thought it essential 

to investigate the level of attention to stem 

cell tourism in the print media and how the 

emerging industry is characterized. Our 

analysis also provides information about the 

individuals accessing these services, where the 

clinics reside and what services are provided. 

We searched the Factiva database for 

newspaper articles about stem cell tourism 

from Canada, the United States, the United 

Kingdom, Australia and New Zealand between 

October 1, 2006, and September 30, 2009. 

Our search used ‘stem cell treatment’ or ‘stem 

cell therapy’ and one or more terms related to 

overseas travel (e.g., overseas, abroad, China 

or India; Supplementary Methods). Our 

data set contained a combined 445 articles 

across countries and outlets (Supplementary 

Table 1). Our coding frame was informed by 

previous work that identified communication 

issues associated with the stem cell tourism 

phenomenon, particularly issues relevant to 

representations of efficacy and risk 3,4 . The 

coding frame included patient demographic 

information, clinic location, treatment details, 

cost, donation information, discussions of 

policy, relevant risks, state of the science, 

presence of patient testimonials and the tone 

of the article. Given the subjectivity of news 

content analysis, we tested 10% of the articles 

for inter-coder reliability using Cohen’s kappa, 

which produced a mean score of k = 0.857 

(Supplementary Methods). 

We found coverage of stem cell tourism 

in each country. The majority of articles 

(234 or 52.6%) stemmed from the United 

Kingdom, followed by the United States with 

99 articles (22.2%), Australia with 74 articles 

(16.6%), New Zealand with 21 articles (4.7%) 

and Canada with 17 articles (3.8%). These 

445 articles discussed 224 different patients 

and, as a result, we were able to obtain some 

interesting patient demographic information 

(Table 1). Of course, there are various 

limitations with these data; newspapers 

represent only one form of media and our data 

only reflect individuals whose stories were 

covered by the press captured in our sample 

currently available in the Factiva database. 

In total, over 19 different countries 

were discussed as treatment destinations, 

supporting the findings of earlier research 

indicating the global nature of this 

phenomenon 4,5 . Destinations in addition to 

those outlined included India, the Dominican 

Republic, the Netherlands, Thailand, Russia, 

Costa Rica, the United States, Portugal, 

the Ukraine, Israel, the United Kingdom, 

Argentina, Ecuador, Chile and Brazil. Our 

results also verified the high financial costs 

associated with these treatments. The average 

cost noted was ~$47,315. The lowest was 

~$3,500 and the highest (paid for multiple 

treatments over time) was $399,687. In 

many cases, articles discussed the intense 

fundraising efforts made by patients, their 

families, friends and supporters. In fact, 134 

(30%) of the articles provided readers with 

information on how they could donate funds. 

Given these treatments’ unproven status and 

the inconsistent information about them 

that is generally available to the public, the 

involvement of the media in helping raise 

funds for these efforts adds another potentially 

concerning element to the dynamic of stem 

cell tourism. 

In addition, the high prevalence of minor 

patients in our data set seems particularly 

worrisome, especially when one considers 

the concerns outlined above, including the 

potential for adverse events and the lack of 

evidence regarding safety or efficacy. 

Examining the articles’ content for 

discussions of policy (e.g., legality of the 

treatment or regulatory environment), 

risk (e.g., tumors or rejection), evidence of 

treatment efficacy (e.g., cautions from stem 

cell scientists or anecdotal reports from 




the clinics) and procedural information 

(that is, types of stem cells and method 

of application) revealed interesting 

jurisdictional differences in the relative 

prominence of these topics (Fig. 1). 

Policy issues arose in just 21% of the 

articles, the majority of which raised them 

only in passing (e.g., “it’s so experimental 

it can’t legally be done in the United States 

because it isn’t approved by the Food and 

Drug Administration,” which appeared 

in an article entitled “Mailia’s Miracle” in 

the 24 August 2009 issue of the Tri-City 

Herald, Kennewick, WA, USA) as opposed 

to including nuanced discussion of policy 

issues or debates. An even smaller proportion 

of articles (13%) mentioned the concept of 

risk. Of those that did, the majority (70.5%) 

did not discuss any specific risks; some of the 

risks that were identified included infection, 

tumors, inflammation, immune rejection 

and inadequate screening (Supplementary 

Discussion 2). This relative lack of focus 

in the print media on policy issues or risks 

associated with stem cell tourism as compared 

with emerging commentary in the academic 

literature on the topic 9,13 reveals an apparent 

disconnect between these two realms. It also 

raises the question of whether the general 

public (which includes prospective stem cell 

tourists) is receiving sufficient information to 

reach effective decisions and judgments on 

the issue. 

In contrast to the comparably lower 

prevalence of discussions regarding policy 

and risk, 225 articles (50.3%) contained 

information about the stem cells being used 

in the treatment, the application procedures 

or both. The following types of stem cells were 

referenced: umbilical cord (107), embryonic 

(27), bone marrow (27), adult (24), stem 

cells from pelvis (2), stem cells from hip 

(2), fetal (2), stem cells from fat tissue (1) 

and a combination of bone marrow and 

umbilical cord blood (5). In total, 20 different 

application procedures were discussed. The 

five most common included the following: 

injection (location not disclosed) (77), 

injection into spinal cord and/or bloodstream 

(including intravenous injection) (69), 

injection into brain (8), intravenous injection 

through hairline towards optic nerve (8) and 

catheter/injection into heart (8). The average 

length of treatment noted was 4–5 weeks, with 

a range of only a few days up to 3 months. It 

is important to note that these simply reflect 

media references. It remains unclear what, 

if any, kind of stem cells were injected in the 

procedures used at clinics. 

It is also surprising that few articles 

contained information regarding the efficacy 

Percentage of articles 

from each country 

80 

70 

60 

50 

40 

30 

20 

10 

0 

United 

States 

Policy 

Canada 

Risk 

Evidence 

United 

Kingdom 

Country 

Australia 

Stem cells 

New 

Zealand 

Figure 1 Percentage of news media articles 

originating for different countries, which focus 

on policy, risk, evidence and the procedure 

associated with stem cell treatments. 

of the stem cell treatment being addressed 

(118 articles, ~25%). In addition, the sources 

of this information varied widely (e.g., stem 

cell scientists and experts, clinics providing 

the therapy, patient groups and family 

members). The evidence or information 

regarding efficacy ranged from being 

supportive of the treatment to cautioning 

against it (Supplementary Discussion 2). It 

is important to note that in many cases, the 

discussions of efficacy did not constitute a 

particularly important aspect of the article as a 

whole, nor were they necessarily predictive of 

its overall tone, discussed below. 

Interestingly, 227 articles (51%) included 

patient testimonials from the patient, friends 

or relatives, other patients with similar 

conditions, the clinic’s previous patients or 

anecdotal reports (e.g., from media sources). 

Considering the prominent role patient 

testimonials have on many clinic websites, 

this confirmation of their significant presence 

in media reports as well further highlights 

the dominance this method of information 

dissemination and promotion may have to the 

growth of this market. 

Finally, and perhaps most important, the 

tone of the discussions is intriguing (Fig. 2). 

Our coders determined whether, overall, 

each article was positive, neutral or negative 

towards stem cell tourism. The results show 

that over time, the tone of articles has become 

increasingly positive (Supplementary 

Table 2). One exception is the period 

between October and December 2008 

(Fig. 2), which may be linked with the release 

of the International Society for Stem Cell 

Research (ISSCR)’s Guidelines for the Clinical 

Translation of Stem Cells in December, 2008 

(ref. 7). Articles coded as positive presented 

stem cell therapy as being a hopeful and/ 

or successful alternative for patients that 

indicated little or no limitations or possible 

risks of the treatment. Given that the news 

media—and newspapers in particular—are 

the dominant source of information about 

stem cell tourism for the public, this apparent 

trend to frame stem cell tourism in terms of 

promise and hope with limited emphasis on 

uncertainty may have important implications 

for public evaluations of clinic claims, 

especially among patients and their families. 

On the whole, these data suggest that print 

media portrayals of stem cell tourism are 

largely and increasingly positive in nature. 

Furthermore, their focus is primarily on the 

individual patient, their hopes and specific 

treatment plans or approach, rather than 

on the potential risks associated with these 

unproven treatments, current scientific 

and clinical limitations, or the various 

policy issues implicated by this emerging 

phenomenon. In many respects, these 

results are unsurprising; the circumstances 

surrounding an individual’s pursuit of stem 

cell tourism often make for a powerful 

personal interest story. It is impossible 

not to empathize with the plights of those 

desperately seeking treatment for themselves 

or their loved ones, and important to be wary 

of overly broad generalizations of a field that 

appears to encompass an incredibly broad 

range of potential treatments and treatment 

providers. Nonetheless, there are serious 

concerns associated with the rise of stem cell 

tourism, including physical and financial 

risks, lack of transparency and appropriate 

review procedures, exploitation of vulnerable 

patients (including minors) and various 

policy issues, which must not be minimized. 

Although some of these issues are making 

their way into the print media, it is clear that 

continued efforts are necessary to improve 

the balance in media reporting of this topic. 

When medical experts and organizations 

do actively seek to engage journalists and 

the public on the uncertainty of stem cell 

tourism and the need for regulation, as was 

the case in 2008 with the ISSCR’s guidelines, 

our findings do point to an influence on 

coverage, with the effort balancing the 

otherwise overwhelmingly positive portrayal 

in the press. This balance is an important 

element in promoting a knowledgeable 

public, necessary both for facilitating 

Percentage of 

10/06–12/06all articles 

100 

90 

80 

70 

60 

50 

40 

30 

20 

10 

0 

01/07–03/07 

04/07–06/07 

Positive 

Neutral 

Negative 

07/07–09/07 

10/07–12/07 

01/08–03/08 

04/08–06/08 

07/08–09/08 

10/08–12/08 

01/09–04/09 

04/09–06/09 

07/09–09/09 

Figure 2 Overall ‘slant’ of different news media 

reports that covered stem cell treatments and 

appeared from October 2006 to September 2009. 




informed debate regarding stem cell tourism 

and for protecting potentially vulnerable 

individuals. 

Note: Supplementary information is available on the 

Nature Biotechnology website. 

Acknowledgments 

We would like to thank Canada’s Stem Cell Network 

for funding and the University of Alberta’s Health Law 

Institute for administrative support. We also gratefully 

acknowledge C. Scott, J. McCormick, T. Bubela and 

C. Murdoch for their helpful suggestions regarding data 

collection. Finally, we thank C. Toole for her research 

support, and G. Barr and T. Adido for their assistance in 

coding the data. 

AUTHOR CONTRIBUTIONS 

All authors contributed to this work. T.C., C.R. and 

A.Z. conceived the research concept. C.R. collected the 

data. C.R. and A.Z. coordinated the coding of the data. 

A.Z. and T.C. prepared the draft manuscript and M.N. 

helped with interpretation and with theoretical analysis. 

All authors discussed the results and implications and 

commented on the manuscript at all stages. 



Amy Zarzeczny 1 , Christen Rachul 1 , 

Matthew Nisbet 2 & Timothy Caulfield 1,3 

1 Health Law Institute, Law Centre, University of 

Alberta, Edmonton, Alberta, Canada. 2 School 

of Communication, American University, 

Washington, DC, USA 3 . Faculty of Law and 

School of Public Health, Law Centre, University 

of Alberta, Edmonton, Alberta, Canada. 

e-mail: tcaulfld@law.ualberta.ca 

1. Qui, J. Nat. Biotechnol. 27, 790–792 (2009) 

2. Lindvall, O. & Hyun, I. Science 324, 1664–1665 

(2009). 

3. Lau, D. et al. Cell Stem Cell 3, 591–594 (2008). 

4. Regenberg, A., Hutchinson, L., Schanker, B. & Matthews, D. 

Stem Cells 27, 2312–2319 (2009). 

5. Ryan, K., Sanders, A., Wang, D. & Levine, A. 

Regenerative Med. 5, 27–33 (2010). 

6. Amariglio, N. et al. PLoS Med 6, e1000029 (2009). 

7. http://www.isscr.org/clinical_trans/pdfs/ISSCRGL 

ClinicalTrans.pdf 

8. MacReady, N. Lancet Oncol. 10, 317–318 (2009). 

9. Kiatpongsan, S. & Sipp, D. Nat. Rep. Stem Cells 

published online 3 December, 2008, doi:10.1038/ 

stemcells.2008.151 (2008). 

10. Zarzeczny, A. & Caulfield, T. Am. J. Bioeth. 10, 1–13 

(2010). 

11. Iyengar, S. & Kinder, D. News that Matters: Television 

and American Public Opinion. (University of Chicago 

Press, Chicago, IL, USA, 1987). 

12. Nisbet, M.C. & Mooney, C. Science, 316, 56 

(2007). 

13. Creasy, G. & Scott, C. Nat. Biotechnol. 27, 21–22 

(2009). 

Tracking and assessing the rise of 

state-funded stem cell research 


The editorial in your October issue 1 

highlights the legal challenges to new 

guidelines issued by the US National 

Institutes of Health (NIH) in July 2009 for the 

federal funding of human embryonic stem 

cell (hESC) research. In the eight-year period 

preceding these most recent NIH guidelines, 

only a small number of cell lines could be 

studied with federal funds. During this time, 

six states—California, Connecticut, Illinois, 

Maryland, New Jersey and New York—took 

on a role typically played by the NIH and 

created funding programs specifically 

designed to support stem cell research, 

including hESC research. These are not the 

first state programs to fund scientific research 

but their commitment to basic research is 

atypical, as most state science and technology 

programs have focused on science closer to 

commercialization 2 . Although the state stem 

cell programs differ, they each share at least 

two goals: advancing promising science, 

including research not eligible for federal 

funding during the Bush Administration, and 

returning economic benefits to their state. 

In this article, we report an initial 

attempt to track and assess the impact of 

these state funding programs. Existing 

work on state stem cell policy has focused 

on identifying policy differences between 

various jurisdictions 3,4 , assessing the impact 

of state decisions to support or restrict hESC 

science 5–7 and examining the role of states 

in governing controversial science 8,9 . The 

analysis reported here extends this literature 

though use of a novel data set of the grants 

these states have awarded. These data provide 

insight into how states have prioritized their 

funding, including the extent to which they 

have supported hESC research generally 

and hESC research not eligible for federal 

funding during the Bush Administration 

more specifically, as well as the extent to 

which these states have drawn new scientists 

into the field. The underlying data have been 

publicly released on a new website (http:// 

www.stemcellstates.net) designed to facilitate 

additional analysis of state-funded stem 

cell science and improve public awareness 

of these programs. Given ongoing legal 

uncertainties surrounding federal funding 

of hESC research and the likelihood that 

voters, at least in California, will be asked to 

approve additional state stem cell funding in 

the future, understanding and evaluating the 

effects of state-funded stem cell research is 

both timely and useful. 

The database that forms the basis for the 

analysis described here contains the title, 

principal investigator, institution, abstract 

and amount for each grant awarded by 

the agency overseeing stem cell research 

funding in these six states (Supplementary 

Methods). In all, between December 2005 

when New Jersey awarded the first state stem 

cell grants and the end of 2009, the six stem 

cell states awarded nearly 750 grants totaling 

just over $1.25 billion. The scale of these 

programs varies substantially, ranging from 

the roughly $15 million awarded by Illinois 

and New Jersey to the $1.02 billion awarded 

by California. On a per capita basis, funding 

awarded through the end of 2009 ranges 

from just over $1 in Illinois to nearly $28 in 

California (Table 1). 

States funding stem cell research 

can choose to support several different 

activities, ranging from investigatorinitiated 

research grants to new facilities 

to workforce development. To investigate 

how states prioritized these various types 

of funding, each grant was classified by 

its primary purpose (Supplementary 

Methods). Research grants and support for 

scientific infrastructure were the two largest 

categories, accounting for more than 90% 

of all state stem cell funding (Table 1). The 

infrastructure category was dominated by 

the $271 million California awarded for the 

construction of 12 major stem cell research 

facilities, although several other states also 

dedicated a substantial portion of their 

funding to infrastructure, such as shared 

equipment or core laboratories. In contrast 

to supporting basic investigator-initiated 

research, spending money on infrastructure 

is a classic state economic development 

approach, but, in these cases, spending 

was motivated, at least in part, by the need 

to create separate laboratories to facilitate 

research on unapproved hESC lines. 

The restrictions on federal funding for 

hESC research instituted by former President 

George W. Bush were an important rationale 

behind the adoption of most state stem cell 

programs, yet it is not clear to what extent 

state programs focused on hESC research 

generally or hESC research not eligible for 

federal funding more specifically. To address 

these questions, each research grant awarded 

through the end of 2009 was analyzed to 

assess if funded research used hESCs and, 

if applicable, appeared ineligible for federal 

funding under the Bush Administration 

rules (Supplementary Methods). The 

percentage of grants that supported hESC 




informed debate regarding stem cell tourism 

and for protecting potentially vulnerable 

individuals. 




We would like to thank Canada’s Stem Cell Network 

for funding and the University of Alberta’s Health Law 

Institute for administrative support. We also gratefully 

acknowledge C. Scott, J. McCormick, T. Bubela and 

C. Murdoch for their helpful suggestions regarding data 

collection. Finally, we thank C. Toole for her research 

support, and G. Barr and T. Adido for their assistance in 

coding the data. 


All authors contributed to this work. T.C., C.R. and 

A.Z. conceived the research concept. C.R. collected the 

data. C.R. and A.Z. coordinated the coding of the data. 

A.Z. and T.C. prepared the draft manuscript and M.N. 

helped with interpretation and with theoretical analysis. 

All authors discussed the results and implications and 

commented on the manuscript at all stages. 



Amy Zarzeczny 1 , Christen Rachul 1 , 

Matthew Nisbet 2 & Timothy Caulfield 1,3 

1 Health Law Institute, Law Centre, University of 

Alberta, Edmonton, Alberta, Canada. 2 School 

of Communication, American University, 

Washington, DC, USA 3 . Faculty of Law and 

School of Public Health, Law Centre, University 

of Alberta, Edmonton, Alberta, Canada. 

e-mail: tcaulfld@law.ualberta.ca 

1. Qui, J. Nat. Biotechnol. 27, 790–792 (2009) 

2. Lindvall, O. & Hyun, I. Science 324, 1664–1665 

(2009). 

3. Lau, D. et al. Cell Stem Cell 3, 591–594 (2008). 

4. Regenberg, A., Hutchinson, L., Schanker, B. & Matthews, D. 

Stem Cells 27, 2312–2319 (2009). 

5. Ryan, K., Sanders, A., Wang, D. & Levine, A. 

Regenerative Med. 5, 27–33 (2010). 

6. Amariglio, N. et al. PLoS Med 6, e1000029 (2009). 

7. http://www.isscr.org/clinical_trans/pdfs/ISSCRGL 

ClinicalTrans.pdf 

8. MacReady, N. Lancet Oncol. 10, 317–318 (2009). 

9. Kiatpongsan, S. & Sipp, D. Nat. Rep. Stem Cells 

published online 3 December, 2008, doi:10.1038/ 

stemcells.2008.151 (2008). 

10. Zarzeczny, A. & Caulfield, T. Am. J. Bioeth. 10, 1–13 

(2010). 

11. Iyengar, S. & Kinder, D. News that Matters: Television 

and American Public Opinion. (University of Chicago 

Press, Chicago, IL, USA, 1987). 

12. Nisbet, M.C. & Mooney, C. Science, 316, 56 

(2007). 

13. Creasy, G. & Scott, C. Nat. Biotechnol. 27, 21–22 

(2009). 

Tracking and assessing the rise of 

state-funded stem cell research 


The editorial in your October issue 1 

highlights the legal challenges to new 

guidelines issued by the US National 

Institutes of Health (NIH) in July 2009 for the 

federal funding of human embryonic stem 

cell (hESC) research. In the eight-year period 

preceding these most recent NIH guidelines, 

only a small number of cell lines could be 

studied with federal funds. During this time, 

six states—California, Connecticut, Illinois, 

Maryland, New Jersey and New York—took 

on a role typically played by the NIH and 

created funding programs specifically 

designed to support stem cell research, 

including hESC research. These are not the 

first state programs to fund scientific research 

but their commitment to basic research is 

atypical, as most state science and technology 

programs have focused on science closer to 

commercialization 2 . Although the state stem 

cell programs differ, they each share at least 

two goals: advancing promising science, 

including research not eligible for federal 

funding during the Bush Administration, and 

returning economic benefits to their state. 

In this article, we report an initial 

attempt to track and assess the impact of 

these state funding programs. Existing 

work on state stem cell policy has focused 

on identifying policy differences between 

various jurisdictions 3,4 , assessing the impact 

of state decisions to support or restrict hESC 

science 5–7 and examining the role of states 

in governing controversial science 8,9 . The 

analysis reported here extends this literature 

though use of a novel data set of the grants 

these states have awarded. These data provide 

insight into how states have prioritized their 

funding, including the extent to which they 

have supported hESC research generally 

and hESC research not eligible for federal 

funding during the Bush Administration 

more specifically, as well as the extent to 

which these states have drawn new scientists 

into the field. The underlying data have been 

publicly released on a new website (http:// 

www.stemcellstates.net) designed to facilitate 

additional analysis of state-funded stem 

cell science and improve public awareness 

of these programs. Given ongoing legal 

uncertainties surrounding federal funding 

of hESC research and the likelihood that 

voters, at least in California, will be asked to 

approve additional state stem cell funding in 

the future, understanding and evaluating the 

effects of state-funded stem cell research is 

both timely and useful. 

The database that forms the basis for the 

analysis described here contains the title, 

principal investigator, institution, abstract 

and amount for each grant awarded by 

the agency overseeing stem cell research 

funding in these six states (Supplementary 

Methods). In all, between December 2005 

when New Jersey awarded the first state stem 

cell grants and the end of 2009, the six stem 

cell states awarded nearly 750 grants totaling 

just over $1.25 billion. The scale of these 

programs varies substantially, ranging from 

the roughly $15 million awarded by Illinois 

and New Jersey to the $1.02 billion awarded 

by California. On a per capita basis, funding 

awarded through the end of 2009 ranges 

from just over $1 in Illinois to nearly $28 in 

California (Table 1). 

States funding stem cell research 

can choose to support several different 

activities, ranging from investigatorinitiated 

research grants to new facilities 

to workforce development. To investigate 

how states prioritized these various types 

of funding, each grant was classified by 

its primary purpose (Supplementary 

Methods). Research grants and support for 

scientific infrastructure were the two largest 

categories, accounting for more than 90% 

of all state stem cell funding (Table 1). The 

infrastructure category was dominated by 

the $271 million California awarded for the 

construction of 12 major stem cell research 

facilities, although several other states also 

dedicated a substantial portion of their 

funding to infrastructure, such as shared 

equipment or core laboratories. In contrast 

to supporting basic investigator-initiated 

research, spending money on infrastructure 

is a classic state economic development 

approach, but, in these cases, spending 

was motivated, at least in part, by the need 

to create separate laboratories to facilitate 

research on unapproved hESC lines. 

The restrictions on federal funding for 

hESC research instituted by former President 

George W. Bush were an important rationale 

behind the adoption of most state stem cell 

programs, yet it is not clear to what extent 

state programs focused on hESC research 

generally or hESC research not eligible for 

federal funding more specifically. To address 

these questions, each research grant awarded 

through the end of 2009 was analyzed to 

assess if funded research used hESCs and, 

if applicable, appeared ineligible for federal 

funding under the Bush Administration 

rules (Supplementary Methods). The 

percentage of grants that supported hESC 




Table 1 The scale and prioritization of state stem cell funding programs a 

research varied substantially among these 

states (Table 1). Large majorities of the 

research grants awarded in Connecticut 

and California supported studies involving 

hESCs, whereas only a minority of grants 

supported hESC research in the other states. 

These disparities likely reflect differences 

in the types of stem cell scientists present in 

these states as well as priorities of the various 

state funding bodies. 

Only a subset of grants for hESC 

research supported science that was 

clearly ineligible for NIH funding during 

the Bush Administration. California and 

Connecticut focused the most on this sort 

of research—which typically involved the 

derivation of new hESC lines or the use 

of newer unapproved cell lines—but even 

in these states fewer than a fifth of grants 

went to clearly ineligible research. Many 

scientists indicated plans to use existing 

hESC lines but did not specify which lines 

they planned to use. Given evidence that a 

handful of approved cell lines account for a 

large proportion of the hESC lines actually 

distributed to scientists and an even larger 

share of published literature 10 , most of these 

projects probably used approved hESC lines. 

In some cases, however, scientists may have 

chosen to use ineligible cell lines but not 

clearly indicated these plans. Thus, the share 

of grants reported here as clearly ineligible 

for NIH funding should be viewed as a lower 

bound on the amount of research each state 

funded that was ineligible for federal funding. 

Several factors could explain the relatively 

small share of grants that went toward clearly 

ineligible research. Some scientists who 

wished to pursue this research may have been 

unable to access the raw materials or acquire 

the intellectual property rights required to do 

so. Alternatively, these findings could simply 

reflect scientific interest. The discovery of 

induced pluripotent stem cells 11 may, for 

instance, have reduced scientific interest in 

the derivation of new hESC lines. Finally, 

these findings may reflect a preference on the 

part of scientists to use well-established and 

well-studied hESC lines. This last explanation 

may be particularly relevant for new scientists 

entering the field of hESC research, as using 

recognized cell lines may give their initial 

research efforts greater credibility. 

In addition to supporting research not 

eligible for federal funding, focused state 

programs might serve to draw new scientists 

into the field of stem cell research. To 

evaluate this potential impact, the recent NIH 

funding portfolio of each scientist receiving a 

state stem cell grant with a primary purpose 

of research was examined (Supplementary 

Methods). Although most scientists had 

received NIH funding, a substantial number 

(ranging from 42% in California to 71% in 

Maryland) had not received NIH funding 

for stem cell research (Table 1). Similar, 

but more pronounced, results are observed 

when the NIH funding portfolio of scientists 

receiving state funding for hESC research is 

examined, as only a small minority of these 

scientists also had NIH grants supporting 

hESC research. Given the importance of 

NIH funding for biomedical research in the 

United States, these results suggest that the 

existence of state funding programs for stem 

cell research has drawn many new scientists 

into the field of stem cell research, or at least 

encouraged scientists to consider how stem 

State program 

Grants and funding California Connecticut Illinois Maryland New Jersey New York 

Year first grants awarded 2006 2006 2006 2007 2005 2008 

Total funding pledged/time period 

$3 billlion/ 

10 years 

$100 million/ 

10 years 

N/A N/A N/A $600 million/ 

11 years 

Number of grants awarded 329 69 17 140 35 158 

Funding awarded ($ millions) 1,024 40 15 54 15 121 

Funding per capita b ($) 28 11 1 10 2 6 

Research prioritization c 

Percentage of funding for research 58% 76% 100% 93% 64% 61% 

Percentage of funding for infrastructure 31% 24% 0% 0% 36% 34% 

Percentage of grants for hESC research 75% 97% 35% 42% 21% 21% 

Percentage of grants clearly not NIH eligible 18% 16% 12% 3% 6% 0% 

NIH funding status of state grant recipients d 

Percentage of state PIs without NIH stem cell funding 42% 61% 65% 71% 61% 49% 

Percentage of state hESC PIs without NIH hESC funding 77% 91% 67% 79% 100% 66% 

a Includes grants awarded through the end of 2009. b Per capita funding based on state population from US Census Bureau 2009 Population Estimates. c hESC prioritization analysis includes 

only grants with a primary purpose of research. d NIH funding was examined from FY2005 to present. PIs, principal investigators. 

cell research could complement their existing 

research programs. 

These data also permit a more nuanced 

comparison between state stem cell funding 

and NIH stem cell funding than has 

previously been available (Fig. 1). Total state 

funding for all types of stem cell research has 

risen rapidly since grants were first awarded 

in 2005, but states still spend less than half of 

what the NIH spends each year on stem cell 

research. The situation is different for hESC 

research, as state funding for hESC research 

grants first exceeded comparable NIH 

funding in 2007 and equaled or exceeded it in 

2008 and 2009. 

Considered together, these data and 

analyses indicate that state funding for stem 

cell research has grown into a substantial 

enterprise that has provided funding on a 

scale comparable to the NIH. Although states 

vary in the degree to which they have focused 

on hESC research, as a whole, state funding 

for hESC research has been substantial, 

exceeding, in cumulative terms, NIH funding 

for this research between 2005 and 2009. 

Most state hESC funding appears to have 

supported research also eligible for federal 

funding during the Bush Administration. 

This finding is surprising, given the 

explicit intent of several state programs to 

preferentially support science not eligible for 

federal funding, but likely reflects the nature 

of the grant proposals state agencies received, 

particularly given the number of grants states 

awarded to scientists relatively new to the 

field of hESC research. 

In the light of the recent change in 

federal stem cell policy and the ongoing 

economic downturn, the future of state 




a 

Funding awarded 

($ millions) 

1,250 

1,000 

750 

500 

250 

0 

609 

2005 

5 

643 

2006 

Figure 1 Comparing state and NIH stem cell funding. (a) Total amount of all NIH stem cell grants and 

all stem cell grants awarded by the six states. (b) Total amount of all NIH and state hESC research 

grants. Only grants with a primary purpose of research are included. State funding is by calendar year. 

NIH funding is by fiscal year. 

stem cell programs, as well as similar state 

programs supporting other areas of science, 

is uncertain. The analysis here suggests 

that state stem cell funding programs 

are sufficiently large and established that 

simply ending the programs, at least in 

the absence of substantial investment in 

the field by other funding sources, could 

have deleterious effects. Such action would 

fail to capitalize on the initial efforts of 

scientists who have been drawn to the field 

of stem cell research by state programs and 

leave many stem cell scientists suddenly 

searching for funding to continue their 

research. 

Large-scale state funding for basic 

research is a relatively new phenomenon, 

and many questions remain about 

the impact of these programs on the 

development of scientific fields and the 

careers of scientists. The influence of state 

funding programs on the distribution 

of research publications, the acquisition 

of future external funding, the creation 

of new companies and the translation 

of basic research into medical practice, 

for instance, are important unanswered 

questions. Similarly, comparing state 

funding programs with federal funding 

programs as well as foundations could offer 

new insight into the relative priorities of 

different funding bodies and the extent 

to which their funding portfolios overlap 

or are distinct. We hope the analysis 

presented here and the public release of the 

underlying database will inspire additional 

analysis of state science funding programs 

generally and state-funded stem cell science 

in particular. 



Acknowledgements 

The authors gratefully acknowledge financial 

support from the Roadmap for an Entrepreneurial 

Economy Program, funded by the Kauffman 

73 

657 

2007 

246 

938 

2008 

426 

1,231 

2009 

519 

NIH grants 

b 


($ millions) 

180 

150 

120 

90 

60 

30 

0 

33 

2005 

1 

State grants 

31 

2006 

21 

Foundation and the Georgia Research Alliance, 

and Georgia Tech. They thank J. Walsh at Georgia 

Tech for helpful comments on an earlier version of 

this manuscript. They also appreciate the assistance 

they received with data collection from officials 

in various state stem cell agencies. A.D.L. would 

also like to thank A. Jakimo, whose comment at 

a meeting of the Interstate Alliance on Stem Cell 

Research inspired collection of these data. 

35 

2007 

76 

2008 

144 

125 


We report on the launch of version 7 of 

the Human Protein Atlas with subcellular 

localization data and expression data 

for all major human tissues and organs. 

A milestone has been achieved with the 

inclusion of expression data for >50% of the 

human protein-coding genes. The main new 

feature of the release is an attempt towards 

a knowledge-based portal, including an 

annotated protein expression feature for 

protein targets analyzed with two or more 

antibodies, and the establishment of the main 

subcellular localization of protein targets. 

In 2005, the first version of the Human 

Protein Atlas (http://www.proteinatlas. 

org/) was released with protein profile 

data based on immunohistochemistry on 

tissue microarrays covering 48 different 

human tissues and organs, including kidney, 

liver, heart, brain and pancreas 1 . The first 

version included data from 718 antibodies 

corresponding to 650 human protein-coding 

genes. High-resolution images were published 

along with annotation of the presence or 

absence of a particular protein target in all 

represented tissues. The 2005 Human Protein 

Atlas also contained information regarding 

protein profiles from 20 different types of 

human cancer, including breast, colorectal, 

155 

76 

2009 



Ruchir N Karmali 1 , Natalie M Jones 1 & 

Aaron D Levine 1,2 

1 School of Public Policy, Georgia Institute of 

Technology, Atlanta, Georgia, USA. 2 Institute of 

Bioengineering & Bioscience, Georgia Institute 

of Technology, Atlanta, Georgia, USA. 

e-mail: aaron.levine@pubpolicy.gatech.edu 

1. Anonymous. Nat. Biotechnol. 28, 987 (2010). 

2. Plosila, W.H. Econ. Dev. Q. 18, 113–126 (2004). 

3. Stayn, S. BNA Med. Law Pol. Rep. 5, 718–725 

(2006). 

4. Lomax, G. & Stayn, S. BNA Med. Law Pol. Rep. 7, 

695–698 (2008). 

5. Levine, A.D. Public Adm. Rev. 68, 681–694 (2008). 

6. Levine, A.D. Nat. Biotechnol. 24, 865–866 (2006). 

7. McCormick, J.B., Owen-Smith, J. & Scott, C.T. Cell 

Stem Cell 4, 107–110 (2009). 

8. Fossett, J.W., Ouellette, A.R., Philpott, S., Magnus, D. 

& Mcgee, G. Hastings Cent. Rep. 37, 24–35 (2007). 

9. Mintrom, M. Publius 39, 606–631 (2009). 

10. Scott, C.T., McCormick, J.B. & Owen-Smith, J. Nat. 

Biotechnol. 27, 696–697 (2009). 

11. Takahashi, K. & Yamanaka, S. Cell 126, 663–676 

(2006). 

Towards a knowledge-based Human 

Protein Atlas 

lung and prostate cancer. The data in the 

portal were made available freely both for 

academia and industry without restrictions or 

password protection. In 2007, the portal was 

extended to also include subcellular profiling 

data 2 using immunofluorescence-based 

confocal microscopy in three human cancer 

cell lines of different (glial, mesenchymal and 

epithelial) origin. More data have been added 

to the portal every year since the first release 3 

and version 6, launched in March 2010, 

contained 11,274 antibodies corresponding to 

8,489 protein-coding genes. This entire effort 

depends heavily on the availability of good 

quality antibodies, and recently a communitybased 

portal, Antibodypedia (http://www. 

antibodypedia.org/), has been launched to 

allow antibodies from different providers to 

be listed and compared 4,5 , although the main 

source of information so far comes from 

the providers’ own validation data, not by 

independent third-party users. At present, 

the Antibodypedia contains close to 100,000 

antibodies, corresponding to >70% of the 

protein-coding genes in humans. 

An important objective has now been 

reached with the inclusion of 10,118 proteincoding 

genes corresponding to >50% of the 

19,559 human entries as defined by UniProt, 

including only entries with evidence at protein 




a 


($ millions) 

1,250 

1,000 

750 

500 

250 

0 

609 

2005 

5 

643 

2006 

Figure 1 Comparing state and NIH stem cell funding. (a) Total amount of all NIH stem cell grants and 

all stem cell grants awarded by the six states. (b) Total amount of all NIH and state hESC research 

grants. Only grants with a primary purpose of research are included. State funding is by calendar year. 

NIH funding is by fiscal year. 

stem cell programs, as well as similar state 

programs supporting other areas of science, 

is uncertain. The analysis here suggests 

that state stem cell funding programs 

are sufficiently large and established that 

simply ending the programs, at least in 

the absence of substantial investment in 

the field by other funding sources, could 

have deleterious effects. Such action would 

fail to capitalize on the initial efforts of 

scientists who have been drawn to the field 

of stem cell research by state programs and 

leave many stem cell scientists suddenly 

searching for funding to continue their 

research. 

Large-scale state funding for basic 

research is a relatively new phenomenon, 

and many questions remain about 

the impact of these programs on the 

development of scientific fields and the 

careers of scientists. The influence of state 

funding programs on the distribution 

of research publications, the acquisition 

of future external funding, the creation 

of new companies and the translation 

of basic research into medical practice, 

for instance, are important unanswered 

questions. Similarly, comparing state 

funding programs with federal funding 

programs as well as foundations could offer 

new insight into the relative priorities of 

different funding bodies and the extent 

to which their funding portfolios overlap 

or are distinct. We hope the analysis 

presented here and the public release of the 

underlying database will inspire additional 

analysis of state science funding programs 

generally and state-funded stem cell science 

in particular. 



Acknowledgements 

The authors gratefully acknowledge financial 

support from the Roadmap for an Entrepreneurial 

Economy Program, funded by the Kauffman 

73 

657 

2007 

246 

938 

2008 

426 

1,231 

2009 

519 

NIH grants 

b 


($ millions) 

180 

150 

120 

90 

60 

30 

0 

33 

2005 

1 

State grants 

31 

2006 

21 

Foundation and the Georgia Research Alliance, 

and Georgia Tech. They thank J. Walsh at Georgia 

Tech for helpful comments on an earlier version of 

this manuscript. They also appreciate the assistance 

they received with data collection from officials 

in various state stem cell agencies. A.D.L. would 

also like to thank A. Jakimo, whose comment at 

a meeting of the Interstate Alliance on Stem Cell 

Research inspired collection of these data. 

35 

2007 

76 

2008 

144 

125 


We report on the launch of version 7 of 

the Human Protein Atlas with subcellular 

localization data and expression data 

for all major human tissues and organs. 

A milestone has been achieved with the 

inclusion of expression data for >50% of the 

human protein-coding genes. The main new 

feature of the release is an attempt towards 

a knowledge-based portal, including an 

annotated protein expression feature for 

protein targets analyzed with two or more 

antibodies, and the establishment of the main 

subcellular localization of protein targets. 

In 2005, the first version of the Human 

Protein Atlas (http://www.proteinatlas. 

org/) was released with protein profile 

data based on immunohistochemistry on 

tissue microarrays covering 48 different 

human tissues and organs, including kidney, 

liver, heart, brain and pancreas 1 . The first 

version included data from 718 antibodies 

corresponding to 650 human protein-coding 

genes. High-resolution images were published 

along with annotation of the presence or 

absence of a particular protein target in all 

represented tissues. The 2005 Human Protein 

Atlas also contained information regarding 

protein profiles from 20 different types of 

human cancer, including breast, colorectal, 

155 

76 

2009 



Ruchir N Karmali 1 , Natalie M Jones 1 & 

Aaron D Levine 1,2 

1 School of Public Policy, Georgia Institute of 

Technology, Atlanta, Georgia, USA. 2 Institute of 

Bioengineering & Bioscience, Georgia Institute 

of Technology, Atlanta, Georgia, USA. 

e-mail: aaron.levine@pubpolicy.gatech.edu 

1. Anonymous. Nat. Biotechnol. 28, 987 (2010). 

2. Plosila, W.H. Econ. Dev. Q. 18, 113–126 (2004). 

3. Stayn, S. BNA Med. Law Pol. Rep. 5, 718–725 

(2006). 

4. Lomax, G. & Stayn, S. BNA Med. Law Pol. Rep. 7, 

695–698 (2008). 

5. Levine, A.D. Public Adm. Rev. 68, 681–694 (2008). 

6. Levine, A.D. Nat. Biotechnol. 24, 865–866 (2006). 

7. McCormick, J.B., Owen-Smith, J. & Scott, C.T. Cell 

Stem Cell 4, 107–110 (2009). 

8. Fossett, J.W., Ouellette, A.R., Philpott, S., Magnus, D. 

& Mcgee, G. Hastings Cent. Rep. 37, 24–35 (2007). 

9. Mintrom, M. Publius 39, 606–631 (2009). 

10. Scott, C.T., McCormick, J.B. & Owen-Smith, J. Nat. 

Biotechnol. 27, 696–697 (2009). 

11. Takahashi, K. & Yamanaka, S. Cell 126, 663–676 

(2006). 

Towards a knowledge-based Human 

Protein Atlas 

lung and prostate cancer. The data in the 

portal were made available freely both for 

academia and industry without restrictions or 

password protection. In 2007, the portal was 

extended to also include subcellular profiling 

data 2 using immunofluorescence-based 

confocal microscopy in three human cancer 

cell lines of different (glial, mesenchymal and 

epithelial) origin. More data have been added 

to the portal every year since the first release 3 

and version 6, launched in March 2010, 

contained 11,274 antibodies corresponding to 

8,489 protein-coding genes. This entire effort 

depends heavily on the availability of good 

quality antibodies, and recently a communitybased 

portal, Antibodypedia (http://www. 

antibodypedia.org/), has been launched to 

allow antibodies from different providers to 

be listed and compared 4,5 , although the main 

source of information so far comes from 

the providers’ own validation data, not by 

independent third-party users. At present, 

the Antibodypedia contains close to 100,000 

antibodies, corresponding to >70% of the 

protein-coding genes in humans. 

An important objective has now been 

reached with the inclusion of 10,118 proteincoding 

genes corresponding to >50% of the 

19,559 human entries as defined by UniProt, 

including only entries with evidence at protein 




or transcript level and proteins inferred 

from homology 6 (Fig. 1). The chromosomal 

coverage of protein-coding genes is shown 

in Figure 1a and the status for a selection 

of important protein classes is reported in 

Figure 1b. Almost 80% of the human kinases 

and Src-homology 2 domain–containing 

proteins and >50% of the transcription factors 

have protein profiling data in the atlas. 

We introduce the concept of annotated 

protein expression for paired antibodies, in 

which two or more independent antibodies 

are used to validate the staining pattern 

of each other. The immunohistochemical 

staining in each tissue or organ by the 

independent antibodies is compared and a 

new annotated protein expression is manually 

curated for each cell type in each tissue or 

organ. A reliability score is generated as an 

estimation of the degree of knowledge-based 

certainty of the reported expression profile. 

The reliability score is based on the similarity 

of the staining for the different antibodies, but 

also takes into account available information 

from literature, bioinformatics predictions 

and additional experimental evidence, such 

as western blots, transcript profiling and/ 

or small interfering RNA knockdowns 

(Supplementary Notes; http://www. 

proteinatlas.org/about/quality+scoring). 

Approximately 2,000 protein-coding genes 

have annotated protein expression patterns 

in this launch and >75% of these have been 

scored with a high or medium reliability score 

(Fig. 1c). It is noteworthy that, at present, 

the majority of the proteins in the atlas have 

only been analyzed with a single antibody 

and thus these genes are reported simply as 

antibody staining. The long-term objective of 

the Human Protein Atlas project is to generate 

at least two antibodies for all human proteincoding 

genes to allow knowledge-based 

annotated protein expression data for the 

complete human proteome. 

The new portal contains a summary page 

for every human gene (Fig. 2) to allow for 

an easy entry point to navigate the detailed 

protein expression data. The summary page 

also serves as a convenient entry page from 

other related databases with a similar genecentric 

format. The results from the curation 

and annotation of the staining patterns from 

one or several antibodies to the same protein 

target are summarized with links to details 

pages, including the original images. A new 

feature is a section for subcellular localization 

and here, the annotated knowledge-base of 

the subcellular distribution reports the main 

and additional subcellular location for each 

protein target in the analyzed cell. Because 

the location of human proteins is known only 

a 

Number of genes 

2,000 No antibody 

Single antibody 

Multiple antibodies 

c 

1,500 

1,000 

500 

0 

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 X Y 

Medium 

36% 

Chromosome 

High 

42% 

Low 

17% 

Very low 

5% 

for a small fraction of the human genes, these 

subcellular localization data are a valuable 

resource for defining the protein content of 

the various compartments of the human cell 

and can thus constitute a starting point for 

further in-depth functional studies. 

A new organ view has been designed to 

allow cell types of similar origin to be easily 

compared. The tissues and organs have 

been divided into 12 functional classes, 

including central nervous system (CNS), 

hematopoietic, cardiovascular and female 

and male reproductive tissues. In this release, 

a total of 66 normal cell types from 46 tissues 

and organs have been scored. As an example, 

in Figure 2, we show part of the protein 

profiles for the human estrogen receptor 1 

(ESR1), including the results from antibody 

staining of three independent antibodies and 

b 

d 

Number of genes 

Fraction of cell types (%) 

1,400 

1,200 

1,000 

800 

600 

400 

200 

0 

100 

80 

60 

40 

20 

Kinases 

CD markers 

Transporters 

0 

0 500 

No antibody 

Single antibody 

Multiple antibodies 

Weak 

Moderate 

Strong 

the resulting annotated protein expression. 

The annotated protein expression in 

the various cell types are obtained by a 

knowledge-based process taking into 

account staining from at least two antibodies, 

literature and additional experimental data. 

For the ESR1 (Fig. 2), one of the antibodies 

shows weak staining in tissues such as those 

of the prostate and urinary bladder, most 

likely due to nonspecific staining, but the 

conflation of data from the three antibodies 

suggests exclusive expression in breast and 

female reproductive tissues. It is important 

to point out that this annotated protein 

expression should be considered as the best 

estimate of the true expression based on 

the experimental data available, but that 

additional data and input from the scientific 

community may lead to subsequent revisions 

Peptidases 

GPCRs (excl. 

olfactory receptors) 

1,000 

Genes 

Figure 1 Overview of the data in the Human Protein Atlas version 7.0. (a,b) The total coverage in terms 

of number of genes analyzed with single or multiple antibodies for each of the human chromosomes (a) 

and a selection of protein classes 3 (b). (c) Reliability scores for the ‘annotated protein expression’ are 

given by a four-graded scale, ranging from ‘very low’ to ‘high’. (d) The percent of cell types with the 

annotated expression levels weak, moderate and strong for each of the 2,000 genes with ‘annotated 

protein expression’. The genes are arranged according to the total abundance in the analyzed cell types, 

with genes detected only in a single cell type to the left and genes detected in all cell types to the far 

right. GPCRs, G protein–coupled receptors. SH2, Src-homology 2 domain. 

Transcription factors 

SH2–domain 

containing proteins 

SH3–domain 

containing proteins 

1,500 



Normal tissues 

Cancer tissues 

Annotated protein expression 

this manner, the antibody-based portal can 

be complemented with other efforts, such 

as mass spectrometry–based proteomics 

and gene fusion technologies, to provide a 

knowledge-base for the human proteins, 

including isoforms, subcellular localization, 

tissue profiles and interaction networks. 

Because all the data on the Human Protein 

Atlas are publicly available, the information 

can be integrated into other databases. The 

availability of the annotated expression 

patterns opens up the possibility for a 

community-based dialog to provide input 

by researchers with specialized knowledge 

about particular protein targets. 


Figure 2 The Human Protein Atlas web portal. The Human Protein Atlas summary view for the human 

estrogen receptor 1 (ESR1), displaying expression data for the target protein in normal and diseased 

tissues. To the right, a detailed organ view of the expression pattern of the ESR1 in normal human 

tissues with the cell types grouped into functional classes, here focused on the female and male tissues 

and the urinary tract. Three different antibodies have been used to generate the staining patterns, from 

which an annotated protein expression has been retrieved. The blue color scale represents annotated 

protein expression, and the red color scale represents relative antibody staining. The intensity of the 

colors indicates the level of expression and/or staining. 

of the interpretation of protein expression in 

individual cells, tissues or organs. 

A new cancer view has been designed, 

allowing an easy overview of the results for 

all antibodies towards a particular protein 

target in tumor tissues from 216 patients 

representing 20 different types of cancer 

(Fig. 2). This overview also contains the 

results for the corresponding normal tissues 

for each cancer form. An advanced search 

function has been created to allow complex 

queries involving both normal and cancer 

tissues, including subcellular localization, 

validation results for the antibody and/or 

target protein classes. An additional feature is 

the possibility to customize the display of the 

results page by adding or deleting columns 

based on interest. An important objective is 

to allow the integration of this database with 

other ’omics databases covering genomics 7 , 

gene variation 8 , transcript profiling 9 , 

proteomics 10 or metabolomics 11 . All the 

expression data and the antibody staining 

data are therefore available for downloads 

to facilitate comparative expression studies 

and systems biology approaches. In the near 

future, we also plan to include additional 

experimental data from mass spectrometry– 

based proteomics 12 and transcriptomics 

profiling using next generation RNA-Seq 13 . 

An issue with the new gene-centric 

design is the presence of protein isoforms, 

including proteolytic variants, posttranslational 

modifications and splice 

variants. Although the antigen sequence 

is reported for most antibodies on the 

detailed antibody validation page, and 

thus makes it possible to determine if a 

particular antibody should recognize all 

splice variants or not, the annotated protein 

expression is based on a consensus model 

in which a representative protein from the 

gene is reported. In the future, it might be 

relevant to move toward separate annotated 

protein expression profiles for each splice 

variant or other major alternative isoforms 

of a particular protein target. 

The annotated protein expression profiles 

allow a refinement of the global analysis 

of tissue specificity recently reported 14 , 

in which the staining of antibodies 

corresponding to one-third of the proteincoding 

genes suggested that a large portion 

of the human proteins are present across 

tissues in a relatively ubiquitous manner. 

An analysis based on the 2,000 protein 

targets with annotated protein expression 

profiles can now be performed and the result 

suggests a similar profile with a large subset 

of proteins detected across various cell types 

(Fig. 1d). 

In conclusion, we here report on a new 

version of the Human Protein Atlas, where 

we have moved towards a knowledge-based 

portal with gene-centric expression profiles 

based on the annotation of several antibodies 

towards the same protein target. The Human 

Protein Atlas effort described here fits well 

into recent suggestions to initiate a genecentric 

Human Proteome Project to map and 

characterize a representative protein from 

every protein-coding gene of humans 15,16 . In 




The authors would like to acknowledge the entire 

staff of the Human Protein Atlas project. This work 

was supported by grants from the Knut and Alice 

Wallenberg Foundation and EU 7 th framework 

program PROSPECTS. 



Mathias Uhlen 1,2 , Per Oksvold 1 , 

Linn Fagerberg 1 , Emma Lundberg 2 , 

Kalle Jonasson 1 , Mattias Forsberg 1 , 

Martin Zwahlen 1 , Caroline Kampf 3 , 

Kenneth Wester 3 , Sophia Hober 1 , 

Henrik Wernerus 1 , Lisa Björling 1 & 

Fredrik Ponten 3 

1 School of Biotechnology, AlbaNova University 

Center, Royal Institute of Technology (KTH), 

Stockholm, Sweden. 2 Science for Life Laboratory, 

Royal Institute of Technology (KTH), Stockholm, 

Sweden. 3 Department of Genetics and Pathology, 

Rudbeck Laboratory, Uppsala University, 

Uppsala, Sweden. 

1. Uhlen, M. et al. Mol. Cell Proteomics 4, 1920–1932 

(2005). 

2. Barbe, L. et al. Mol. Cell Proteomics 7, 499–508 

(2008). 

3. Berglund, L. et al. Mol. Cell Proteomics 7, 2019–2027 

(2008). 

4. Bjorling, E. & Uhlen, M. Mol. Cell Proteomics 7, 2028– 

2037 (2008). 

5. Kiermer, V. Nat. Methods 5, 860 (2008). 

6. The UniProt Consortium. Nucleic Acids Res. 38, 

D142–148 (2010). 

7. Flicek, P. et al. Nucleic Acids Res. 38, D557–D562 

(2010). 

8. The International HapMap Project. Nature 426, 789– 

796 (2003). 

9. Lukk, M. et al. Nat. Biotechnol. 28, 322–324 

(2010). 

10. Mathivanan, S. et al. Nat. Biotechnol. 26, 164–167 

(2008). 

11. Wishart, D.S. et al. Nucleic Acids Res. 35, D521–526 

(2007). 

12. Ong, S.E. et al. Mol. Cell Proteomics 1, 376–386 

(2002). 

13. Sultan, M. et al. Science 321, 956–960 (2008). 

14. Ponten, F. et al. Mol. Syst. Biol. 5, 337 (2009). 

15. Anonymous. Nat. Methods 7, 661 (2010). 

16. Anonymous. Mol. Cell Proteomics 9, 427–429 

(2010). 


commentary 

Anchoring gene patent eligibility to its 

constitutional mooring 

Kenneth G Chahine 


The antiquated legal standard that natural laws and products are not eligible for patent protection is ill-suited for gene 

and diagnostics patents. Here, I propose a new, technology-agnostic framework for determining patent eligibility that 

is tailored to the meet the US Constitutional objective of promoting innovation. 

The escalating debate over the validity of gene 

and diagnostic patents in the United States 

threatens to strip the incentives necessary to 

translate scientific findings from the bench to 

the bedside. Although a strong aversion to patents 

in new technological areas is nothing new, 

what sets the current debate apart is the extent 

of the misinformation, the personal nature of 

the subject matter, the support of powerful 

organizations and the potential damage an illinformed 

decision would have on the future 

of medicine and healthcare reform. The latest 

ruling in Association for Molecular Pathology 

v. US Patent & Trademark Office 1 (The Myriad 

Genetics Case) that invalidated patents covering 

the isolated breast cancer genes (BRCA1 

and BRCA2) and methods for their use in 

assessing the risk of breast and ovarian cancer 

provides an example of a decision based on 

faulty scientific reasoning, murky legal precedent 

and unsupported policy arguments. 

Any judge-made doctrine for assessing patent 

eligibility should be based on sound legal 

principles that are closely aligned with the constitutional 

purpose to “promote the Progress of 

Science and the Useful Arts” 2 . The challenge 

lies in finding the proper balance between 

encouraging innovation by awarding limited 

exclusivity to inventors and stymieing basic 

research through overprotection. Refocusing 

the legal reasoning for the natural products 

doctrine on this constitutional language offers 

a fresh perspective on the debate and opens the 

door to a new proposal for determining patent 

Kenneth G. Chahine is professor of law and 

member of the BioLaw Project, S.J. Quinney 

College of Law at University of Utah, 

Salt Lake City, Utah, USA. 

e-mail: chahinek@law.utah.edu 

eligibility, which is closely aligned to the goal 

of promoting innovation. 

In this article, I provide five criteria that 

would be useful for courts, legislatures and 

businesses when confronted with a question 

of subject matter eligibility. Taken together, 

these criteria incorporate many of the principles 

previously announced by the courts. They 

also allow courts to build a solid evidentiary 

foundation from which to decide the scope 

of an invention and its eligibility for patent 

protection. Importantly, the criteria may provide 

greater predictability to the legal, investment 

and research community by creating a 

legal framework for patent eligibility with one 

unequivocal and constitutionally consistent 

objective—promoting innovation. 

The natural products doctrine 

US courts have long held that “laws of nature, 

natural phenomena, and abstract ideas” are 

not eligible for patent protection 3 . Under 

this ‘natural laws’ or ‘natural products’ doctrine, 

the eligibility of two classes of patents 

are being questioned: first, patents covering 

isolated gene sequences; and second, methods 

of comparing genetic sequences, proteins and 

metabolites to diagnose disease or individualize 

treatment 4,5 . 

It has become increasingly clear, however, 

that advances in medicine and other information-based 

technologies are calling into 

question the applicability of the natural products 

doctrine. During the oral arguments in 

Bilski v. Kappos, US Supreme Court Justice 

Breyer voiced his own uncertainty on where 

to draw the line when he asked counsel—“All 

right, so what do I do?” 6 . Although the Bilski 

decision was hailed as a victory by the biotech 

community, it was based on the court’s 

reluctance to make a sweeping decision that 

might cause collateral damage to other industries 

7 . In rejecting the US Federal Circuit’s 

rigid test for determining patent eligibility, 

the Supreme Court once again affirmed the 

need for other criteria to establish patent eligibility 

that was consistent with the ‘purpose’ 

of our patent system 8 . 

In examining the legal precedence for the 

natural products doctrine, I found there is 

little consistency in the manner in which the 

doctrine is applied. Because at a basic level all 

innovations comprise products of nature, even 

proponents of the doctrine have difficulty 

Judge Robert Sweet of the NY federal circuit 

court, who in March invalidated Myriad Genetics’ 

BRCA1 and BRCA2 patents. 

©Joe Lawton 


COMMENTARY 


defining when an invention crosses the line 

and becomes ineligible for patent protection. 

In distinguishing the patent ineligibility of 

isolated genes and the patent eligibility of 

purified adrenaline the Department of Justice 

argues that “patent eligibility may arise when 

a natural compound has been so refined and 

purified through human intervention as to 

become a substance different in kind from the 

natural product” 9 . The Department, however, 

offers no suggestion for determining when a 

product has been simply purified and remains 

patent ineligible (e.g., isolated genes) and 

when it has been purified to the extent that 

it now has changed ‘in kind’ and is patent eligible 

(e.g., adrenaline). 

Early decisions boldly declared without 

much support that natural products are “free 

to all men and reserved exclusively to none” 10 . 

More recently, the courts have relied on the 

justification that §101 of the US Patent Act 

excludes products of nature. As the Supreme 

Court has admitted, however, “[t]he plain language 

of §101 does not answer the question” 11 

and Chief Judge Rader of the Federal Circuit 

Court has called it a blunt tool for defining 

patent eligibility (http://www.patentlyo.com/ 

patent/2010/08/gene-patents-on-appeal-aclusrecusal-motion.html). 

Indeed, §101 broadly 

embraces “any new and useful” invention or 

discovery, causing courts to strain the plain 

language of the statute to justify their exclusion 

of laws and products of nature 12 . Notably, 

both the US Constitution and the Patent Act 

explicitly embrace not only inventions, but also 

discoveries. The plain language of the statute 

and US Constitution, therefore, discredits the 

often-used argument that natural products are 

ineligible because they have been discovered 

rather than invented and cast doubt over the 

Department of Justice’s claim that patent law 

embraces only “human-made inventions” 13 . 

Thus, the natural products doctrine seems 

to be based on a legal house of cards. Decisions 

in the US courts are best rationalized based 

on how judges perceive the invention rather 

than any available legal theory. The discourse 

between Dan Ravicher representing Association 

for Molecular Pathology and Chief Judge Rader 

at a conference at the Fordham Law School in 

April succinctly captures nearly 150 years of 

legal precedence on the issue. Ravicher (pointing 

to a bottle of water), asked “was that [purification] 

sufficient intervention between what 

God gave us...and what man created to merit a 

patent?” Chief Judge Rader responded, “how 

many people have died of water pollution over 

the course of human events? Probably billions.” 

Ravicher is narrowly focusing on the structural 

differences between water and purified 

water without regard to its new utility (that 

is, potability); Judge Rader is less concerned 

with the magnitude of the structural differences 

and instead focuses on the impact of its 

new utility (that is, saving lives) 14 . Similarly, 

looking at the structural differences between a 

gene in one’s body and a patented isolated gene 

has led many to conclude that gene patents are 

not deserving of protection. This is the logic 

Judge Sweet employed in his reasoning in the 

Myriad Genetics (Salt Lake City, Utah, USA) 

case, which emphasized that isolated and complementary 

DNA is not “markedly different” 

from native DNA 15 . 

On the other hand, those focused on the 

novel medical uses of an isolated gene to diagnose 

disease come to a different conclusion. 

This is consistent with the Federal Circuit’s reasoning 

in Prometheus Laboratories, Inc. v. Mayo 

Collaborative Services in which the court stated 

that “methods of treatment … are always” eligible 

for patent protection 16 . Regardless of the 

ideological view held, it seems clear that we 

need a more objective tool for making decisions 

on the eligibility of inventions. That tool 

should be closely aligned with an objective we 

can all embrace—encouraging scientific innovation 

and bringing new technology to market 

that improves living conditions around the 

world now and for future generations. 

Do we even need a natural products 

doctrine? 

The distinction between patent eligibility and 

patentability is often blurred, yet it remains an 

important distinction. An invention is patentable 

if it is useful, novel and not obvious to 

one in the field provided that other sections 

of the Patent Act are met (e.g., it is adequately 

described and enabled). An invention that is 

declared ineligible cannot be patented, even 

if it would have met all of the patentability 

requirements of the Patent Act. As discussed in 

greater detail below, this distinction is critical 

to building a sound legal argument on which 

to rest a decision on the eligibility of gene and 

diagnostic patents. 

The natural products doctrine for patent 

ineligibility is a judicially created doctrine 

without any direct support in the Patent Act 

or US Constitution. The doctrine was created 

to prevent patents being given on fundamental 

discoveries that might stymie innovation. 

It is not entirely clear, however, whether this 

doctrine is a necessary legal tool to reign in 

patents whose breadth threatens innovation. 

The examples used by the courts to justify the 

natural products doctrine are not convincing, 

especially when one considers the less controversial 

patentability requirements of the 

Patent Act that can adequately address similar 

concerns. Take, for example, the Supreme 

Court’s often quoted statement that “Einstein 

could not patent his celebrated law that E = mc 2 ; 

nor could Newton have patented the law of 

gravity” 17 . Rather than arguing that gravity is 

not eligible for patent protection because it is 

a “law of nature,” one can persuasively argue 

instead that gravity was “known or used by 

others…before the invention thereof by the 

applicant for patent” and, therefore, is not novel 

under section 102 of the Patent Act 18 . Put differently, 

Newton could not have patented gravity 

because it would have restricted public use 

of something the public was enjoying before 

his discovery—an explanation for how something 

already in the public domain works has 

never been per se patentable (for a review of 

law of anticipation, see ref. 19). Discovering the 

mechanism of action of aspirin is not patentable 

if the public has, albeit unknowingly, been 

enjoying those benefits. 

Similarly, Samuel Morris’s attempt to broadly 

patent all forms of communication that used 

“electomagnetism” has been used to support 

the idea that laws of nature are not patent eligible. 

The court, however, also engaged in a 

lengthy discussion that the claim was invalid 

because it did not teach all methods of communicating 

by means of electromagnetism, only 

now by the telegraph 20 . The court’s reasoning 

is now codified as § 112 of the Patent Act and 

forms a sound basis for invalidity in instances 

when the patentee attempts to capture more 

intellectual estate than taught in the patent. 

Even in the Myriad case, Christopher 

Holman and Robert Cook-Deegan have 

pointed to vulnerabilities in the patent claims 

and suggest that “other doctrines of patentability 

can limit gene-based patent claims to an 

appropriate scope, rather than using the patent 

eligibility doctrine that would unnecessarily 

invalidate all DNA-based patents indiscriminately 

and regardless of their merit” 21 . 

By using other legal grounds, such as novelty 

and enablement, the court would obviate 

the need to clearly and predictably define what 

constitutes laws of nature, natural phenomena, 

and abstract ideas—a task it has struggled with 

unsuccessfully for over a century. It would also 

allow courts to make a decision on a case-bycase 

basis rather than invalidate an entire class 

of patents and risk hobbling the diagnostic 

industry by inadvertently eliminating the 

incentive to invent and invest. 

A new proposal for assessing patent 

eligibility 

Judicial reliance on §101 of the Patent Act to 

justify patent ineligibility is misplaced and 

serves only to confound the issue. Section 8 of 

Article 1 in the Constitution gives Congress 

plenary power to define what is and is not 


COMMENTARY 


patentable subject matter under one fundamental 

condition—it must “promote the 

Progress of Science and the Useful Arts” 22 . 

Therefore, a more sound legal justification is 

that an invention is not eligible for patent protection 

if it attempts to patent subject matter 

so fundamental that granting even a limited 

monopoly would impede scientific progress 23 . 

If so, courts could then logically hold that the 

patenting of fundamental products of nature 

are unconstitutional for failing to “promote the 

Progress of Science and useful Arts” even if the 

Patent Act does not specifically proscribe this 

class of invention or discovery 2 . 

Focusing on the constitutional underpinning 

for patent eligibility is not merely an exercise in 

legal gymnastics. Pushing past the arguments 

stemming from the legal precedent focused on 

§101 toward agreement on why certain inventions 

should not be eligible for patent protection 

may pave the way for a balanced and reasoned 

approach to determine when an invention or 

discovery is or is not promoting innovation. 

The current doctrine could then be viewed 

as one of several factors in considering patent 

eligibility. Just as a physician would never rely 

on one abnormal cholesterol score to determine 

whether to conduct open-heart surgery, 

courts should not rely solely on this doctrine to 

invalidate countless biotech patents. The court 

should incorporate into the analysis additional 

factors that can better define the breadth of an 

invention and its threat to our constitutional 

mandate to promote innovation. 

Using a balanced, multipronged approach 

to defining the scope of intellectual property 

rights is not new. In copyright law, a 

four-pronged approach is used to determine 

whether a reproduction is made for the purposes 

of “criticism, comment, news reporting, 

teaching … scholarship, or research,” and thus, 

exempt from infringement under the fair use 

doctrine. The statute specifically states: “In 

determining whether the use made of a work… 

is a fair use the factors to be considered shall 

include—(1) the purpose and character of the 

use, including whether such use is of a commercial 

nature or is for nonprofit educational purposes; 

(2) the nature of the copyrighted work; 

(3) the amount and substantiality of the portion 

used in relation to the copyrighted work as 

a whole; and (4) the effect of the use upon the 

potential market for or value of the copyrighted 

work.” What is appealing about these factors 

is how it balances the copyright holders’ economic 

interests with First Amendment rights 

to free speech and basic research. Over time, 

the court has applied these factors and brought 

clarity and predictability to the field. 

I propose that the US Congress and courts 

should take a similar approach to determining 

patent eligibility using criteria that refocus the 

determination on factual considerations that 

bear directly on the objective of promoting 

innovation. In developing this framework, I 

have reviewed three areas: first, Supreme Court 

precedence, which clarified the factors and 

rationales that guided the Court when making 

patent eligibility determinations; second, 

factual circumstances, case studies and market 

forces relevant to determining the risks of 

overprotection, including the specific contribution 

of the patent to the overall concerns; 

and finally, the roles of existing laws that may 

limit the abuse by patent holders. This review 

allowed the formulation of five criteria that I 

believe would be useful for courts, legislatures 

and business when confronted with a question 

of subject matter patentability. 

1. Does the patented invention cover a 

product or law of nature unaltered by 

humans? 

2. Do other limitations in the claims and 

prosecution history confer meaningful 

limitations on the scope of the patented 

claims? 

3. Are there reasonable approaches to 

design around or circumvent the patented 

invention? 

4. Is there evidence that basic research is 

being hindered, and if so, is the patent 

solely or substantially responsible? 

5. And are there other laws in place that 

would mitigate the risk of overprotection? 

Taken together, these criteria incorporate 

many of the principles previously adhered to 

by the courts. They also allow courts to build 

a solid evidentiary foundation from which to 

decide the scope of an invention and its eligibility 

for patent protection. Importantly, the 

criteria may provide greater predictability by 

creating a legal framework for patent eligibility 

with one unequivocal and constitutionally consistent 

objective—promoting innovation. I discuss 

each of the criteria in more detail below. 

Does the patented invention cover a product 

or law of nature unaltered by humans? This 

first criterion largely preserves the US courts’ 

previously announced laws and products of 

nature doctrine. The US Supreme Court has 

stated that “anything under the sun made by 

man was patentable” 23 . Therefore, as the US 

Department of Justice and others agree, all 

human-made inventions are per se patent eligible. 

In cases where there is a question whether 

the invention is human-made as with isolated 

DNA, or the discovery employs a correlation 

or algorithm as with the diagnostic claims, 

additional factors should be considered when 

determining patent eligibility of the invention 

as a whole. 

For more information on the natural products 

doctrine, readers are referred to relevant 

reviews in the literature 17 . 

Do other limitations in the claims and prosecution 

history confer meaningful limitations 

on the scope of the patented claims? 

This second criterion considers the impact of 

claim limitations on the scope of the exclusivity 

sought and the risk of overprotection. After 

all, an invention is nothing more than the use 

of natural laws and products in an application 

not found in nature. A mixture of naturally 

occuring elements that make a new alloy with 

unique properties is patent eligible. Likewise, 

inventions do not become ineligible for patent 

protection because they employ a natural law, 

even if the law in central to its utility. A novel 

airplane wing design that improves fuel efficiency 

is patent eligible, even if central to its 

operation is the Bernoulli principle that creates 

lift and permits flight. In both instances, others 

are free to use the individual elements and 

the Bernoulli principle in applications outside 

those claimed. What is not eligible, therefore, 

is the individual elements and the Bernoulli 

principle per se because exclusivity to those 

products and laws without limits would grant 

an unreasonable broad monopoly. 

Indeed, the Supreme Court has recognized 

that “an application of a law of nature…may 

well be deserving of patent protection” 24 , 

and at times has even used that reasoning to 

uphold patent eligibility. In Diamond v. Diehr, 

the Supreme Court held that although the 

Arrhenius equation alone was not patent eligible, 

the scope of a claim for curing rubber 

that used the equation was eligible because it 

didn’t foreclose the use of the equation in other 

processes 25 . 

Likewise, in gene and diagnostic patents 

the court should look at limitations, including 

the following: the gene covered, the mutations 

recited, the diseases it claims to diagnose and 

the tissue source. The Myriad Genetics patents 

do not seek to exclude all genetic correlations, 

of all genes, to diagnose all diseases—they 

claim two genes for diagnosing breast and ovarian 

cancer. As such, these limitations should 

be considered in a patentability determination, 

especially if the justification for the doctrine is 

to reign in monopolies that unreasonably preempt 

future progress. 

Are there reasonable approaches to design 

around or circumvent the patented invention? 

This third criterion considers the ability 

to design around the invention. As Justice 


COMMENTARY 


Breyer notes in his decent to the revocation of a 

grant of certiori in Laboratory Corp. of America 

Holdings v. Metabolite Labs., Inc., “[p]atent law 

[also]…seeks to avoid the diminished incentive 

to invent that underprotection can threaten” 26 . 

If patents do not place meaningful barriers to 

ward off imitators the entire constitutional 

purpose of promoting the progress of science 

would be subverted. 

It is important, therefore, for the court to 

consider alternatives that exist, or could be 

developed, which would reasonably circumvent 

the patent, avoid underprotection and 

promote the progress of science. This factor 

is particularly useful when applied to other 

inventions the courts have previously held 

ineligible. It is not possible to design around a 

truly fundamental product or principle such as 

a new element, Einstein’s law of relativity and 

Newton’s law of universal gravitation 27 . 

History instructs us that this inquiry should 

not be narrowly limited to the precise technology 

claimed. In the music industry, for example, 

the biggest threats have not come from improving 

the vinyl records but from the evolution 

of disruptive technologies, such as the cassette 

player, CDs, DVDs and digital recordings that 

were developed over a 20-year span. 

In gene and diagnostic patents the inquiry 

is too narrowly focused on circumventing 

the precise invention. In a genetic test where, 

for example, mutations or polymorphisms 

are a proxy for deficient protein activity, it is 

entirely possible that another assay could be 

developed—a protein, metabolite or functional 

assay—that would better predict a predisposition 

to the disease. In the Myriad case, 

a functional assay for BRCA1 and BRCA2 proteins 

would obviate the need to examine the 

gene sequences and circumvent the patents. 

Evidence of a potential design-around, therefore, 

would weigh heavily against finding the 

claim overly broad and unconstitutional. 

Is there evidence that basic research is being 

hindered, and if so, is the patent solely or 

substantially responsible? This fourth criterion 

considers whether the patent in question 

is solely or substantially responsible for the 

alleged industry concerns. As a recent Nature 

Biotechnology article by Carbone et al. 28 suggests, 

“the potential obstacles to innovation 

that patents cause in diagnostics may not be as 

high…as some had feared.” The authors’ policy 

changes in licensing practices would solve many 

of the problems that patent ineligibility is meant 

to address. This approach is supported by a case 

study by Fore et al. 29 on the polymerase chain 

reaction (PCR) technology. At the time PCR 

was patented, a strong argument could have 

been made that it was a product of nature and 

patenting its use would foreclose basic research 

and stymie innovation—after all, the enzyme 

in a test tube merely does what it naturally 

evolved to do. However, the authors found 

that despite “heavy patent protection” PCR 

was “disseminate[d] broadly in the molecular 

biology world, becoming an indispensable 

research tool employed in nearly every biological 

field” 29 . This was due to deliberate licensing 

and business practices on the part of the patent 

holders, including ‘rational forbearance’ where 

companies refrain from suing researchers for 

patent infringement 30 . 

The Carbone et al. 28 thesis, that licensing 

practices will resolve most of the issues, should 

be tempered for several reasons. First, licensing 

practices may increase the number of labs that 

offer the diagnostic test but it does not guarantee 

lower pricing, reimbursement and broader 

accessibility. As another paper 31 by several of 

the authors in Carbone et al. reported, the per 

unit price of the exclusively licensed BRCA1/ 

BRCA2 diagnostic is lower than a non-exclusively 

licensed colorectal cancer test. There are, 

therefore, other market forces outside patents 

that conspire to sustain higher prices and limit 

accessibility. 

Second, an issue not often discussed is the 

impact of regulatory hurdles. Carbone et al. 28 

suggest that unlike the diagnostic industry, 

exclusive licenses to therapeutic proteins, 

such as erythropoietin, growth hormone and 

interferon, are “commercially significant” and 

justified. Although true, the reason exclusive 

licensing of therapeutics is commercially significant 

(and why erythropoietin at the time 

was likely not challenged as a natural product) 

is the requirement for US Food and Drug 

Administration (FDA; Silver Spring, MD) 

approval. The lack of FDA oversight over many 

diagnostics not only lowers the barriers to entry 

but also substantially reduces the development 

costs. If the FDA began regulating all diagnostics 

as it currently does for therapeutics, this 

debate would likely become mute as most of 

the laboratories and associations opposing 

gene and diagnostic patents would be unable 

(or unwilling) to develop FDA-approved diagnostic 

tests, even in the absence of patents and 

exclusive license arrangements. Most hospital 

and university-based labs draw a bright line 

between developing tests that can be offered 

by the Commercial Laboratory Improvement 

Act (CLIA) and FDA 510k or premarketing 

approval (PMA). Given the FDA’s increased 

scrutiny over these so-called laboratory developed 

tests, we should not ignore that a change 

in the regulatory winds may fundamentally 

alter market forces and take the wind out of 

the sails of the most vocal opponents to gene 

and diagnostic patents. 

And finally, most analysis starts with the 

assumption that there is a robust reimbursement 

policy in place for diagnostics. This gives the 

false impression of lower barriers to entry, especially 

for academic and hospital labs. A role of 

the exclusive licensee in a disruptive technology 

is to create a market and pioneer reimbursement. 

Establishing a commercially viable 

market for the product is a costly undertaking 

that all subsequent providers benefit from at a 

fraction of the investment. It is worth considering 

what the availability of the BRCA1/BRCA2 

diagnostic would have been in the absence of 

Myriad Genetics having pioneered the market 

and established a reimbursement policy. 

To ensure that patents are not unjustly 

blamed for the ills of the diagnostic industry, 

this criterion considers other factors that may 

impact dissemination and access. Overreacting 

to the public outcry concerning gene patents 

without considering additional factors may 

fundamentally cause a change in patent law 

without resolving the present concerns of the 

industry. Moreover, even assuming patents currently 

do not play a significant role in the diagnostic 

industry, market and regulatory forces 

may shift in the future necessitating once again 

strong incentives to invent and invest to bring 

diagnostics to market. 

Are there other laws in place that would mitigate 

the risk of over-protection? This final 

criterion takes into consideration other mechanisms 

that substantially mitigate the risk of 

over-protection and abuse by patent holders. For 

example, under the Bahy-Dole Act the government 

retains a “nontransferrable, irrevocable, 

paid-up license…to practice the invention… 

throughout the world” 32 . The Act further 

specifies that the government may, under certain 

conditions, exercise “March-in-Rights” to 

make the invention available to the public. In 

federally funded research, therefore, the courts 

should consider the government’s rights to reign 

in any potential abuse by a patent holder. 

Other laws could be revisited or created to 

address other concerns. There is, for example, a 

reasonable public policy argument that an individual 

should have a right to obtain a second 

opinion before undergoing a substantial medical 

procedure. If so, the courts could examine 

an extension of the doctrine of patent exhaustion. 

Under the doctrine (also referred to as 

the first sale doctrine), the first unrestricted 

sale of a patented item or method exhausts 

the patentee’s control over that particular item 

or method. This compromise would uphold 

the patentee’s rights and economic incentives 

by retaining exclusivity over the first sale, yet 

allow second opinions to be rendered without 

the threat of infringement. Similarly, concerns 


COMMENTARY 


that a physician’s fear of infringement may chill 

the practice of medicine could be addressed 

by US Congress as it has by excluding medical 

practitioners from liability for infringing a 

patent by performing of a medical or surgical 

procedure on a body 33 . 

Conclusions 

In the ongoing debate, both sides are talking 

over each other in part because there is a lack 

of well-articulated criteria for what constitutes 

a law of nature, a natural phenomena and an 

abstract idea, as well as the nexus between the 

doctrine and its putative objective of promoting 

innovation. Focusing the analysis on the 

constitutional objective that inventions and discoveries 

must promote the progress of science 

provides the clearest framework for determining 

what inventions should be eligible for patent 

protection. The proposal made for using a 

balanced, multipronged approach that emphasizes 

a factual determination directly bearing 

on this constitutional mandate is intended as 

the first step on the road to a workable compromise 

and a predictable legal framework. 

By mirroring the approach to determining 

copyright fair use, this proposal allows courts 

to thoughtfully assess patent eligibility on a 

case-by-case basis and build, over time, a predictable 

legal framework for different classes of 

inventions and discoveries. 

Without clear evidence that DNA patents 

are the sole or significant factor in disrupting 

basic research and dissemination, US courts 

should be loath to invalidate an entire class 

of patents. One misstep may have unforeseeable 

negative consequences on the future of 

personalized medicine and cripple the biotech 

industry. Moreover, a flexible approach 

to patent eligibility is more likely to withstand 

Supreme Court scrutiny, which has repeatedly 

emphasized its dislike for rigid tests previously 

proposed by the Federal Circuit. Although 

there will no doubt be a healthy debate as to 

where to draw the line between overprotection 

and underprotection, both the scientific 

and commercial communities should recognize 

that neither can advance if a single 

party owns exclusive and absolute rights to 

an invention involving a truly fundamental 

law or product of nature. 

ACKNOWLEDGMENTS 

The author is grateful to J. Mixco for his contribution 

and assistance. The views expressed are solely those of 

the author. 



1. No. 09 Civ. 4515 (RWS), 2010 U.S. Dist. LEXIS 35418, 

at *2 (S.D.N.Y. Mar. 29, 2010, revised Apr. 2, 2010). 

2. U.S. Const. art. I, § 8, cl. 8. 

3. Diamond v. Diehr, 450 U.S. 175 (1981). 

4. See Association for Molecular Pathology, supra note 2. 

5. See Prometheus Laboratories, Inc. v. Mayo Collaborative 

Services, 581 F.3d 1336 (Fed. Cir. 2009). 

6. Transcript of Oral Argument at 20, Bilski v Kappos, 130 

S. Ct. 3218 (2010) (No. 08–964). 

7. Bilski v. Kappos, 130 S. Ct. 3218 (2010). 

8. Ibid. 

9. Brief For The United States as Amicus Curiae in 

AMP v. USPTO (No. 2010–1406) at 20. (emphasis 

in original) 

10. Funk Brothers Seed Co. v. Kalo Inoculant Co., 333 U.S. 

127 (1948). 

11. Parker v. Flook, 437 U.S. 584 (1978). 

12. “Whoever invents or discovers any new and useful process, 

machine, manufacture, or composition of matter, 

or any new and useful improvement thereof, may 

obtain a patent therefore, subject to the conditions and 

requirements of this title.” 35 U.S.C. §101. 

13. Supra note 15 at 12. 

14. Motion by Plaintiffs-Appellees for Recusal of Chief 

Judge Randall R. Rader, AMP v. USPTO, No. 09-CV- 

4515 (Fed. Cir. June 29, 2010). 

15. See Association for Molecular Pathology, supra note 

2 at 114. 

16. Prometheus Labs, Inc v. Mayo Collaborative Services, 

581 F.3d 1336, 1346 (Fed. Cir. 2009) 

17. Diamond v. Chakrabarty, 447 U.S. 303 (1980). 

18. 35 U.S.C. §102(a) (2010). 

19. Dan, L. Burk & Mark A. Lemley, Inherency, 47. William 

Mary Law Rev., 371 (2005). 

20. O’Reilly v. Morse, 56 U.S. 62, 117–18 (1853). 

21. Brief of Amici Curiae Christopher M. Holman and 

Robert Cook-Deegan in Support of Neither Party in 

AMP v. USPTO (No. 2010-1406) at 24. 

22. “...the reason for exclusion [of laws and products of 

nature] is that sometimes too much patent protection 

can impede rather than ‘promote the Progress of 

Science and useful Arts,’ the constitutional objective 

of patent... protection.” Laboratory Corp. of America 

Holdings v. Metabolite Labs., Inc., cert. dismissed as 

improvidently granted, 548 U.S. 124, 126–27 (2006) 

(Breyer, J., dissenting). 

23. Chakrabarty, supra note 17. 

24. Diamond v. Diehr, supra note 3. 

25. Ibid. 

26. Laboratory Corp. supra note 22. 

27. Chakrabarty, supra note 17 at 309. 


(2010). 

29. Fore, J. Jr. et al. J. Biomed. Discov. Collab. 1, 7 

(2006). 

30. Ibid. 

31. Cook-Deegan, R. et al. Genet. Med. 12 Suppl, S15–S38 

(2010). 

32. 35 U.S.C. §§200–12 (2010). 

33. 35 U.S.C. §287(c) (2010). 


commentary 

The environmental impact subterfuge 

Gregory Conko & Henry I Miller 

Action needs to be taken to prevent anti-biotech activists from co-opting environmental law to derail the planting of 

transgenic crops that have already received regulatory approval. 


The latest weapon used in the misinformation 

war against recombinant DNA technology 

and its agricultural applications is an obscure 

environmental law from the seventies. Green 

activists and organic farmers are exploiting 

the National Environmental Policy Act of 1970 

(NEPA) to convince courts that inconsequential 

paperwork oversights by regulators at the 

US Department of Agriculture warrant the revocation 

of two final approvals for recombinant 

DNA–modified crop varieties and of the issuance 

of permits to test several others. At least 

one more case is pending. 

Under NEPA, all US federal government 

agencies are required to consider the effects 

that any “major actions” they take may have on 

the “human environment.” Agencies can exempt 

whole categories of routine or repetitive activities 

but most other decisions—such as the issuance 

of a new regulation, the location of a new bridge 

or the approval of a new agricultural technology—trigger 

the NEPA obligation to evaluate 

environmental impacts. If the agency concludes 

that the action will have “no significant impact” 

(a legal term of art), it issues a relatively brief 

Environmental Assessment explaining the basis 

for that decision. If significant effects are likely, 

though, the agency must prepare a comprehensive 

Environmental Impact Statement (EIS), 

which typically requires thousands of hours of 

work, details every imaginable effect and runs to 

hundreds (or even thousands) of pages. 

The obligation under NEPA is wholly procedural, 

which means that even significant environmental 

effects do not prohibit the agency 

Gregory Conko is a senior fellow at the 

Competitive Enterprise Institute in Washington, 

DC, and Henry I. Miller, a physician and fellow 

at Stanford University’s Hoover Institution, 

was the founding director of the FDA’s Office of 

Biotechnology. 

e-mail: henry.miller@stanford.edu 

In August, a federal judge revoked the USDA’s approval of Roundup Ready sugar beets, which represent 

95% of the crop now grown in the United States. 

from ultimately taking the proposed action. 

Its purpose is solely to force government agencies 

to in fact consider possible environmental 

effects. However, courts have interpreted the law 

broadly by requiring a comprehensive review of 

every imaginable effect on the “human environment.” 

This category now encompasses not only 

harm to the natural ecology but also economic, 

social and even aesthetic impacts. 

Thus, if agencies miss some tangential or 

speculative issue, they can be tripped up by an 

irresponsible litigant who alleges that the environmental 

review was incomplete. Even when 

regulators actually do consider a potential 

impact but reject the concern owing to its unimportance 

or improbability, they can run afoul of 

NEPA by failing to extensively and comprehensively 

document their reasoning. This latter phenomenon 

has lately plagued USDA approvals of 

recombinant DNA–modified crops. 

NEPA and biotech 

The USDA has approved 74 different recombinant 

DNA–modified crop varieties, or 

transformation events, for commercial-scale 

cultivation, including varieties of corn, canola, 

squash, soybean, potato, tomato and other 

species. In each case, the department’s Animal 

and Plant Health Inspection Service (APHIS) 

reviewed copious amounts of data from several 

years’ worth of controlled field experiments 

to evaluate the variety’s agronomic and 

environmental effects. And because each one 

of these plants is highly similar to conventionally 

bred varieties already grown throughout 

the United States—differing only in the addition 

of a small number of well characterized 

genes that introduce useful traits—APHIS 

concluded that they would have no significant 

environmental impact. 

But some anti-biotech activists are unwilling 

to permit facts or the judgments of experts to 

get in the way of their antagonism, so a group 

of environmental organizations led by the antibiotech 

Center for Food Safety joined with a 

handful of organic farmers to sue in federal 

court to halt the planting of alfalfa, sugar beets 

and turf grass modified for resistance to the 

David R. Frazier/DanitaDelimont.com/Newscom 


COMMENTARY 


herbicide glyphosate (better known by its 

trade name Roundup (Monsanto Company; 

St. Louis, MO)). The plaintiffs made no substantive 

arguments that these crops were 

unsafe, but instead claimed that APHIS’s 

Environmental Assessment was procedurally 

insufficient and that a comprehensive EIS 

should have been prepared. 

What were the supposedly significant environmental 

impacts that APHIS allegedly failed 

to account for in its environmental analysis? 

The plaintiffs made two claims: first, that crosspollination 

by recombinant DNA–modified 

crops would ‘contaminate’ non-genetically engineered 

alfalfa and produce negative economic 

and socioeconomic effects, such as the loss of 

certification for organic crops and an inability 

to produce non-genetically engineered seed; 

and second, that the widespread adoption of 

Roundup Ready crops and increased use of 

glyphosate could accelerate the development of 

glyphosate-resistant weeds and force farmers to 

switch to other, less convenient herbicides. Both 

claims are dubious. 

Cross-pollination by recombinant DNA– 

modified plants does not, in fact, jeopardize the 

organic certification of those crops because the 

USDA’s rules for organic production are based on 

process, not outcome. As long as organic growers 

adhere to permissible practices, take reasonable 

steps to prevent unintentional contact with transgenes 

and do not intentionally plant recombinant 

DNA–modified seeds, unintentional crosspollination 

does not cause those crops to lose 

their organic status. Furthermore, organic rules 

require growers to create distinct buffer zones 

to prevent unintended contact with prohibited 

substances, such as pollen from recombinant 

DNA–modified plants and synthetic pesticides. 

Seed breeders are also accustomed to using isolation 

distances as well as physical and biological 

buffer zones to maintain genetic purity, and the 

net effect of approving these recombinant plants 

would be, at worst, minimal. 

Similarly, the development of herbicideresistant 

weeds is not unique to recombinant 

DNA–modified varieties, but occurs commonly 

in all crops exposed to herbicides. APHIS is well 

aware of the phenomenon and considers it routinely 

when making approval decisions. The 

agency made clear to the court that good stewardship 

is the only reasonable defense against 

the development of herbicide-resistant weeds, 

but the court criticized APHIS for not evaluating 

the cumulative extra impact of approving new 

Roundup Ready varieties. 

From a NEPA perspective, what makes recombinant 

DNA–modified herbicide-resistant plants 

fair game is that they are regulated by a federal 

government agency and that approval to commercialize 

them constitutes, in NEPA-speak, a 

“major action.” Even though plant scientists are 

virtually unanimous that recombinant DNA is 

an extension, or refinement, of earlier techniques 

and that it is safer and more predictable than 

many other conventional breeding methods, 

such as wide-cross hybridization and mutation 

breeding, only recombinant DNA–modified 

varieties are subject to the kind of government 

approval that triggers the EIS obligation. 

It is ironic that recombinant DNA–modified 

crop varieties, whose development and environmental 

impacts are carefully scrutinized by 

the USDA, are subject to the extra burden of a 

compelled EIS, whereas varieties developed with 

cruder conventional methods are subject to neither 

premarket assessment nor the NEPA rules. 

Indeed, the EIS requirement for recombinant 

DNA–modified crop approvals could have been 

avoided if the USDA and other federal regulatory 

agencies had heeded the advice of the scientific 

community some 25 years ago and chosen not to 

subject the products of recombinant DNA technology 

to special, discriminatory government 

regulation. Had the regulators instead followed 

the scientific consensus, they would not have 

been tripped up by NEPA’s paperwork requirement 

because, in the absence of an approval 

process, there would be no “major actions” to 

trigger the EIS obligation. 

Environmental creep 

Since its enactment, the scope of NEPA’s coverage 

has experienced continual creep. Through 

a combination of prodding by environmental 

activists and judicial overreach, the meaning 

of the term “human environment” has been 

expanded to include not just tangible ecological 

harms that affect people and human communities 

but also impacts that are economic, social, 

cultural, historic and aesthetic. A decision by a 

federal district court in Minnesota illustrates 

how liberally the statute has been interpreted: 

“Relevant as well is whether the project 

will affect the local crime rate, present fire 

dangers, or otherwise unduly tap police 

and fire forces in the community…the project’s 

impact on social services, such as the 

availability of schools, hospitals, businesses, 

commuter facilities, and parking…harmonization 

with proximate land uses, and a blending 

with the aesthetics of the area…[and a] 

consideration of the project’s impact on the 

community’s development policy…” 

In other words, almost any possible effect 

that a clever plaintiff can imagine may constitute 

a significant impact under NEPA if a 

sympathetic judge agrees. 

In the first ever NEPA challenge to a product 

made with recombinant DNA technology, 

environmental activist Jeremy Rifkin successfully 

overturned a 1984 decision by the US 

National Institutes of Health (NIH) to permit 

the field testing of recombinant DNA–modified 

Pseudomonas syringae bacteria, engineered to 

help protect crop plants from frost damage. 

Researchers at the University of California, 

Berkeley, discovered that P. syringae contains 

an ‘ice nucleation’ protein that helps initiate 

the growth of ice crystals that damage growing 

crops. These scientists deleted the gene sequence 

coding for the ice-nucleation protein in the hope 

that spraying the resulting ‘ice minus’ variants 

on plants could inhibit ice crystal formation and 

thereby reduce frost damage. 

Rifkin’s Foundation on Economic Trends, 

Bethesda, sued to stop the tests and argued that 

the NIH had not sufficiently considered, among 

other things, the possibility that spraying the 

modified bacteria might disrupt wind circulation 

patterns and affect aircraft flying overhead. 

Not surprisingly, the NIH had dismissed such a 

ridiculous theory out of hand because P. syringae 

bacteria with a missing or non-functioning 

ice-nucleation gene were known to arise spontaneously 

in nature owing to natural mutations, 

and the permit was for a single small-scale field 

trial. Astonishingly, however, both a federal district 

court and federal appeals court allowed the 

challenge and overturned the NIH decision. It 

is little wonder, then, that federal courts would 

recognize the possibility of agronomic impacts 

as the basis to challenge the approval of recombinant 

DNA–modified crop plants. 

NEPA’s effect on farmers 

Although unsurprising, the resolution of these 

lawsuits has been a nightmare for US plant 

breeders and farmers. In one case, the plaintiffs 

challenged APHIS’s decision to permit 

the Scotts, Marysville, Ohio, garden products 

company to conduct field trials of glyphosateresistant 

turf grass. Another case involved APHIS 

field trial permits to four different seed companies 

to cultivate small, geographically isolated 

plots of corn and sugarcane genetically modified 

to produce proteins that would be used in medical 

products. The cases of glyphosate-resistant 

alfalfa and sugar beet were particularly vexing 

to farmers, however, because the lawsuits were 

filed after the USDA had approved the varieties 

for commercial release and farmers had already 

begun to cultivate the seeds. 

The alfalfa case involves the 2004 approval 

of a glyphosate-resistant variety sold under 

the trade name Roundup Ready by its codevelopers 

Monsanto and Forage Genetics 

(Nampa, ID). To secure approval, the firms 

conducted almost 300 government-monitored 

field trials over a period of eight years. 

APHIS scientists evaluated the data generated 


COMMENTARY 


from these trials to determine the likely environmental 

impacts of an approval—called ‘deregulation’ 

in agency parlance—and of widespread 

use of Roundup Ready alfalfa by US farmers. 

Because the same genetic trait had already 

been incorporated into dozens of approved varieties 

of corn, soybeans, canola and other crop 

plants grown in the United States since 1996, 

APHIS was quite familiar with the glyphosateresistance 

trait’s likely effects—not just from the 

field tests needed to secure approval, but also 

from nearly a decade of real-world experience. 

Regulators naturally concluded that deregulating 

the crop would not have any significant environmental 

impact because there is no evidence that 

the glyphosate-resistance gene harms humans 

or other animals and it was already in common 

use in US agriculture. In turn, the USDA 

prepared a shorter Environmental Assessment 

explaining its “finding of no significant impact” 

and deregulated Roundup Ready alfalfa. 

Roughly 5,500 farmers across the United 

States had planted more than a quarter million 

acres of Roundup Ready alfalfa by 2007 when 

a federal district judge in San Francisco ruled 

that the USDA’s Environmental Assessment 

was legally insufficient. The court issued an 

injunction revoking the approval and prohibiting 

new seeds from being sold until 

the USDA completed a full EIS to evaluate 

concerns about glyphosate-resistant weeds 

and the possible impacts on organic farmers. 

Biotech advocates hailed the decision in 

June of the US Supreme Court reversing the 

blanket injunction on sales of new Roundup 

Ready alfalfa seed, but the actually decision 

was a purely procedural one that reinforces the 

USDA’s obligation to complete an EIS before 

permitting more seed sales. 

It took two years, but by November of 2009 

APHIS had published a draft EIS and solicited 

public comments. Unsurprisingly, the EIS concluded 

what agency scientists and the scientific 

community already knew: the potentially 

negative impacts of approving Roundup Ready 

alfalfa are minimal and manageable, and they 

are in any event far outweighed by the crop’s 

many substantial benefits. The USDA must 

now publish a final version of the EIS and 

reapprove the crop; however, although the substantive 

work is complete, the administrative 

process of moving to final reapproval could 

take many more months and probably will not 

be finalized in time for new Roundup Ready 

alfalfa seed to be planted in 2011. Fortunately, 

the farmers who planted Roundup Ready 

alfalfa when it was first approved were permitted 

to continue growing and then harvest that 

initial crop, so the overall impact will be limited. 

A far worse fate is in store for America’s 

sugar beet growers. 

In August, another federal district judge 

revoked the USDA’s approval of Roundup Ready 

sugar beets. That decision will sow monumental 

confusion because an estimated 95% of the 

sugar beets currently grown in the United States 

are of the Roundup Ready variety. As was the 

case with Roundup Ready alfalfa, the court’s 

injunction will permit the already planted crops 

to be harvested and processed, but before the 

USDA may re-approve the variety, a full EIS will 

be necessary. The process will take years, leaving 

farmers in regulatory limbo in the meantime. 

Perhaps the biggest short-term problem arising 

from the decision, however, is a looming shortage 

of conventional sugar beet seeds for next 

spring’s planting. 

Because the Roundup Ready variety so quickly 

became popular with American beet growers, 

seed companies cut back on their production of 

non-genetically engineered seed. According to 

Duane Grant, chairman of the Snake River Sugar 

(Boise, ID, USA), which produces about 20% of 

the nation’s beet sugar, many growers fear they 

will have nothing to plant in 2011. Consumers 

will surely feel the pinch of sharply rising prices, 

and farmers and others who care deeply about 

protecting the environment will be denied the 

proven beneficial effects of these crops until they 

are restored to the marketplace. 

Time for NEPA reform 

Remarkably, because the NEPA obligation is 

purely procedural, US courts are not permitted 

to consider the fact that recombinant DNA– 

modified herbicide-resistant crop varieties have 

offsetting benefits to farmers, consumers and 

the natural environment. The mere fact that the 

USDA did not properly document its evaluation 

of potential negative effects is sufficient grounds 

for revoking the approval. 

The only function of NEPA is to ensure 

that agencies do in fact consider whether their 

actions may harm the environment. This is a 

laudable goal, but it is manifestly not the intention 

of most NEPA litigation. Instead, the statute 

has been hijacked by environmental activists to 

slow down or prevent government agencies from 

taking actions the activists do not like. Because 

the law requires agencies to consider almost 

any conceivable impact, the statute offers fertile 

ground for bad-faith, obstructionist litigation no 

matter how meticulous the agency is in preparing 

an Environmental Assessment or EIS. 

NEPA is therefore a recipe for stagnation, a 

particular problem for ‘gatekeeper’ regulatory 

agencies that must grant approvals before a 

product can be tested or commercialized. 

Something must be done to change the system. 

But what? Short of substantive reform 

of the underlying statute by Congress—the 

preferable and definitive solution—agencies 

themselves can take some minor steps to 

mitigate the Act’s worst effects. 

Under the NEPA statute, every agency may 

establish a set of ‘categorical exclusions’ that 

exempt whole classes or types of activities from 

the EIS obligation. These may include routine or 

repetitive actions that, on the basis of past experience, 

do not involve significant impacts on 

natural, cultural, recreational, historic or other 

resources; and also those that do not otherwise, 

either individually or cumulatively, have any significant 

environmental impacts. Because they 

fall into those categories, APHIS has already categorically 

excluded most small-scale field trials 

of recombinant DNA–modified plants. 

The exclusion stipulates that all large-scale 

field tests, as well as any field release of recombinant 

DNA–modified organisms involving 

unusual species or novel modifications, still generally 

require an Environmental Assessment or 

EIS. But the list of excluded or included activities 

can be modified through notice-and-comment 

rulemaking, in which the agency sets forth the 

complete analysis and rationale for excluding 

the activity. At the very least, APHIS should 

consider categorically excluding some of the 

classes of recombinant DNA–modified crops 

with which it now has more than two decades’ 

worth of precommercial and commercial experience, 

including herbicide-tolerant varieties of 

common crop species. 

The most rational and definitive approach, 

however, would be to eliminate the agency 

action that triggers the NEPA obligation initially—namely, 

case-by-case reviews of virtually 

all field trials and the commercialization 

of recombinant DNA–modified plant varieties. 

That would offer the dual advantages of relieving 

the USDA’s NEPA difficulties and also making 

regulators’ approach to recombinant DNA 

technology more scientifically defensible and 

risk based. As the scientific community has 

explained for over two decades, the decision to 

subject recombinant and conventional organisms 

to different regulatory standards cannot be 

justified scientifically. The increasing prevalence 

of obstructionist litigation now shows that doing 

so is also wholly impractical. 

Recent NEPA lawsuits have prevented the 

marketing of products that offer palpable, demonstrated 

benefits to farmers, consumers and 

the environment. Nuisance litigation intended 

to slow the advance of socially responsible technologies 

are abusive, irresponsible and antisocial. 

And so are those who file them. It is long 

past time for NEPA’s burdensome paperwork 

requirements to be lifted from such an important 

and beneficial technology. 




FEATURE 

To selectivity and beyond 

George S Mack 

First-generation epigenetic drugs have proven clinically useful in several hematological cancers. But newer enzyme 

inhibitors in the pipeline aim to be more selective and promise to broaden the portfolio of therapeutic uses. 


Fifty years after covalent chromatin modifications 

were first discovered, four drugs that 

modulate the cellular epigenetic machinery 

are now registered by the US Food and Drug 

Administration (FDA) for use in prolonging 

the lives of people afflicted with blood and lymphoid 

cancers. Although clinical responses to 

these first-generation products have been reasonable—given 

that they act on a pan-genomic 

scale—drug developers are now aiming to produce 

a second generation of agents that target 

newly recognized enzymes involved in DNA and 

chromatin modification. Because these enzymes 

alter expression of a narrower, more specific set 

of modifications, it is hoped the new experimental 

agents will not only have an improved 

therapeutic index but also open up epigenetic 

therapy to indications outside of cancer. 

Early validation 

Among the current approved chromatin modulating 

agents are drugs that were first investigated 

in the 1970s and early 1980s (Table 1). 

In a few cases, their clinical value in treating 

malignancies only became apparent decades 

later. In the intervening years, advances in our 

understanding of the mechanisms of chromatin 

remodeling and DNA modification have 

lent support to the notion that in certain diseases, 

the dysregulation of gene expression 

arising from alterations in the epigenetic ‘software’ 

may be corrected by pharmacological 

intervention. By correcting aberrant cytosine 

methylation in the DNA sequence, or incorrect 

acetylation, methylation, phosphorylation, 

ubiquitination or sumoylation modifications 

of histone proteins, such agents should, in 

principle, mitigate abnormal patterns of gene 

expression associated with disease. 

In the past 15 years, researchers have 

begun to identify a battery of enzymes, which 

George S. Mack is a freelance writer based in 

Columbia, South Carolina. 

Unmarking the genome. An artist rendition of HDAC inhibitors removing marks from chromatin. 

(Source: Marc Phares, Photo Researchers) 

promote or reverse modifications to histones, 

and chemical changes to DNA, which 

do not alter the nucleotide sequence (Fig. 1). 

It is now known that many of these marks 

are heritable from gametes to offspring and 

from somatic to daughter cells, but at the 

same time, demonstrate a level of plasticity 

that could be exploited in therapeutic settings. 

Epigenetic modifications can also be 

environmentally (and perhaps behaviorally) 

Table 1 Approved drugs that target epigenetic mechanism 

induced, and these, too, may be passed down 

through meiosis and mitosis. 

Although chromatin interactions affecting 

transcription are known to occur at even 

greater chromosomal scales, until now, much 

of the research on epigenetic machinery has 

focused on nucleosomes—nuclear structures 

in which the DNA is wrapped around a core 

of two copies each of the H2A, H2B, H3 and 

H4 histone proteins. It is now known that the 

Company Target Agent Indications 

Merck HDAC Zolinza (vorinostat) (suberoylanilide 

hydroxamic acid, aka SAHA) 

Celgene HDAC Istodax (romidepsin) (formerly FK228, a 

cyclic peptide principally active against 

class 1 HDACs); nanomolar potent 

Cutaneous T-cell 

lymphoma (CTCL) 

CTCL 

Development 

stage 

FDA approved 

Oct. 2006 

FDA approved 

Nov. 5, 2009 

Celgene DNMT Vidaza (5-azacitidine) MDS FDA approved 

May 2004 

Eisai 

Tokyo; 

sublicensed to 

Johnson & Johnson 

DNMT Dacogen (decitabine) MDS FDA approved 

May 2006 


feature 


Figure 1 The ever-expanding 

family of enzymes that control 

the structure of chromatin 

and gene transcription. 

Modifications of the histone 

protein tail and DNA are 

shown: changes in acetylation 

(Ac) by acetyltransferases and 

deacetylases, phosphorylation 

(PO 4 ) by kinases, ubiquitylation 

(Ub) by ligases and changes 

in methylation (Me) by 

methyltransferases and 

demethylases. Methylation of 

CpG on DNA is also shown. 

(Source: Epizyme) 

Ac 

Ub 

stability of nucleosome occupancy at specific 

genomic loci is affected by the presence of particular 

variants of core histone proteins, such 

as H3.3 and H2A.Z. 

Whereas our knowledge of some of the 

chromatin modifications associated with 

certain diseases is progressing, little is yet 

understood about the mechanisms by which 

such marks influence pathogenesis of disease. 

Epigenetic modifications can either upregulate 

or downregulate genes; for example, it 

has been assumed with some clinical vindication 

that deacetylated histone conditions 

bring some tumor suppressor genes to 

unfavorably diminished states of expression, 

thereby preventing apoptosis and leading to 

cell proliferation and advancing disease. In 

addition, hypermethylation with 5-methylcytosine 

of DNA within CpG islands is seen in 

many cancers, and this phenomenon invites 

repressive complexes, consisting of methyl- 

CpG-binding-domain proteins and DNA 

methyltransferases (DNMTs) for maintaining 

DNA methylation, resulting in a relatively 

durable silencing of the same gene region 1 . 

First-generation DNMT and histone 

deacetylase (HDAC) inhibitors are largely 

indiscriminate in their activity. DNMTs apply 

methylation marks to any cytosine on a DNA 

strand that is accessible, but the factors governing 

which methyl groups are removed by 

DNA demethylases remain unclear. It is also 

now recognized that cytosine can be marked 

with hydroxymethyl groups as well, but the 

significance of this modification is currently 

only conjecture. 

HDAC inhibitors are classified into four 

categories: hydroxamic acid derivatives 

(including the fungal antibiotic trichostatin 

A), short chain fatty acids (e.g., valproic acid), 

cyclic peptides (e.g., depsipeptides) and phenylene 

diamines. Pharmacological inhibition 

of HDAC activity can affect many cellular 

processes other than histone function. For 

example, HDAC classes I and II target many 

K 

K 

Me 

R 

K 

S 

PO 4 

96 Histone methyltransferases (HMTs) 

44 Arginine methyltransferases (RMTs) 

52 Lysine methyltransferases (KMTs) 

Me 

CpG 

nonhistone proteins, including transcription 

factors and proteins that regulate cell division, 

migration and death, all of which are 

presumably affected by HDAC inhibitors. 

Class III HDACs make up the so-called sirtuin 

family, which has also been the subject 

of intense focus in designing treatments for 

age-associated diseases, such as activation of 

SIRT1 in type 2 diabetes. 

Histone acetylation takes place on lysine 

residues that reside on the histone tails, and 

as far as researchers can tell the histone acetyltransferases 

that apply these acetyl marks and 

the HDACs that take them off don’t show any 

preference for which lysine residue or which 

histone is affected, or even for histones, for that 

matter. Some researchers now refer to them as 

lysine deacetylases rather than HDACs, which 

better describes them by their function, as 

opposed to defining them by a target. 

A greater level of specificity might be 

achievable by targeting histone methyltransferases 

(HMTs) and demethylases 

(HDMs), where a great deal of new research 

has been focused over the past eight years. 

Nessa Carey, scientific director at epigenetics 

startup company CellCentric (Cambridge, 

UK), notes that, compared with HDACs 

and DNMTs, some HMTs “are very, very 

precise.” For example, the histone lysine 

Box 1 Noncoding RNAs as targeted epigenetic drugs? 

Evidence is growing that natural antisense transcripts and other long, interdispersed, 

noncoding RNAs (lincRNAs) are involved in inducing chromatin modifications (and to a 

much lesser extent, alteration of sense promoter DNA methylation). The exact mechanisms 

by which noncoding RNAs mediate these activities are being worked out and may involve 

promotion of DNA methylation by binding to the promoter region of a corresponding gene 

sequence (in the sense orientation) in the genome or providing a template onto which 

histone-modifying enzymes are then recruited. 

Although mechanisms remain unclear, the association of lincRNAs with disease, 

particularly cancer, is convincing. The lincRNA H19, one of the first to be discovered, 

is associated with several tumor types, including breast, bladder and liver. Furthermore, 

under conditions favoring tumorigenesis, such as hypoxia, knocking H19 down affects 

the expression of genes involved in angiogenesis, survival and tumorigenesis, suggesting 

a central role for the RNA in tumorigenesis 4 . Similarly, recent studies have shown that a 

specific lincRNA called HOTAIR, a part of the HOX locus, which is dysregulated in breast 

cancer and foreshadows metastasis, appears to do so by reprogramming cells to a more 

embryonic state. Overexpression of HOTAIR in epithelial cells causes repositioning of the 

Polycomb repressive complex 2 and alterations of methylation at H3 lysine 27 (ref. 5). 

Startup CURNA (Jupiter, FL, USA) is attempting to translate this knowledge into 

oligonucleotide therapies based on work from the laboratory of Claes Wahlestedt at 

the Scripps Research Institute (Jupiter, FL, USA). According to Wahlestedt, of the two 

possible natural antisense mechanisms—concordant, in which inhibiting an antisense 

transcript directed against a lincRNA has the effect of repressing gene expression, and 

discordant, where it has the opposite effect of derepressing or activating genes—CURNA 

is focused on discordant effects (that is, upregulation of gene expression). These are 

more attractive from a commercial standpoint because most conventional oligonucleotide 

therapeutics, monoclonal antibodies or small molecules act by inhibiting the target 

rather than activating it. Upregulating proteins in vivo has other advantages in that it gets 

around the problem of having to produce proteins synthetically and of delivering them 

intracellularly, says Wahlestedt. 

CURNA has an exclusive license for the technology from Scripps on natural antisense 

and noncoding RNA as drug targets and has filed additional patents on roughly 100 genes 

and inhibitor sequences for upregulating pharmaceutically important targets, seven of 

which they have now tested in vivo. In all cases, according to Wahlestedt, the oligos had 

the desired effect of upregulating their target gene. 


feature 


Table 2 Deals centering around epigenetics products or technology 

Epigenetics company Partnering company (location) Date Agreement 

MethylGene EnVivo Pharmaceuticals March 2004 EnVivo entered into an exclusive agreement to develop an isotypic HDAC inhibitor for 

neurodegenerative diseases for $1.1 million upfront 

Syndax Bayer Schering Pharma April 2007 Syndax acquired worldwide development and commercial rights to entinostat 

(SNDX-275, formerly MS-275) 

Celgene Pharmion November 2007 Celgene acquired rights to Vidaza (DNMT inhibitor) by acquiring Pharmion for $2.9 billion 

MGI Pharma Eisai January 2008 Eisai acquired rights to Dacogen by acquiring MGI Pharma for $3.9 billion. MGI had 

licensed Dacogen from SuperGen in 2004 

4SC 

Nycomed 

(Planegg-Martinsried, (Zurich) 

Germany) 

Cronos Therapeutics 

(Oxfordshire, UK) 

Aphios 

(Woburn, MA, USA) 

Celleron Therapeutics 

(Oxon, UK) 

SuperGen 

(Dublin, CA, USA) 

ValiRx 

(London) 

VivaCell Biotechnology España 

(Cordoba, Spain) 

AstraZeneca 

(London) 

GSK 

(Brentford, UK) 

methyltransferase KMT1E will only methylate 

H3K9 (lysine nine on the H3 tail); so 

far, it has not been determined to methylate 

any of the other possible lysines. “To a 

greater or lesser extent,” says Carey, “these 

histone methylases and demethylases will 

only look at very specific residues on a very 

specific histone tail.” Even so, such enzymes 

are still working genome-wide rather than 

targeting DNA or chromatin modifications 

at particular genes. This is because chromatin 

modification enzymes are recruited to 

specific genes by DNA-binding proteins or 

co-factors or RNAs that localize the enzyme 

complex to a specific stretch of sequence. In 

themselves, they have no inherent specificity 

for a particular stretch of DNA. 

Hence, one last intriguing possibility for 

epigenetic therapeutic intervention is RNA or 

cofactors that direct traffic to specific genes. 

New evidence continues to be found that antisense 

RNAs transcribed from the genome play 

a role in DNA methylation, chromatin modifications 

and monoallelic expression (e.g., 

in X-chromosome-inactivation–linked or 

imprinting disorders). Such transcripts might 

be targeted using oligos or oligos designed to 

mimic natural antisense mechanisms (Box 1). 

July 2008 

July 2008 

March 2009 

May 2009 

October 2009 

4SC acquired eight oncology programs from Nycomed for €14 million ($17.8 million in 

today’s dollars), including the HDAC inhibitors resminostat (4SC-201) and 4SC-202, 

plus two other HDAC inhibitor programs 

Cronos became a wholly owned subsidiary of ValiRx 

Collaborative R&D agreement to develop a combination therapy for treatment of HIV 

latency, consisting of bryostatins (PKC modulators) and HDAC inhibitors 

Celleron licensed the HDAC inhibitor AZD9468 (now CDX101) from AstraZeneca. 

AstraZeneca was granted the option to discuss reacquisition of commercialization rights 

in the future 

GSK and SuperGen entered into a collaborative agreement where GSK could license 

or acquire as many as four discovery-stage and preclinical agents following proof-ofconcept 

experiments. Upfront payment of $5 million with potential of $375 million in 

royalties, milestones 

Celgene Gloucester Pharmaceuticals December 2009 Celgene acquired rights to Istodax (HDAC inhibitor) by acquiring the company for $340 

cash plus potential milestones and royalties worth $300 million 

Spectrum 

Pharmaceuticals 

(Irvine, CA, USA) 

Topotarget February 2010 Co-development deal for belinostat (HDAC inhibitor). Spectrum licensed rights for North 

America and India for upfront payment of $30 million and potential milestones up to 

$320 million 

CellCentric Takeda February 2010 CellCentric exclusively outlicensed one of its cancer programs to Takeda 

Pharmaceutical, which could potentially be worth more than $200 million to CellCentric 

over its course 

Cellzome GSK March 2010 GSK gains exclusive license to Cellzome’s Episphere technology as applied to immunoinflammatory 

disease for an upfront payment of €33 million, with royalties that could equal 

an additional €475 million 

Greater specificity? 

One key question for epigenetic therapy is, 

What actually provides target specificity for 

a particular HMT, HDM or, for that matter, 

any other chromatin-altering enzyme? “That’s 

the question I asked when I first looked at 

this job,” says Mark Goldsmith, who joined 

Constellation Pharmaceuticals (Cambridge, 

MA, USA) as CEO in August 2009. “The 

complexity of factors upregulating or downregulating 

genes is daunting,” he says, “but the 

problem is at least becoming more approachable.” 

A few puzzle pieces are now starting to 

come together to produce the beginnings of a 

blueprint that describes where certain marks 

cluster within or around genes according to 

their activity status. For example, if a histone is 

methylated at position H3K9, it tends to be on 

genes that are inactive; however, H3K4 methylation 

seems to be on active genes. 

Some unpublished research in the laboratories 

at Constellation Pharmaceuticals has 

shown a change of expression in “tens of genes” 

when histone methylation enzymes were 

inhibited, “not hundreds or thousands” as with 

HDAC inhibitors, according to Goldsmith. 

“The fundamental premise and strategy of our 

company and the field of epigenetics right now 

is the belief that these enzymes are much more 

selective,” says Goldsmith, whose company is 

currently surveying an array of more than 50 

HMTs and more than 30 HDMs. “They are 

much more selective in what their substrates 

are as well as how they are regulated,” he says. 

“Most importantly they are selective in their 

cumulative effect on target genes. But we are 

still in very early-stage discovery.” 

Up until now no company has been willing 

to show its hand on exactly which enzymes 

might be targeted next, but it appears that 

one of the first of the new-generation epigenetic 

agents advancing to clinical trials will be 

a histone demethylator. “I think that over the 

next 18 months you’ll see some histone methyltransferase 

inhibitors getting into the clinic,” 

says Will West of CellCentric. “We know of a 

number of parties who are actively working on 

that, and we ourselves are, of course, working 

in that space as well.” West won’t confirm any 

specific targets his company may be going after, 

but in late February CellCentric licensed one of 

its cancer programs to Asia’s largest drug maker 

Takeda Pharmaceutical (Osaka, Japan), which is 

looking to develop new oncology products out 

of this deal, which could be worth more than 

$200 million to CellCentric (Table 2). 


feature 


Histones 

DNA 

PRC1 

PRC2/3 EZH2 

DNMTs 

Methylation of 

histone tails 

Methylation 

of DNA 

Repressed genes 

Repressive 

complexes 

Takeda oncology research division leader 

Osamu Nakanishi says his company is looking 

into other unnamed epigenetic enzymes 

as well as the histone methylases CellCentric 

has identified. “Epigenetics controls so many 

things,” he says. “And it has decisive power 

for inducing cancer.” Although his work at 

Takeda is limited to oncology, Nakanishi ponders 

other possibilities, and he suggests that 

it would be interesting to look into targeting 

epigenetic enzymes for indications beyond 

cancers, particularly in metabolic disease, 

such as diabetes, and central nervous system 

(CNS) disorders, such as schizophrenia and 

depression (Box 2). The CNS connection is 

noteworthy because neurons are rarely mitotic 

after they are in place, and thus gene mutation 

would not be a likely culprit in disease etiology 

as it often is in cancers. “But epigenetic 

mutation or changes are certainly a possibility 

in CNS disease,” says Nakanishi. He sums up 

Takeda’s perspective on next-generation epigenetic 

therapies: “HDACs are so nonselective 

that even if we had a very specific HDAC 

inhibitor, we would still see toxicity, and that 

would limit dose,” he says. “Also, it is still not 

clear that even very selective HDAC inhibitors 

could be applied to specific diseases.” 

Although the new crop of epigenetics companies 

has been understandably coy about 

what newly identified targets they might be 

pursuing, some candidates have been receiving 

Figure 2 Making marks. The Polycomb group protein EZH2 adds methyl groups to histone proteins, 

primarily to lysine 27 of histone H3 (small blue circles). EZH2 recruits DNA methyltransferases 

(DNMTs, small green circles) to certain target genes, either setting up new methyl marks during 

development or maintaining the marks after each round of cell division. These marks could recruit 

further repressive complexes, such as PRC1 or HDAC co-repressors, through methyl-binding 

domain proteins. (Reprinted with permission from Taghavi, P. and van Lohuizen, M., Nature 439, 

794–795, 2006) 

wide attention. In the histone methylation pathway, 

there is a group of polycomb proteins, one 

of which is the enhancer of zeste homolog 2 

(EZH2), which trimethylates H3K27 and is 

currently drawing lots of interest from investigators 

(Fig. 2). Certain mutations occurring 

in the gene encoding the HDM UTX result in 

copious EZH2 expression, and there is strong 

in vivo evidence that inhibiting EZH2 function 

reduces histone methylation at the one particular 

lysine residue (H3K27) and significantly 

reduces tumor cell proliferation in various 

models 2 . Furthermore, when EZH2 expression 

is high in solid tumors, it tends to be a poor 

indicator of prognosis. 

Aside from these observations, one other 

reason for the interest is that EZH2 overexpression 

seems to cause gene silencing 

in parallel with DNA hypermethylation 3 . 

Translational investigator and medical 

oncologist Jean-Pierre Issa of the University 

of Texas M.D. Anderson Cancer Center 

(Houston) has been working with EZH2, and 

he believes it’s going to be an important target 

if an agent can be developed to counteract it. 

“We have evidence that the EZH2 mechanism 

is used in the same way DNA methylation is 

used to turn off tumor suppressor genes,” he 

says. “If you inhibit its expression, it abolishes 

the growth of the tumor and induces senescence 

in some cancers. I suspect that this will 

be the first target that companies are developing 

drugs for. 

Box 2 Marks in the brain 

Epigenetics company Epizyme (Cambridge, 

MA, USA) may have a leg up with EZH2 

because it holds an exclusive license for this as 

well as a second HMT, DOT1L, based on work 

done by Yi Zhang at the University of North 

Carolina at Chapel Hill. According to Zhang— 

who in 2007, along with Robert Horvitz of MIT, 

co-founded Epizyme—the company has rights 

to the two best characterized HMTs, EZH2 

and HOT1L. Epizyme CEO Robert Gould says 

that RNA interference (RNAi) screens carried 

out in the company to downregulate HMTs 

associated with cancer have demonstrated that 

such changes influence both epigenetic marks 

and invasive phenotypes. Epizyme is currently 

screening for small molecules that recapitulate 

the same effect as RNAi and optimizing the 

resulting candidate drugs in-house. 

Expanding indications 

There has been considerable debate as to why 

epigenetic inhibitors have been more successful 

in hematologic cancers than in other 

malignancies; indeed, solid tumor results with 

HDAC and DNMT inhibitors have been lackluster 

at best. Investigators have been looking 

to the future, anticipating that more selective 

enzyme antagonists would improve results. 

Researchers suspect that there are likely hundreds 

of epigenetic modifiers, and therefore 

the surface has hardly been scratched in determining, 

which are the best enzymes to inhibit 

or modulate in some way. 

Several neurodegenerative diseases have their etiology linked to epigenetics. Rett’s 

syndrome, the first for which the link was made in 1999, is caused by a mutation in 

methyl-CpG-binding-domain protein. Several syndromes that have an expansion of 

repeats, such as Fragile X, in which a CGG triplet is repeated, and Friedrich’s ataxia, in 

which GAA X TCC triplets are expanded within an intron in the gene encoding Frataxin, 

show silencing of critical genes due to hypermethylation. In the case of Friedrich’s 

ataxia, silencing of Frataxin is accompanied by hypoacetylation of histones 3 and 4 and 

trimethylation of H3L9. Researchers at Scripps Clinic (La Jolla, CA, USA) developed 

a class of HDAC inhibitors that reversed the Frataxin silencing 6 , which Repligen 

(Waltham, MA, USA) has licensed and is taking to the clinic. The company filed an 

investigational new drug application for one drug, RG2833, and was granted orphan 

drug status by the FDA in May of this year. 

Based on a Drosophila model of neurodegenerative disease, EnVivo (Watertown, MA, 

USA) claims to have developed the only CNS-penetrant HDAC inhibitor, which it reported 

in 2008. According to its website, the company is exploring this drug for CNS indications, 

like Alzheimer’s and Parkinson’s disease. 


feature 

One reason why HDAC and DNMT inhibitors 

may have been less successful in solid 

tumors is that, like any other new, experimental 

therapy, they have often been tested in individuals 

with advanced disease, who have already 

been through several cycles of treatment with 

chemotherapies or with targeted cytostatic 

agents and are therefore very sick. In solid 

tumors, HDAC inhibitors have been tested as 

fifth- and sixth-line therapies, in patients who 

have already become multi-drug resistant. In 

contrast to solid tumors, there were no good 

first-line therapies available for myelodysplastic 

syndromes (MDS) and, as a result, DNMT 

inhibitors were used in people naive to therapy. 

“This was just an accident of drug development,” 

says Jean-Pierre Issa who was principle 

investigator for the pivotal trial that preceded 

FDA approval of DNMT inhibitor Dacogen 

(decitabine) in May 2006. “To my knowledge, 

no one has really tested these drugs [DNMT 

inhibitors] in solid tumors as first-line therapies 

to determine if their activity is actually substantially 

lower than it is in MDS.” Issa designs 

clinical trials, and sees patients who have newly 

diagnosed disease that is far advanced. He says 

he would like to see oncologists speak frankly 


Table 3 Select epigenetic inhibitors in various stages of development 

Company Target Agent Indications Development stage 

HDAC inhibitors 

Merck HDAC Zolinza (vorinostat; suberoylanilide hydroxamic 

acid, SAHA). Nonselective, micromolar potent 

NSCLC 

Multiple myeloma 

Mesothelioma 

GBM 

Latent HIV infection in T cells 

Celgene HDAC Istodax (romidepsin) (formerly FK228) Peripheral T-cell lymphoma 

Selective for class I HDACs; nanomolar potent Multiple myeloma 

Pfizer 

New York 

HDAC1, 

3, 6 and 8 

CI-994 (N-acetyldinaline, also tacedinaline) 

Originally developed as an anti-convulsive 

NSCLC 

Pancreatic cancer 

MM 

Topotarget HDAC Belinostat (PXD101). Hydroxamic acid-type Peripheral T-cell lymphoma (monotherapy) 

CUP (BelCaP) 

Ovarian cancer (mono and with carboplatin) 

Hepatocarcinoma (monotherapy) 

Soft tissue sarcoma (mono and with doxorubicin) 

NSCLC (mono and in combination) 

AML and MDS 

Novartis HDAC Panobinostat (LBH589). Nonselective, very 

potent. Discovered in house by Peter Atadja 

Italfarmaco 

Milan 

Syndax Pharmaceuticals 

MethylGene 

Montreal 


CTCL 

CML 

Hodgkin’s lymphoma 

Metastatic melanoma 

Prostate cancer 

HDAC Givinostat (ITF2357) Juvenile idiopathic arthritis 

Hodgkin’s lymphoma 

Polycythemia vera 

HDAC1, 

2 and 3 

HDAC 

Hos2/ 

HDAC 

Entinostat (SNDX-275; formerly MS-275) 

Selective for HDACs 1, 2 and 3 

Mocetinostat (MGCD0103) 

Lymphoma 

Selective for class I HDACs 

AML 

MGCD290 selective for fungal HDAC enzymes Fungal infections 

EnVivo Pharmaceuticals HDAC EVP-0334 (central nervous system 

(CNS) penetrant) 

Sirtris 

Cambridge, MA, USA 

(Unit of GlaxoSmithKline) 

Curis 

Cambridge, MA, USA 


New Brunswick, NJ, USA 

Pharmacyclics 

Sunnyvale, CA, USA 

HDAC 

(SIRT1) 

HDAC 

EGFR/ 

ErbB1 

HER2/ 

neu 

HDAC 

HDAC 

HDAC8 

SRT501 

Selective for Class III HDACs 

Phase 3 

Phase 3 

Phase 2 

Discovery 

Phase 2 

Phase 2 

Phase 3 

Phase 2 

Phase 2 

Phase 2 

Phase 2 

Phase 2 

Phase 2 

Phase 1/2 

Phase 1/2 

Phase 1/2 

NSCLC (in combination with Tarceva (erlotinib)) Phase 2 

Breast cancer (in combination with aromatase Phase 2 

inhibitors) 

Leukemia and MDS (in combination with Vidaza Phase 2 

(5-azacitidine)) 

Alzheimer’s disease and other CNS indications 

Schizophrenia 

Colon cancer 


Phase 3 

Phase 3 

Phase 3 

Phase 3 (not open yet) 

Phase 2 

Phase 2 

Phase 2 

Phase 2 

Phase 2 

Phase 2 

Phase 2 

Phase 1 

Phase 1 

Preclinical 

Phase 2 

Phase 2 

CUDC-101 

(a hydroxamic acid moiety integrated into a 

quinazoline receptor tyrosine kinase inhibition 

pharmacophore) 

Advanced refractory tumors Phase 1 

JNJ-26481585 

Selective for class I HDACs 

Potential for enhanced HSP70 upregulation 

and Bcl-2 downregulation 

PCI24781 (broad range multi-isoform HDAC 

inhibitor) 

PCI-34051 

Selective for HDAC8 

MDS 

CML 

CLL 

B-cell lymphomas 

BCR-ABL positive leukemia 

Sarcomas 

Lymphomas, MM, CLL 

T-cell lymphomas 

Pediatric neuroblastoma 

Autoimmune disease 

Phase 1 

Phase 1 and 2 

Phase 1 

Preclinical 

(continued) 


feature 


Table 3 Select epigenetic inhibitors in various stages of development (continued) 

Company Target Agent Indications Development stage 

DNMT inhibitors 

Celgene DNMT Vidaza (5-azacitidine) AML Phase 2 

Eisai 

DNMT Dacogen (decitabine) CML 

Phase 3 

Tokyo 

AML 

Phase 3 

Sublicensed by Eisai to 

AML (elderly) 

Phase 2 


Ovarian cancer 

Phase 2 

HMT inhibitors 

Takeda Pharmaceutical 

Osaka 

(Licensed from CellCentric) 

HMTs 

Histone 

ubiquitinrelated 

enzymes 

with patients about using investigational drugs 

in late-stage disease before traditional cytotoxic 

agents have been tried. 

There is a generation of newer HDAC 

inhibitors coming online, including Celgene’s 

(Summit, NJ, USA) Istodax (romidepsin), 

which shows highest activity against class 1 

and was approved in November 2009. There 

is also the very potent nonselective HDAC 

HMT inhibitors Cancers Discovery 

CellCentric 

HMTs HMT inhibitors Cancers Discovery 

Histone 

ubiquitinrelated 

enzymes 

HDM 

Cellzome HMTs HMT inhibitors Inflammatory and autoimmune diseases, Discovery 

such as rheumatoid arthritis, multiple sclerosis, 

and inflammatory bowel disease 

Epizyme HMTs HMT inhibitors Cancers Discovery 

Constellation 

HMTs HMT inhibitors Cancers Discovery 


HDAC, histone deacetylase; DNMT, DNA methyltransferase; HMT, histone methyltransferase; HDM, histone demethylase; CUP, cancers of unknown primary origin; GBM, 

glioblastoma multiforme; MM, multiple myeloma; NSCLC, non-small cell lung cancer; CLL, chronic lymphocytic leukemia; CML, chronic myelogenous leukemia; MDS, 

myelodysplastic syndromes; AML, acute myeloid leukemia. 

inhibitor panobinostat (LBH589) being developed 

by Novartis (Basel), which is in phase 3 

trials for cutaneous T-cell lymphoma (CTCL), 

as well phase 2 for prostate cancer and metastatic 

melanoma. “I don’t think we’ve heard the 

end of histone deacetylase inhibitors, perhaps 

not as single agents but in combination with 

other drugs,” says Issa (Box 3). “I’m confident 

we’re going to find more applications for them.” 

Takeda’s Nakanishi doesn’t want to shut the door 

on existing drugs either. He proposes that there 

may be a way to make HDAC inhibitors work 

more effectively by combination with other epigenetic 

modulators. “That would be a potential 

strategy for regulating more specific, narrower 

gene sets,” he says (Table 3). 

Several biotechs as well as some pharma 

companies have programs in clinical trials 

Box 3 It takes two (or maybe more) 

Several companies are looking beyond single agents to drug 

combinations. Topotarget (Copenhagen) has a pan-HDAC inhibitor, 

belinostat, in multiple clinical trials as a single agent (Table 3), 

and in combination with a number of chemotherapeutic agents, 

including carboplatin, a DNA-modifying agent, and Taxol (paclitaxel), 

which interacts with microtubules. The rationale is that if you open 

up the chromatin with an HDAC inhibitor, there will be greater 

access for DNA-damaging agents. In the case of the belinostat and 

Taxol combination, the two drugs have shown synergy in preclinical 

studies that looked at tubulin acetylation and antiproliferative 

activity. Acetylation of tubulin stabilizes microtubule formation and 

thus adversely affects the ability of cancer cells to divide. Henri 

Lichtenstein, chief development officer at Topotarget, says their 

dose escalation studies in solid tumors show that belinostat could be 

combined safely with therapeutic doses of the two chemotherapies. 

They looked further into ovarian and bladder cancer and had good 

results even with platinum-resistant tumors, which has gotten the 

attention of the gynecological group at the US National Cancer 

Institute (Bethesda, MD, USA), which is sponsoring further clinical 

studies in gynecological cancers. The company is also completing 

a proof-of-concept study with belinostat plus carboplatin and Taxol 

(BelCaP) in cancers of unknown primary origin (CUP). “What’s also 

very interesting [with CUP] is that we can go into first-line patients 

since there is no effective treatment,” says Lichtenstein. 

Elsewhere, quasi-virtual biotech company Syndax (Waltham, 

MA, USA) is focused entirely on combination therapies in solid 

tumors, where the need for good therapies is the greatest. Based on 

work from Ron Evans’ laboratory at the Salk Institute (La Jolla, CA, 

USA) on regulating the expression of nuclear hormone receptors, 

Syndax is attempting to reverse drug resistance, in particular 

resistance to hormone therapy that develops in some breast cancers. 

According to CEO Joanna Horobin, “[Combining HDAC inhibitors 

with conventional chemotherapy] has a well-trodden regulatory 

path, but it was not clear to us that it was playing to the specific 

science.” Instead, they are trying combinations that “leverage the 

role of modulating genes that interfere with cell growth in cancer, 

particularly in [a] solid tumor setting,” she says. The company has 

two phase 2 programs in metastatic breast cancer in combination 

with aromatase inhibitors, and in lung cancer combining Tarceva 

(erlotinib) and Vidaza (5-azacitidine). 


feature 


targeting HDACs (Table 3). Furthermore, 

the number of HDAC inhibitor programs in 

all stages of development greatly outnumbers 

any other class of epigenetic targets (Fig. 3). 

MethylGene (Montreal), which was among 

the first companies to focus on epigenetic 

targets and the first to commit to identifying 

isoform-selective inhibitors, according to 

Jeffrey Besterman, CSO at MethylGene, currently 

has a class 1 ‘selective’ HDAC inhibitor, 

mocetinostat (MGCD-0103), an orally active 

phenylene diamine derivative currently in 

phase 2 trials for follicular lymphoma. The 

company also has another HDAC inhibitor, 

MGCD-290, that is a specific inhibitor for the 

fungal HDAC HOS2. Results of a phase 1 clinical 

trial of this compound were reported at the 

Interscience Conference on Antimicrobial 

Agents and Chemotherapy in September and 

Box 4 Diagnostic test development 

There has been explosive growth in 

commercial operations offering tests 

for detecting methylation patterns in 

certain malignancies. Basic epigenetic 

knowledge can be more rapidly 

translated into diagnostic products 

because the path to marketing is 

abbreviated compared with drug 

development, and usually only Clinical 

Laboratory Improvement Amendments 

certification is required for companies 

to set up assays in reference pathology 

laboratories in the United States. 

On the other hand, tests often have 

shortened life cycles due to the 

introduction of new and improved 

versions by competitors. 

Two colorectal cancer (CRC) blood 

assays based on DNA hypermethylation 

are in late-stage development and 

clinical use (Table 4) to noninvasively 

check for nascent disease or to see if 

an individual is at risk for cancer. One 

is being developed to determine loss of 

imprinting of insulin-like growth factor 2 

(IGF2) as a potential risk factor for 

colorectal cancer in younger people by 

Orion Genomics (St. Louis), based on 

technology Orion exclusively licensed 

from Johns Hopkins University (JHU; 

Baltimore). Loss of imprinting creates 

imbalance through increased expression, 

and indeed, loss of IGF2 imprinting 

has been shown to occur in nearly every 

human tumor that has been examined. 

Colon cancer is a particular instance 

where both copies of the gene are 

being expressed inappropriately. “There 

probably should be no IGF2 production 

Figure 3 Programs in epigenetics by phase of 

development. (Source: CellCentric) 

showed the drug was well tolerated, with no 

drug-drug interactions at the highest dose. 

Looking forward 

Researchers anticipate that greater selectivity 

will be a major factor for improvements 

in clinical responses to epigenetic inhibitors. 

Although drug development programs 

often seek to increase therapeutic index by 

increasing target selectivity, the latter doesn’t 

always yield the desired results. 

In a collaboration that took place during 

2005 and 2006 between Boehringer Ingelheim 

(Ingelheim am Rhein, Germany) and the 

Research Institute of Molecular Pathology 

Number of programs 

Table 4 Select epigenetics diagnostics companies 

60 

40 

20 

0 

Discovery 

Company Technology Products or development stage 

Epigenomics 

Epiontis 

(Berlin) 

Spin-off from 

Epigenomics in 

July 2003 

Exact Sciences 

(Madison, WI, USA) 

Orion Genomics 

Oncomethylome 

Sciences 

(Liege, Belgium) 

Diagnostics and 

biomarkers by 

means of DNA 

methylation 

analysis 


biomarkers via 

DNA methylation 

analysis 

Multiplex DNA 

assay 




analysis 




analysis 

Phase 1 

HDAC 

DNMT 

Histone demethylase 

Histone methyltransferase 

Histone ubiquitin ligase 

Histone-ubiquitin-specific proteases 

Phase 2 

Phase 3 

Launched 

(IMP, Vienna), which is principally supported 

by Boehringer, more than 125,000 

compounds were assayed by means of highthroughput 

screens to find suitable inhibitors 

Main product is mSept9, a test based on methylation of SEPT9 for 

early detection of colorectal cancer via blood sample. CE marked 

in October 2009 and marketed in the EU. Licensed to Abbott 

Molecular for its m2000 real-time PCR automated system. Abbott 

will submit a PMA in late 2010. Quest Diagnostics will offer the test 

in the US 

Lung cancer via bronchial lavage sample expected to be launched in 

EU in 2010 

Prostate cancer via urine sample in development 

Foxp3 marker for regulatory T cells for cancer diagnostics and 

screening in development 

T-cell marker for cancer screening in development 

CD4 T cell (naive and memory) assay in development for immune 

monitoring 

Collaboration with Genzyme’s cartilage repair product, Carticel, to 

develop biomarker assays for cell purity and quality of product 

Stool-based DNA test that combines mutation testing, methylation 

marker analysis and immunochemical tests 

Blood test for detection of loss of imprinting of the pro-growth, 

anti-apoptotic IGF2 gene as an assay for colorectal cancer risk. 

Technology is licensed from Johns Hopkins University and is in 

development 

Bladder, lung and ovarian cancer assays in development 

Prostate cancer via urine and tissue samples. Partnered with 

LabCorp and Johnson & Johnson. Late-stage development 

Other cancers partnered with LabCorp, Schering-Plough (now Merck 

& Co.) and Merck in development. 

• Colorectal cancer screening via blood and stool in development, 

partnered with LabCorp 

• Bladder cancer testing via urine sample in development 

• Lung cancer in early stage development 

• Pharmacogenomic tests in early stage development with partners 

Abbott Molecular and GSK Biologicals 

Valirx (London) Cancer diagnostics • Nucleosomics cancer detection test measures specific histone 

modifications that correlate with certain cancers 

Sequenom 

(San Diego) 

Enzo Life Sciences 

(Farmingdale, NY, 

USA) 

Tools for research, 

including software 

Tools for research 

• EpiPanels for cancers and imprinting; marketed 

Tools for histone- and DNA-modifying enzymes, for protein methylation 

and demethylation, and development of kits for HDACs and 

lysine-specific demethylases 

PMA, premarket approval; CE, meets requirements of applicable European directive. 

(continued) 


feature 

Box 4 Diagnostic test development (continued) 


in colon cells—or at least you should have production from 

one allele only,” says epigenetics researcher and gene imprint 

expert Randy Jirtle of Duke University (Durham, NC, USA). 

Researchers from Orion and JHU are collaborating to improve 

Orion’s IGF2 assay, and independent of Orion, JHU is conducting 

large-scale prospective trials to establish the lifetime cancer 

risk. Prior published retrospective studies involving more than 

200 patients have demonstrated that CRC patients are 21 times 

more likely to harbor the IGF2 biomarker in their blood compared 

with appropriate cancer-free subjects 7 . “When clinical trials are 

completed, we will market the Orion CRC Risk Test to the 81.6 

million people in their twenties and thirties in the United States 

and to the primary care physicians who treat them,” says Nathan 

Lakey, Orion’s CEO. 

Another CRC test based on loss of imprinting of SEPT9 is 

already being marketed in the European Union (Brussels) by 

Epigenomics (Berlin and Seattle). The Epigenomics CRC test 

has not yet been approved for use in the United States, but the 

company’s nonexclusive licensee Quest Diagnostics (Madison, 

NJ, USA) is set to market those laboratory services to physicians 

and patients once the test has been approved. If CRC can be 

identified at an early stage while still localized, the 5-year 

survival rate is high, ~90%. Yet only 40% of colorectal cancers 

are diagnosed as early stage disease, one reason being that there 

is resistance to colonoscopy exams due to fear of discomfort as 

well as economic concerns. The SEPT9 biomarker accurately 

picks up 62.75% of tumors, which is more precise than current 

methods of testing fecal samples. 

for the H3K9me2 HMT G9a, which is implicated 

in such cancers as liver and prostate. 

Investigators ended up with seven confirmed 

hits and ultimately identified a single smallmolecule 

compound BIX-01294 (a diazepinquinazolin-amine 

derivative), which was 

specific for G9a. “It had only modest potency,” 

says Boehringer’s medicinal chemist and 

senior principal scientist Michael August. 

“At the time, the project didn’t fit any of our 

therapeutic program objectives,” he says, but 

Boehringer did make the compound available 

to academic and translational researchers 

through a material transfer agreement. 

“There has been a lot of interest in the compound 

from the academic community,” he 

says. “We sent out a lot of samples to various 

groups.” Post-doctoral fellow Stefan Kubicek 

of the Broad Institute (Cambridge, MA, USA) 

was principal laboratory investigator on the 

screening project at IMP when he was a graduate 

student; he has found BIX-01294 to be 

“very selective,” with regulation of as few as 50 

genes, 20 of which are upregulated “in a common 

biochemical pathway.” But toxicity has 

been “a major problem in preclinical mouse 

models,” he says. 

The future will undoubtedly bring greater 

selectivity and precision in epigenetic therapies. 

In the meantime, there are nearer-term 

commercial opportunities in molecular diagnostics 

(Box 4 and Table 4). What is clear is 

that advances in our understanding of chromatin 

remodeling processes and their control 

is unlikely to translate into panaceas, especially 

in cancers that seem to escape and go around 

Early diagnosis can give patients the best prognosis, no matter 

the disease. With autism spectrum disorders (ASDs), there is no 

specific blood test or imaging study that can deliver a definitive 

diagnosis so that young children can begin treatment when it is 

likely to be most effective. Because autism spectrum disorders 

(ASD) occur three to six times more frequently in males than in 

females, clinical genetics researchers Julie Jones and 

Michael Friez at the Greenwood Genetic Center (GGC; Greenwood, 

SC, USA) are working on the hypothesis that the syndrome is 

X linked 8 . With chip developer Oxford Genome Technologies 

(OTG; Oxford, UK), they have developed a microarray system to 

analyze expression at GGC and so far have identified a subset 

of eight genes on the X chromosome that are differentially 

overexpressed or underexpressed in a cohort of 17 boys with 

autism. A larger study is under way, and so far, the trend of 

dysregulation continues in the new data set, but the eight genes 

do not stand out to the degree they did in the first round of data. 

Given the heterogeneity of patients, screening so few genes 

to obtain a definitive diagnosis might not be possible. What is 

apparent is that there is a greater degree of gene dysregulation in 

the autism cohort compared with the controls. “We are working 

with OGT to come up with the most appropriate methylation 

array design based on this most recent data set,” says Friez. The 

etiology of ASDs has been daunting for investigators. “You could 

argue that it’s really a chromosome- or genome-wide dysregulation 

of methylation and that it may just be a vulnerable location in 

males only because there’s not a second copy of the 

X chromosome,” says Friez. “We just don’t know yet.” 

other modalities. “So far, these drugs [HDAC 

and DNMT inhibitors] have been helpful but 

not curative,” says Issa. “I don’t believe there is 

anything magical about epigenetic therapy as 

opposed to other kinds of therapies. Resistance 

often develops, and this is the nature of cancer,” 

he says. “I think we just have to deal with 

it and look for other epigenetic approaches 

because multiple modality therapy is the way 

to overcome resistance in general.” 

1. Bostick, M. et al. Science 317, 1760–1764 (2007). 

2. Sooryanarayana, V. Nature 419, 624–629 (2002). 

3. Kondo, Y. et al. Nat. Genet. 40, 741–750 (2008). 

4. Matouk, I.J. et al. PLoS ONE 2, e845 (2007). 

5. Gupta, R.A. et al. Nature 464, 1071–1076 (2009). 

6. Herman, D. et al. Nat. Chem. Biol. 2, 551–558 

(2006). 

7. Cui, H. et al. Science 299, 1753–1755 (2003). 

8. Jones, J.R. et al. Am. J. Med. Genet. A. 146A, 2213– 

2220 (2008). 


patents 

On firm ground: IP protection of therapeutic 

nanoparticles 

Paul Burgess, Peter Barton Hutt, Omid C Farokhzad, Robert Langer, Scott Minick & Stephen Zale 

The development of generic therapeutic nanoparticles faces a combination of scientific, patent and regulatory 

challenges. 


Over the past decade, nanotechnology has 

shown the potential to dramatically affect 

several fields. In the healthcare sector, there has 

been an intense investment in the development 

of therapeutic nanoparticles for treatment of 

common diseases such as cancer, inflammatory 

disorders, infectious disease and cardiovascular 

disease. Despite the significant potential 

benefits of these products to patients and their 

families, no authoritative article has addressed 

the intellectual property (IP), regulatory, scientific, 

clinical and commercial advantages for 

therapeutic nanoparticles. This article outlines 

the reasons why innovative pharmaceutical 

products based on therapeutic nanoparticles 

offer particular advantages in IP protection 

from generic competition. 

What is a therapeutic nanoparticle? 

Parenterally administered therapeutic nanoparticles 

typically comprise an active ingredient 

together with organic or inorganic biomaterials, 

and range from 50 to 200 nm in diameter. 

Several classes of therapeutic nanoparticles, 

commercially available or in development, 

include liposomes, albumin-drug complexes 

and polymeric nanoparticles. The earliest therapeutic 

nanoparticles were liposomal drugs. For 

example, Doxil, which was approved by the US 

Food and Drug Administration (FDA) in 1995 

for treatment of Kaposi’s sarcoma, is a long- 

Paul Burgess, Scott Minick and Stephen 

Zale are at BIND Biosciences, Cambridge, 

Massachusetts, USA; Peter Barton Hutt is at 

Covington & Burling, Washington, DC, USA; 

Omid C. Farokhzad is at Brigham and Women’s 

Hospital and Harvard Medical School, Boston, 

Massachusetts, USA; and Robert Langer is 

at Massachusetts Institute of Technology, 

Cambridge, Massachusetts, USA. 

e-mail: paulburgess@bindbio.com 

Formulation patents 

Conventional 

drug product 

circulating liposome containing doxorubicin 1 . 

More recently, there has been interest in therapeutic 

nanoparticles composed of biodegradable 

polymers, which offer long circulation 

times characteristic of liposomes together with 

controllable drug-release kinetics mediated 

by polymer biodegradation 2 . After parenteral 

administration, these therapeutic nanoparticles 

can selectively accumulate in particular tissues 

or body locations, thereby enhancing the delivery 

of the drug payload to the site of disease. 

Many therapeutic nanoparticles access diseased 

tissue through the enhanced permeability and 

retention (EPR) effect. The EPR effect occurs 

in tumors, sites of inflammation and other diseased 

body tissues where the blood vessels are 

either disrupted or not fully formed, and as a 

result are leakier than normal vessels. The EPR 

effect allows nanoparticles to pass readily from 

the blood vessels into the tumor or inflamed 

tissue and to be differentially retained while the 

drug is released from the particles. It has been 

demonstrated that, as a result of this effect, therapeutic 

nanoparticles accumulate in a particular 

target tissue location and deliver more of the 

drug to the disease site over a longer period of 

time than a conventional drug product administered 

as a solution. 

Currently marketed therapeutic nanoparticles 

use passive targeting; that is, they rely exclu- 

Nanoparticle 

therapeutic 

Platform IP 

Process patents 

Formulation patents 

Manufacturing trade secrets 

PK study does not assess 

drug at site of action 

Figure 1 Market exclusivity: therapeutic nanoparticle versus conventional drug product. 

PK, pharmacokinetic. 

sively on the EPR effect to enhance delivery of 

drugs to the site of disease. Actively targeted 

nanoparticles now under development are 

surface-functionalized with targeting ligands 

that impart even greater specificity by binding 

to receptors located on diseased cells or tissues. 

The targeting ligand enhances the binding and 

retention of the nanoparticle to the target tissue 

and can also mediate the uptake or ingestion of 

the nanoparticle (and its therapeutic payload) 

by a target cell type. Whether a disease site is 

actively or passively targeted, therapeutic nanoparticles 

can increase the concentration of drugs 

in diseased cells or tissues and reduce concentration 

where the drugs can cause undesirable 

side effects. Moreover, therapeutic nanoparticles 

enable the drug developer to modulate the trafficking 

of the active pharmaceutical ingredient 

within the body without altering its molecular 

structure, and hence its inherent pharmacological 

activity. The decoupling of drug disposition 

from molecular structure affords the possibility 

of developing better drugs with fewer tradeoffs 

than is possible with conventional drugs, where 

both biodistribution and pharmacological 

activity depend on molecular structure and 

cannot be manipulated independently. This 

interdependence often forces undesirable compromises, 

or worse, abandonment of otherwise 

promising new drug candidates. 


patents 


Table 1 Currently approved nanoparticle or liposomal products on the US market 

Trade name Generic name Date approved Patents listed Par IV filing Generic equivalent 

Doxil Doxorubicin 1995 No No No 

Depocyt Cytarabine 1999 Yes No No 

Ambisome Amphotericin B 1997 Yes No No 

Abelcet Amphotericin B 1995 No No No 

Amphotec Amphotericin B 1996 No No No 

DepoDur Morphine 2004 Yes No No 

Daunoxome Daunorubicin 1996 No No No 

Abraxane Paclitaxel 2005 Yes No No 

A successful therapeutic nanoparticle must 

be precisely optimized to prevent its removal 

from the circulation by the body’s defense systems 

before it has a chance to reach its target, 

maximize trafficking to the desired location and 

minimize accumulation at sites where the drug 

may cause side effects. Like conventional pharmaceutical 

products, it is necessary to develop 

a robust and well-controlled manufacturing 

process that produces nanoparticles of uniform 

quality from batch to batch; however, nanoparticle 

manufacturing has additional complexities 

to sort through to uniformly and precisely 

control the critical parameters that lead to their 

therapeutic benefits. 

Therapeutic nanoparticles can encompass 

both existing and new active pharmaceutical 

ingredients. Indeed, when manufacturers use 

validated active pharmaceutical ingredients, the 

risk of drug development can be substantially 

lower as compared to development of a new 

chemical entity. The end product will still be a 

substantially differentiated product and retain 

market exclusivity as well as or better than a new 

chemical entity (Fig. 1). This exclusivity is likely 

to extend beyond patent expiration as a result 

of additional barriers to generic competition 

discussed below that are unique to therapeutic 

nanoparticles, thereby potentially reducing or 

avoiding the sudden revenue fall-off for successful 

proprietary drugs that confronts the 

pharmaceutical industry today (Fig. 2). 

Patenting of nanoparticles 

A therapeutic nanoparticle is a complex, yet 

uniformly precise, platform for developing 

new drugs. Because of the complexity of these 

nanoparticles, opportunities abound for obtaining 

meaningful patent protection. For example, 

a polymeric nanoparticle can vary by polymer 

type, polymer size, drug, mix of polymers, surface 

characteristics, targeting ligands, content 

of drug encapsulated, concentration of nanoparticles 

or manufacturing process. Each one of 

those factors can have a considerable effect on 

the behavior of the nanoparticle in a biological 

system. Polymeric nanoparticle characteristics 

critical to its safety and effectiveness, such as its 

target tissue accumulation, pharmacokinetics 

(clearance and volume of distribution), stability 

and drug release kinetics, are dependent on 

numerous formulation and process parameters. 

The combination of many factors leading to 

many outcomes provides numerous grounds 

for patentability. Opportunities exist to patent 

(i) particular aspects of therapeutic nanoparticles 

critical for effective use, (ii) classes of 

therapeutic nanoparticle compositions, (iii) 

specific therapeutic nanoparticle compositions, 

(iv) methods of using therapeutic nanoparticles, 

(v) methods of manufacturing therapeutic 

nanoparticles, (vi) therapeutic nanoparticles of 

particular drugs and (vii) the pharmacokinetic 

profile of particular nanoparticles. 

As they do for new chemical entities, patenting 

opportunities exist in the discovery phase of 

identifying new therapeutic nanoparticles. Later 

in the product development cycle, additional 

opportunities exist to patent clinical-stage uses 

and dosing schedules. These patent opportunities 

can be combined with existing IP to protect 

a new drug entity or can be the basis of a new 

patent estate protecting an existing active pharmaceutical 

ingredient or product. In addition, 

many aspects of developing and manufacturing 

nanoparticle products involve extensive knowhow 

to optimize the product. 

Manufacturing challenges 

Therapeutic nanoparticles present many of 

the same scientific challenges for generic drug 

entry as complex biological drugs. Very small 

changes to the manufacturing process of a 

therapeutic nanoparticle can result in a different 

nanoparticle altogether, both with respect to 

its chemical composition and structure as well 

as to its ultimate therapeutic effects. Changes 

to the manufacturing process of a therapeutic 

nanoparticle can affect the drug content of 

the particle and the size and chemical makeup 

of the polymer component of the particle. 

Structural parameters, including the nanoparticle 

surface properties and size, and the spatial 

arrangement of the polymer, drug and targeting 

ligand, are determined by the manner in which 

the nanoparticle assembles. Changing any of 

these factors may alter critical properties, such 

as the distribution of the particles in the body or 

the rate at which the drug is released from the 

particle, thereby resulting in a different clinical 

outcome 3 . Because the analytical tools necessary 

to characterize therapeutic nanoparticles 

and their specific biological effects are not yet 

well established, a generics company may not 

be able to demonstrate the degree of sameness 

of a nanoparticle product required for FDA or 

other regulatory approval. As such, the manufacturer 

of the trade or ‘pioneer’ drug could 

have great exclusivity advantages by keeping 

their final manufacturing process a trade secret 

or by obtaining process patents. If a generics 

company cannot practice the same manufacturing 

process, the generic product probably 

does not have the same therapeutic nanoparticle. 

Simply put, because the FDA requires that 

a generic product be the “same,” then without 

the same process, the generics company will 

face considerable difficulties producing the 

same product. 

Regulatory challenges 

Stringent regulatory requirements exist for 

the entry of generic therapeutic nanoparticles. 

Table 1 lists the currently approved nanoparticle 

or liposomal products on the US market. 

Despite the fact that many of these products 

lack patent protection today, and that several 

generate worldwide sales of $250–750 million, 

none of these products has a generic equivalent 

4 . In fact, the FDA has never approved a 

parenterally administered generic therapeutic 

nanoparticle. No generics company has even 

attempted to seek approval for a generic form 

of a patent-protected nanoparticle product, 

and thus there has not been any patent litigation 

resulting from a generics challenge (that 

is, a “paragraph IV challenge”). In essence, 

the technical and manufacturing complexity, 

difficulties in demonstrating bioequivalence 

and regulatory challenges result in longer and 

more expensive development pathways, which 

are incompatible with most generic drug 

business models. 

The two major regulatory difficulties for the 

development of generic therapeutic nanoparticles 

are (i) showing bioequivalence and (ii) 

FDA requirements for parenteral drug products. 

For a generic drug company to avoid 

doing full clinical trials of safety and effectiveness, 

they must show that, among other things, 

the generic product is bioequivalent to the 

pioneer product that it references and relies on 

for approval. For typical, oral, small-molecule 

products, this bioequivalence is easily shown by 

dosing healthy volunteers and measuring and 

comparing the mean area under the plasma 

concentration curve and maximum plasma 


patents 


and complex analytical methods. Even with all 

that data, the FDA might still require a clinical 

trial. The FDA discussed some of these complex 

comparison issues in 2001 (ref. 6) and 2007 

(ref. 7) and has yet to decide how to address or 

solve these issues for therapeutic nanoparticles 

as a whole. A recent draft guidance relating specifically 

to abbreviated new drug applications 

for Doxil, a passively targeted nanoparticle, sets 

a very exacting standard 8 . The guidance states 

that the generics company should conduct 

clinical studies in cancer patients and assess 

bioequivalence based on analysis of both free 

doxorubicin and liposome-encapsulated doxorubicin. 

It further states that the pivotal clinical 

study should use test product produced using 

the proposed commercial scale manufacturing 

process—a far more stringent requirement 

than the 1:10 scale conventionally employed 

for bioequivalence studies. Moreover, it is recommended 

that the proposed generic product 

contain the same lipid excipients produced by 

the same synthetic route as the reference product, 

and that the generic product be manufactured 

using the same process as the reference 

product. Lastly, the generic product should 

be equivalent with respect to a broad array of 

physicochemical characteristics, including 

liposome composition, physical state of the 

encapsulated drug, the liposome internal environment, 

morphology, lipid bilayer fluidity, 

size distribution, surface chemistry, electrical 

potential and charge, and in vitro drug leakage 

under a variety of conditions 9 . The FDA will 

likely review each therapeutic nanoparticle on 

its own and determine the recommended tests 

on an individual basis. For therapeutic nanoconcentration 

of the drug and then calculating 

the 90% confidence interval for the ratio of 

those mean responses 5 . If that confidence interval 

is within 80–125% of the pioneer product 

for those parameters, the generic product is said 

to be bioequivalent 5 . 

In the case of therapeutic nanoparticles 

and other locally acting drugs, the standard 

approach to the evaluation of bioequivalence 

is not sufficient to show sameness with respect 

to rate and extent of absorption of the active 

pharmaceutical ingredient at the site of action. 

A therapeutic nanoparticle circulates in the 

plasma and targets specific tissue locations. It 

is this tissue localization that is responsible for 

nanoparticles’ enhanced effectiveness and/or 

safety. A standard bioequivalence test suitable 

for most small-molecule generic products only 

assesses plasma drug concentrations in healthy 

volunteers, and does not reveal whether a 

generic product has the same tissue distribution 

and drug concentration in disease sites as a pioneer 

product. To directly demonstrate equivalent 

drug concentration at the site of action, a 

prospective generic company would have to 

confront several issues. First, it would need to 

test the generic product in individuals with the 

target indication because the EPR effect and the 

product biodistribution would be different in 

healthy individuals. Moreover, for most anticancer 

drugs, testing in healthy volunteers is not 

feasible because of the drugs’ toxicity. Second, 

noninvasive assays for quantifying active pharmaceutical 

ingredients or nanoparticles in most 

tissues do not exist. No assay currently exists to 

measure the drug release that occurs at the tissue 

site as opposed to other areas in the body. In 

principle, the possibility exists of sampling the 

target tissue and analyzing it for the concentration 

of nanoparticles and drug. This possibility 

is only theoretical because the FDA has never 

accepted this type of data for a generic product. 

Even if a company did conduct a study testing a 

generic product’s localization by conducting an 

invasive procedure, there would be a low likelihood 

the company would successfully identify 

a generic product with comparative localization 

on the first attempt. Many tissues simply would 

not be available for sampling without an invasive 

procedure such as surgery or percutaneous 

biopsy, which may create an ethical dilemma. 

Assuming that the necessary techniques to 

do this type of work could be developed (as 

these techniques do not currently exist), the 

cost and timing to conduct such testing could 

be prohibitive for most generic manufacturers 

or create a cost structure for a generic drug that 

would make it difficult to set the price below the 

proprietary medicine. A company would have 

to recruit patients, possibly conduct an invasive 

procedure on the patients, and develop new 

Cash flow 

Profit 

Loss 

Shorter, less expensive 

path to approval 

and launch 

Traditional drug economics 

Nanoparticle drug economics 

Longer-term peak sales 

Figure 2 Drug development economics: traditional versus therapeutic nanoparticle drugs. 

particles that localize to a greater extent than 

Doxil, the FDA may also require a showing 

of drug and nanoparticle concentration at the 

site of action. The combination of these factors 

makes for an uncertain, expensive and daunting 

development route for generic equivalents 

to therapeutic nanoparticles. 

A generic therapeutic nanoparticle would 

also have to meet the FDA’s stringent requirements 

for drug products intended for parenteral 

use. The FDA regulations state: 

“Generally, a drug product intended for 

parenteral use shall contain the same inactive 

ingredients and in the same concentration 

as the reference listed drug identified by the 

applicant under paragraph (a)(3) of this section. 

However, an applicant may seek approval 

of a drug product that differs from the reference 

listed drug in preservative, buffer, or antioxidant 

provided that the applicant identifies 

and characterizes the differences and provides 

information demonstrating that the differences 

do not affect the safety or efficacy of the proposed 

drug product” 9 . 

Essentially, this requirement means that 

a generics company seeking to make a drug 

product for parenteral use, such as a therapeutic 

nanoparticle, must develop the same 

formulation, both qualitatively and quantitatively, 

as that of the pioneer drug; that is, with 

certain exceptions, it must contain precisely 

the same ingredients in precisely the same 

amounts. The recent guidance from the FDA 

on liposomal doxorubicin suggests that even 

changes in preservative, buffer or antioxidant 

may not be accepted for a generic therapeutic 

nanoparticle. 


patents 


This level of sameness contrasts with typical 

oral formulations. For generic oral products, a 

generics company can substitute a wide variety 

of excipients and change their concentrations 

as compared to the pioneer product. Generics 

companies often change the excipients in a 

formulation because the manufacturer of the 

pioneer drug has a formulation patent and the 

generics company is looking to circumvent this 

patent. A generics company’s ability to switch 

excipients in oral products often limits the value 

of formulation patents for oral products. The 

issue for the pioneer drug company is typically 

that broad formulation claims can be invalidated 

in litigation and narrow formulation 

claims are often circumventable. 

In contrast, because of the FDA’s requirements 

for parenteral drug products, patents 

for a therapeutic nanoparticle have a much 

greater value than formulation patents for 

an oral product. For example, a pioneer drug 

company could have a patent claim narrowly 

covering a nanoparticle formulation, which 

a generics company could not circumvent. 

Importantly, however, if this claim covered 

an FDA-approved therapeutic nanoparticle, a 

generics company would have to develop an 

identical product with the exact same polymers 

and the exact same drug, all in the exact same 

proportions. Based on the FDA guidance on 

liposomal doxorubicin, it is now apparent that, 

at a minimum, the generic nanoparticle would 

also have to match the pioneer product in state 

of encapsulated drug, nanoparticle internal 

environment, nanoparticle morphology, nanoparticle 

size distribution, polymer orientation, 

electrical surface potential and drug release 9 . 

Regulations for more complex nanoparticles, 

such as polymeric nanoparticles, could have 

even more rigorous regulatory requirements 

including clinical outcome data. An attempt to 

circumvent the claim by substituting a different 

polymer or slightly reducing the concentration 

of a particular polymer would not be allowed 

for a generic therapeutic nanoparticle. Thus, the 

combination of even a very narrow composition 

or process patent claim with the FDA’s general 

requirements for parenteral drug products and 

the heightened requirements for therapeutic 

nanoparticles can result in a significant and 

long exclusivity period. Even broader patent 

claims could also be obtained for additional 

market protection, as described above. 

A generics company could attempt to circumvent 

the FDA’s requirements for parenteral 

drug products by filing a Section 505(b) 

(2) new drug application (NDA). Such a filing 

would not require the ‘same’ formulation. Of 

course by switching formulations, a generics 

company would not be able to fully rely on the 

pioneer drug company’s previous clinical findings. 

At best, a 505(b)(2) NDA would be able 

to rely on the toxicology data for the active 

pharmaceutical ingredient. A 505(b)(2) applicant 

would need to conduct new effectiveness 

trials in order to ensure that the new formulation 

would deliver the same quantity of drug to 

disease sites over the same time course, or produce 

the same drug exposure to nondiseased 

tissues. Because conducting effectiveness 

trials is the most expensive part of product 

development, this 505(b)(2) regulatory route 

is unlikely to be economically viable for the 

generics drug industry. In addition, products 

approved through the 505(b)(2) route are not 

directly substitutable in the pharmacy for the 

branded product. Hence, a 505(b)(2) generic 

product would need to be marketed and sold 

by a sales force, which most generics companies 

do not have. 

Conclusions 

Therapeutic nanoparticles offer the potential 

to dramatically improve the effectiveness and 

side-effect profile of new and existing drugs. In 

addition, this new drug class presents equivalent 

or possibly greater challenges for generic drug 

entry than other types of pharmaceutical products, 

including biologics. The combination of scientific, 

patent, know-how and regulatory 

issues makes the development of a generic 

therapeutic nanoparticle very difficult and 

inconsistent with the generics business model. 

Pioneer drug companies have the opportunity 

to obtain strong patent protection, which can 

maintain market exclusivity for the full patent 

term and potentially beyond patent expiration 

due to the scientific, manufacturing and regulatory 

hurdles confronting the development of 

generic therapeutic nanoparticles. If successful, 

this opportunity will fundamentally change the 

pharmaceutical business model by reducing or 

avoiding the sudden revenue fall-off for successful 

proprietary drugs that confronts the 

industry today (Fig. 2). 


The authors declare competing financial interests: 

details accompany the full-text HTML version of the 

paper at http://www.nature.com/naturebiotechnology/. 

1. Doxil Product Information. Ortho Biotech Products, LP 

(2008). 

2. Farokhzad O. & Langer, R. ACS Perspective 3, 16–20 

(2009). 

3. Davis, M. Nanotechnol. Law & Bus. 255–262 

(September 2006). 

4. US Food and Drug Administration. Orange Book: 

Approved Drug Products with Therapeutic Equivalence 

Evaluations 

5. US Food and Drug Administration. Guidance 

for Industry: Bioavailability and Bioequivalence 

Studies for Orally Administered Drug Products 

(March 2003). 

6. US Food and Drug Administration. Advisory Committee 

For Pharmaceutical Science Proceedings 

(July 20, 2001). 

7. US Food and Drug Administration. Critical Path 

Opportunities for Generic Drugs 

(May 1, 2007). 

8. US Food and Drug Administration. Draft guidance on doxorubicin 

hydrochloride (February 2010). 

9. 21 CFR 314.94(a)(iii). 


patents 


Recent patent applications in pharmacogenomics 

Patent number Description Assignee Inventor 

WO 2010115154, 

US 20100273219 

US 20080038810, 

US 7816121 

WO 2010115044 

US 20100210025 

WO 2008144345, 

EP 2145021 

US 20100178655, 

WO 2010083250 

WO 2010075570 

WO 2010048497 

KR 2010011719 

JP 4403376 

US 20090117583 

A method of amplifying target nucleic acids in 

samples involving preparing an amplification 

mixture of forward and reverse primers comprising 

the target-specific portion and a barcode 

primer, comprising nucleotides, and subjecting 

the mixture to amplification. 

A droplet microactuator comprising a substrate 

with electrodes for conducting droplet operations 

and temperature control elements arranged near 

the electrodes to transport nucleic acid droplets on 

electrodes; useful for nucleic acid amplification. 

A method for depleting a nucleic acid sample 

of nontarget nucleic acids involving denaturing 

the sample nucleic acids in a reaction mixture, 

contacting the denatured nucleic acids with at 

least one target-specific primer pair under suitable 

annealing conditions, conducting a first cycle of 

extension of any annealed target-specific primer 

pairs and conducting nuclease digestion of singlestranded 

nucleic acid sequences in the reaction 

mixture. 

A system comprising assigning modules to a 

genome, assigning a value or weight to a module 

for a given profile, analyzing a genomic sequence 

to identify modules and assigning a profile to the 

sequence; useful, e.g., for profiling a genomic 

sequence. 

A method of predicting the likelihood of a 

therapeutic response to insulin growth factor-1 

receptor (IGF1R) modulator as a cancer treatment, 

comprising measuring levels of biomarker genes or 

proteins before and after exposing sample to IGF1R 

modulator. 

Genotyping a single cell, comprising preamplifying 

an amplified genome to produce a 

pre-amplification reaction mixture comprising 

amplicons specific for target nucleic acids, and 

amplifying and detecting the amplicons. 

A computer-accessible medium used for 

assembling at least one haplotype sequence or 

genotype sequence of at least one genome, with 

stored computer-executable instructions. 

A method of diagnosing predisposition to, or 

progression of, Alzheimer’s disease, comprising 

determining a genetic profile comprising at least 

one marker in a specified candidate region and 

correlating the genetic profile with a reference 

profile. 

A single-nucleotide polymorphism provided for 

calretinin gene, retinoic acid receptor beta gene 

and pre-B-cell leukemia transcription factor 1 

gene; for use in biochips for predicting the risk 

of obesity. 

A unit for detecting interaction between 

substances (e.g., hybridization) that has a 

reaction region, water-absorptive gel in the 

reaction region, an electrode and water-holding 

portion; useful in a substrate for bioassays for 

analyzing mutations of genes. 

A new epidermal growth factor–like variant in 

skin (ELVIS)-2 polypeptide; useful for screening 

assays, detection assays, predictive medicine and 

methods of treatment. 

Fluidigm 

(S. San Francisco, CA, 

USA) 

Anderson M, Chen P, 

Kaper F, May A, Wang J 

Paik PY, Pamula VK, Paik PY, Pamula VK, 

Pollack MG, 

Pollack MG 

Duke University 

(Durham, NC, USA), 

Advanced Liquid Logic 

(Research Triangle Park, 

NC, USA) 

Fluidigm (S. San 

Francisco, CA, USA) 

Priority 

application date 

Publication date 

4/2/2009 10/7/2010, 

10/28/2010 

4/18/2006 2/14/2008, 

10/19/2010 

Zimmermann BG 4/2/2009 10/7/2010 

Victor Chang Cardiac George R, Wouters M 8/15/2006 8/19/2010 

Research Institute 

(Darlinghurst, Australia) 

Bristol-Myers Squibb 

(Princeton, NJ, USA) 

Fluidigm 

(S. San Francisco, CA, 

USA) 

New York University 

(New York) 

Genizon Biosciences 

(St. Laurent, 

Quebec, Canada), 

TechnoSynapse 

(Blainville, Quebec, 

Canada) 

Attar RM, Carboni JM, 

Chen J, Dongre AR, 

Gottardis MM, 

Hafezi R, Han X, 

Huang F, Hurlburt W, 

Robinson DM, 

Wittenberg GM 

Hamilton A, Lin M, 

Mir A, Pieprzyk M 

5/17/2007 11/27/2008, 

7/22/2010 

1/13/2009 7/15/2010, 

7/22/2010 

Mishra B, Narzisi G 12/24/2008 7/1/2010 

Belouchi A, Croteau P, 

Debrus S, Kebache S, 

Keith T, Little RD, 

Paquin B, Poirier J, 

Raelson JV, Segal J, 

van Eerdewegh P 

Jeong J Gang S, Jeong J, 

Jeong M, Kim S 

Sony (Tokyo) Mamine T, Sakamoto Y, 

Segawa Y, Yuo K 

Busfield SJ, 

Gearing DP, 

Millennium 


(Cambridge, MA, USA) 

Busfield SJ, 

Gearing DP 

10/24/2008 4/29/2010 

7/25/2008 2/3/2010 

10/9/2003 1/27/2010 

11/19/1998 5/7/2009 

Source: Thomson Scientific Search Service. The status of each application is slightly different from country to country. For further details, contact Thomson Scientific, 

1800 Diagonal Road, Suite 250, Alexandria, Virginia 22314, USA. Tel: 1 (800) 337-9368 (http://www.thomson.com/scientific). 


news and views 

Megabases for kilodollars 

Mikkel Algire, Radha Krishnakumar & Chuck Merryman 

Two approaches for selecting oligonucleotides from complex mixtures improve the fidelity and scalability of 

DNA synthesis. 


The potential applications of synthetic DNA 

are virtually unlimited, ranging from the 

design of new genetic circuits and biosynthetic 

pathways to the creation of complete 

bacterial ge nomes 1–3 . Two reports in this issue 

from Church and colleagues bring us a step 

closer to an era of abundant synthetic DNA by 

addressing the hurdles of high cost and error 

rates. Kosuri et al. 4 and Matzas et al. 5 describe 

approaches that allow relatively inexpensive 

DNA oligo nucleotides (oligos), synthesized 

on a microarray, to be used to assemble long 

DNA molecules encoding genes and other 

genomic elements (Fig. 1). 

Currently, the price of long synthetic DNA 

is dominated by the price of the initial oligos, 

~$0.10 per nucleotide. The second most 

costly aspect of constructing synthetic DNA is 

the identification of assembled products that 

have the correct sequence, a step that requires 

sequencing. Sequencing costs vary tremendously 

depending on the degree of automation 

and other factors, but under the best of 

circumstances, they are 10–20% of oligo costs. 

The methods of Kosuri et al. 4 and Matzas et al. 5 

produce synthetic DNA at a cost of



considerably. Presumably, assembly is more 

robust because longer tracts of DNA are left 

after the amplification primers are removed. 

Thus, fewer starting fragments are needed to 

assemble each product. The method of Kosuri 

et al. 4 is still subject to the error rate of input 

oligos, although errors are reduced by using 

new, more accurate microarrays and by performing 

an enzymatic error correction step. 

The article from Matzas et al. 5 directly 

addresses the problem of errors in oligonucleotide 

sequences and the labor needed to deal 

with them. They use a 454 sequencing instrument 

to identify microarray oligos with the 

correct sequence and then recover these oligos 

from the sequencer. In essence, this provides 

cheap oligos preverified by cheap sequencing. 

454-beads bearing correct sequences are 

located with microscope cameras and picked 

using a micropipette. The oligos are subsequently 

assembled into larger products. Because 

the DNA assembly step can be virtually error 

free (even during the construction of whole 

genomes 1,2 ), presequenced and clonal oligos 

have the potential to produce near-perfect 

assemblies. Even though Matzas et al. 5 find an 

error in one of the eight ~220-bp constructs 

made from their microarray, this error rate 

is likely to decrease upon further optimization. 

Although the starting oligos, polymerase 

fidelity and accuracy of 454 sequencing all 

contribute errors, the experiments indicate 

that major gains are to be had by improving 

bead localization and retrieval. Automated 

bead extraction is under development, and the 

authors hope to achieve recovery rates of two 

or three beads per minute. The error rate of 

the current system is estimated to reach about 

1 in 21,000 bp. 

These papers demonstrate that the use of 

inexpensive microarray oligonucleotides to 

produce long synthetic DNA is becoming more 

practical. Assembly technology is also undergoing 

rapid development; for example, oligos 

produced by any method can be assembled 

into larger sequences by a fast and essentially 

labor-free method 3 . At the same time, projects 

involving intensive DNA synthesis are 

gene rating advances such as empirically determined 

codon-optimization rules 8 , new insights 

into host metabolism 9 and the first microbe 

controlled by a synthetic genome 2 . 

It is reasonable to wonder about how to 

make effective use of megabases of userdefined 

sequence. But with low-cost oligos 

and a simple means of assembling them, we 

can safely predict that our ability to understand 

and manipulate biology will only continue 

to improve, perhaps leading eventually 

to advances such as microbes engineered to be 

attenuated vaccines or to synthesize biofuels. 



1. Gibson, D.G. et al. Science 319, 1215–1220 (2008). 


3. Gibson, D.G. et al. Nat. Methods 7, 901–903 (2010). 

4. Kosuri, S. et al. Nat. Biotechnol. 28, 1295–1299 (2010). 

No refuge for insect pests 

Kongming Wu 

Transgenic cotton and corn expressing insecticidal 

proteins from the bacterium Bacillus 

thuringiensis (Bt) have been cultivated on >200 

million ha worldwide over the past 15 years 1 , 

reducing the use of chemical insecticides and 

increasing farmers’ profits 2,3 . But the environmental 

and economic advantages of these crops 

are threatened by the tendency of insect pests 

to develop resistance to Bt toxins. A report in 

this issue by Tabashnik et al. 4 demonstrates a 

new approach for suppressing the emergence 

of resistance. Field trials in Arizona over four 

growing seasons show that releasing sterile 

male pink bollworm moths, which can mate 

with resistant females, succeeded in almost 

completely eradicating a major cotton pest. The 

study establishes that sterile insect release is a 

viable alternative to the so-called refuge strategy 

for preventing the emergence of resistance and 

can enhance the sustainability of Bt crops. 

Continuous monoculture of crop varieties 

producing Bt toxins provides strong selective 

pressure for Bt-resistant insect pests (Fig. 1a). 

Indeed, although Bt crops remain effective 

against most targeted insect populations, several 

pests have evolved resistance 5 . The most 

promising strategy for delaying the emergence 

of resistance involves planting Bt crops in close 

proximity to ‘refuges’, which contain plants that 

do not express Bt toxin. Such refuges maintain 

populations of insects susceptible to Bt toxin, 

which are likely to mate with the rare, resistant 

insects. If the mode of inheritance of resistance 

is recessive, Bt plants will kill the hybrid progenies 

produced by such mating and thereby 

delay the evolution of Bt resistance 6 . 

Certain governments limit the proportion 

of crop that each farmer can plant to Bt 

Kongming Wu is at the State Key Laboratory 

for Biology of Plant Diseases and Insect Pests, 

Institute of Plant Protection, Chinese Academy 

of Agricultural Sciences, Beijing, China. 

e-mail: kmwu@ippcaas.cn 

5. Matzas, M. et al. Nat. Biotechnol. 28, 1291–1294 (2010). 

6. LeProust, E.M. et al. Nucleic Acids Res. 38, 

2522–2540 (2010). 

7. Xu, Q. et al. Proc. Natl. Acad. Sci. USA 106, 

2289–2294 (2009). 

8. Welch, M. et al. PLoS ONE 4, e7002 (2009). 

9. Warner, J. et al. Nat. Biotechnol. 28, 856–862 (2010). 

The sterile insect technique offers an alternative to the refuge strategy for 

managing resistance to Bt toxins. 

crops and require a minimum proportion of 

non-Bt crop to serve as a refuge. For example, 

the United States and Australia mandate this 

approach for Bt cotton that produces only one 

Bt toxin 5 . Resistance monitoring data from 

these countries have suggested that the refuge 

strategy can significantly delay the evolution of 

insect resistance to Bt crops 7 . It is also beneficial 

to non-Bt crops in the refuges, which are 

exposed to fewer pests owing to the control 

exerted by neighboring Bt crops 2 . Moreover, 

as nontransgenic seed is cheaper, there is an 

economic incentive for farmers to adopt the 

refuge strategy. 

Despite the benefits of the refuge strategy, it 

has proved difficult to implement in developing 

countries such as China and India, primarily 

because of the challenges associated with training 

and monitoring millions of smallholder farmers. 

During 2009, China planted 3.7 million ha 

(~70% of total cotton production) and India 

planted 8.4 million ha (~90% of total cotton 

production) to Bt cotton. In certain cases, however, 

it has been put in place unintentionally. In 

China, the primary pest targeted by Bt cotton 

is Helicoverpa armigera, which also feeds on 

other crops, including corn, soybean, vegetables 

and peanuts. These nontransgenic host plants 

are frequently planted near Bt cotton and serve 

as refuges for H. armigera. Thus far, continuous 

resistance monitoring has not detected any resistance 

of H. armigera populations to Bt toxin 7 . 

In contrast to the situation with H. armigera 

in China, reliance on noncotton host plants as 

refuges is not an option for controlling pink 

bollworm (Pectinophora gossypiella), which 

feeds almost exclusively on cotton in some 

regions 4 . This pest has evolved resistance 

to Bt cotton producing one toxin in western 

India, where farmers did not follow regulations 

requiring them to plant non-Bt cotton refuges 8 . 

In Arizona, where pink bollworm is an alien 

species first detected about a century ago, the 

efficacy of Bt cotton has been sustained for 




considerably. Presumably, assembly is more 

robust because longer tracts of DNA are left 

after the amplification primers are removed. 

Thus, fewer starting fragments are needed to 

assemble each product. The method of Kosuri 

et al. 4 is still subject to the error rate of input 

oligos, although errors are reduced by using 

new, more accurate microarrays and by performing 


The article from Matzas et al. 5 directly 

addresses the problem of errors in oligonucleotide 

sequences and the labor needed to deal 

with them. They use a 454 sequencing instrument 

to identify microarray oligos with the 

correct sequence and then recover these oligos 

from the sequencer. In essence, this provides 

cheap oligos preverified by cheap sequencing. 

454-beads bearing correct sequences are 

located with microscope cameras and picked 

using a micropipette. The oligos are subsequently 

assembled into larger products. Because 

the DNA assembly step can be virtually error 

free (even during the construction of whole 

genomes 1,2 ), presequenced and clonal oligos 

have the potential to produce near-perfect 

assemblies. Even though Matzas et al. 5 find an 

error in one of the eight ~220-bp constructs 

made from their microarray, this error rate 

is likely to decrease upon further optimization. 

Although the starting oligos, polymerase 

fidelity and accuracy of 454 sequencing all 

contribute errors, the experiments indicate 

that major gains are to be had by improving 

bead localization and retrieval. Automated 

bead extraction is under development, and the 

authors hope to achieve recovery rates of two 

or three beads per minute. The error rate of 

the current system is estimated to reach about 

1 in 21,000 bp. 

These papers demonstrate that the use of 

inexpensive microarray oligonucleotides to 

produce long synthetic DNA is becoming more 

practical. Assembly technology is also undergoing 

rapid development; for example, oligos 

produced by any method can be assembled 

into larger sequences by a fast and essentially 

labor-free method 3 . At the same time, projects 

involving intensive DNA synthesis are 

gene rating advances such as empirically determined 

codon-optimization rules 8 , new insights 

into host metabolism 9 and the first microbe 

controlled by a synthetic genome 2 . 

It is reasonable to wonder about how to 

make effective use of megabases of userdefined 

sequence. But with low-cost oligos 

and a simple means of assembling them, we 

can safely predict that our ability to understand 

and manipulate biology will only continue 

to improve, perhaps leading eventually 

to advances such as microbes engineered to be 

attenuated vaccines or to synthesize biofuels. 





3. Gibson, D.G. et al. Nat. Methods 7, 901–903 (2010). 

4. Kosuri, S. et al. Nat. Biotechnol. 28, 1295–1299 (2010). 

No refuge for insect pests 

Kongming Wu 

Transgenic cotton and corn expressing insecticidal 

proteins from the bacterium Bacillus 

thuringiensis (Bt) have been cultivated on >200 

million ha worldwide over the past 15 years 1 , 

reducing the use of chemical insecticides and 

increasing farmers’ profits 2,3 . But the environmental 

and economic advantages of these crops 

are threatened by the tendency of insect pests 

to develop resistance to Bt toxins. A report in 

this issue by Tabashnik et al. 4 demonstrates a 

new approach for suppressing the emergence 

of resistance. Field trials in Arizona over four 

growing seasons show that releasing sterile 

male pink bollworm moths, which can mate 

with resistant females, succeeded in almost 

completely eradicating a major cotton pest. The 

study establishes that sterile insect release is a 

viable alternative to the so-called refuge strategy 

for preventing the emergence of resistance and 

can enhance the sustainability of Bt crops. 

Continuous monoculture of crop varieties 

producing Bt toxins provides strong selective 

pressure for Bt-resistant insect pests (Fig. 1a). 

Indeed, although Bt crops remain effective 

against most targeted insect populations, several 

pests have evolved resistance 5 . The most 

promising strategy for delaying the emergence 

of resistance involves planting Bt crops in close 

proximity to ‘refuges’, which contain plants that 

do not express Bt toxin. Such refuges maintain 

populations of insects susceptible to Bt toxin, 

which are likely to mate with the rare, resistant 

insects. If the mode of inheritance of resistance 

is recessive, Bt plants will kill the hybrid progenies 

produced by such mating and thereby 

delay the evolution of Bt resistance 6 . 

Certain governments limit the proportion 

of crop that each farmer can plant to Bt 

Kongming Wu is at the State Key Laboratory 

for Biology of Plant Diseases and Insect Pests, 

Institute of Plant Protection, Chinese Academy 

of Agricultural Sciences, Beijing, China. 

e-mail: kmwu@ippcaas.cn 

5. Matzas, M. et al. Nat. Biotechnol. 28, 1291–1294 (2010). 

6. LeProust, E.M. et al. Nucleic Acids Res. 38, 

2522–2540 (2010). 

7. Xu, Q. et al. Proc. Natl. Acad. Sci. USA 106, 

2289–2294 (2009). 

8. Welch, M. et al. PLoS ONE 4, e7002 (2009). 

9. Warner, J. et al. Nat. Biotechnol. 28, 856–862 (2010). 

The sterile insect technique offers an alternative to the refuge strategy for 

managing resistance to Bt toxins. 

crops and require a minimum proportion of 

non-Bt crop to serve as a refuge. For example, 

the United States and Australia mandate this 

approach for Bt cotton that produces only one 

Bt toxin 5 . Resistance monitoring data from 

these countries have suggested that the refuge 

strategy can significantly delay the evolution of 

insect resistance to Bt crops 7 . It is also beneficial 

to non-Bt crops in the refuges, which are 

exposed to fewer pests owing to the control 

exerted by neighboring Bt crops 2 . Moreover, 

as nontransgenic seed is cheaper, there is an 

economic incentive for farmers to adopt the 

refuge strategy. 

Despite the benefits of the refuge strategy, it 

has proved difficult to implement in developing 

countries such as China and India, primarily 

because of the challenges associated with training 

and monitoring millions of smallholder farmers. 

During 2009, China planted 3.7 million ha 

(~70% of total cotton production) and India 

planted 8.4 million ha (~90% of total cotton 

production) to Bt cotton. In certain cases, however, 

it has been put in place unintentionally. In 

China, the primary pest targeted by Bt cotton 

is Helicoverpa armigera, which also feeds on 

other crops, including corn, soybean, vegetables 

and peanuts. These nontransgenic host plants 

are frequently planted near Bt cotton and serve 

as refuges for H. armigera. Thus far, continuous 

resistance monitoring has not detected any resistance 

of H. armigera populations to Bt toxin 7 . 

In contrast to the situation with H. armigera 

in China, reliance on noncotton host plants as 

refuges is not an option for controlling pink 

bollworm (Pectinophora gossypiella), which 

feeds almost exclusively on cotton in some 

regions 4 . This pest has evolved resistance 

to Bt cotton producing one toxin in western 

India, where farmers did not follow regulations 

requiring them to plant non-Bt cotton refuges 8 . 

In Arizona, where pink bollworm is an alien 

species first detected about a century ago, the 

efficacy of Bt cotton has been sustained for 




more than a decade. From 1996 to 2005, farmers 

who grew Bt cotton also planted non-Bt cotton 

in compliance with the refuge strategy, and the 

susceptibility of pink bollworm to Bt did not 

decrease 4 . A disadvantage of this approach, 

however, is that populations of the insect pest 

must be maintained. 

Tabashnik et al. 4 report that superimposing 

the sterile insect technique (SIT) on Bt cotton 

plants offers a compelling alternative to the 

refuge strategy while helping to eradicate 

pink bollworm. The authors’ program, which 

spanned ~100,000 ha in Arizona from 2006 to 

2009, showed that susceptibility to Bt cotton 

did not decrease, pink bollworm population 

density declined dramatically and insecticide 

sprays against this pest were eliminated. The 

SIT, a method of pest control using area-wide 

inundative releases of sterile insects to reduce 

fertility of a field population of the same species, 

was first developed in the United States 

more than 50 years ago 9 . The repeated release 

of sterile insects into the environment in numbers 

10- to 100-fold in excess of the size of the 

native population can eventually drive the 

native population to extinction if the majority 

of native female insects mate with sterile males. 

Although the SIT has been successful in some 

cases, such as the screwworm fly (Cochliomyia 

hominivorax) eradication program, its application 

has been limited 9 . One challenge has 

been the cost of generating a sufficiently high 

ratio of sterile to wild insects at the start of a 

SIT program (Fig. 1b). Using another control 

method to lower the pest’s population density 

can make it easier to attain a suitable ratio, but 

intensive use of conventional insecticides raises 

concerns about harm to nontarget organisms, 

including people. 

The report of Tabashnik et al. 4 reveals how 

the SIT and Bt transgenic technology can be 

used in a complementary manner (Fig. 1c). 

On the one hand, expression of the Bt toxin 

suppresses the size of the native insect population 

enough to jump-start the efficacy of the 

SIT while increasing its economic feasibility. 

On the other hand, when female moths that 

have evolved resistance to the Bt toxin mate 

with sterile insects, the failure to produce 

fertile progeny delays the establishment of 

resistance based on dominant inheritance. More 

importantly, the combination of the SIT and 

Bt crops allows farmers to reduce or eliminate 

planting of non-Bt refuges. Meanwhile, because 

mating with sterile insects does not produce 

fertile progeny, the approach could delay pest 

resistance to Bt crops that is based on either 

dominant or recessive inheritance (Fig. 1c). 

The stunning success of the Arizona 

program can be attributed to several favorable 

factors, including long-term public 

a 

Bt cotton 

Low level of damage with high risk of 

resistance in the absence of refuges 

b 

c 

Non-Bt cotton 

Severe damage from 

wild insects 

Bt cotton 

Low level of damage with high risk of 

resistance in the absence of refuges 

investment in SIT against pink bollworm. Since 

1967, sterile pink bollworm moths have been 

released over cotton fields in the San Joaquin 

Valley of California to prevent establishment of 

this pest by moth immigration from southern 

California 9 . The Arizona program also benefits 

from the extremely high efficacy of Bt cotton 

against pink bollworm. Results from Arizona 

show that Bt cotton producing either one or 

two Bt toxins kills virtually 100% of susceptible 

pink bollworm larvae 4 . Coordinated contributions 

from cotton farmers, the US Department 

Emergence of 

resistance alleles in the 

absence of refuges 

Release of 

sterile moths 

Release of 

sterile moths 

Severe change from 

resistant insects 

Low level of damage with costly 

release of large numbers of sterile moths 

Extremely low level of damage with 

reduced chance of resistance and less 

extensive release of sterile insects 

Figure 1 Use of the SIT together with transgenic cotton expressing the Bt transgene suppresses the 

growth of the pink bollworm population and facilitates management of resistance to Bt toxin. Pink 

bollworm feeds only on cotton bolls and does not damage other tissues. (a) Sustainable use of Bt cotton 

to control pink bollworm populations is threatened by the emergence of resistance. (b) Although costly, 

repeated release of sterile pink bollworm moths (red) in vast excess to the number of wild moths (brown) 

can suppress the growth of pink bollworm populations. (c) Combined use of Bt cotton and SIT ensures 

that the release of fewer sterile moths can suppress the growth of pink bollworm populations while 

preventing the emergence of resistance to Bt toxin. 

of Agriculture and university scientists have 

also been critical. 

Despite the impressive results reported by 

Tabashnik et al. 4 and the potential of this combinatorial 

strategy for many situations, several 

alternatives to conventional Bt crops also show 

considerable promise. For instance, new insectresistant 

transgenic crops may decrease native 

insect populations more efficiently than current 

Bt crops. They include proteins modified to 

counteract resistance development or the expression 

of two or more distinct toxins targeting 

Katie Vicari 




the same pest species 10 . In another approach, 

the US Environmental Protection Agency has 

approved sales of mixtures of corn seeds with 

and without Bt toxins that kill corn rootworms. 

This ensures that farmers comply with the 

refuge strategy. However, its use is limited to 

the pest species whose larvae do not migrate 

between plants. Finally, there are also transgenic 

strategies to improve SIT efficiency by 

expressing dominant lethal genes, promoting 

refractoriness of the sterile insects to disease or 

facilitating strain marking, genetic sexing and 

molecular sterilization 11 . There thus seems 

considerable scope to further enhance the sustainability 

of Bt crops and develop new strategies 

for integrated pest management. 

Nanoparticles in the lung 

Wolfgang G Kreyling, Stephanie Hirn & Carsten Schleh 

Little is known about the fate of nanoparticles 

that enter the lungs, either deliberately through 

medical treatments 1 or incidentally through 

air pollution 2,3 and occupational exposure 

in the workplace. As inhaled nanoparticles 

can cross the air-blood barrier into the circulation 

and accumulate in secondary organs 

and tissues, a biokinetic analysis would be the 

first step in a dose-response study as part of 

a comprehensive risk assessment (Fig. 1). In 

this issue, Frangioni, Tsuda and colleagues 4 

undertake such a biokinetic analysis in rats, 

using near-infrared imaging to follow the 

fate of intratracheally instilled nanoparticles 

that are varied systematically in size, surface 

modification and core composition. The 

authors study the parameters that control 

whether nanoparticles remain in the lung, 

are transported to regional lymph nodes and 

the bloodstream or are cleared from the body 

through the kidneys. Their findings should 

benefit several areas of research, including 

the design of nanoparticles for drug delivery 

and the evaluation of the toxicity of particulate 

air pollution. 

Wolfgang G. Kreyling, Stephanie Hirn and 

Carsten Schleh are at the Comprehensive 

Pneumology Center, Institute of Lung Biology 

and Disease, and Focus Network Nanoparticles 

and Health, Helmholtz Zentrum München, 

German Research Center for Environmental 

Health, Neuherberg/Munich, Germany. 

e-mail: kreyling@helmholtz-muenchen.de 



1. James, C. Global Status of Commercialized Biotech/ 

GM Crops: 2009 (ISAAA 41, Ithaca, New York, 

USA, 2009). 

2. Hutchison, W. et al. Science 330, 222–225 (2010). 

3. Wu, K. et al. Science 321, 1676–1678 (2008). 

4. Tabashnik, B. et al. Nat. Biotechnol. 28, 1304–1307 

(2010). 

5. Tabashnik, B.E. et al. J. Econ. Entomol. 102, 2011– 

2025 (2009). 

6. Gould, F. Annu. Rev. Entomol. 43, 701–726 (1998). 

7. Liu, C. et al. Sci. China Life Sci. 53, 934–941 

(2010). 

8. Bagla, P. Science 327, 1439 (2010). 

9. Klassen, W. & Curtis, C. in Sterile Insect Technique 

(eds. Dyck, V., Hendrichs, J. & Robinson, A.) 3–36 

(Springer, The Netherlands, 2005). 

10. Tabashnik, B. Science 330, 189–190 (2010). 

11. Alphey, N., Bonsall, M.B. & Alphey, L. J. Econ. Entomol. 

102, 717–732 (2009). 

An imaging study begins to define the parameters that control the 

biodistribution of nanoparticles after pulmonary delivery. 

The lungs represent a promising route for 

drug delivery because of their accessible, 

large surface area and because the air-blood 

barrier is rather thin 1 . After crossing the airblood 

barrier, inhaled nanoparticles have been 

shown to reach detectable levels in the blood 

circulation and to enter specific cells and 

tissues 5 (Fig. 1). Translocation and accumulation 

in tissues seem to depend on physicochemical 

properties of the nanoparticle core 

and surface 5–7 . Thus far, however, few studies 

have described nanoparticle transport across 

the lungs quantitatively 5,6 . 

Frangioni, Tsuda and colleagues 4 find that 

noncationic nanoparticles smaller than ~34 nm 

in diameter that do not bind serum proteins 

reach the regional lymph nodes within 30 min. 

Nanoparticles larger than ~34 nm are consistently 

retained in the lungs. When the diameter 

falls below ~6 nm and the charge is zwitterionic, 

about half of the nanoparticles rapidly 

enter the bloodstream from the alveolar airspaces 

and are mostly cleared from the body by 

means of renal filtration. The authors also show 

that nanoparticle behavior depends strongly on 

the surface coating, which affects binding to 

proteins in body fluid. 

Biodegradable nanoparticles are currently 

considered to be the best choice for targeted 

drug delivery, but biopersistent nanoparticles 

might also be used if they are cleared from the 

body quickly enough to minimize exposure 

and retention 8 . As demonstrated previously 

by the Frangioni laboratory, intravenously 

administered nanoparticles can be rapidly 

excreted through the kidneys if their physicochemical 

properties are carefully chosen 9,10 . 

Efficient renal clearance represents one elegant 

way of keeping the accumulation and, hence, 

the dose to secondary organs, low 9,10 . The present 

study 4 indicates how renal clearance could 

be achieved following deposition in the lungs. 

The new work 4 also defines the requirements 

for targeting nanoparticles to regional 

lymph nodes, which may be useful for therapies 

directed to the immune system. However, 

further quantitative investigation is needed to 

differentiate between transport to the blood 

through lymphatic drainage and direct translocation 

across the air-blood barrier to the 

blood. Additional challenges will arise in extrapolating 

from the rather simple nanoparticles 

used in this study to highly functionalized 

nanosystems bearing targeting, therapeutic 

and diagnostic molecules. 

Finally, the results of Frangioni, Tsuda and 

colleagues 4 may help in understanding the 

Interstitium 

Regional 

lymph 

nodes 

Kidneys 

Urine 

Epithelium 

lining fluid 

Nano 

particles 

AM 

Fetus 

Circulation 

Heart 

Liver 

Bronchial 

epithelium 

Alveolar 

epithelium 

Neutrophil 

Monocyte 

Immune 

system 

Brain 

Figure 1 Schematic of nanoparticle translocation 

from the lung epithelium to regional lymph 

nodes and blood circulation. Once circulating 

in blood, nanoparticles may accumulate in various 

secondary organs of the body. The study by 

Frangioni, Tsuda and colleagues 4 highlights 

two important pathways of nanoparticle 

biodistribution: the pathway of lymphatic drainage 

toward regional lymph nodes and renal clearance 

from blood to urine. AM, alveolar macrophages. 

Katie Vicari 




the same pest species 10 . In another approach, 

the US Environmental Protection Agency has 

approved sales of mixtures of corn seeds with 

and without Bt toxins that kill corn rootworms. 

This ensures that farmers comply with the 

refuge strategy. However, its use is limited to 

the pest species whose larvae do not migrate 

between plants. Finally, there are also transgenic 

strategies to improve SIT efficiency by 

expressing dominant lethal genes, promoting 

refractoriness of the sterile insects to disease or 

facilitating strain marking, genetic sexing and 

molecular sterilization 11 . There thus seems 

considerable scope to further enhance the sustainability 

of Bt crops and develop new strategies 

for integrated pest management. 

Nanoparticles in the lung 

Wolfgang G Kreyling, Stephanie Hirn & Carsten Schleh 

Little is known about the fate of nanoparticles 

that enter the lungs, either deliberately through 

medical treatments 1 or incidentally through 

air pollution 2,3 and occupational exposure 

in the workplace. As inhaled nanoparticles 

can cross the air-blood barrier into the circulation 

and accumulate in secondary organs 

and tissues, a biokinetic analysis would be the 

first step in a dose-response study as part of 

a comprehensive risk assessment (Fig. 1). In 

this issue, Frangioni, Tsuda and colleagues 4 

undertake such a biokinetic analysis in rats, 

using near-infrared imaging to follow the 

fate of intratracheally instilled nanoparticles 

that are varied systematically in size, surface 

modification and core composition. The 

authors study the parameters that control 

whether nanoparticles remain in the lung, 

are transported to regional lymph nodes and 

the bloodstream or are cleared from the body 

through the kidneys. Their findings should 

benefit several areas of research, including 

the design of nanoparticles for drug delivery 

and the evaluation of the toxicity of particulate 

air pollution. 

Wolfgang G. Kreyling, Stephanie Hirn and 

Carsten Schleh are at the Comprehensive 

Pneumology Center, Institute of Lung Biology 

and Disease, and Focus Network Nanoparticles 

and Health, Helmholtz Zentrum München, 

German Research Center for Environmental 

Health, Neuherberg/Munich, Germany. 

e-mail: kreyling@helmholtz-muenchen.de 



1. James, C. Global Status of Commercialized Biotech/ 

GM Crops: 2009 (ISAAA 41, Ithaca, New York, 

USA, 2009). 

2. Hutchison, W. et al. Science 330, 222–225 (2010). 

3. Wu, K. et al. Science 321, 1676–1678 (2008). 

4. Tabashnik, B. et al. Nat. Biotechnol. 28, 1304–1307 

(2010). 

5. Tabashnik, B.E. et al. J. Econ. Entomol. 102, 2011– 

2025 (2009). 

6. Gould, F. Annu. Rev. Entomol. 43, 701–726 (1998). 

7. Liu, C. et al. Sci. China Life Sci. 53, 934–941 

(2010). 

8. Bagla, P. Science 327, 1439 (2010). 

9. Klassen, W. & Curtis, C. in Sterile Insect Technique 

(eds. Dyck, V., Hendrichs, J. & Robinson, A.) 3–36 

(Springer, The Netherlands, 2005). 

10. Tabashnik, B. Science 330, 189–190 (2010). 

11. Alphey, N., Bonsall, M.B. & Alphey, L. J. Econ. Entomol. 

102, 717–732 (2009). 

An imaging study begins to define the parameters that control the 

biodistribution of nanoparticles after pulmonary delivery. 

The lungs represent a promising route for 

drug delivery because of their accessible, 

large surface area and because the air-blood 

barrier is rather thin 1 . After crossing the airblood 

barrier, inhaled nanoparticles have been 

shown to reach detectable levels in the blood 

circulation and to enter specific cells and 

tissues 5 (Fig. 1). Translocation and accumulation 

in tissues seem to depend on physicochemical 

properties of the nanoparticle core 

and surface 5–7 . Thus far, however, few studies 

have described nanoparticle transport across 

the lungs quantitatively 5,6 . 

Frangioni, Tsuda and colleagues 4 find that 

noncationic nanoparticles smaller than ~34 nm 

in diameter that do not bind serum proteins 

reach the regional lymph nodes within 30 min. 

Nanoparticles larger than ~34 nm are consistently 

retained in the lungs. When the diameter 

falls below ~6 nm and the charge is zwitterionic, 

about half of the nanoparticles rapidly 

enter the bloodstream from the alveolar airspaces 

and are mostly cleared from the body by 

means of renal filtration. The authors also show 

that nanoparticle behavior depends strongly on 

the surface coating, which affects binding to 

proteins in body fluid. 

Biodegradable nanoparticles are currently 

considered to be the best choice for targeted 

drug delivery, but biopersistent nanoparticles 

might also be used if they are cleared from the 

body quickly enough to minimize exposure 

and retention 8 . As demonstrated previously 

by the Frangioni laboratory, intravenously 

administered nanoparticles can be rapidly 

excreted through the kidneys if their physicochemical 

properties are carefully chosen 9,10 . 

Efficient renal clearance represents one elegant 

way of keeping the accumulation and, hence, 

the dose to secondary organs, low 9,10 . The present 

study 4 indicates how renal clearance could 

be achieved following deposition in the lungs. 

The new work 4 also defines the requirements 

for targeting nanoparticles to regional 

lymph nodes, which may be useful for therapies 

directed to the immune system. However, 

further quantitative investigation is needed to 

differentiate between transport to the blood 

through lymphatic drainage and direct translocation 

across the air-blood barrier to the 

blood. Additional challenges will arise in extrapolating 

from the rather simple nanoparticles 

used in this study to highly functionalized 

nanosystems bearing targeting, therapeutic 

and diagnostic molecules. 

Finally, the results of Frangioni, Tsuda and 

colleagues 4 may help in understanding the 

Interstitium 

Regional 

lymph 

nodes 

Kidneys 

Urine 

Epithelium 

lining fluid 

Nano 

particles 

AM 

Fetus 

Circulation 

Heart 

Liver 

Bronchial 

epithelium 

Alveolar 

epithelium 

Neutrophil 

Monocyte 

Immune 

system 

Brain 

Figure 1 Schematic of nanoparticle translocation 

from the lung epithelium to regional lymph 

nodes and blood circulation. Once circulating 

in blood, nanoparticles may accumulate in various 

secondary organs of the body. The study by 

Frangioni, Tsuda and colleagues 4 highlights 

two important pathways of nanoparticle 

biodistribution: the pathway of lymphatic drainage 

toward regional lymph nodes and renal clearance 

from blood to urine. AM, alveolar macrophages. 

Katie Vicari 




health effects of air pollution as the nanoparticles 

studied may share common properties with the 

nano-fraction of particulate air pollutants. Most 

previous work in this field has used inhalation 

of nanoparticle aerosols as the ‘gold standard’ 

method of exposure because the particles will 

deposit throughout the entire bronchiolar and 

alveolar epithelium in a physiological manner. 

However, such studies are rather elaborate and 

expensive, and the deposited nanoparticle dose 

is usually difficult to determine. Intratracheal 

instillation of small nanoparticle suspension 

volumes, as used in the present study 4 , is widely 

accepted as an alternative method of delivery 

to the lungs. Although the dose administered 

by this method is well controlled, lung deposition 

is less uniform and far from physiological 7 . 

Moreover, the dose is frequently high in order 

to observe adverse short-term effects. 

The daily dose of inhaled nanoparticles in 

humans is limited to a maximal nanoparticle 

aerosol concentration of 10 12 m –3 (beyond that, 

coagulation changes the concentration rapidly 

within one minute under ambient air conditions), 

and the daily volume of air inhaled by 

an adult is ~15 m 3 . Assuming a nanoparticle 

deposition probability of 0.3, the daily deposition 

amounts to ~5 × 10 12 nanoparticles. 

Frangioni, Tsuda and colleagues 4 used doses of 

~10 15 nanoparticles per animal, which exceed 

human daily doses to the lungs. Their results 

must therefore be interpreted with caution 

when attempting to extrapolate to the effects 

of particulate air pollutants in humans. 

In the context of a field that lacks elementary 

knowledge about the short- and longterm 

transport of nanoparticles delivered to 

the lungs, the present study 4 represents an 

important advance. Ultimately, a comprehensive 

mechanistic understanding of the interactions 

of nanoparticles with biomolecules 

and proteins in body fluids or with cellular 

membranes and of the biokinetics across the 

air-blood barrier, the blood-brain barrier, the 

placenta, etc. can be expected to allow predictable, 

safe and sustained application of nanoparticles 

in biomedicine. 



1. Mansour, H.M., Rhee, Y.S. & Wu, X. Int. J. 

Nanomedicine 4, 299–319 (2009). 

2. Oberdörster, G., Oberdörster, E. & Oberdörster, J. 

Environ. Health Perspect. 113, 823–839 (2005). 

3. Maynard, A.D. et al. Nature 444, 267–269 (2006). 

4. Choi, H.S. et al. Nat. Biotechnol. 28, 1300–1303 

(2010). 

5. Semmler-Behnke, M. et al. Small 4, 2108–2111 

(2008). 

6. Geiser, M. & Kreyling, W. Part. Fibre Toxicol. 7, 

2 (2010). 

7. Oberdorster, G. J. Intern. Med. 267, 89–105 

(2010). 

8. Kostarelos, K. Nanomed. 5, 341–342 (2010). 

9. Choi, H.S. et al. Nat. Nanotechnol. 5, 42–47 (2010). 

10. Choi, H.S. et al. Nat. Biotechnol. 25, 1165–1170 

(2007). 


esearch highlights 


Rett’s-like neurons from iPS cells 

Induced pluripotent stem (iPS) cells offer 

the possibility of creating disease-specific 

cells that model pathology and can be used 

for screening therapies. Using skin biopsies 

from Rett syndrome patients, Marchetto 

et al. create iPS cells, from which they are 

able to create both neuronal progenitor 

cells and mature neurons. Whereas the 

progenitor cells appear normal, the Rett’slike 

neurons are both morphologically and 

functionally abnormal. This recapitulates 

the onset of disease, as symptoms don’t 

appear until as long as 18 months after 

birth (when progenitors have differentiated). 

The authors show that knockdown of methyl-CpG–binding protein 

(MeCpG), which is mutated in Rett’s patients, leads to cells with 

fewer synapses than neurons derived from iPS cells generated from 

normal individuals. Finally, the application of two drugs—insulinlike 

growth factor, which reverses the phenotype in a Rett’s 

mouse model, and gentamicin, which reverses premature protein 

termination—partially rescues the Rett’s-like neurons. By staining 

cell populations for proteins involved with X inactivation—of 

interest as MeCP2 is X linked—the authors could select iPS cells 

that had two active X chromosomes. Even so, during neuronal 

differentiation, re-inactivation of the X chromosomes was not 

random, suggesting the presence of some remnant of X inactivation 

after reprogramming. (Cell 143, 527–539, 2010) 

LD 

Personalizing epigenetic therapy 

Unlike the genetic changes that contribute to the progression of liver 

cancer, the epigenetic modifications responsible for driving the development 

of the disease can be reversed by pharmacological intervention. 

Nonetheless, there is little understanding of the factors that influence 

whether the therapeutic effects of broad-spectrum chemical inhibitors 

of DNA methylation (e.g., reactivation of epigenetically silenced tumor 

suppressor genes) over-ride any detrimental consequences. Working with 

the cytidine analog zebularine, Anderson et al. use transcriptomic and 

epigenomic profiling to reveal a response signature that classifies liver 

cancer cell lines as being either sensitive or resistant to the drug. The ability 

of zebularine to promote apoptosis in sensitive cell lines correlates with 

reduced growth and metastasis of drug-sensitive tumors and prolonged 

survival of mice bearing xenografts of sensitive human tumors. In contrast, 

the drug increased tumor growth rates and decreased survival rates 

of mice bearing resistant xenografts—consistent with zebularine’s ability to 

upregulate oncogenic pathways in cell lines predicted to be resistant to the 

drug. The ability of the signature to predict clinical outcome with 84–96% 

accuracy in a relatively small cohort of liver cancer patients suggests its 

value for identifying individuals most likely to benefit from methyltransferase 

inhibitors. (Sci. Transl. Med. 2, 54ra77, 2010) 

PH 

The ’omics of protein–small molecule 

interactions 

Metabolites are important regulators of many proteins. Nonetheless, the 

systematic analysis of small molecule–protein interactions has lagged 

Written by Kathy Aschheim, Laura DeFrancesco, Markus Elsner, 

Peter Hare & Craig Mak 

behind efforts to characterize, for example, the entirety of proteinprotein 

or protein-DNA interactions. Now, Li et al. perform a largescale 

analysis of hydrophobic metabolites that bind the enzymes of the 

ergosterol synthesis pathway and protein kinases in yeast. They affinity 

purify tagged versions of proteins of interest and identify proteinbound 

molecules by mass spectrometry after methanol extraction. In 

the ergosterol pathway, the authors uncover many previously unknown 

interactions between intermediate products and enzymes at other levels 

of the synthetic cascade. This suggests more integrated control of sterol 

synthesis than previously appreciated, although a detailed functional 

analysis of the significance of the interactions was not performed. 

Among the 103 kinases studied, 21 bind a total of ten different metabolites. 

The authors go on to demonstrate that the binding of ergosterol 

stabilizes Ssk22 and activates the highly conserved Ypk1 kinase. It seems 

likely that the approach could be adapted for hydrophilic molecules. 

(Cell 143, 639–650, 2010) 

ME 

Efficient discovery of rare genetic variants 

With sequencing single genomes now commonplace, efforts are shifting 

to gather data from many individuals. The resulting increase in statistical 

power should identify rare genetic variants, which may underlie 

disease. Initial results toward this goal are reported in Nature by the 

1000 Genomes Project Consortium, with analytic algorithms detailed in 

Genome Research. Three pilot sequencing efforts demonstrate efficient 

uses of sequencing to survey genetic variation. First, by reducing the average 

number of times any given region of the genome was sequenced, called 

‘low coverage’ sequencing, the project analyzed 179 genomes. This identified 

most common single-nucleotide variants present in >5% of individuals 

and many less-common variants. Second, targeted sequencing 

of a subset of the genome allowed 8,140 exons to be analyzed across 697 

individuals. Third, sequencing related individuals—in this case a mother, 

father and child from two families—enabled de novo germline mutations 

to be identified. In Science, Sudmant et al. used single, unique nucleotides 

found within repetitive regions of the genome to discover remarkable 

plasticity in copy number variation in duplicate genes, thereby making 

them amenable to genetic association studies. (Nature 467, 1061–1073, 

2010; Genome Res., published online 27 October 2010, doi:10.1101/ 

gr.1123267.110, doi:10.1101/gr.111120.110, doi:10.1101/gr.113084.110; 

Science 330, 641–646, 2010) 

CM 

lincRNAs in reprogramming 

The reprogramming of somatic cells to induced pluripotent stem cells 

(iPSCs) is accompanied by widespread changes in gene expression and 

epigenetic marks. A recent study by Loewer et al. explores whether it 

also involves changes in the expression of long intervening noncoding 

RNAs (lincRNAs), a class of RNAs with diverse roles, including 

regulation of gene expression and of the epigenome. The authors began 

by comparing the expression of ~900 lincRNAs in human fibroblasts, 

in iPSCs derived from the fibroblasts and in human embryonic stem 

cells. This analysis identified 133 upregulated and 104 downregulated 

lincRNAs in the pluripotent cells compared with the fibroblasts. 

Twenty-eight of the upregulated lincRNAs were expressed more highly 

in iPSCs than in embryonic stem cells, suggesting a possible involvement 

in reprogramming. To identify a more generic association with 

reprogramming, the authors looked for lincRNAs upregulated in iPSCs 

derived both from fibroblasts and from CD34 + cells. This approach 

yielded ten candidate lincRNAs, one of which was shown experimentally 

to influence reprogramming. (Nat. Genet. published online, 

7 November 2010; doi:10.1038/ng.710) KA 


A n a ly s i s 

Large-scale in silico modeling of metabolic interactions 

between cell types in the human brain 

Nathan E Lewis 1 , Gunnar Schramm 2,5 , Aarash Bordbar 1 , Jan Schellenberger 3 , Michael P Andersen 1 , Jeffrey K Cheng 1 , 

Nilam Patel 1 , Alex Yee 1 , Randall A Lewis 4 , Roland Eils 2,5 , Rainer König 2,5 & Bernhard O Palsson 1 


Metabolic interactions between multiple cell types are difficult 

to model using existing approaches. Here we present a workflow 

that integrates gene expression data, proteomics data and 

literature-based manual curation to model human metabolism 

within and between different types of cells. Transport reactions 

are used to account for the transfer of metabolites between 

models of different cell types via the interstitial fluid. We apply 

the method to create models of brain energy metabolism that 

recapitulate metabolic interactions between astrocytes and 

various neuron types relevant to Alzheimer’s disease. Analysis 

of the models identifies genes and pathways that may explain 

observed experimental phenomena, including the differential 

effects of the disease on cell types and regions of the brain. 

Constraint-based modeling can thus contribute to the study 

and analysis of multicellular metabolic processes in the human 

tissue microenvironment and provide detailed mechanistic 

insight into high-throughput data analysis. 

Constraint-based reconstruction and analysis of genome-scale 

microbial metabolic networks has matured over the past decade. 

This approach has a wide range of applications, such as providing 

insight into evolution, aiding in metabolic engineering, and providing 

a mechanistic bridge between genotypes and complex phenotypes 1,2 . 

Computational methods 3 and detailed standard operating procedures 4 

have been outlined for the reconstruction of high-quality prokaryotic 

metabolic networks, and many methods can be deployed for their 

analysis 5,6 . Constraint-based modeling of metabolism entered a new 

phase with the publication of the human metabolic network (Recon 1) 7 , 

based on build 35 of the human genome. Methods allowing tissuespecific 

model construction have followed 8–10 . 

Many tissue metabolic functions rely on interactions between cell 

types. Thus, methods are needed that integrate the metabolic activities of 

multiple cell types. Here, using Recon 1, we analyze and integrate ’omics 

1 Department of Bioengineering, University of California, San Diego, La Jolla, 

California, USA. 2 Department of Bioinformatics and Functional Genomics, 

Institute of Pharmacy and Molecular Biotechnology and Bioquant, University 

of Heidelberg, Heidelberg, Germany. 3 Bioinformatics Program, University of 

California, San Diego, La Jolla, California, USA. 4 Department of Economics, 

Massachusetts Institute of Technology, Cambridge, Massachusetts, USA. 

5 Department of Theoretical Bioinformatics, German Cancer Research Center 

(DKFZ), Heidelberg, Germany. Correspondence should be addressed to B.O.P. 

(palsson@ucsd.edu). 

Published online 21 November 2010; doi:10.1038/nbt.1711 

data with information from detailed biochemical studies to build multicellular, 

constraint-based models of metabolism. We demonstrate this 

process by constructing and analyzing models of human brain energy 

metabolism, with an emphasis on central metabolism and mitochondrial 

function in astrocytes and neurons. Moreover, we provide three detailed 

examples, demonstrating the use of models to guide experimental work 

and provide biological insight into the metabolic mechanisms underlying 

physiological and pathophysiological states in the brain. 

RESULTS 

Building metabolic models of multiple cell types 

‘Omics data sets can be difficult to analyze owing to their size. 

However, such data sets can be used to construct large mechanistic 

models for specific tissues and cell types 8,9 that serve as a context for 

further analysis. The workflow for generating multicellular models 

(Fig. 1) consists of the following four steps. 

Step 1. Reconstruct a metabolic network for the organism of interest 

from genome annotation, lists of biomolecular components and 

the literature 4 . Metabolic pathways and associated gene products 

are not completely known for any species. Thus, a reconstruction is 

refined through iterations of manual curation, hypothesis generation, 

experimental validation and incorporation of new knowledge. Recon 1 

has been through five iterations 7 . 

Step 2. Identify reactions specific to the tissue of interest. Many gene 

products are not expressed in all cells at any given time 11 . Therefore, 

the presence of a gene product, inferred from ’omics data, such as gene 

expression or mass spectrometry, is mapped to the metabolic network 

using associations linking genes to proteins and to reactions. This 

process may be performed manually or algorithmically 8,9 . 

Step 3. Partition the reconstruction, using manual curation of the 

literature, into compartments representing different cell types and 

organelles. The different cell types are linked with transport reactions, 

as supported by literature and experimental data. Initial context-specific 

reconstructions are incomplete and may contain false positives owing 

to contamination from proximal tissue. Moreover, few high-throughput 

data sets are cell-type specific. Thus, the initial reconstruction represents 

the union of metabolic networks from various cell types. After the 

reconstruction is partitioned using manual curation, it is converted into 

a model by specifying inputs, outputs and relevant parameters, and by 

representing the network mathematically 12 . Details of proper manual 

curation have been described 4 . 

Step 4. Simulation and analysis 1,2 . Once the network is accurately 

reconstructed and converted into an in silico model, it is used to 

nature biotechnology VOLUME 28 NUMBER 12 DECEMBER 2010 1279



Figure 1 A workflow for bridging the genotypephenotype 

gap with the use of high-throughput 

data and manual curation for the construction of 

multicellular models of metabolism. The models 

can be used to (i) predict disease-associated 

genes, such as glutamate decarboxylase; 

(ii) analyze high-throughput data in the network 

context to identify sets of genes that change 

together and affect specific pathways, such as 

the brain-region-specific suppression of central 

metabolism in Alzheimer’s disease patients; 

(iii) analyze physiological data in the context of 

the model, thereby enabling, for example, the 

identification of tissue properties relevant to 

disease treatment, such as the calculation of 

the percentage of the brain that is cholinergic. 

generate hypotheses and to obtain insight into 

systems-level biological functions. 

In contrast to previous approaches that 

reconstructed tissue–specific models 8–10 , the 

approach described here accounts for interactions 

between cell types within the tissue. First, 

published experimental results (e.g., immunohistochemistry 

staining of brain tissue 11 ) are 

used to refine the reconstructed network into 

individual cell type networks. Subsequently, interactions between 

these cell types are included as transporters that transfer metabolites 

between the different cell type models. This process results in a 

tissue-specific model that accounts for metabolic interactions between 

cells and the separation of metabolic functions into the relevant cell 

types. Thus they provide a more accurate view into tissue metabolism. 

Previous work had not taken these steps to generate multicellular 

tissue-specific models of this scale. 

This workflow was used to build three different multicellular models 

of brain energy metabolism. Each model represents one canonical 

neuron type (that is, glutamatergic, γ-aminobutarate (GABA)ergic 

or cholinergic), its interactions with the surrounding astrocytes and 

the transport of metabolites through the blood-brain barrier (Fig. 2). 

This reconstruction focuses on the core of cerebral energy metabolism, 

including central metabolism, mitochondrial metabolic pathways, and 

pathways relevant to anabolism and catabolism of the neurotransmitters 

glutamate, GABA and acetylcholine. Briefly, we began with the manually 

curated human metabolic reconstruction Recon 1 (ref. 7) (step 1). 

Next, we extracted relevant brain-specific reactions by mapping them 

to proteins expressed or localized to the brain as reported in the Human 

Protein Reference Database (release 5) 13 , H-inv (version 4.3) (HINV) 14 

or The Human Proteome Organisation (HUPO) brain proteome 

project 15 ; additional reactions were added as dictated by biochemical 

data from the literature (step 2). Reactions and pathways were manually 

Measured CMR of metabolites 

Endothelium/blood 

Genome 

Step 1: Build species-specific 

reconstruction 

Experimental 

data 

EC 

HIP 

SFG 

VCX 

Brain region 

Literature 

Add new 

knowledge 

Relative 

intensity 

Step 2: Find context-specific 

subnetwork 

m/z 

Step 3: Construct a 

manually curated model 

Compartmentalization 

and manual 

curation 

Formulation 

into model 

Step 4: Simulation and analysis for predictions and biological insight 

(i) Gene identification 

(ii) 'Omics data analysis 

(iii) Physiological data analysis 

GAD1 

GAD2 

Context-specific 

data 

Suppressed 

pathways 

Genes 

EC 

HIP 

MTG 

PC 

SFG 

VCX 

Down 

Up 

14 CO 2 release 

curated to verify their presence in the human brain, to determine celltype 

specificity and to add reactions unique to the different neuron 

types (step 3) (see Online Methods for complete details). Thus, the 

three models contain the high-flux pathways and important reactions 

in neuron and astrocyte metabolism. To our knowledge, these models 

currently represent the largest and most detailed models of brain 

energy metabolism 16–18 (Supplementary Notes). 

Our models of glutamatergic, GABAergic and cholinergic neurons 

contain 1,066, 1,067 and 1,070 compartment-specific interactions (that 

is, reactions, transformations and exchanges), respectively, involving 983, 

983 and 987 compartment-specific metabolites. In total, the three models 

are associated with 403 genes. Lists of reactions, genes, citations and 

parameters used to constrain the models are detailed in Supplementary 

Tables 1–5. The validity of these models is demonstrated through comparisons 

to physiological data. Specifically, our models predict ATP production 

rates within 8% of the average published value, and internal 

flux measurements are consistent with experimentally measured values 

(Supplementary Notes). Moreover, three analyses using the models are 

detailed here (step 4). Most of these analyses cannot be done on previous 

brain models or on Recon 1 (ref. 7) as published. Thus, our models can 

provide novel insight into brain energy metabolism. 

Identifying genes behind cell type–specific metabolic phenotypes 

Alzheimer’s disease is characterized by dementia and is diagnosed postmortem 

by histopathological features such as neurofibrillary tangles and 

β-amyloid plaques. Notably, metabolic rates of various brain regions 

decrease years before the onset of dementia 19 . Experiments have suggested 

that glutamatergic and cholinergic neurons are more affected in 

moderate stages of Alzheimer’s disease 20 , whereas most GABAergic cells 

ChAT flux 

Exp. data 

Simulation 

Astrocyte 

Mito 

Int 

Mito 

Glutamatergic 

neuron 

GABAergic 

neuron 

Cholinergic 

neuron 

Figure 2 General structure of the models. Three models were built from the 

brain reconstruction. Each model consists of various compartments including 

the endothelium/blood, astrocytes, astrocytic mitochondria, neurons, neuronal 

mitochondria and an interstitial space between the cell types. Each neuron 

metabolic network was tailored to represent a specific neuron type, containing 

genes and reactions generally accepted to be unique to the neuron type. Mito, 

mitochondrion; Int, interstitial space; CMR, cerebral metabolic rate. 

1280 VOLUME 28 NUMBER 12 DECEMBER 2010 nature biotechnology

a n a ly s i s 

are relatively unaffected until later stages 21 . Genes responsible for these 

differences are not known. Thus, we first analyzed the models to identify 

genes that potentially underlie these cell type–specific effects. 

Analysis of our models yielded results consistent with known 

metabolic changes in Alzheimer’s. Several central metabolic enzymes 

exhibit altered expression or activity in Alzheimer’s disease, such as 

pyruvate dehydrogenase (PDHm), α-ketoglutarate dehydrogenase 

(AKGDm) and cytochrome c oxidase 22–24 . The activities of these 

enzymes are affected by the Alzheimer’s disease–related proteins 

β-amyloid and Tau kinase 25,26 . In silico, as the activities of PDHm 

and cytochrome c oxidase decrease, neurons demonstrate impaired 

metabolic capacity (Supplementary Notes and Fig. 1a,b therein), and 

deficiencies in PDHm activity leads to a decreased cholinergic neurotransmission 

capacity (Supplementary Notes and Fig. 1c therein). 

The in vitro AKGDm activity shows the greatest impairment in brains 

from individuals with Alzheimer’s disease examined postmortem 

(57% decrease compared to normal) 23 . An in silico analysis shows that 

this deficiency impairs the metabolic rate in glutamatergic and cholinergic 

neurons (Fig. 3a) because it limits oxidative phosphorylation 

capacity in these neurons (Fig. 3b,c). Such impairment of oxidative 

phosphorylation leads to neuronal apoptosis 27 . However, oxidative 

phosphorylation is not impaired in the GABAergic neuron model 

(Fig. 3d), consistent with the different phenotypes of GABAergic 

compared with glutamatergic and cholinergic neurons. Therefore, 

the models were further interrogated to identify a mechanism that 

allows GABAergic neurons to absorb the perturbation, thereby leading 

to the cell type–specific effects. 

Simulations show that GABAergic neurons absorb the AKGDm 

perturbation through the GABA shunt (Fig. 3e,f), a pathway that 

uses 4-aminobutyrate transaminase and succinate-semialdehyde 


a 

b 

c 

d 

e 

0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 

Rate of cerebral CO 2 release (µmol min −1 g wet brain −1 ) 

0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 

0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 

0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 

Cytochrome c oxidase flux (µmol min −1 g wet brain −1 ) 

10 −4 10 −3 10 −2 10 −1 10 0 

GABA shunt flux (µmol min −1 g wet brain −1 ) 

Normal glutamatergic 

AKGDm inhib. glutamatergic 

Normal cholinergic 

AKGDm inhib. cholinergic 

Normal GABAergic 

AKGDm inhib. GABAergic 

g 

Normalized non-AD GAD NMN expression 

f 

coa[m] 

pyr[m] 

atp[m] 

coa[m] 

amp[m] 

ppi[m] 

CO [m] 2 

nad[m] nadh[m] 

PDHm 

1.0 

0.8 

0.6 

0.4 

0.2 

0 

ac[m] 

cit[m] ACSm 

CSm 

accoa[m] 

h[m] 

ACONTm 

TCA cycle 

nad[m] 

icit[m] 

coa[m] 

h[m] 

H 2 

O[m] 

oaa[m] 

mal-L[m] 

EC HIP MTG PC SFG VCX 

Brain region 

ICDHxm 

nadp[m] 

nadh[m] 

MDHm 

nad[m] 

FUMm 

ICDHyrm 

H 2 

O[m] 

fum[m] 

fadh 2 

[m] 

nadh[m] 

CO 2 

[m] 

nadph[m] 

glu-L[m] 

GABA 

shunt 

h[m] 

nadh[m] 

akg[m] 

CO 2 

[m] 

sucsal[m] 

fad[m] 

SUCD1m 

h 

Brain region 

succ[m] 

atp[m] 

EC 

HIP 

MTG 

PC 

SFG 

VCX 

nad[m] 

coa[m] 

4abut[m] 

ABTArm 

nad[m] 

h2o[m] 

SSALxm 

SUCOASm 

coa[m] 

GAD1 

adp[m] 

AKGDm 

nadh[m] 

succoa[m] 

GAD2 

pi[m] 

DLX 

co 2 

[m] 

1.5 

1.0 

0.5 

0 

−0.5 

−1.0 

−1.5 

1.0 

0.5 

0 

−0.5 

−1.0 

log 2 (Flux fold-change in AD) 

log 2 (GAD NMN expression change in AD) 

Figure 3 Decrease in AKGDm activity associated with Alzheimer’s disease (AD) shows cell-type and regional effects in silico consistent with 

experimental data. (a–e) Kernel density plots show the distribution of feasible fluxes for various reactions. An in silico reduction of AKGDm flux from 

normal activity (solid lines) to Alzheimer’s disease brain activity (dashed) decreases the oxidative metabolic rate for glutamatergic and cholinergic 

neurons, but not GABAergic neurons (a). This results from a decrease in the feasible fluxes for oxidative phosphorylation (e.g., cytochrome c oxidase) 

for both glutamatergic (b) and cholinergic neurons (c), but not GABAergic cells (d). (e,f) This cell-type-specific protection from the AKGDm deficiency 

results from an increased flux through the GABA shunt in GABAergic cells (e), by bypassing the damaged AKGDm (f). GABAergic cells maintain a higher 

GABA shunt flux because of the expression of glutamate decarboxylase (GAD). Neuroprotective properties of GAD are supported by gene expression. 

(g) Severely damaged brain regions in Alzheimer’s disease patients have lower GAD NMN expression in control brain, whereas high GAD NMN regions 

(SFG and VCX) show little damage. (h) In Alzheimer’s disease brain, severely affected regions (HIP and entorhinal cortex) show an increase in 

GAD NMN and the GAD-inducing DLX family, suggesting that non-GAD expressing neurons may be lost in Alzheimer’s disease. EC, entorhinal cortex; 

HIP, hippocampus; MTG, middle temporal gyrus; PC, posterior cingulate cortex; SFG, superior frontal gyrus; VCX, visual cortex; NMN, neuron marker 

normalized; inhib., inhibited. All reaction and metabolite abbreviations are defined in Supplementary Tables 1 and 2. 




dehydrogenase to bypass part of the tricarboxylic acid (TCA) cycle. 

However, our models suggest that glutamatergic and cholinergic neurons 

cannot, despite carrying a small flux through the shunt enzymes 

(Fig. 3e). Support for these results includes recent evidence that suggests 

that cerebellar granule neurons, which have higher levels of 

GABA, can absorb perturbations to AKGDm through this shunt 28 . 

To identify the mechanism allowing only GABAergic neurons to 

use the GABA shunt to absorb the AKGDm perturbation, an in silico 

analysis was performed to identify contributing genes. Briefly, each 

reaction was removed from the GABAergic model, followed by an 

assessment of the correlation of flux between AKGDm and oxidative 

phosphorylation (Supplementary Notes). This analysis suggested 

that the two isoforms of glutamate decarboxylase (GAD) could provide 

the cell type–specific neuroprotection. 

GAD allows the GABA shunt to carry a higher flux following the 

AKGDm perturbation in GABAergic neurons; however, the lack of 

GAD in other neuron types greatly limits the use of the GABA shunt 

in silico (Fig. 3e). Therefore, by fueling the GABA shunt, GAD may 

play a neuroprotective function, thus sparing of GABAergic systems 

in earlier Alzheimer’s disease 21 . 

Certain populations of glutamatergic and cholinergic cells tend to 

be lost earlier in Alzheimer’s disease but others survive. Interestingly, 

although GAD is canonically encoded by a GABAergic gene, it occasionally 

shows low expression in other neuron types, including 

glutamatergic and cholinergic cells 29 . Therefore, such populations of 

non-GABAergic, GAD-expressing neurons would also be protected. 

Thus, we follow with an analysis of the correlation of GAD expression 

and Alzheimer’s disease pathology for further validation. 

If GAD has neuroprotective capacity in vivo, we would expect that 

brain regions with less GAD per neuron will be more affected in 

Alzheimer’s disease, whereas regions with abundant GAD will be 

spared. To test this hypothesis, we used a compendium of published 

microarrays of neurons without tangles from six brain regions in 

Alzheimer’s patients and age-matched, non-Alzheimer’s controls 30 . 

In controls, GAD expression levels among the brain regions is consistent 

with the extent of neuron loss found in Alzheimer’s disease 

patients; that is, brain regions with more neuron loss in Alzheimer’s 

disease (e.g., the entorhinal cortex and hippocampus) have lower 

GAD expression in control patients, whereas relatively unaffected 

regions in Alzheimer’s disease (e.g., superior frontal gyrus and visual 

cortex) show much higher levels of GAD expression (Fig. 3g). 

In addition, if GAD is neuroprotective, neurons with low GAD 

expression should be lost in Alzheimer’s disease, resulting in an 

increase in expression of GAD per neuron in histopathologically 

affected regions. In microarrays, we observed a significant increase 

in the expression of the brain-specific GAD2 in the entorhinal cortex 

and hippocampus in Alzheimer’s disease (P = 0.0050 and 0.018, 

respectively; SAM test) (Fig. 3h). Expression of all other neuronspecific 

genes were tested as controls and showed no correlation with 

Alzheimer’s disease pathology (Supplementary Notes and Fig. 2 

therein), except the genes encoding DLX2 and DLX5, which induce 

GAD expression in the brain (Fig. 3h and Supplementary Notes and 

Fig. 3 therein) 31 . Therefore, these results lend additional support to 

the possibility that GAD is providing a neuroprotective effect and 

that this effect is correlated with the regional specificity of Alzheimer’s 

disease. Moreover, the model was able to guide the identification of a 

gene and the mechanism for its role in Alzheimer’s disease. 

Pathway-based analysis of multicellular models 

Atrophy and the cell type–specific effects investigated above cannot 

fully explain the decreased metabolic rate in Alzheimer’s disease in 

many brain regions 32 . Therefore, using the models as a context for analyzing 

gene expression data from Alzheimer’s patients and age-matched 

controls, we searched for downregulated metabolic pathways within 

surviving cells in metabolically suppressed brain regions. This was done 

using PathWave 33 , a method that takes each model and maps microarray 

data to the metabolic pathways, which have been optimally arranged on 

a two-dimensional grid. Haar Wavelet transforms detected concerted 

regulation of neighboring enzymes in the models. Thus, we could identify 

groups of differentially expressed genes that are not solely related 

by annotation but that are connected by metabolic pathway functions, 

thereby providing a mechanistic view of the effects of differential gene 

expression (see Online Methods and Supplementary Notes). 

This analysis revealed that brain regions showed distinct changes 

in metabolic pathways (Supplementary Table 6). The visual cortex 

and superior frontal gyrus lack any differentially expressed pathways, 

consistent with previous work that shows little change in metabolic 

rate in Alzheimer’s disease in these regions 30 . However, the posterior 

cingulate cortex (PC) and middle temporal gyrus (MTG) have the 

most significantly differentially expressed pathways (23 and 18, respectively). 

These two regions show significantly decreased metabolic rates 

in Alzheimer’s disease but show fewer histopathological effects 30 . Both 

the entorhinal cortex and hippocampus also show decreases in expression 

of nine metabolic pathways, though the number may be lower 

since these regions suffer a high amount of neuron loss and only histopathologically 

healthy neurons were expression profiled. Therefore, 

more affected neurons may already have been lost or not profiled. 

PathWave also revealed that the four brain regions showing substantially 

lower metabolic rates in Alzheimer’s disease (PC, MTG, 

hippocampus and entorhinal cortex) 32 also show a significant suppression 

of glycolysis and the TCA cycle (P < 5 × 10 −4 ) (Fig. 4). In 

addition, the hippocampus, MTG and PC show a suppression of the 

malate-aspartate shuttle and oxidative phosphorylation. Individual 

regions also show a suppression of other pathways, such as heme 

biosynthesis (MTG, PC), ethanol metabolism (entorhinal cortex, PC) 

and several amino acid metabolism pathways. Thus, using pathway 

topology of our models as a context for microarray analysis, we find 

that the decreased metabolic rate in specific regions in Alzheimer’s 

disease is associated with the downregulation of central metabolic 

gene expression in histopathologically normal neurons. 

Identifying metabolic properties relevant to treatment 

Deficiencies in cholinergic neurotransmission have long been believed 

to contribute to Alzheimer’s disease. Thus, treatment of Alzheimer’s 

patients often includes efforts to enhance the cholinergic system. 

Although relevant to treatment, it is still not clearly understood which 

pathways are used to synthesize acetylcholine and what percentage of 

the brain participates in cholinergic neurotransmission. 

Studies have demonstrated that cytosolic acetyl-CoA, which is 

used to synthesize the neurotransmitter acetylcholine, comes from 

the acetyl-CoA formed in the mitochondria. This tight coupling of 

acetylcholine to mitochondrial metabolism allows treatments that 

increase glucose uptake in the brain to improve cognitive functions 

in rats 34 and humans with severe cholinergic cognitive pathologies, 

such as Alzheimer’s disease and trisomy 21 (ref. 35). Pathways that 

transport acetyl-CoA carbon to the cytosol have been suggested; however, 

the mechanism is still not clear 36 . 

Constraint-based modeling helped identify two pathways that could 

indirectly transport acetyl-CoA into the cytosol and provided insight into 

needed complementary pathways. To identify these pathways, reaction 

sets were identified by randomly removing reactions from Recon 1 until a 

minimum set was determined that couples the mitochondrial and cytosolic 




a 

GLCt2r 

G3PD2m 

HEX1 

PGI 

PFK 

TPI 

FBA 

GAPD 

PGK 

ACYP 

EC HIP MTG PC SFG VCX EC HIP MTG PC SFG VCX 

ACYP 

ABTArm 

DPGM 

ACONTm 

DPGase 

AKGDm 

ENO 

PDHm 

CSm 

PCm 

FBA 

FUMm 

G3PD2m 

CSm 

GLUDxm 

GAPD 

ACONTm 

GLUDym 

GLCt1r 

MDHm 

ICDHxm 

GLCt2r 

ICDHxm ICDHyrm 

HEX1 

ICDHyrm 

MALtm 

PFK 

MDHm 

PGI 

SUCD1m 

ME2m 

PGK 

PDHm 

PGM 

SSALxm 

PYK 

SSALxm ABTArm 

SUCCt2m 

TPI 

SUCOAS1m 

SUCD1m 

Regulation of surrounding reactions SUCOASm 

SUCOAS1m 

Down Up 

AKGDm 

SUCOASm 

PGM ENO PYK 

>6 3 0 3 >6 

Regulation of surrounding reactions 

–log 10 (P value) 

GLUDxm GLUDym 

Down Up 

acetyl-CoA pools. This was repeated until more than 21,000 unique minimal 

reaction sets were identified. Singular value decomposition was then 

used to identify dominant sets of reactions that frequently co-occur. 

The first singular vector is dominated by reactions that frequently 

co-occur in the reaction sets (e.g., water transport across cell membranes). 

However, the second and third singular vectors are dominated 

by reactions that usually co-occur or never co-occur (Fig. 5). 

These reactions cluster into three distinct pathways, providing 

hypotheses of pathways coupling acetylcholine synthesis and mitochondrial 

metabolism. After re-examining the literature and the 

’omics data used in the reconstruction, we were able to eliminate 

the pathway using cytosolic acetyl-CoA synthetase (Fig. 5a) and 

b 

MALtm 

ME2m 

SUCCt2m 

FUMm 

ME1m 

>6 3 0 3 >6 

–log 10 (P value) 

Figure 4 Analysis of metabolic pathways in different regions of brains affected by Alzheimer’s. (a,b) Pathway analysis demonstrates that 

histopathologically normal cells from regions of the brain that are metabolically affected (EC, hippocampus, MTG, and PC) demonstrate a significant 

suppression of central metabolic pathways, such as glycolysis (a) and the TCA cycle and surrounding reactions (b). Regions that are metabolically less 

affected (SFG and VCX) show no significant suppression. Reaction suppression shown here is a composite expression of the reaction-associated genes 

and the genes of closely connected reactions. Only significantly changed reactions are shown (FDR = 0.05). EC, entorhinal cortex, HIP, hippocampus, 

MTG, middle temporal gyrus, PC, posterior cingulate cortex, SFG, superior frontal gyrus, VCX, visual cortex. All reaction and metabolite abbreviations 

are defined in Supplementary Tables 1 and 2. 

validate pathways using ATP-citrate lyase (ACITL) or cytosolic 

acetyl-CoA C-acetyltransferase (ACACT1r), which transport 

acetyl-CoA from the mitochondria to the cytosol on citrate or 

acetoacetate, respectively (Fig. 5b,c and Supplementary Notes). 

These two pathways were included in our cholinergic model, contributing 

to a correlation between the flux through mitochondrial 

pyruvate dehydrogenase and choline acetyltransferase (r = 0.45, 

P = 3 × 10 −247 ), consistent with the experimentally observed coupling 

between mitochondrial metabolism and acetylcholine production. 

ACITL and ACACT1r also correlate with choline acetyltransferase 

flux (P < 3 × 10 −90 ). Moreover, it has been reported that the inhibition 

of ACITL reduces the acetylcholine production rate by 30% 36 . 

Third singular vector 

0.4 

0.3 

0.2 

0.1 

a 

ASPGLUm 

b 

ACS 

ASPNATm 

GLUt2m NACASPtm 

NACASPAH 

AKGMALtm 

0 

ASPTAm 

ASPTA 

SUCOASm 

cHMGLm/HMGCOASim 

−0.1 CSm 

OCOAT1m ACACT1rm 

ACITL 

CITtam 

ACACT1r AACOAT 

ACACt2m 

−0.2 

−0.3 −0.2 −0.1 0 0.1 0.2 0.3 0.4 

Second singular vector 

accoa 

d e 

accoa 

f 

accoa Mitochondrion 

H2O oaa 

asp-L 

h coa H2O ACACT1rm 

asp-L 

ASPTAm 

coa 

CSm 

HMGCOASim 

aacoa 

ASPNATm 

atp 

coa glu-L akg 

hmgcoa 

succ coa 

coa 

cit h 

HMGLm OCOAT1m SUCOASm 

h glu-L 

akg 

acac 

adp 

Nacasp 

h CITtam 

mal-L 

h glu-L 

h 

pi 

succoa 

NACASPtm ASPGLUm h GLUt2m 

AKGMALtm 

h 

Nacasp 

ACACt2m 

mal-L 

h glu-L 

acac atp 

glu-L 

akg 

coa 

NACASPAH H2O 

ASPGLUm 

cit 

AACOAT 

asp-L 

coa 

amp 

ac atp 

glu-L akg 

ppi 

atp 

coa 

aacoa 

ACITL 

ACS 

adp 

coa 

amp 

ppi 

pi oaa 

ASPTA 

ACACT1r 

asp-L 

accoa 

accoa 

accoa 

Cytosol 

Figure 5 Singular value decomposition (SVD) of feasible pathways elucidates potential pathways that allow for coupling of mitochondria acetyl-CoA 

metabolism and cytosolic acetylcholine production. We computed 21,000 unique feasible reaction sets, each showing transport of mitochondrial acetyl-CoA 

carbon to the cytosol in human metabolism. (a–c) Dominant loading from the SVD of a matrix of all 21,000 pathways represented the reactions making 

up three primary pathways (d–f) that allow this coupling of mitochondrial metabolism to acetylcholine production, by carrying the acetyl-CoA carbon on 

N-acetyl-l-aspartate (a,d), citrate (b,e) or acetoacetate (c,f). As shown by the second singular vector, reactions in the pathway with citrate tend to be missing 

from pathways when the reactions for the acetoacetate pathway are included. The third singular vector shows a similar relationship of the N-acetyl-laspartate 

pathway. The ’omics data and known enzyme localization only support the usage of citrate and acetoacetate as potential carriers in neurons. 



a 

Cholinergic percentage of brain 

5% 

4% 

3% 

2% 

1% 

[1− 14 C] 

[2− 14 C] 

Carbon label 

position 

b c d 

ChAT flux 

(µmol * g wet brain −1 * min −1 ) 

4.0 × 10−3 × 10 −3 × 10 −3 

2.2 

Exp. data 

0 mM 

3.5 Simulation 

2.6 

3.0 

1.8 

2.2 

2.5 

0.1 mM 

2.0 

0.2 mM 

1.8 

1.4 

1.5 

0.3 mM 

1.0 

0.4 mM 

1.4 

1.0 

Exp. data 

0.5 1.0 mM 

Bromopyruvate 1.0 

0.6 

Simulation 

0.5 mM 

0 

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.42 0.46 0.50 0.54 0.58 

0.14 0.18 0.22 0.26 0.30 

14 CO 2 release (µmol * g wet brain −1 * min −1 ) 

ChAT flux 


2-oxo-3-methyl 

butanoate 


pentanoic acid 

Leucine 


ChAT flux 



pentanoic acid 


butanoate 

Leucine 



Figure 6 Model-aided prediction of cholinergic contribution is consistent with experimental acetylcholine production. Percent brain cholinergic 

neurotransmission was predicted based on 14 sets of experimental data in which brain minces were fed [1- 14 C]-pyruvate or [2- 14 C]-pyruvate, followed 

by measurement of 14 C-labeled CO 2 and acetylcholine. (a) For each experiment, the feasible amount of the brain that can generate the experimental 

response was computed, centering at 3.3%. (b) This parameter was used in the analysis, and the updated model predictions were consistent with 

experimental data, such as seen in the case of treating the brain minces with [1- 14 C]-pyruvate and increasing levels of the pyruvate-dehydrogenase 

inhibitor bromopyruvate. (c,d) Moreover, the updated model predictions were consistent with measured 14 C-labeled CO 2 and acetylcholine production 

for brain minces that were treated with three PDHm inhibitors withheld from previous computations for both supplementation with [1- 14 C]-pyruvate 

(c) and [2- 14 C]-pyruvate (d). Error bars on the simulation results represent 25 th and 75 th percentiles. ChAT, choline acetyltransferase. 

The in silico inhibition of ACITL reduces acetylcholine production 

by 7.3%. It is possible that the in silico decrease is smaller because the 

model can immediately adapt to the perturbation, whereas in vivo 

regulatory responses would take time to adapt. Notably, the in silico 

inhibition of ACACT1r reduces acetylcholine production by 39%. 

Thus, cholinergic neurotransmission depends on redundant pathways, 

and acetoacetate may play a more dominant role in transporting 

mitochondrial acetyl-CoA to the cytosol. 

Our analysis provides further insight into the coupling of mitochondrial 

metabolism to acetylcholine synthesis, which aids in the 

treatment of cholinergic disorders. However, knowledge of the abundance 

of cholinergic neurotransmission also aids in this purpose. It is 

difficult to identify cholinergic neurons, based solely on cell morphology, 

because cholinesterases and immunohistochemical markers for 

cholinergic neurons are also found in noncholinergic neurons and 

other tissues 37 . Therefore, it is unknown what percentage of all neurotransmission 

is cholinergic. 

Using our cholinergic model, we compute the percent contribution 

of cholinergic neurotransmission based on published data 38 . The 

data were obtained from rat brain minces, incubated in solutions 

containing [1- 14 C]pyruvate or [2- 14 C]pyruvate. Both acetylcholine 

and radiolabeled CO 2 were measured at various titrations of several 

pyruvate dehydrogenase inhibitors. 

The cholinergic model was subjected to similar levels of pyruvate 

dehydrogenase inhibition. The simulations successfully reproduced 

the experimental linear relationship between acetylcholine production 

and metabolic rate, and acetylcholine production was correlated 

with CO 2 release (r = 0.68). 

The fraction of cholinergic neurotransmission for the brain was 

computed by randomly choosing points from both the distributions 

of experimental data and distributions predicted by the simulations. 

A scaling factor was subsequently found that reconciles the two. This 

was repeated for 14 different combinations of pyruvate labeling and 

pyruvate dehydrogenase inhibitors 38 , yielding a median predicted 

cholinergic portion of total brain neurotransmission of 3.3% (Fig. 6a). 

After adding this new parameter to the model, the predictions correspond 

well with experimental data sets (Fig. 6b), including data 

representing three pyruvate dehydrogenase inhibitors withheld from 

the previous computations (Fig. 6c,d). Thus, the model was used in 

conjunction with experimental data to gain insight into physiological 

observations and derive physiological parameters, which are dependent 

on systems-level activity and are relevant to treatment. 

DISCUSSION 

In this study we presented a workflow for generating tissue-specific, 

multicellular metabolic models. Through the analysis and integration 

of ’omics data, followed by manual curation, we used this workflow to 

build a first-draft, manually curated, multicellular, metabolic reconstruction 

of brain energy metabolism. Three models were generated 

from this reconstruction, representing different types of neurons coupled 

to astrocytes. We used these models in three distinct analyses, 

in which we made predictions and gained systems-level insights into 

Alzheimer’s disease and cholinergic neurotransmission. 

As experimental methods and data resolution improve, the accuracy 

of these models and their predictions should also improve. 

Improvements in neuroimaging and metabolomics will allow for 

more precise quantification of metabolite flow through the bloodbrain 

barrier, which is of interest because dysfunction of this system 

accompanies many neurological disorders and injuries 39 . In addition, 

improvements in transcriptomics and proteomics will provide higherresolution 

quantification of cell- and organelle-specific genes and 

proteins. These data will allow models to account for neuron groups 

in specific brain regions, subcellular heterogeneity within cells and the 

inclusion of less abundant glial cells. For example, higher-resolution 

models may provide insight into metabolic changes in specific cell 

populations, such as the structures closely related to the olfactory 

system, which are affected early on in Alzheimer’s disease 40 . 

Insight into mammalian tissue-specific metabolism may be gained 

as more multicellular models are constructed. Our models demonstrate 

metabolic coupling and synergistic activities that more coarse-grained 

models miss, as the three analyses presented here were not possible 

using Recon 1 or the previous brain metabolism models. The compartmentalization 

of metabolic processes within cells 41 , between cells 42 

and in host-pathogen interactions 43 has an important role in normal 

physiology. Therefore, such models may provide greater insight and 

more accurately predict both true cellular functions and responses to 

medical interventions 44 . This insight will be attained as many analytical 

methods, in addition to those demonstrated in this work, will be 




developed and deployed in multicellular reconstructions ranging from 

constraint-based analyses 5 to topological studies 45 . 

This study serves as an example of how mechanistic relationships between 

genotype and phenotype can be built through the difficult task of multi-omic 

data integration 46 . From the genotype one can begin to reconstruct the network 

for an organism. The integration of high-throughput data and careful 

manual curation can add context-specific mechanistic network structure 

to genomics information. Thus, this network becomes a representation 

of complex genetic interactions and biochemical mechanisms underlying 

observed phenotypes. This complex, but mechanistic, relationship between 

the genotype and phenotype can be used as a foundational structure upon 

which additional high-throughput data can be analyzed and predictive 

simulations can be conducted, thus leading to improved understanding, 

testable hypotheses and increased knowledge 1,2,44 . 

Methods 

Methods and any associated references are available in the online version 

of the paper at http://www.nature.com/naturebiotechnology/. 

Note: Supplementary information is available on the Nature Biotechnology website. 


The authors thank G. Gibson at Cornell University, I. Thiele at the University of 

Iceland and M. Abrams, M. Mo and C. Barrett at UCSD for suggestions pertaining to 

this work. This work was funded in part by a Fulbright fellowship, a National Science 

Foundation IGERT Plant Systems Biology training grant (no. DGE-0504645), 

US National Institutes of Health grants 2R01GM068837_05A1 and RO1 GM071808 

and the Helmholtz Alliance on Systems Biology and the BMBF by the NGFN+ 

neuroblastoma project ENGINE. 


N.E.L., J.K.C., A.Y., N.P., M.P.A. and B.O.P. conceived and designed the model. 

N.E.L., J.K.C., G.S., R.K., R.E., J.S., A.B. and R.A.L. performed data analyses. The 

manuscript was written by N.E.L., G.S., J.S., A.B. and B.O.P. 



Published online at http://www.nature.com/naturebiotechnology/. 

Reprints and permissions information is available online at http://npg.nature.com/ 

reprintsandpermissions/. 

1. Feist, A.M. & Palsson, B.O. The growing scope of applications of genome-scale metabolic 

reconstructions using Escherichia coli. Nat. Biotechnol. 26, 659–667 (2008). 

2. Oberhardt, M.A., Palsson, B.O. & Papin, J.A. Applications of genome-scale metabolic 

reconstructions. Mol. Syst. Biol. 5, 320 (2009). 

3. Breitling, R., Vitkup, D. & Barrett, M.P. New surveyor tools for charting microbial 

metabolic maps. Nat. Rev. Microbiol. 6, 156–161 (2008). 

4. Thiele, I. & Palsson, B.O. A protocol for generating a high-quality genome-scale 

metabolic reconstruction. Nat. Protoc. 5, 93–121 (2010). 

5. Lewis, N.E., Jamshidi, N., Thiele, I. & Palsson, B.O. Metabolic systems biology. in 

Encyclopedia of Complexity and Systems Science (ed. Meyers, R.A.) 5535–5552 

(Springer, New York, 2009). 

6. Orth, J.D., Thiele, I. & Palsson, B.O. What is flux balance analysis? Nat. Biotechnol. 

28, 245–248 (2010). 

7. Duarte, N.C. et al. Global reconstruction of the human metabolic network based on 

genomic and bibliomic data. Proc. Natl. Acad. Sci. USA 104, 1777–1782 (2007). 

8. Becker, S.A. & Palsson, B.O. Context-specific metabolic networks are consistent 

with experiments. PLOS Comput. Biol. 4, e1000082 (2008). 

9. Shlomi, T., Cabili, M.N., Herrgard, M.J., Palsson, B.O. & Ruppin, E. Network-based 

prediction of human tissue–specific metabolism. Nat. Biotechnol. 26, 1003–1010 

(2008). 

10. Jerby, L., Shlomi, T. & Ruppin, E. Computational reconstruction of tissue-specific 

metabolic models: application to human liver metabolism. Mol. Syst. Biol. 6, 401 

(2010). 

11. Ponten, F. et al. A global view of protein expression in human cells, tissues, and 

organs. Mol. Syst. Biol. 5, 337 (2009). 

12. Palsson, B.O. in Systems Biology: Properties of Reconstructed Networks 322 

(Cambridge University Press, Cambridge/New York, 2006). 

13. Mishra, G.R. et al. Human protein reference database–2006 update. Nucleic Acids 

Res. 34, D411–D414 (2006). 

14. Fujii, Y., Imanishi, T. & Gojobori, T. H-Invitational Database: integrated database 

of human genes. Tanpakushitsu Kakusan Koso 49, 1937–1943 (2004). 

15. Reidegeld, K.A. et al. The power of cooperative investigation: summary and 

comparison of the HUPO Brain Proteome Project pilot study results. Proteomics 6, 

4997–5014 (2006). 

16. Chatziioannou, A., Palaiologos, G. & Kolisis, F.N. Metabolic flux analysis as a tool 

for the elucidation of the metabolism of neurotransmitter glutamate. Metab. Eng. 

5, 201–210 (2003). 

17. Cakir, T., Alsan, S., Saybasili, H., Akin, A. & Ulgen, K.O. Reconstruction and flux 

analysis of coupling between metabolic pathways of astrocytes and neurons: 

application to cerebral hypoxia. Theor. Biol. Med. Model. 4, 48 (2007). 

18. Occhipinti, R., Puchowicz, M.A., LaManna, J.C., Somersalo, E. & Calvetti, D. 

Statistical analysis of metabolic pathways of brain metabolism at steady state. Ann. 

Biomed. Eng. 35, 886–902 (2007). 

19. Reiman, E.M. et al. Functional brain abnormalities in young adults at genetic risk for 

late-onset Alzheimer’s dementia. Proc. Natl. Acad. Sci. USA 101, 284–289 (2004). 

20. Ginsberg, S.D., Che, S., Counts, S.E. & Mufson, E.J. Single cell gene expression 

profiling in Alzheimer’s disease. NeuroRx 3, 302–318 (2006). 

21. Lai, M.K.P., Ramirez, M.J., Tsang, S.W.Y. & Francis, P.T. Alzheimer’s disease as a 

neurotransmitter disease in Neurobiology of Alzheimer’ Disease (eds. Dawbarn, D. 

& Allen, S.J.) 245–281 (Oxford University Press, New York, 2007). 

22. Fukui, H., Diaz, F., Garcia, S. & Moraes, C.T. Cytochrome c oxidase deficiency in 

neurons decreases both oxidative stress and amyloid formation in a mouse model 

of Alzheimer’s disease. Proc. Natl. Acad. Sci. USA 104, 14163–14168 (2007). 

23. Bubber, P., Haroutunian, V., Fisch, G., Blass, J.P. & Gibson, G.E. Mitochondrial 

abnormalities in Alzheimer brain: mechanistic implications. Ann. Neurol. 57, 

695–703 (2005). 

24. Gibson, G.E. et al. Alpha-ketoglutarate dehydrogenase in Alzheimer brains bearing 

the APP670/671 mutation. Ann. Neurol. 44, 676–681 (1998). 

25. Casley, C.S., Canevari, L., Land, J.M., Clark, J.B. & Sharpe, M.A. Beta-amyloid 

inhibits integrated mitochondrial respiration and key enzyme activities. 

J. Neurochem. 80, 91–100 (2002). 

26. Hoshi, M. et al. Regulation of mitochondrial pyruvate dehydrogenase activity by tau 

protein kinase I/glycogen synthase kinase 3beta in brain. Proc. Natl. Acad. Sci. 

USA 93, 2719–2723 (1996). 

27. Gorman, A.M., Ceccatelli, S. & Orrenius, S. Role of mitochondria in neuronal 

apoptosis. Dev. Neurosci. 22, 348–358 (2000). 

28. Santos, S.S. et al. Inhibitors of the alpha-ketoglutarate dehydrogenase complex 

alter [1–13C]glucose and [U-13C]glutamate metabolism in cerebellar granule 

neurons. J. Neurosci. Res. 83, 450–458 (2006). 

29. Hassel, B., Johannessen, C.U., Sonnewald, U. & Fonnum, F. Quantification of the 

GABA shunt and the importance of the GABA shunt versus the 2-oxoglutarate 

dehydrogenase pathway in GABAergic neurons. J. Neurochem. 71, 1511–1518 (1998). 

30. Liang, W.S. et al. Altered neuronal gene expression in brain regions differentially 

affected by Alzheimer’s disease: a reference data set. Physiol. Genomics 33, 240–256 (2008). 

31. Stuhmer, T., Anderson, S.A., Ekker, M. & Rubenstein, J.L. Ectopic expression of 

the Dlx genes induces glutamic acid decarboxylase and Dlx expression. Development 

129, 245–252 (2002). 

32. Ibanez, V. et al. Regional glucose metabolic abnormalities are not the result of 

atrophy in Alzheimer’s disease. Neurology 50, 1585–1593 (1998). 

33. Schramm, G. et al. PathWave: discovering patterns of differentially regulated 

enzymes in metabolic pathways. Bioinformatics 26, 1225–1231 (2010). 

34. Ragozzino, M.E., Pal, S.N., Unick, K., Stefani, M.R. & Gold, P.E. Modulation of 

hippocampal acetylcholine release and spontaneous alternation scores by 

intrahippocampal glucose injections. J. Neurosci. 18, 1595–1601 (1998). 

35. Watson, G.S. & Craft, S. Modulation of memory by insulin and glucose: 

neuropsychological observations in Alzheimer’s disease. Eur. J. Pharmacol. 490, 

97–113 (2004). 

36. Cooper, J.R. Unsolved problems in the cholinergic nervous system. J. Neurochem. 

63, 395–399 (1994). 

37. Karczmar, A.G. Cholinergic cells and pathways in Exploring the Vertebrate Cholinergic 

Nervous System 686 (Springer, New York, 2006). 

38. Gibson, G.E., Jope, R. & Blass, J.P. Decreased synthesis of acetylcholine 

accompanying impaired oxidation of pyruvic acid in rat brain minces. Biochem. J. 

148, 17–23 (1975). 

39. Abbott, N.J., Ronnback, L. & Hansson, E. Astrocyte-endothelial interactions at the 

blood-brain barrier. Nat. Rev. Neurosci. 7, 41–53 (2006). 

40. Thompson, M.D., Knee, K. & Golden, C.J. Olfaction in persons with Alzheimer’s 

disease. Neuropsychol. Rev. 8, 11–23 (1998). 

41. Schryer, D.W., Peterson, P., Paalme, T. & Vendelin, M. Bidirectionality and 

compartmentation of metabolic fluxes are revealed in the dynamics of isotopomer 

networks. Int. J. Mol. Sci. 10, 1697–1718 (2009). 

42. Serres, S., Raffard, G., Franconi, J.M. & Merle, M. Close coupling between astrocytic 

and neuronal metabolisms to fulfill anaplerotic and energy needs in the rat brain. 

J. Cereb. Blood Flow Metab. 28, 712–724 (2008). 

43. Bordbar, A. et al. Insight into human alveolar macrophage and M. tuberculosis 

interactions via metabolic reconstructions. Mol. Syst. Biol. 6, 422 (2010). 

44. Chang, R.L., Xie, L., Xie, L., Bourne, P.E. & Palsson, B.O. Drug off-target effects 

predicted using structural analysis in the context of a metabolic network model. 

PLoS. Comput. Biol. 6, e1000938 (2010). 

45. Lee, D.S. et al. The implications of human metabolic network topology for disease 

comorbidity. Proc. Natl. Acad. Sci. USA 105, 9880–9885 (2008). 

46. Palsson, B.O. & Zengler, K. The challenges of integrating multi-omics data sets. 

Nat. Chem. Biol. 6, 787–789 (2010). 



ONLINE METHODS 

Reconstruction of iNL403. This work focuses on the core of cerebral energy 

metabolism and the pathways that play a critical role in cell type–specific 

functions in the brain. The pathways in this work include mitochondrial metabolic 

pathways, central metabolic pathways closely tied to mitochondrial 

function and additional pathways that are needed for modeling neuron and 

astrocyte functions. To reconstruct these pathways, a list of known human 

mitochondrial, glycolytic and transport reactions were extracted from the 

manually curated human metabolic reconstruction Recon 1 (ref. 7). From 

this list, reactions were directly added to the brain reconstruction if brainlocalized 

protein or gene expression was suggested by the Human Protein 

Reference Database (release 5) 13 or HINV 14 , both of which provide tissue 

expression presence calls for each gene. Proteomics data from live human 

brain, acquired for the HUPO brain proteome project 15 , were also used (see 

Supplementary Table 7 for accession numbers). Additional reactions were 

added as dictated by biochemical data from the literature (Supplementary 

Table 1). This reconstruction is designated iNL403 because it contains 

403 genes. Reactions and pathways were manually curated to verify their presence 

in the human brain and to determine cell-type localization, thus yielding 

a first-draft metabolic reconstruction of the brain metabolic network. 

Reactions unique to the different neuron types were determined from the 

literature (see notes in Supplementary Table 1), and consist largely of the 

reactions needed to make and metabolize their associated neurotransmitters. 

A list of all reactions, supporting data, citations and a comparison with previous 

brain metabolism models can be found in Supplementary Tables 1–4 and 8. 

Three SBML models representing the union of reactions from the neuron-type 

specific models can be found as Supplementary Models 1–3. Neuron-type 

specific models in SBML format and model updates can be obtained from 

http://systemsbiology.ucsd.edu/In_Silico_Organisms/Brain. 

Constraint-based modeling. Constraint-based modeling and analysis of 

metabolic networks have been previously described 5,12 . Briefly, all of the 

reactions are described mathematically by a stoichiometric matrix, S, of size 

m × n, where m is the number of metabolites and n is the number of reactions, 

and each element is the stoichiometric coefficient of the metabolite in 

the corresponding reaction. The mass balance equations at steady state are 

represented as 

S • v = 0, 

where v is the flux vector 12 . Maximum and minimum fluxes and reaction 

reversibility, when known, are placed on each reaction, further constraining 

the system as follows 

vmin ≤ v ≤ vmax . 

(1) 

(2) 

At this point the model can then be used with many constraint-based methods 5 

to study network characteristics. 

The S matrix was constructed with the mass and charge balanced reactions 

from the reconstruction. Select metabolites, known to cross the blood-brain 

barrier, were added as exchange reactions, allowing those metabolites to leave 

or enter the extracellular space in the model. A few metabolites from network 

gaps were allowed to enter or leave the system from the cytosol or mitochondria. 

This was only done when transporter mechanisms or subsequent pathway 

steps were not known, and when their entrance or removal from the system 

was necessary for model function. When available, cerebral metabolic rates 

were used from published data to constrain the upper and lower bounds of the 

exchange reactions 45,46 . All parameters are detailed in Supplementary Tables 1 

and 5. Parameters derived from rat data were varied to demonstrate that results 

presented here were robust and therefore relevant to human metabolism, and 

constraint-based methods were further employed to assess and compare the 

model to metabolic functions of the brain (Supplementary Notes). 

Monte Carlo sampling. Monte Carlo sampling was used to generate a set of 

feasible flux distributions (points). The method is based on the artificially 

centered hit-and-run algorithm with slight modifications. Initially, a set of 

nonuniform pseudo-random points, called warm-up points, is generated. In a 

series of iterations, each point is randomly moved, always remaining within the 

feasible flux space. This is done by (i) choosing a random direction, (ii) computing 

the limits of how far one can travel in that direction and (iii) choosing 

a new random point along this line. After many iterations, the set of points is 

mixed and approaches a uniform sample of the solution space, thus providing 

a distribution for each reaction that represents the range and probability of 

the flux for each reaction, given the network topology and model constraints. 

For more detail, see the Supplementary Notes. 

Simulating enzyme deficiencies. Enzyme deficiencies were obtained from the 

literature 23 . To simulate each deficiency, the distribution for all candidate flux 

states was determined using Monte Carlo sampling. From this distribution, the 

most probable flux was found and the reaction upper bound was reduced by the 

fraction reported in the literature. All candidate states were then recomputed 

and compared with normal candidate flux states. 

Alzheimer’s disease microarray analysis. Microarrays were obtained from 

the Gene Expression Omnibus (GSE5281). Arrays consist of 161 Affymetrix 

Human Genome U133 Plus 2.0 Arrays that profile the gene expression from 

laser-capture, microdissected, histopathologically normal neurons from six 

different brain regions of Alzheimer’s disease patients and age-matched controls. 

These arrays were not used in model construction. 

Arrays were normalized using the GC Robust Multi-array Average (gcrma) 

normalization function in the bioconductor package for R. Pearson’s correlation 

coefficients were computed for all array pairs, and arrays with r < 0.8 were 

discarded (that is, GSM119643, GSM119661, GSM119666 and GSM119676). 

Different arrays had different levels of glial contamination. Therefore, to 

assess the amount of GAD (neuron-specific), the GAD1 and GAD2 levels on 

each array were normalized as follows. For each array, the relative amount of 

neuron material was determined by computing a ratio for four neuron-specific 

genes to the median level across all arrays. Neuron-specific genes were chosen to 

represent different neuron parts, including the soma, axon and synaptic bouton 

(TUBB3 (ref. 47), NeuN 48 , SYN1 (ref. 49) and ACTL6B 50 ). These were summed 

to compute a relative amount of neuron material (NM) for each array, j, 

gi j 

NM j = ∑ , , 

g 

i i 

(3) 

for each neuron marker gene g i . Because GAD genes are neuron specific in the 

central nervous system, these were normalized for each array by the associated 

relative amount of neuron material, thus termed GAD NMN for neuron-marker 

normalized GAD. It is assumed in this study that the four neuron marker 

transcripts used here do not change their expression level between Alzheimer’s 

patients and age-matched controls, in the cell populations sampled for microarray 

analysis. This assumption is made since there are no published studies 

that demonstrate that these genes change expression in healthy cells through 

the progression of Alzheimer’s disease and because efforts were made to only 

expression profile histopathologically normal neurons. It is possible that there 

is downregulation of some neuron marker transcripts among neurons bearing 

neurofibrillary tangles, as synapse loss is a hallmark of Alzheimer’s disease 20 . 

However, the arrays used in this study profile histopathologically normal neurons 

and the surrounding glial cells. Therefore, it is not expected that there will 

be significant changes in the expression of these key neuronal genes in the data 

used here. Lastly, the inclusion of multiple genes from different cell regions 

aims to minimize the effects from expression changes not attributable to glial 

cell contamination. The results presented in this work are robust to the removal 

of each neuron marker gene (Supplementary Notes and Fig. 4 therein). 

PathWave analysis. PathWave allows for the elucidation of pathways that 

significantly change together. Its advantage over other methods, such as Gene 

Set Enrichment Analysis, is that it takes metabolic network connectivity into 

account to identify changes in pathways, thereby identifying significantly 

differentially regulated pathways in which the gene products are mechanistically 

connected. 

PathWave was used as published previously 33 . The reactions in each model 

were subdivided into biologically relevant functional pathways. Reactions that 


doi:10.1038/nbt.1711


were involved in multiple pathways were added to each associated pathway. 

Reactions and their pathways are shown in Supplementary Table 9. 

PathWave analyzes microarray data using metabolic reaction connectivity. 

For each model, pathways were simplified by removing all metabolites 

with connectivity >8 in the metabolic network. Exceptions are listed in 

Supplementary Table 10. 

For each of these simplified metabolic pathways from each model, reactions 

were laid into a two-dimensional, regular square lattice grid. To optimally preserve 

neighborhood relations of the reactions, adjacent nodes of the network 

were placed onto the grid as close to each other as possible. We mapped each 

expression data set, obtained from the Gene Expression Omnibus (GSE5281), 

onto the corresponding reactions of the transcribed enzymes. If a reaction 

was catalyzed by a complex of proteins, the average expression was taken. The 

resulting expression values of each reaction were z-transformed. Haar wavelet 

transforms on the optimized grid representation of each pathway were performed 

to explore every possible expression pattern of neighboring reactions 

and to define groups of reactions within a pathway that showed significant 

differences between samples of different conditions. 

To obtain significance values, the sample labels were permutated (n = 10,000) 

and scores were calculated for each wavelet and permutation. The scores represent 

the absolute value of the logarithm of the P-value for each wavelet feature, 

calculated by t-tests. For the best hit (highest score of the nonpermutated wavelet 

features) a P-value was obtained from the reference distributions and represented 

the significance for the corresponding pathway. The P-value for each pathway 

was corrected for multiple testing (false-discovery rate (FDR) = 0.05) 51 . Only 

pathways with more than three significantly differentially regulated reactions were 

further considered (FDR = 0.05). To obtain local patterns in the pathways, all 

wavelet features were statistically tested applying t-tests and corrected for multiple 

testing. Statistically significant features contained those subgraphs of the metabolic 

network that showed differentially regulated patterns. Reconstructing these 

subgraphs allowed us to directly detect the regions of interest in the metabolic 

network (Supplementary Table 11). 

Identifying pathways for acetylcholine synthesis. A flux balance analysisderived 

approach was used to identify all possible pathways coupling the mitochondrial 

and cytosolic acetyl-CoA pools using known reactions in human 

metabolism. First, the potential pathways were identified using Recon 1 

(ref. 7). A reaction that supplies mitochondrial acetyl-CoA was added to the 

model. A second reaction was added to remove cytosolic acetyl-CoA from the 

model. Lastly, all other metabolite uptake and secretion constraints were opened. 

Reactions were randomly removed until a minimum pathway was identified, 

capable of carrying flux between mitochondrial and cytosolic acetyl-CoA. 

This was repeated until >21,000 unique sets of reactions were identified. An 

r × p binary matrix was then built with the p unique reaction sets consisting of 

r reactions. Each element (i,j) of this matrix was 0 if reaction i was absent from 

pathway j or 1 if reaction i was in pathway j. Rows for all reactions that were 

never necessary were subsequently removed from the matrix. Singular value 

decomposition was then used, followed by varimax factor rotation of the first 

five singular vectors. Singular vector loadings demonstrated the dominant sets 

of reactions, and their major dependencies, that could be used to couple mitochondrial 

acetyl-CoA metabolism and cytosolic acetylcholine metabolism. 

Predicting cholinergic neurotransmission. The percentage cholinergic neurotransmission 

was computed based on published data 38 . The previously published 

data were obtained from rat brain minces that were incubated in solutions containing 

[1- 14 C]pyruvate or [2- 14 C]pyruvate. Both acetylcholine and radiolabeled 

CO 2 were measured at various titrations of several different inhibitors of pyruvate 

dehydrogenase (PDHm) (see Supplementary Notes for all inhibitors). 

Simulations were conducted using the cholinergic model. The models were 

allowed to take up the same substrates provided experimentally 38 , at rates 

consistent with the data (Supplementary Table 5). Monte Carlo sampling 

was used to identify all feasible flux states. This was done for various levels of 

PDHm inhibition, ranging from 0 to 90% inhibition. The percentage cholinergic 

neurotransmission was computed by randomly selecting a feasible flux 

state from each level of PDHm inhibition and computing the slope of the 

sum of labeled CO 2 -producing fluxes and choline acetyltransferase for the 

different simulations. A similar slope was computed from randomly sampled 

points from the reported experimental distributions. The ratio of these 

slopes represents a feasible percentage cholinergic neurotransmission. This 

was repeated 1,000 times and the median value was reported. Comparisons 

with the experimental data were done by suppressing the in silico pyruvate 

dehydrogenase flux until the measured CO 2 release rate was obtained. At this 

level of suppression, the resulting predicted acetylcholine production rate was 

compared with the experimentally measured rates. See Supplementary Notes 

for more details. 

47. Lying-Tunell, U., Lindblad, B.S., Malmlund, H.O. & Persson, B. Cerebral blood flow 

and metabolic rate of oxygen, glucose, lactate, pyruvate, ketone bodies and amino 

acids. Acta Neurol. Scand. 62, 265–275 (1980). 

48. Lying-Tunell, U., Lindblad, B.S., Malmlund, H.O. & Persson, B. Cerebral blood flow 

and metabolic rate of oxygen, glucose, lactate, pyruvate, ketone bodies and amino 

acids. Acta Neurol. Scand. 63, 337–350 (1981). 

49. Tischfield, M.A. et al. Human TUBB3 mutations perturb microtubule dynamics, 

kinesin interactions, and axon guidance. Cell 140, 74–87 (2010). 

50. Kim, K.K., Adelstein, R.S. & Kawamoto, S. Identification of neuronal nuclei (NeuN) 

as Fox-3, a new member of the Fox-1 gene family of splicing factors. J. Biol. Chem. 

284, 31052–31061 (2009). 

51. De Camilli, P., Cameron, R. & Greengard, P. Synapsin I (protein I), a nerve terminalspecific 

phosphoprotein. I. Its general distribution in synapses of the central and 

peripheral nervous system demonstrated by immunofluorescence in frozen and 

plastic sections. J. Cell Biol. 96, 1337–1354 (1983). 

52. Olave, I., Wang, W., Xue, Y., Kuo, A. & Crabtree, G.R. Identification of a polymorphic, 

neuron-specific chromatin remodeling complex. Genes Dev. 16, 2509–2517 (2002). 

53. Benjamini, Y. & Yekutieli, D. The control of the false discovery rate in multiple 

testing under dependency. Ann. Stat. 29, 1165–1188 (2001). 

doi:10.1038/nbt.1711 

nature biotechnology

B r i e f c o m m u n i c at i o n s 


A robust system for production 

of minicircle DNA vectors 

Mark A Kay 1,2 , Cheng-Yi He 1,2 & Zhi-Ying Chen 1,2 

Minicircle DNA vectors allow sustained transgene expression in 

quiescent cells and tissues. To improve minicircle production, 

we genetically modified Escherichia coli to construct a 

producer strain that stably expresses a set of inducible 

minicircle-assembly enzymes, ΦC31 integrase and I-SceI 

homing endonuclease. This bacterial strain produces purified 

minicircles in a time frame and quantity similar to those of 

routine plasmid DNA preparation, making it feasible to use 

minicircles in place of plasmids in mammalian transgene 

expression studies. 

A minicircle episomal DNA vector is a circular expression cassette 

devoid of the bacterial plasmid DNA backbone. These vectors have 

been used for years in preclinical gene transfer research because of 

their 10- to 1,000-fold enhancement compared with regular plasmids 

in long-term transgene expression in quiescent tissues in vivo 1,2 and 

in vitro 3 . The mechanism of enhanced transgene expression is unclear 

but may result from eliminating heterochromatin formation induced 

by the plasmid backbone 4 and/or from reducing the death of transfected 

cells from inflammation due to CpG responses when plasmids 

are delivered by lipid carriers 5 . 

The major obstacle to widespread use of minicircles has been their 

time-consuming, labor-intensive production. In our previous minicircle 

production schemes (Fig. 1a), the minicircle producer plasmid 

contained a transgene expression cassette flanked with attB and attP, 

a set of inducible enzyme genes (a gene encoding homing endonuclease 

I-SceI and two copies of the gene encoding ΦC31 integrase) and 

an I-SceI recognition site 6 . The attB and attP sites are the bacterial 

and phage attachment sites of ΦC31 integrase, and the ΦC31 and 

I-SceI genes are regulated by the l-arabinose–inducible araCBAD 

system. Minicircle DNA is generated by recombination between the attB 

and attP sites, and I-SceI initiates the destruction of the plasmid DNA 

backbone circle by cutting through the engineered I-SceI site (Fig. 1a). 

Although the yields from this protocol were ~1 mg of minicircle 

DNA from 1 liter of overnight culture, the preparations still contained 

~3–15% of the input minicircle producer plasmid plus the 

plasmid backbone circle as contaminants. Including CsCl equilibrium 

gradient centrifugation to remove these unwanted DNAs, 

the production procedure is four labor-intensive days longer than 

routine plasmid production protocols. Other groups have made 

minicircle DNA vectors using different recombinases, such as the 

bacteriophage λ integrase 7 or Cre recombinase 8 . Limitations of these 

approaches include low yields, high contamination or the need for 

expensive, labor-intensive techniques to isolate minicircles from 

bacterial lysates 9 . 

We present a system that allows simple, rapid and inexpensive production 

of a high-quality form of minicircles (Fig. 1b,c). We reasoned 

that the impurity plasmid DNAs in our earlier minicircle production 

schemes (Fig. 1a) resulted largely from an all-or-none phenomenon 10 , 

such that in a subset of bacterial cells, the l-arabinose transporter 

AraE is not expressed, resulting in the failure of ΦC31 integrase– 

directed formation of minicircle and I-SceI–mediated degradation of 

the plasmid backbone circle. To overcome this limitation, we replaced 

the E. coli ‘Top10’ cells with the BW27783 bacterial strain 10 , in which 

the araE gene was driven by the constitutive promoter, cp8. When 

used in combination with previous minicircle producer plasmids, 

the conversion to minicircle DNA was more complete (Fig. 2) even 

with the addition of a very small amount (0.001%) of l-arabinose 

to the culture (Fig. 2a). There was variable but substantial plasmid 

DNA degradation in these preparations (Fig. 2a,c). We hypothesized 

this was due to the presence of endonuclease A, which was confirmed 

when we knocked the gene out (Fig. 2b,c), resulting in strain 

BWΔendA (Fig. 2b and Supplementary Fig. 1). 

Trace amounts of impurity DNAs became visible after knocking 

out the gene encoding endonuclease A (Fig. 2c), suggesting that 

there was still a lack of complete recombination after l-arabinose 

induction. Thus, we engineered the bacterial cells to simultaneously 

express a second l-arabinose transporter, LacY A177C 11 and created 

the strain 2T (Supplementary Figs. 1 and 2a–d). Its wild-type counterpart, 

LacY, is a glucose transporter, which gains an l-arabinose 

transporter function with the described missense mutation. This 

further reduced but did not completely eliminate the contaminating 

DNAs (data not shown). 

To eliminate any possible contamination of the genes encoding 

ΦC31 integrase and I-SceI from the small amount of parental plasmids 

used to generate minicircle DNAs, we relocated both genes from the 

minicircle producer plasmid to the bacterial genome. Accordingly, we 

integrated three copies of the BAD.I-SceI cassette using the bacteriophage 

λ Red homologous recombination system 12 and created the 

strain 3S2T (Supplementary Figs. 1 and 3a). We found that the integrated 

BAD.I-SceI was fully functional, as evident by the efficient 

destruction of a plasmid with eight consecutive I-SceI sites after 

l-arabinose induction (Supplementary Fig. 3b,c). 

Integration of the BAD.ΦC31 gene was less straightforward. First, 

the BAD.ΦC31 cassette could not be integrated using our original Red 

system without destroying the integrated BAD.I-SceI gene. Second, we 

attempted to integrate the BAD.ΦC31 cassette from a plasmid through 

ΦC31 integrase–mediated recombination between the attB site in the 

plasmid and attP pre-integrated into the genome. However, this integrant 

was unstable (data not shown), most likely because of a reverse 

1 Departments of Pediatrics and Genetics, Stanford University School of Medicine, Stanford, California, USA. 2 The Center for Clinical Science Research, Stanford 

University School of Medicine, Stanford, California, USA. Correspondence should be addressed to M.A.K. (markay@stanford.edu) or Z.-Y.C. (zychen98@stanford.edu). 

Received 21 June; accepted 15 October; published online 21 November 2010; doi:10.1038/nbt.1708 


i e f c o m m u n i c at i o n s 


Figure 1 Comparison of the present and previous 

minicircle systems. (a) An earlier version of the 

minicircle production system. (i) Structure of 

the previous minicircle producer plasmid. BAD 

and araC, the promoter and the repressor gene 

of the inducible l-arabinose-araC.BAD system; 

ΦC31, bacteriophage ΦC31 integrase gene; attB 

and attP, the bacterial and phage attachment 

sites of the ΦC31 integrase; I-SceI, I-SceI 

homing endonuclease gene; I-SceIs, the I-SceI 

recognition site; AmpR, ampicillin resistance 

gene; ColE1, DNA replication origin. (ii) E. coli 

strain Top10, original strain used to produce 

minicircle. (iii) Flow chart showing the minicircle 

production protocol. Each box represents a 

major step and the starred boxes represent the 

steps required in addition to a routine plasmid 

production protocol. (b) The present minicircle 

system. (i) Diagram of a minicircle producer 

plasmid and its conversion to minicircle DNA. 

pMC.ApoE.hFIX, minicircle producer plasmid; 

hFIX, human factor IX; sApoE, promoter/ 

enhancer, as described previously 1 ; KanR, 

kanamycin resistance gene. Upon l-arabinose 

induction, ΦC31 is expressed to mediate the 

formation of minicircle and plasmid backbone 

circle and I-SceI to induce the destruction 

of plasmid backbone circle. (ii) The genetic 

Transgene attP 

p2ΦC31.Transgene 

2BADΦC31 araC AmpR ColE1 I-SceIs 

modifications of the minicircle producing bacterial strain ZYCY10P3S2T. 10P3S2T stands for (1) ten copies of BAD.ΦC31 cassette, that were integrated in 

three loci of the bacterial genome: two tandem copies at the ΔendA locus (Supplementary Fig. 4b), and four copies at the araD (Supplementary Fig. 5a) and 

galK (Supplementary Fig. 6a) each; (2) three tandem copies of BAD.I-SceI cassette, which were integrated at UMU locus (Supplementary Fig. 3a) and (3) 

two genes constitutively expressing l-arabinose transporter, one was araE gene driven by an artificial promoter cp8, which presented in strain BW27783 

(ref. 10); the other was the bla-lacY A177C cassette, which was integrated at the lacY locus (Supplementary Fig. 2); bla, beta-galactosidase gene promoter; 

lacY A177C, the missense mutant of lacY gene. (iii) Flow chart showing the present minicircle production protocol. (c) Stepwise genetic modification of the 

bacterial genome to make the current ZYCY10P3S2T strain. 

reaction mediated by leaky ΦC31 integrase expression in concert with 

an uncharacterized cofactor 13 . Third, because many more copies of 

the BAD.ΦC31 were needed to efficiently generate minicircles, we 

designed a strategy using ΦC31 integrase and a second recombinase. 

After making the desired integrant by the ΦC31 integrase–mediated 

reaction, the resulting attL site was removed by the second recombinase, 

eliminating the possibility of the reverse excision reaction 

(Supplementary Figs. 4a–c, 5a,e and 6a,d). 

Using the above approach, we generated a bacterial host containing 

two copies of the BAD.ΦC31 inserted into the deleted endA site 

(ΔendA), called the 2P3S2T strain (Supplementary Figs. 1 and 4a–c). 

We found that ΦC31 integrase mediated the formation of the minicircle, 

and I-SceI mediated destruction of the unrecombined minicircle 

producer plasmid and plasmid backbone circle in a minicircle producer 

plasmid encoding a 2.2-kb transgene and 32 tandem copies of 

the I-SceI site in the plasmid backbone (Supplementary Fig. 4d,e). 

However, the minicircle yield was low, suggesting that most of the 

minicircle producer plasmid was destroyed by I-SceI before minicircle 

formation, and that additional ΦC31 integrase activity was needed to 

mediate greater minicircle formation. 

We therefore used another targeting plasmid to integrate four copies 

of the BAD.ΦC31 cassette into the araD locus (Supplementary 

Figs. 1 and 5), and four additional copies into the gene encoding 

galK (Supplementary Figs. 1 and 6), resulting in strains 6P3S2T and 

ZYCY10P3S2T carrying six and ten copies of BAD.ΦC31, respectively. 

These two present strains, in concert with the minicircle producer 

plasmid described earlier (Supplementary Fig. 4d), enabled three 

improvements over the previous system (Fig. 1a) 6 . First, the procedure 

was greatly simplified and, compared to a routine plasmid preparation, 

required only an additional temperature change and 5-h incubation 

a 

b 

(i) 

(iii) 

(i) 

(ii) 

(iii) 

attB 

Overnight 

culture 

attB 

Restrict 

contaminated 

plasmid 

backbone circle 

Overnight 

culture 

Pellet the 

bacteria 

via centrifugation 

* 

ApoE.hFIX attP 

pMC.ApoE.hFIX 

32I-SceIs KanR 

ColE1 

BADI-SceI 

* Resuspend * * 

bacteria in LB 

containing 

1% L-arabinose 

Isolate 

minicircle via 

CsCl gradient 

centrifugation 

Add minicircle 

induction mix 

comprising 1 volume 

of LB, 0.04 volume of 

1N NaOH and 

0.02% L-arabinose 

ApoE.hFIX 

* 

* 

MC.ApoE.hFIX + 

attR 

(ii) 

E. coli strain Top 10 

Incubate at 

32 °C, 250 

r.p.m. for 2 h 

Remove ethidium 

bromide via 

isopropane 

extraction 

The genetic modifications in bacterial strain ZYCY10P3S2T: 

Incubate 

at 32 °C/ 

250 r.p.m. 

for 5 h 

Plasmid backbone circle 

32I-SceIs KanR ColE1 

* 

Isolate 

minicircle 

by plasmid 

purification 

column 

* * 

Isolate 

minicrcle by 

plasmid 

purification 

column 

Remove 

isopropane 

via ethanol 

precipitation 

Cp8.araE ∆endA Bla.lacY A177C 3BAD.I-SceI 2BAD.ΦC31 4BAD.ΦC31 4BAD.ΦC31 

attL 

Degradation 

after addition of l-arabinose (Fig. 1b). Second, the yield of three 

minicircles, with a size range of 2.2–6.0 kb, was 3.4–4.8 mg/1,000 ml 

of overnight culture, making it ~3- to 5-times higher (Fig. 2e) than our 

previous minicircle producing system (Fig. 1a). In addition, compared 

with the previous minicircle production protocol, there were 10-times 

fewer contaminating plasmid DNAs, ranging from 0.4% to 1.5% of the 

input minicircle preparation as determined by qPCR (Online Methods, 

qPCR section) 6 . On a molar scale, the yield of minicircle was 20–70% 

higher than the minicircle producer plasmid (Fig. 2e). Third, the cost 

of minicircle production was similar to that of a standard plasmid. 

These production improvements, together with their superior 

expression profiles, make it feasible for minicircle DNA vectors to be 

used in place of plasmid DNAs in mammalian expression studies. 

Our efforts to optimize minicircle production led us to develop an 

improved bacterial genome modification strategy. The enhancements 

include: (i) the use of a circular integrating plasmid instead of linear 

DNA, allowing repeated integration of the same or different DNA 

sequences of up to 10 kb in selected targets (Supplementary Figs. 4b, 

5a and 6a); (ii) inclusion of a second recombinase, allowing selective 

removal of unwanted sequences, such as the hybrid sequences responsible 

for the reverse recombinase reaction, along with the useless or 

harmful plasmid backbone DNAs, from the genome (Supplementary 

Figs. 4b, 5a and 6a); and (iii) the use of the TPin/9attB.9attP recombination 

system (Supplementary Figs. 5a and 6a), eliminating the inherent 

problems with the FLP/FRT system, in which the uncontrolled reaction 

between the substrate and product FRTs makes repeated integration 

virtually impossible. Although these bacterial gene modifications 

would be difficult to achieve with conventional homologous recombination 

methods, there are other technologies using dual site-specific 

recombinases for targeted integration of circular DNA into the bacterial 

c 

Strain BW27783 

Figure 2 

Knockout endA, 

resulting in strain 

BW∆endA 

Supplementary 

Figure 2 

Integrate lacY A177C, 


2T 

Supplementary 

Figure 3 

Integrate 3XBADI-Scel, 


3S2T 

Supplementary 

Figure 4 

Integrate 2xBADΦC31, 


2P3S2T 

Supplementary 

Figure 5 



6P3S2T 

Supplementary 

Figure 6 



10P3S2T 


i e f c o m m u n i c at i o n s 


Figure 2 Improvement in minicircle quality 

and quantity. (a) Strain BW27783 produced 

minicircle (MC) with enhanced purity. Minicircles 

were produced according to the protocol 

described previously 6 . 2Φ31.hFIX, minicircle 

producer plasmid (PP); BglII+EcoN1, two 

restriction enzymes used to cleave MC before 

electrophoresis; l-arab(%), percent of l-arabinose 

in the minicircle induction reaction; PB, plasmid 

backbone circle. (b) Strategy for inactivation 

of endA. pK a nR.endA, the plasmid used to 

generate the Pme1-restricted targeting DNA 

fragment; KanR, kanamycin-resistance gene; attB 

and attP, the bacterial and phage attachment 

sites of bacteriophage ΦC31 integrase; boxed 

endA, PCR-generated 329- and 754-bp end A 

fragments; pBAD.Red, a plasmid expressing the 

bacteriophage λ homology recombination complex 

(Red) under the control of araC.BAD (BAD); 

p2ΦC31, a complementing plasmid encoding 

two copies of BAD.ΦC31 gene, one copy of BAD. 

I-SceI gene and one I-SceI site; BWΔendA, a 

strain derived from BW27783 with the endA 

interrupted. (c) Minicircle DNA integrity before 

and after disruption of the endA gene. Before 

disrupting endA, we observed repeatedly large 

variations in the degree of plasmid degradation 

as shown in a and c. Because the endonuclease 

A is a membrane-bound enzyme, it was possible 

that its membrane release and activation varied 

during plasmid preparation. 32 °C and 37 °C, the 

incubation temperature. All reactions contained 

1% l-arabinose. (d) Quality of the minicircle 

determined by gel analyses. Minicircle was 

made according to the simplified protocol outlined in Figure 1b and Supplementary Protocol; DNAs were cleaved before electrophoresis. (e) Yield of 

minicircle producer plasmids and minicircle vector DNAs. The yield was derived from triplicate 400-ml overnight cultures; PP, minicircle producer plasmid; 

MC, minicircle. Wilcoxon rank sum test comparing the yield of minicircle and its minicircle producer plasmid: (I), P < 0.05; (II), P > 0.05. We used the 

following formula to convert the yield from mg/l to mol/l: mol/l = [yield (mg/l) × 10E-3 g/l]/[size (kb) × 1,000 × 330 × 2 g/mol], where 330 is the average 

molecular weight of dNTP. The minicircle producer plasmid, pMC.RSV.hAAT is schematically illustrated in Supplementary Figure 4d; the pMC.CMV.LGNSO 

is described in the Constructs section of Online Methods. 

genome and creation of a marker-less bacterial strain 14 . However, in our 

approach we were able to remove one of the two recombination hybrids 

(either attL or attR; Supplementary Figs. 5 and 6), disabling the reverse 

reaction and resulting in a stable genomic insertion. Recent studies 

have shown that repeated gene sequences inserted into the bacterial 

genome are stable for at least 80 generations 15 . This type of site-specific 

integration will broaden the ability to make stable genetic modifications 

in prokaryotic and eukaryotic genomes. 

Methods 

Methods and any associated references are available in the online 

version of the paper at http://www.nature.com/naturebiotechnology/. 



The authors would like to thank P. Valdmanis for critical review of the manuscript. 

This work was supported by the US National Institutes of Health - HL064274 

(M.A.K.). Bacterial strain BW27783 was a gift of Jay D. Keasling of the University 

of California at Berkeley. Plasmid placY A177C was obtained from John E. Cronan 

at the University of Illinois. 


Z.-Y.C. planned and carried out the experiments. C.-Y.H. conducted a number 

of the experiments. M.A.K. and Z.-Y.C. discussed and planned the experimental 

strategies. M.A.K. and Z.-Y.C. wrote the manuscript. 

a 

c 

e 

Parental p2Φ31.hFIX 

Enzyme Bgl II+EcoNI 

Strain Top10 

BW27783 

L-arab (%) 0 1.0 0.1 0.01 0.001 0 1.0 0.1 0.01 0.001 

Plasmid p2Φ31.hFIX 

Enzyme Bgl II + EcoNI 

Strain BW∆endA1 BW27783 

32 °C (h) 0 2 2 0 2 

37 °C (h) 0 0 2 0 2 

PP 

PB 

MC 

d Producer pMC 

pMC. 

pMC. 

Plasmid RSV.hAAT ApoE.hFIX CMV.LGNSO 

DNA type PP MC PP MC PP MC 



b 

ColE1 AmpR F1 

pK a nR.endA 

endA attB KanR attP endA 

PP 

PB Pme I 

Pme 1- Pme I 

MC 

cutting 


+ 

EndA Strain BW27783 

pBAD.Red 


p2øC31 

Targeting 

DNA fragment 

Strain 

BW∆endA.KanR 

endA attR endA Strain BW∆endA 

Size (kb) Yield (mg/liter) Weight Yield (E-9 mole/liter) Molar 

Minicircle producer 

ratio 

ratio 

plasmid 

PP MC PP 

MC (MC/PP) PP MC (MC/PP) 

pMC.RSV.hAAT 6.1 2.2 5.84 ± 0.41 3.42 ± 0.87(I) 0.59 1.60 ± 0.11 2.59 ± 0.66(I) 1.63 

pMC.ApoE.hFIX 8.1 4.2 5.45 ± 0.38 4.83 ± 0.60(II) 0.89 1.12 ± 0.08 1.91 ± 0.24(I) 1.71 

pMC.CMV.LGNSO 9.8 6.0 5.57 ± 0.15 3.40 ± 0.35(I) 0.61 0.95 ± 0.03 1.07 ± 0.11(II) 1.18 




1. Chen, Z.Y., He, C.Y., Ehrhardt, A. & Kay, M.A. Mol. Ther. 8, 495–500 (2003). 

2. Huang, M. et al. Circulation 120, S230–S237 (2009). 

3. Jia, F.J. et al. Nat. Methods 7, 197–199 (2010). 

4. Riu, E., Chen, Z.Y., Xu, H., He, C.Y. & Kay, M.A. Mol. Ther. 15, 1348–1355 

(2007). 

5. Tan, Y., Li, S., Pitt, B.R. & Huang, L. Hum. Gene Ther. 10, 2153–2161 

(1999). 

6. Chen, Z.Y., He, C.Y. & Kay, M.A. Hum. Gene Ther. 16, 126–131 (2005). 

7. Darquet, A.M., Cameron, B., Wils, P., Scherman, D. & Crouzet, J. Gene Ther. 4, 

1341–1349 (1997). 

8. Bigger, B.W. et al. J. Biol. Chem. 276, 23018–23027 (2001). 

9. Mayrhofer, P., Blaesen, M., Schleef, M. & Jechlinger, W. J. Gene Med. 10, 

1253–1269 (2008). 

10. Khlebnikov, A., Datsenko, K.A., Skaug, T., Wanner, B.L. & Keasling, J.D. Microbiology 

147, 3241–3247 (2001). 

11. Morgan-Kiss, R.M., Wadler, C. & Cronan, J.E. Jr. Proc. Natl. Acad. Sci. USA 99, 

7373–7377 (2002). 

12. Yu, D. et al. Proc. Natl. Acad. Sci. USA 97, 5978–5983 (2000). 

13. Thorpe, H.M. & Smith, M.C. Proc. Natl. Acad. Sci. USA 95, 5505–5510 

(1998). 

14. Minaeva, N.I. et al. BMC Biotechnol. 8, 63 (2008). 

15. Sallam, K.I., Tamura, N., Imoto, N. & Tamura, T. Appl. Environ. Microbiol. 76, 

2531–2539 (2010). 




Materials. Bacterial strain BW27783 was a gift of J.D. Keasling 10 . Luria- 

Bertani broth (LB) powder was purchased from MP Biomedicals and Terrific 

Broth (TB) powder from Invitrogen. Plasmid DNA purification kits were 

from Qiagen. 

Constructs. Plasmid pK a nR.endA (Fig. 2b and Supplementary Glossary) was 

made by inserting the following DNA elements: the kanamycin-resistance gene 

(KanR) flanked by the bacterial (attB) and phage (attP) attachment sites of the 

Streptomyces bacteriophage ΦC31 integrase and PCR-generated 329- and 734-bp 

fragments of endonuclease A gene (endA) into the pBlueScript (Stratagene) 

plasmid. The plasmid p2ΦC31 (Fig. 2b and Supplementary Figs. 1, 2b, 3a 

and Supplementary Glossary) was made by removing the human factor IX 

(hFIX) transgene and flanking attB and attP sites from p2ΦC31.hFIX 6 , 

the BAD.I-SceI cassette and I-SceI site were retained. To make the plasmid 

p3BAD.I-SceI (Supplementary Fig. 3a and Supplementary Glossary), 

the araC repressor gene together with three tandem copies of BAD.I-SceI gene 

were inserted downstream of the attB site of the plasmid pK a nR.endA (Fig. 2b 

and Supplementary Glossary), followed by replacement of endA with the 

737- and 647-bp UMU fragment generated by PCR. The plasmid pBS.8I-SceIs 

(Supplementary Fig. 3b and Supplementary Glossary) was constructed by 

inserting eight consecutive copies of the 18-bp I-SceI site into pBlueScript. The 

plasmids pBAD.Red (Fig. 2b and Supplementary Figs. 1, 2a,c, 3a, 4a, 5a and 

6a and Supplementary Glossary) 16 and pcI587.FLP (Supplementary Figs. 

1, 4a,b and Supplementary Glossary) 17 , which both carry the temperaturesensitive 

A101 origin of replication, were obtained from the E. coli Genetic 

Resource Center of Yale University. Plasmid pFRT.KanR.attB (Supplementary 

Figs. 1, 4a and Supplementary Glossary) was generated by inserting the attB 

site and the KanR flanked with FRT sites of flipase into pBlueScript using DNA 

oligonucleotides. Plasmid p2ΦC31.attP.FRT (Supplementary Figs. 1, 4b and 

Supplementary Glossary) was constructed using our previous minicircleproducing 

plasmid p2ΦC31.hFIX 6 as starting material. The sApoE.hFIX and 

BAD.I-SceI cassettes and the I-SceI site were eliminated, the ColE1 origin 

and the AmpR were replaced with the KanR, the FRT and attP sites and the 

plasmid replication origin R6K plus zeocin-resistance gene (Zeo) derived from 

the plasmid pCpG-mcs (InvivoGen). The plasmid pMC.ApoE.hFIX (Figs. 1b, 

2d,e and Supplementary Glossary) was made by stepwise replacement of the 

AmpR and the F1 origin in the plasmid pBS.8I-SceI (Supplementary Fig. 3b 

and Supplementary Glossary) with KanR and sApoE.hFIX and the flanking 

attB and attP sites from the plasmid p2ΦC31.hFIX 6 , followed by the insertion 

of an additional 24 consecutive I-SceI recognition sites. The plasmid 

pMC.CMV.LGNSO (Fig. 2d,e) was made by replacing the attB.RSV.hAAT.attP 

fragment in the plasmid pMC.RSV.hAAT (Fig. 2d,e, Supplementary Fig. 4d 

and Supplementary Glossary) with the fragment flanked with the attB and 

attP derived from the plasmid p2ΦC31.LGNSO, as described previously 3 . The 

plasmid placY.TetR (Supplementary Figs. 1, 2a and Supplementary Glossary) 

was made by inserting the TetR sequence from pACYC184 (New England 

Biolabs) and flanked with PCR-generated 425 and 227 bp of the z and a 

lactose operon genes, respectively. Plasmid pbla.lacY A177C (Supplementary 

Figs. 1, 2a and Supplementary Glossary) was made by inserting the betalactosidase 

gene promoter (bla) derived from pBlueScript fused with the lacY 

A177C cDNA derived from plasmid placY A177C obtained from J.E. Cronan 11 . 

Plasmid p9attP.TetR.attB (Supplementary Figs. 5a, 6a and Supplementary 

Glossary) was made by replacing the z and a fragments in the plasmid 

placY.TetR with the attachment site (9attP) from bacteriophage TP901-1 

(ref. 18) and attB. Plasmid p4ΦC31.attP.9attB (Supplementary Figs. 5a, 

6a and Supplementary Glossary) was generated by replacing the FRT and 

ZEO.R6K sequences of p2ΦC31.attP.FRT (Supplementary Figs. 1, 4b and 

Supplementary Glossary) with the bacterial attachment site 9attB of bacteriophage 

TP901-1 and A101 from plasmid pBAD.Red 16 , followed by insertion 

of two additional copies of the BAD.ΦC31 sequence. 

Insertional inactivation of the endonuclease A gene in BW27783 (Fig. 2b). 

This was achieved by Red-mediated homologous recombination between a 

linear DNA and the endA gene. To do this, BW27783 cells were transformed 

with pBAD.Red. Cells from one transformed colony were used to make competent 

cells, as described 16 . Briefly, the cells were cultured in 25 ml of low salt 

LB containing ampicillin (50 μg/ml) and 1% l-arabinose and incubated at 

32 °C with shaking at 250 r.p.m. until the OD600 was 0.5. The competent cells 

were immediately transformed with 50-ng of the Pme1-restricted targeting 

DNA fragment prepared from pK a nR.endA, cultured at 32 °C for 1 h, spread 

onto a plate containing kanamycin (Kan, 25 μg/ml) and incubated at 43 °C 

overnight. To select the KanR + colonies free of pBAD.Red, eight KanR + colonies 

were cultured in 500 μl LB with Kan at 43 °C for 30 min and 1 μl each 

was loaded onto the antibiotic-free, Kan + , and Amp + plates, which were incubated 

at 43 °C overnight. To remove the KanR gene in the strain BWΔendA. 

KanR, competent cells were prepared from one KanR + -colony, transformed 

with p2ΦC31 and spread onto a Amp + plate; subsequently, eight cultures were 

begun from eight p2ΦC31-transformed colonies in 500-μl antibiotic-free LB 

containing 1% l-arabinose at 32 °C for 2 h to induce the removal of KanR via 

ΦC31 integrase-mediated recombination between the attB and attP flanking 

the KanR, followed by 4 additional hours at 37 °C to induce I-SceI–mediated 

destruction of p2ΦC31. The cells were spread onto an antibiotic-free plate. 

Cells from eight colonies were cultured in 500 μl antibiotic-free LB at 37 °C 

for 30 min, and 1 μl each was loaded onto the antibiotic-free, Kan + and Amp + 

plates to select the AmpR − /KanR − colonies. This resulted in the BWΔendA 

strain. The disruption of the endA gene was confirmed by DNA sequencing of 

the locus-specific PCR products generated from selected colonies. 

Integration of the 2 nd l-arabinose transporter lacY A177C and three copies of 

the BAD.I-SceI gene (Supplementary Fig. 2). Replacement of the y gene (lacY) 

of the lactose operon with the mutant lacY A177C (Supplementary Fig. 2) and 

integration of three copies of the I-SceI gene into the UMU locus (Supplementary 

Fig. 3) were achieved by the same Red-mediated homology recombination protocol 

as described for the endA gene disruption (Fig. 2b). To integrate the lacY A177C, 

however, knockout of the wild-type lacY was performed before knocking in the 

lacY A177C. Both were accomplished using the same Red-mediated homologous 

recombination. We used a DNA fragment containing TetR flanked with fragments 

of the z and a genes to knock out lacY, and selected the intermediate, BWΔendA. 

TetR on the Tet + (6 μg/ml)-plate. This allowed the selection of the next integrant 

using differential KanR/TetR selection (Supplementary Figs. 2a, 5a and 6a). 

Integration of the BAD.ΦC31 cassettes. Three experiments were conducted 

to integrate ten copies of BAD.ΦC31 into the bacterial genome: two, four 

and four copies into the ΔendA locus (Supplementary Fig. 4), araD gene 

(Supplementary Fig. 5) and galK gene (Supplementary Fig. 6), respectively. 

All three integration events were achieved by ΦC31 integrase-mediated, sitespecific 

recombination between the attB or attP site in an integrating plasmid, 

and the attB or attP in the targeted genomic sites as described for the endA 

gene disruption (Fig. 2b). 

In an earlier attempt, we found that the integrated BAD.ΦC31 was lost soon 

after its integration. This is probably the result of a reverse reaction mediated by 

an uncharacterized excisionase or cofactor that worked in concert with ΦC31 

integrase produced by the leaky expression from the integrated BAD.ΦC31. 

To stabilize the integrant, we designed a double recombination strategy. The 

ΦC31 integrase was used to mediate an integration event followed by FLP- or 

TPin-mediated recombination to remove the attL. This would eliminate the 

possibility of a reverse reaction between the attL and attR. For integration of the 

2BAD.ΦC31 (Supplementary Fig. 4), we used the Red-mediated homologous 

recombination strategy with a linear DNA containing an attB site, a KanR 

gene and two flanking FRTs for insertion into the ΔendA locus. To remove the 

KanR, cells from one 3S2T.KanR colony were transformed with pcI587.FLP, 

and cells from a transformed colony were grown in 5 ml of antibiotic-free LB at 

42 °C for 8 h. This induced the expression of FLP to mediate the recombination 

between the two FRTs, resulting in the elimination of KanR followed by the loss 

of the temperature-sensitive pcI587.FLP. Subsequently, the cells named 3S2T. 

attB were transformed with p2ΦC31.attP.FRT. Cells from one transformed 

colony were grown in 2-ml of LB containing 1% l-arabinose at 32 °C for 2 h, 

which induced the expression of ΦC31 integrase to mediate the recombination 

between the attB in the ΔendA locus and the attP in the plasmid. The 

same FLP-mediated FRT-FRT recombination was conducted to mediate the 

removal of the attL and R6K.Zeo and KanR, resulting in the strain of 2P3S2T 

(Supplementary Fig. 4b). The integrated 2BAD.ΦC31 was confirmed by 

DNA sequencing of the locus-specific PCR product (Supplementary Fig. 4c) 


doi:10.1038/nbt.1708


and by demonstrating the function in mediating minicircle formation 

(Supplementary Fig. 4e). Integration of the four copies of the BAD.ΦC31 

gene into the araD (Supplementary Fig. 5) and galK (Supplementary 

Fig. 6) loci, respectively, was also mediated by ΦC31 integrase; however, 

the removal of the attL was mediated by bacteriophage TP901-1 integrase 

(TPin). To do this, we used the Red recombination system to integrate a DNA 

fragment encoding TetR flanked with an attB and a phage attachment site of 

TPin (9attP) into the araD or galK site and confirmed the integrant by DNA 

sequencing of the integrant-specific PCR product (Supplementary Figs. 5b 

and 6b). We integrated p4ΦC31.attP.9attB into the modified araD or galK 

locus following the same procedure as used for integrating the p2ΦC31.attP. 

FRT (Supplementary Fig. 4) with modifications. We selected the colonies with 

integrants using plates containing both Tet and Kan, transformed the selected 

intermediates 6P3S2T.KanR.TetR or 10P3S2T.KanR.TetR with plasmid pBAD. 

TPin, and screened the resulting colonies using triple antibiotic resistance (Tet, 

Kan and Amp). We grew the cells from one colony in 2 ml of antibiotic-free 

LB containing 1% l-arabinose, with shaking (250 r.p.m.) at 43 °C for 6–8 h 

before spreading onto an antibiotic-free plate and incubated at 43 °C overnight. 

l-arabinose induced expression of both recombinases, however, the incubation 

temperature allowed significant TPin integrase activity 18 but little ΦC31 

integrase activity, resulting in the TPin integrase-mediated removal of attL 

and selection of colonies with the desired integrant, 6P3S2T (Supplementary 

Fig. 5a,d) and ZYCY10P3S2T (Supplementary Fig. 6a,c). 

Minicircle production protocol. Our present minicircle producing system, 

the strain ZYCY10P3S2T plus the minicircle producer plasmid pMC.hFIX 

or pMC.RSV.hAAT, allowed for a greatly simplified minicircle production 

protocol (Fig. 1b). On day one, we inoculated cells from one transformed 

colony in 5 ml of TB (pH 7.0) with Kan (50 μg/ml) and incubated at 

37 °C with shaking at 250 r.p.m. Later that day, we amplified the bacteria by 

combining 100 μl of culture to every 400 ml TB containing Kan (50 μg/ml) and 

continued incubation for 16 to 18 h. For the yield comparison study (Fig. 2e), 

a 400-ml overnight culture was used to prepare intact plasmid DNA. At the 

end of the culture period the A 600 was 3.5– 4.2 with a pH of ~6.5. On day 2, 

we prepared a minicircle induction mix comprising 400 ml fresh LB, 16 ml 

1N sodium hydroxide and 0.4 ml 20% l-arabinose, and combined it with a 

400 ml overnight culture, and incubated the culture at 32 °C with shaking at 

250 r.p.m. for an additional 5 h. We used Qiagen plasmid purification kits to 

isolate minicircle from bacterial lysates following the manufacturer’s protocol 

with modifications. For every 400-ml overnight culture, we used a 2,500 

column and 100-ml each of buffers P1, P2 and P3 to ensure complete resuspension 

and lysis of the bacteria and a high yield of minicircle DNA vector. 

A step-by-step protocol is provided in Supplementary Protocol. 

qPCR determination of impurity DNAs. The presence of impurity DNAs, the 

unrecombined minicircle producer plasmid and plasmid backbone circle, in 

the three minicircle preparations was determined as described in the legends 

of Figures 2d,e by qPCR. To optimize DNA-polymerase activity, all the DNA 

templates were digested with AflII plus XhoI flanking the ColE1 origin before 

qPCR. The minicircle producer plasmid (Supplementary Fig. 7) without 

a transgene expression cassette was used as a standard. The PCR primers, 

5′-TCCTGTTACCAGTGGCTGCT and 5′-AGTTCGGTGTAGGTCGTTCG, 

were specific for ColE1 DNA origin shared by all templates. qPCR was 

performed using the Qiagen Quant SYBR Green RT-PCR kit with a RCorbett 

system (Corbett). The program included a denaturing step of 94 °C for 10 min, 

followed by 25 cycles at 94 °C for 20 s, 55 °C for 15 s, and 68 °C for 20 s each. 

16. Datsenko, K.A. & Wanner, B.L. Proc. Natl. Acad. Sci. USA 97, 6640–6645 

(2000). 

17. Cherepanov, P.P. & Wackernagel, W. Gene 158, 9–14 (1995). 

18. Stoll, S.M., Ginsburg, D.S. & Calos, M.P. J. Bacteriol. 184, 3657–3663 (2002). 

doi:10.1038/nbt.1708 


l e t t e r s 

High-fidelity gene synthesis by retrieval of 

sequence-verified DNA identified using 

high-throughput pyrosequencing 

Mark Matzas 1,5 , Peer F Stähler 1,5 , Nathalie Kefer 1 , Nicole Siebelt 1 , Valesca Boisguérin 1 , Jack T Leonard 1 , Andreas Keller 1 , 

Cord F Stähler 1 , Pamela Häberle 1 , Baback Gharizadeh 2 , Farbod Babrzadeh 2 & George M Church 3,4 


The construction of synthetic biological systems involving 

millions of nucleotides is limited by the lack of high-quality 

synthetic DNA. Consequently, the field requires advances in 

the accuracy and scale of chemical DNA synthesis and in the 

processing of longer DNA assembled from short fragments. 

Here we describe a highly parallel and miniaturized method, 

called megacloning, for obtaining high-quality DNA by using 

next-generation sequencing (NGS) technology as a preparative 

tool. We demonstrate our method by processing both chemically 

synthesized and microarray-derived DNA oligonucleotides with 

a robotic system for imaging and picking beads directly off of 

a high-throughput pyrosequencing platform. The method can 

reduce error rates by a factor of 500 compared to the starting 

oligonucleotide pool generated by microarray. We use DNA 

obtained by megacloning to assemble synthetic genes. In 

principle, millions of DNA fragments can be sequenced, 

characterized and sorted in a single megacloner run, enabling 

constructive biology up to the megabase scale. 

Current de novo gene construction 1–4 rests on 1990’s technology for 

chemical oligonucleotide synthesis, which is costly and has error rates 

of 1 in 300 base pairs (bp). Errors are typically avoided by manually 

selecting the best Sanger sequences using electrophoretic automation. 

Recent innovations in programmable array technology 5–8 offer the 

possibility to synthesize pools of thousands to millions of sequences 

per array with lengths comparable to conventional synthesis. The 

technology thus provides an extremely rich source of DNA oligonucleotides 

with great flexibility and superior efficiency regarding 

throughput and cost per bp. However, the error rate of microarrayderived 

oligonucleotides is typically higher compared to conventional 

synthesis, making error avoidance or correction necessary. 

Furthermore it is challenging to divide the derived oligonucleotide 

pools, containing vast amounts of species, into subpools—necessary, 

for example, to guide the assembly of synthetic genes, chromosomal 

regions or whole pathways in synthetic biology. 

Megacloning turns NGS from a previously purely analytical 

method into a preparative tool, and represents a tremendous source 

of sequence-verified DNA where the yield from one NGS run is 

equivalent to that from hundreds to thousands of Sanger-sequence 

runs. It therefore addresses the challenge of error reduction for both 

conventional and microarray-derived DNA oligonucleotides. The 

method yields high-quality DNA libraries containing perfect parts 

with desired and correct sequences in adjustable ratios useful for a 

wide range of (bio-)technological applications. 

Here we present a proof-of-concept study aimed at the retrieval 

of clonal DNA with known sequence from an NGS platform after 

sequencing (Fig. 1). The workflow comprises the input of DNA of 

short length, an NGS run to generate sequence-verified DNA clones, 

the identification of DNA with desired sequence on the sequencer’s 

substrate and the retrieval of the clones of choice. The sources for 

the input DNA are for the most part independent of the megacloning 

step. For the present work, input DNA was derived from conventional 

oligonucleotide synthesis and from DNA microarrays. We used 

the NGS platform GS FLX from Roche 454 Life Sciences 9,10 . Owing 

to its open-top architecture, accessibility of the beads and the bead 

size, this platform is well suited for a pick-and-place approach using 

micropipettes to retrieve specific beads from the 454-Picotiterplate 

(PTP) and transfer them into conventional multi-well plates for further 

processing. 

First, we established a technical setup for the controlled extraction 

of beads. The PTP at this stage contained a natural sample from 

human DNA, and extraction was done using a micropipette controlled 

by a microactuator device (Supplementary Data). To assess the fidelity 

of our setup, we compared the reads coming from the GS FLX 

platform with Sanger-derived sequences of DNA amplified from 

extracted beads. The alignment of Sanger sequences to the NGS reads 

matched 99.9%. Only two mismatches were obtained in 2,410 bp. 

Both were putative insertions in the GS FLX reads occurring at 

homopolymer stretches and therefore have a high likelihood of being 

platform-specific, base-calling artifacts 9 (Supplementary Data). 

Next we collected a set of 319 beads with DNA clones from a microarray-derived 

pool initially containing 3,918 sequences. The beads for 

extraction were selected to ensure that their GS FLX reads perfectly 

matched sequences in our starting pool. The obtained DNA and the 

1 febit group, Heidelberg, Germany. 2 Stanford Genome Technology Center, Stanford University, Palo Alto, California, USA. 3 Harvard Medical School, Boston, 

Massachusetts, USA. 4 Wyss Institute for Biologically Inspired Engineering, Boston, Massachusetts, USA. 5 These authors contributed equally to this work. 

Correspondence should be addressed to G.M.C. (gmc@harvard.edu). 

Received 8 June; accepted 19 October; published online 28 November 2010; doi:10.1038/nbt.1710 


l e t t e r s 


Data 

Data 

Oligonucleotide design 

Write 

(oligonucleotide synthesis) 

Microarrays 

Natural sources 

Conventional 

oligonucleotide synthesis 

Microarray synthesis 

-Oligo elution 

-Amplification 

Read 

Next-generation 

sequencing 

Roche/454 

ABI/SOLiD 

Illumina 

GS FLX 

Roche/454 Sort 

& 

select 

Next-generation 

sequencing 

Pick 

& 

place 

untreated pool were compared after being sequenced independently 

on a Genome Analyzer II (Illumina GAII). We mapped 3.1% of reads 

from the initial (nonenriched) DNA pool without errors to the set 

of 319 selected sequences. In the enriched pool the fraction of reads 

mapping perfectly to the target sequences was 84.3%. The increase 

by a factor of 27.2 shows clearly a successful enrichment of selected 

and correct sequences (Fig. 2a,b). Also the analysis of reads on the 

level of single-target sequences shows that for 94% of the sequences 

in the selected pool, 50% or more of the reads were correct (Fig. 2c). 

Error-prone sequences contained a high number of different species 

likely to be caused by known sequence variations on the GAII, as 

reported previously 11 . 

To test the assembly of gene fragments based on megacloned 

oligonucleotides stemming from a microarray, we assembled two 

gene fragments, each ~220 bp in length, combining either nine 

or ten megacloned, bead-derived amplicons in a PCR-based gene 

assembly reaction 12,13 . The obtained assemblies were cloned and 

Sanger sequenced. Seven out of eight clones matched the target 

sequence perfectly. Interestingly, one clone showed insertions and 

deletions all located within a region 23 bp wide. Errors in assemblies 

originating from inaccuracies in the starting material could 

be expected to be distributed evenly over the entire construct. As 

this sequence was otherwise free of errors, these defects were likely 

caused by misassembly rather than errors in the building blocks used 

(Supplementary Data). 

To further evaluate the capabilities of the megacloning approach to 

generate biologically functional genes, we applied the method to DNA 

fragments 274–394 bp in length and extracted 32 beads from the PTP 

carrying putatively correct sequences. These DNA fragments were 

the product of gene assembly reactions 12 using overlapping 40-mer 

oligonucleotides synthesized using conventional phosphoramidite 

chemistry and could be assembled into a model gene encoding β-dglucuronidase 

(uidA) 14 (2,080 bp). 

Three Sanger sequences obtained from the bead DNA were totally 

unrelated to the expected sequence and were probably caused by 

wrong bead extraction or contamination. The remaining 29 sequences 

PTP 

Select 

Pick & place 

Optical release 

Copy off surface 

Multiwell plate 

Amplify 

PCR 

Amplify 

PCR 

Perfect part 

Figure 1 Coalescence of DNA reading and writing. The general approach begins with DNA from 

a variety of sources. Here we used oligonucleotides synthesized from microarrays as well as from 

conventional sources. Then, next-generation sequencing is used to read and identify oligonucleotides 

with desired sequences. Here we used the GS FLX platform (454/Roche). Finally, the DNA is sorted 

and retrieved selectively, in this case with a microactuator-controlled micropipette guided by two 

microscope cameras. The technologies used for retrieval depend on the sequencing platform. 

covered 7,195 bp and matched without 

errors to the expected target sequences 

(Supplementary Data). 

We then assembled the model gene out of 

nine DNA fragments from the set of 29 matching 

beads. The full-length gene construct 

was again checked by Sanger sequencing for 

absence of errors, and the biological functionality 

of the gene was tested in an enzymatic 

assay based on the conversion of X-Glc 

(5-bromo-4-chloro-3-indolyl-β-glucoside) 

substrate into blue dye 15 (Supplementary 

Data). Besides the proof of feasibility of 

generating biological functional genes, this 

experiment further mimics other applications 

of our technology, such as the use of sheared 

natural DNA and its subsequent sorting 

and reordering. 

The absence of errors in 7,195 bp of DNA 

obtained from 29 extracted beads raised the 

question of achievable error rates from the 

megacloner process. Therefore we explored 

the potential of megacloning using a statistical 

model. This model considers two main sources 

of error—namely, wrong sequencing calls and 

polymerase errors during DNA amplification 16 . The calculations estimated 

the chance of finding one error in our extracted sequence space 

of ~7,200 bp to be 29%, which is in line with our experimental findings. 

The theoretical error rate of bead amplicons after megacloning using 

the setup employed in this study was estimated to be 1 error in 21 kbp 

(Supplementary Data). Compared with the error rate in the starting 

material of 1 error in 40 bp (determined from GAII data of the initial 

microarray pool), this equals a 500-fold error reduction. 

We further calculated the expected amount of reads from NGS 

that match the target sequences of a given pool without errors. These 

numbers are crucial to estimate the complexity of pools that can be 

processed in one megacloner run. The resulting efficiency and cost 

structure are influenced mainly by three parameters: the error rate of 

the starting pool, the sequencing accuracy and the length of the variable 

sequence (Supplementary Data). With an error rate of 1 error 

in 40 bp and an average sequencing accuracy of 99.9% in the GS FLX, 

we expect a five- to tenfold cost reduction in producing DNA fragments 

(compared to conventional oligonucleotide synthesis) that can 

be achieved now with the prototype device (Supplementary Data). 

Because these fragments are largely free of errors, further savings can 

be expected in gene synthesis because the cost of subsequent sequencing 

for final quality control will be lower. 

In this work we demonstrated the targeted retrieval of bead-bound 

DNA from a high-throughput sequencer without major modifications 

to the sequencing process. Previous methods for error correction in 

DNA pools 7,17–21 do not adequately handle collections of closely related 

oligonucleotide sequences that occur during assembly of repetitive 

sequences or multi-gene family libraries. They also do not enable hierarchical 

assembly strategies, which are made possible by the ordered 

selection and physical separation of clonal DNA described here. 

The megacloner process has been proven to be useful for retrieval 

and sorting of correct and functional sequences and to increase the 

portion of error-free sequences in a sample substantially. This technology 

allows the processing of DNA from microarrays but also from 

a variety of other sources, such as conventional oligonucleotide synthesis 

or natural DNA fragments. 


l e t t e r s 

Megacloning could be optimized beyond the estimates in this work 

of one error in 21 kbp from input DNA having an error rate of 1 in 

40 bp. Although such raw material can be obtained by state-of-the-art 

microarray technologies, the quality of input DNA could be increased 

further by addressing the amplification step of bead-bound DNA—for 

example, with higher fidelity polymerases, as the predicted contribution 

of the polymerase to the error rate is 4.7-fold higher than the 

expected error rate of the megacloner itself (Supplementary Data). 

Another accessible parameter for optimizing the overall process in 

terms of error rates is improvement in the quality of the DNA starting 

material. Also, optimization of sequencing accuracy could be a way to 

improve the ability to select correct parts after NGS. This is, however, 

the subject of ongoing optimization in the scope of NGS development, 

including ligase-based methods with improved accuracy 22 . 

The pool used in our conceptual study contained ~4,000 

sequences. According to our results and extrapolations, this can be 

increased to ~30,000 sequences per pool with the described setup. 

As the bead extraction is generally independent of the pool complexity, 

it is mainly limited by the NGS platform and the quality of 

the starting material (Supplementary Data). More advanced microarray 

formats are able to deliver libraries with even higher complexity 

and of sufficient quality to fit into a gene assembly process 23 . 

Therefore, with an appropriate degree of automation that reaches 

an extraction frequency of two or three beads per minute, which is 

achievable with state-of-the-art robotics, the work-up of one PTP 

becomes possible within days, resulting in > 10 6 bp per plate. Hence, 

the downstream process (amplification, cleanup, assembly) will 

represent the next bottleneck. 


a 

c 

Percent of total read count 

Percent of correct matching reads 

100 

80 

60 

40 

20 

100 

90 

80 

70 

60 

50 

40 

30 

20 

10 

0 

Set 1 Set 2 Set 3 Set 4 

98.45 

96.53 

85.33 

84.28 

69.38 

Reads matching with 

max 3 errors in 

whole pool 

Before megacloning 

33.51 

Reads with 

perfect match 

in whole pool 

6.07 

Reads matching 

with max 3 errors 

in selected 

sequences 

After megacloning 

3.07 

Reads with 

perfect match 

in selected 

sequences 

b 

Number of sequences 

90 

80 

70 

60 

50 

40 

30 

20 

10 

0 

0 

1–10 

11–20 

21–30 

31–40 

41–50 

51–60 

61–70 

71–80 

81–90 

91–100 

Before megacloning 

After megacloning 

Reads/filtered reads in parts per million 

101–200 

201–300 

301–400 

401–500 

501–600 

601–700 

701–800 

801–900 

901–1,000 

1,001–2,000 

2,001–3,000 

3,001–4,000 

4,001–5,000 

5,001–6,000 

6,001–7,000 

7,001–8,000 

8,001–9,000 

9,001–10,000 

More 

0 

0 50 100 150 200 250 300 

100 

Percent of total read count 

80 

60 

40 

20 

0 

0 50 100 150 

Selected oligos 

200 250 300 

Figure 2 NGS-based comparison of untreated and megacloned oligonucleotide pools from microarray. (a) Comparison of the initial microarray 

oligonucleotide pool (blue) and the pool enriched with the megacloner technology (red) based on the results of the Illumina GAII runs. The bars in 

set 1 represent the fraction of reads that could be mapped allowing up to three errors. Bars in set 2 show the fractions of perfectly matching reads to 

the sequence set of the initial pool (3,918 sequences). The difference between the blue and the red bar in set 2 represents the enrichment of correct 

sequences by megacloning. The bars in set 3 and set 4 show the fractions of reads mapping to sequences from the selected pool of 319 sequences. 

The difference between blue and red bars in set 3 shows the enrichment of a selected 319 sequences before megacloning compared with after. Blue 

and red bars in set 4 represent the enrichment of sequences that are in the set of 319 selected sequences and that are correct. (b) Histogram of read 

counts in the Illumina GAII data of the initial pool (blue) and the enriched megacloned sample (red). Only reads mapping without errors to one of the 

319 selected target sequences have been taken into account. To compare the two NGS runs on the basis of read counts, we converted the numbers 

into parts-per-million (p.p.m.) from the total number of filtered reads. (c) Composition of reads from the Illumina GAII data including 319 selected 

sequences in the initial pool (top) and the enriched pool (bottom). The oligonucleotides are sorted by the fraction of correct reads. Green, correct reads; 

red, error-prone reads (compartments in the red bars represent single sequences with a read count of 0.1% or more of total reads for the particular 

sequence); light blue, sum of nonunique error-prone reads where each sequence represents less than 0.1% of total reads for the particular sequence; 

blue, unique reads. In the Illumina GAII data set from the enriched sample, just 315 out of 319 selected sequences could be detected. 


l e t t e r s 


Our next focus in the present context is improvement and 

automation of physical bead extraction. The workflow used in this 

study still involved a considerable number of manual steps and some 

human intervention, which was identified as the most important 

source of error in terms of extraction of unwanted beads. Therefore, 

the success rate of ~90% (29 beads out of 32) has to be increased for 

the bead localization and retrieval process. 

The method described here holds the potential to decrease production 

cost for synthetic DNA by one or more orders of magnitude. This 

source of high-quality DNA could aid the field of synthetic biology, as 

well as the production of libraries for antibodies or enzyme variants. 

In addition to synthetic sources, the sorting of natural DNA could 

enable the quick reconstruction or combination of DNA fragments 

to assemble genes, chromosomes or genomes, while simultaneously 

including synthetic parts of DNA. 

The principle that we applied here using the GS FLX technology 

should also be generally applicable to other available NGS platforms 

such as Illumina’s GAII, SOLiD, the Polonator or others. In the 

present context, the advantage of the GS FLX platform is the robotaccessible 

platform architecture and the comparably large size of the 

beads. Owing to different architectures of the other platforms, such 

as partially closed systems and substantially smaller DNA carriers, 

harvesting DNA from those will require a different mechanism, such 

as optical approaches including photosensitive and cleavable linkermolecules. 

The advantage of these platforms is a considerably higher 

number of DNA clones, which potentially could increase the capacity 

and throughput of the technology up to the gigabase level. 

Methods 





We thank B.A. Roe, F.Z. Najar and D.D. White for sequencing support, J. Jäger for 

technical consulting, and D. Summerer, T. Brefort, S. Kosuri and D. Levner for 

discussions and comments. 


M.M., P.F.S. and G.M.C. conceptualized the megacloning method and wrote the 

manuscript; M.M. designed and lead the study, wrote all algorithms for sequence 

design, data analysis, image conversion, image processing and microactuator 

control; M.M., N.K., N.S. acquired the used technology, set up the microactuator 

device and optical systems; N.S. designed the uidA genetic model; M.M., N.K., 

N.S., V.B. and P.H. designed and optimized molecular biological methods; C.F.S. 

and J.T.L. contributed to bead picking and engineering concepts; A.K. set up the 

statistical models and calculations; J.T.L. contributed to the design of molecular 

biological steps and the acquisition of sequencing samples; B.G. and F.B. evaluated 

and implemented necessary changes into the sample preparation and the 

sequencing process on the 454/Roche platform. 


The authors declare competing financial interests: details accompany the full-text 

HTML version of the paper at http://www.nature.com/naturebiotechnology/. 




1. Endy, D. Foundations for engineering biology. Nature 438, 449–453 (2005). 

2. Menzella, H.G. et al. Combinatorial polyketide biosynthesis by de novo design and 

rearrangement of modular polyketide synthase genes. Nat. Biotechnol. 23, 

1171–1176 (2005). 

3. Gibson, D.G. et al. Creation of a bacterial cell controlled by a chemically synthesized 

genome. Science 329, 52–56 (2010). 

4. Carr, P.A. & Church, G.M. Genome engineering. Nat. Biotechnol. 27, 1151–1162 

(2009). 

5. Gao, X. et al. A flexible light-directed DNA chip synthesis gated by deprotection 

using solution photogenerated acids. Nucleic Acids Res. 29, 4744–4750 (2001). 

6. Singh-Gasson, S. et al. Maskless fabrication of light-directed oligonucleotide 

microarrays using a digital micromirror array. Nat. Biotechnol. 17, 974–978 

(1999). 

7. Tian, J. et al. Accurate multiplex gene synthesis from programmable DNA microchips. 

Nature 432, 1050–1054 (2004). 

8. Porreca, G.J. et al. Multiplex amplification of large sets of human exons. Nat. 

Methods 4, 931–936 (2007). 

9. Margulies, M. et al. Genome sequencing in microfabricated high-density picolitre 

reactors. Nature 437, 376–380 (2005). 

10. Wicker, T. et al. 454 sequencing put to the test using the complex genome of 

barley. BMC Genomics 7, 275 (2006). 

11. Willenbrock, H. et al. Quantitative miRNA expression analysis: comparing microarrays 

with next-generation sequencing. RNA 15, 2028–2034 (2009). 

12. Stemmer, W.P., Crameri, A., Ha, K.D., Brennan, T.M. & Heyneker, H.L. Single-step 

assembly of a gene and entire plasmid from large numbers of oligodeoxyribonucleotides. 

Gene 164, 49–53 (1995). 

13. Richmond, K.E. et al. Amplification and assembly of chip-eluted DNA (AACED): a method 

for high-throughput gene synthesis. Nucleic Acids Res. 32, 5011–5018 (2004). 

14. Jefferson, R.A., Burgess, S.M. & Hirsh, D. beta-Glucuronidase from Escherichia 

coli as a gene-fusion marker. Proc. Natl. Acad. Sci. USA 83, 8447–8451 

(1986). 

15. Couteaudier, Y., Daboussi, M.J., Eparvier, A., Langin, T. & Orcival, J. The GUS gene 

fusion system (Escherichia coli beta-D-glucuronidase gene), a useful tool in studies 

of root colonization by Fusarium oxysporum. Appl. Environ. Microbiol. 59, 

1767–1773 (1993). 

16. Cline, J., Braman, J.C. & Hogrefe, H.H. PCR fidelity of pfu DNA polymerase and 

other thermostable DNA polymerases. Nucleic Acids Res. 24, 3546–3551 (1996). 

17. Carr, P.A. et al. Protein-mediated error correction for de novo DNA synthesis. Nucleic 

Acids Res. 32, e162 (2004). 

18. Smith, J. & Modrich, P. Removal of polymerase-produced mutant sequences from 

PCR products. Proc. Natl. Acad. Sci. USA 94, 6847–6850 (1997). 

19. Bang, D. & Church, G.M. Gene synthesis by circular assembly amplification. 

Nat. Methods 5, 37–39 (2008). 

20. Fuhrmann, M., Oertel, W., Berthold, P. & Hegemann, P. Removal of mismatched 

bases from synthetic genes by enzymatic mismatch cleavage. Nucleic Acids Res. 

33, e58 (2005). 

21. Binkowski, B.F., Richmond, K.E., Kaysen, J., Sussman, M.R. & Belshaw, P.J. 

Correcting errors in synthetic DNA through consensus shuffling. Nucleic Acids Res. 

33, e55 (2005). 

22. McKernan, K.J. et al. Sequence and structural variation in a human genome 

uncovered by short-read, massively parallel ligation sequencing using two-base 

encoding. Genome Res. 19, 1527–1541 (2009). 

23. Kosuri, S. et al. Scalable gene synthesis by selective amplification of DNA pools from 

high-fidelity microchips. Nat. Biotechnol. advance online publication, doi:10.1038/ 

nbt.1716 (28 November 2010). 




Oligo synthesis, sequence design, adaptors. Oligonucleotides used for this 

work were synthesized on programmable microarray synthesizers using lightdirected 

synthesis methods 5 . Conventional oligonucleotides used for gene 

assembly were obtained from Sigma Aldrich. Harvesting of oligonucleotides 

from microarray surfaces was performed by chemical cleavage of succinateester 

bonds using ammonia hydrochloride solution. 

Amplification of microarray-derived oligonucleotide pools by emulsion 

PCR. Microarray-derived oligonucleotide pools were amplified before NGS 

using emulsion PCR 24 . Therefore universal terminal sequences were attached 

during synthesis and served as primer regions. Amplification primers contained 

adaptors for sequencing on the Illumina GAII platform and/or the 454 

GS FLX (Supplementary Data). 

Sequencing on the 454 GS FLX. The sample preparation for the PCRamplified 

oligonucleotides was done according to the manufacturer’s protocols 

(Roche/454). To keep the DNA intact after sequencing, we exchanged the 

bleaching cleaning buffer with TE buffer before the sequencing run to avoid 

degradation of DNA during the final cleaning steps of the Roche sequencer. 

Data analysis of 454 data and image conversion. NGS reads obtained from 

the GS FLX sequencer were aligned to the target sequences in the oligonucleotide 

pool to find the best matching sequence for every read and to perform 

further analysis, such as error rate estimation. Perfect matching sequences 

were selected and localized in the sequencer image by using the coordinates 

attached to every read sequence. For sequence data analysis, we used various 

Python scripts using the BioPython package. The images from the GS FLX 

sequencer were converted into the TIFF standard format using the Python 

Imaging Library. 

Bead localization and extraction. After aligning the GS FLX reads to the 

set of target sequences, we selected reads that perfectly matched one of the 

desired oligonucleotide sequences in the pool. For localization of beads we 

located the corresponding chemiluminescent signals in the converted raw 

image from the GS FLX platform using the x- and y-coordinates that were 

included in the NGS raw data. To locate beads in the PTP, we identified reference 

points in the raw image and their corresponding positions in the PTP 

using suitable patterns of light signals. Based on these reference points the 

bead positions on the PTP were calculated using an algorithm for scaling and 

rotation. The extraction was performed with a micropipette with an outer 

diameter of 28 μm. For pipette handling we used a three-axis microactuator 

(Supplementary Data). Before extraction of beads the PTP was stored under 

a water layer to prevent desiccation and shrinking of beads. After picking, the 

beads were transferred immediately into a PCR vial and stored under water 

until further processing. 

Amplification of DNA from beads. Amplification of bead-bound DNA 

was performed with the primers 454-A and 454-B, targeting the Roche/454 

adaptors, or ‘slx-fw-long’ and ‘slx-rev-long’ for Illumina adaptors. For amplification 

of fragments with 40-mer variable regions, primers were 5′-biotinylated 

to facilitate subsequent removal of primer regions on a streptavidin 

matrix. PCR conditions: 20 mM Tris-HCl (pH 8.8), 10 mM ammoniumsulfate, 

10 mM potassium chloride, 2 mM magnesium-sulfate, 0.1% 

Triton X-100, 200 μM each dNTP, 2% (vol/vol) DMSO, 1 μM each primer, 

50 U/ml native pfu polymerase (Fermentas). Cycling: initial denaturation 

96 °C (2 min); then 30 cycles of 96 °C (30 s), 63 °C (30 s), 72 °C (30 s) and 

final elongation 72 °C (3 min). After amplification, all PCR products were 

analyzed on PAGE (Supplementary Data) to check specificity and yield. 

For generation of the subpool containing 319 sequences, we estimated the 

concentration on the basis of the gel analysis and mixed the amplicons in 

equimolar concentrations. 

Illumina sequencing and data analysis. As the sample contained suitable 

adaptors all steps regarding adaptor ligation have been omitted. All other steps 

were done according to the protocols from Illumina. 

The NGS raw data obtained from Illumina GAII were processed by the 

following steps. 

1. Truncation of reads to the length of the variable regions (40 bp). 

2. Filtering out reads containing ambiguities (filtered reads). 

3. Group reads with similar sequences (bins). 

Subsequently for each read we identified the best matching target sequence 

from the oligonucleotide pool by mapping all reads to a pseudo-genome using 

rapid alignment of small RNA reads (razerS) (http://www.seqan.de/projects). 

The pseudo-genome was generated by concatenation of the variable parts 

of pool sequences separated by 40-mer poly-T stretches. The corresponding 

target sequence could then be determined by the matching position in the 

pseudo-genome. Alignments from the razerS output were used to determine 

insertions, deletions and substitutions. To compare the two GAII runs based 

on the number of correct reads, we converted the read counts into parts-permillion 

units (p.p.m.), taking the number of filtered reads before the matching 

procedure (after step 2) as a basis. 

Assembly of gene fragments from conventional oligonucleotides. Gene fragments 

> 200 bp were assembled from conventionally synthesized 40-mer oligonucleotides 

having a constant overlap region of 20 nucleotides to the adjacent 

oligomer. Primer regions for 454 sequencing and restriction sites for primer 

removal were included during assembly. The assembly reaction contained 

5 nM of each construction oligonucleotide and 200 nM of terminal primers. 

PCR conditions: 1× KOD polymerase buffer (Novagen), 1.25 mM MgSO 4 , 

40 μM each dNTP, 5 U/ml KOD Hot Start Polymerase (Novagen). Cycling 

for gene assembly: initial denaturation 96 °C (4 min); then 30 cycles of 96 °C 

(10 s), 55–40 °C touchdown (30 s), 72 °C (10 s). For subsequent amplification: 

96 °C (10 s), 55 °C (30 s), 72° (30 s), final elongation 72 °C (3 min). 

Assembly of genes from >200 bp fragments. Gene assembly up to 2 kbp 

were performed according to the protocol used for assembly of > 200 bp from 

oligonucleotides. 

Primer removal and cleanup of bead amplicons before gene assembly. For 

removal of primer regions amplicons were incubated with LguI restriction endonuclease 

in 1× Tango buffer (Fermentas) for 3 h at 37 °C. For > 200 bp fragments, 

small restriction fragments containing primer regions were removed by PCR 

purification columns (GenElute PCR Clean-Up, Sigma Aldrich). For cleanup 

of microarray-derived fragments, we used 40-mer variable region biotinylated 

primers during bead DNA amplification and removed restriction products containing 

biotin residues using streptavidin matrix. The 40-mer fragments were 

ethanol precipitated and dissolved in water before further processing. 

Assembly of genes from 40-mer double-stranded DNA fragments. For the 

assembly of genes from 40-mer dsDNA we used a two-stage assembly protocol 

including a primerless PCR followed by a PCR for amplification of the 

resulting products described previously 13 . 

24. Williams, R. et al. Amplification of complex gene libraries by emulsion PCR. 

Nat. Methods 3, 545–550 (2006). 

doi:10.1038/nbt.1710 


l e t t e r s 

Scalable gene synthesis by selective amplification 

of DNA pools from high-fidelity microchips 

Sriram Kosuri 1,2,6 , Nikolai Eroshenko 1,3,6 , Emily M LeProust 4 , Michael Super 1 , Jeffrey Way 1 , 

Jin Billy Li 2,5 & George M Church 1,2 


Development of cheap, high-throughput and reliable gene 

synthesis methods will broadly stimulate progress in biology 

and biotechnology 1 . Currently, the reliance on columnsynthesized 

oligonucleotides as a source of DNA limits further 

cost reductions in gene synthesis 2 . Oligonucleotides from 

DNA microchips can reduce costs by at least an order of 

magnitude 3–5 , yet efforts to scale their use have been largely 

unsuccessful owing to the high error rates and complexity of 

the oligonucleotide mixtures. Here we use high-fidelity DNA 

microchips, selective oligonucleotide pool amplification, 

optimized gene assembly protocols and enzymatic error 

correction to develop a method for highly parallel gene 

synthesis. We tested our approach by assembling 47 genes, 

including 42 challenging therapeutic antibody sequences, 

encoding a total of ~35 kilobase pairs of DNA. These 

assemblies were performed from a complex background 

containing 13,000 oligonucleotides encoding ~2.5 megabases 

of DNA, which is at least 50 times larger than in previously 

published attempts. 

The synthesis of DNA encoding regulatory elements, genes, pathways 

and entire genomes provides powerful ways to both test biological 

hypotheses and harness biology for our use. For example, from the 

use of oligonucleotides in deciphering the genetic code 6,7 to the recent 

complete synthesis of a viable bacterial genome 8 , DNA synthesis has 

engendered tremendous progress in biology. Currently, almost all 

DNA synthesis relies on the use of phosphoramidite chemistry on 

controlled-pore glass (CPG) substrates. The synthesis of gene-sized 

fragments (500–5,000 base pairs (bp)) relies on assembling many 

CPG oligonucleotides together using a variety of gene synthesis techniques 

2 . Technologies to assemble verified gene-sized fragments into 

much larger synthetic constructs are now fairly mature 8–12 . 

The price of gene synthesis has fallen drastically over the last decade. 

However, the current commercial price of gene synthesis, ~$0.40– 

1.00/bp, has begun to approach the relatively stable cost of the CPG 

oligonucleotide precursors (~$0.10–0.20/bp) 1 , suggesting that oligonucleotide 

cost is limiting. At these prices, the construction of large gene 

libraries and synthetic genomes is out of reach to most. There are many 

ongoing efforts to lower the cost of gene synthesis that focus on reducing 

the cost of the oligonucleotide precursors. For example, microfluidic 

oligonucleotide synthesis can reduce reagent cost by an order of magnitude 

and has been used for proof-of-concept gene synthesis 13 . 

Another promising route is to harness existing DNA microchips, 

which can produce up to a million different oligonucleotides on a single 

chip, as a source of DNA. Previous efforts have demonstrated that 

genes can be synthesized from DNA microchips 3–5,14 . Thus far it has 

not been possible to scale up these approaches for at least three reasons. 

First, the error rates of oligonucleotides from DNA microchips 

are higher than traditional column-synthesized oligonucleotides. 

Second, the assembly of gene fragments becomes increasingly difficult 

as the diversity of the oligonucleotide mixture becomes larger. Finally, 

the potential for cross-hybridization between individual assemblies 

imposes strong constraints on the sequences that can be constructed 

on an individual microchip. 

Recently, the quality of microchip-synthesized oligonucleotides was 

improved by controlling depurination during the synthesis process 15 . 

These arrays produce up to 55,000 200-mer oligonucleotides on a 

single chip and are sold as a ~1–10 picomole pools of oligonucleotides, 

termed OLS pools (oligo library synthesis). Several groups have 

used OLS pools in DNA capture technologies, promoter analysis and 

DNA barcode development 16–20 . We have previously shown that individual 

oligonucleotides in a 55,000 150-mer OLS pool were evenly 

distributed 18 . We reanalyzed this data set to provide an estimate of 

the frequency of transitions, transversions, insertions and deletions 

in this OLS pool (Online Methods) and found the overall error rate 

to be ~1/500 bp both before and after PCR amplification, suggesting 

that OLS pools can be used for accurate large-scale gene synthesis 

(Supplementary Table 1). 

To test whether OLS pools could be used for DNA microchipbased 

gene synthesis, we designed two pools (OLS pools 1 and 2) 

of different lengths, each containing ~13,000 130-mer or 200-mer 

oligonucleotides, respectively. Figure 1 is a general schematic of our 

methods for using OLS pools to perform gene synthesis. Briefly, we 

designed oligonucleotides that were then printed on DNA microchips 

and recovered as a mixed pool of oligonucleotides (OLS pool). 

Next, we took advantage of the long oligonucleotide lengths to 

1 Wyss Institute for Biologically Inspired Engineering, Boston, Massachusetts, USA. 2 Department of Genetics, Harvard Medical School, Boston, Massachusetts, USA. 

3 Harvard School of Engineering and Applied Sciences, Cambridge, Massachusetts, USA. 4 Agilent Technologies, Santa Clara, California, USA. 5 Present address: 

Department of Genetics, Stanford University, Stanford, California, USA. 6 These authors contributed equally to this work. Correspondence should be addressed to 

S.K. (sri.kosuri@wyss.harvard.edu) or N.E. (eroshenk@wyss.harvard.edu). 

Received 3 August; accepted 25 October; published online 28 November 2010; doi:10.1038/nbt.1716 


l e t t e r s 

a 

b 

c 

DNA microchip 

OLS pool 

Figure 1 Schematic for scalable gene synthesis from OLS pool 2. 

(a,b) Pre-designed oligonucleotides (no distinction is made between 

dsDNA and ssDNA in the figure) are synthesized on a DNA microchip (a) 

and then cleaved to make a pool of oligonucleotides (b). (c) Plate-specific 

primer sequences (yellow or brown) are used to amplify separate plate 

subpools (only two are shown), which contain DNA to assemble different 

genes (only three are shown for each plate subpool). (d) Assembly-specific 

sequences (shades of blue) are used to amplify assembly subpools that 

contain only the DNA required to make a single gene. (e) The primer 

sequences are cleaved using either type IIS restriction enzymes (resulting 

in dsDNA) or by DpnII/USER/λ exonuclease processing (producing 

ssDNA). (f) Construction primers (shown as white and black sites flanking 

the full assembly) are then used in an assembly PCR reaction to build 

a gene from each assembly subpool. Depending on the downstream 

application, the assembled products are then cloned either before or after 



d 

e 

f 

Plate subpools 

Assembly subpools 

Processed subpools 

Assembled genes 

independently PCR amplify and process only those oligonucleotides 

required for a given gene assembly. For the 200-mer OLS pool 2, we 

first amplified a ‘plate subpool’ that contained DNA to construct 

up to 96 genes, and then amplified individual ‘assembly subpools’ 

to separate the oligonucleotides for an individual gene. For the 

130-mer OLS pool 1, we directly amplified assembly subpools, 

foregoing the plate subpool step. Next, the primers used for these 

amplification steps were removed by either type IIS restriction endonucleases 

to form double-stranded DNA (dsDNA) fragments (OLS pool 2), 

or a combination of enzymatic steps to form single-stranded DNA 

(ssDNA) fragments (OLS pool 1). Finally, we constructed full-length 

genes using PCR assembly, performed enzymatic error correction to 

improve error rates if necessary, and, finally, cloned and characterized 

the constructs. 

Obtaining subpools of only those DNA fragments required for any 

particular assembly is crucial for robust gene synthesis in very complex 

DNA backgrounds. In addition, isolating subpools relieves constraints 

on sequence similarity inherent in past approaches. To facilitate the 

partitioning of OLS pools into smaller subpools, we designed 20-mer 

PCR primer sets with low potential cross-hybridization (‘orthogonal’ 

primers) derived from a set of 244,000 25-mer orthogonal sequences 

developed for barcoding purposes 21 . Two separate orthogonal primer 

sets were constructed for the different OLS pools because of their 

varying requirements for downstream processing. Both sets were 

screened for potential cross-hybridization, low secondary structure 

and matched melting temperatures to construct large sets of orthogonal 

PCR primer pairs. 

To construct genes from the OLS pools, we developed algorithms to 

split the sequence into overlapping segments with matching melting 

temperatures such that they could be later assembled by PCR. Genes 

on OLS pool 1 and 2 were designed differently to test the effect of 

different overlap lengths. We designed genes on OLS pool 1 such 

that the processed ssDNA pools fully overlapped to form a complete 

dsDNA sequence. In OLS pool 2, the processed dsDNA fragments 

partially overlapped by ~20 bp and could be assembled into a contiguous 

gene sequence using PCR. We initially constructed a set of 

fluorescent proteins to test the efficacy of the gene synthesis methods 

on both OLS pools (Fig. 2). 

For OLS pool 1, we designed two independent ‘assembly subpools’ 

that encoded GFPmut3b plus flanking orthogonal primer sequences 

that were later used for PCR assembly (construction primers). The 

two assembly subpools, GFP43 and GFP35, differed in the average 

overlap length (43 and 35 bp, respectively), total length (82–90 and 

64–78 bases, respectively) and number of oligonucleotides (18 and 

22, respectively). We also designed two subpools, control subpools 1 

and 2, containing ten and five 130-mer oligonucleotides, respectively, 

to test amplification efficacy. The other eight subpools, containing a 

total of 12,945 130-mer sequences, were constructed on the same chip 

but were not used in this study except to provide potential sources 

of cross-hybridization. Each of these 12 subpools was flanked with 

independent orthogonal primer pairs (assembly-specific primers). 

As a control, we used these same algorithms to design a set of shorter 

column-synthesized oligonucleotides (20 bp average overlap; 35–45 

bases in length; and 39 total oligonucleotides) encoding GFPmut3b 

and obtained them from a commercial provider (IDT). These oligonucleotides 

were combined to form a third pool (GFP20) that was 

also tested. (All synthesized oligonucleotides used in the study can 

be found in Supplementary Sequences). 

Each of the four subpools (GFP43, GFP35, control 1 and control 2) 

were PCR amplified from the synthesized OLS pool using modified 

primers that facilitated downstream processing (Supplementary 

Figs. 1 and 2a) 18 . The oligonucleotides were then processed to remove 

primer sequences (Supplementary Figs. 2b and 3). Briefly, lambda 

exonuclease was used to make the PCR products single stranded, and 

then uracil DNA glycosylase, endonuclease VIII and DpnII restriction 

endonuclease were used to cleave off the assembly-specific primers. 

The resultant gel shows that although the reaction was efficient, 

unprocessed oligonucleotide still remained. In addition, we observed 

spurious cleavage by DpnII that was likely due to the extensive overlap 

within the subpool that is inherent in the gene synthesis process. 

We assembled the GFP43, GFP35 and GFP20 subpools using PCR, 

which resulted in GFP-sized products as well as many incorrect low 

molecular weight products (Fig. 2a). 


l e t t e r s 


Figure 2 Gene synthesis products. 

(a) Results of PCR assembly of GFPmut3 

from two different assembly subpools 

(GFP43 and GFP35) that were amplified 

from OLS pool 1. Full-length GFPmut3 is 

expected to be 779 bp and is indicated 

with an asterisk (*). Other bands show lower 

molecular weight misassembled products. 

(b) Gel purification and re-amplification 

of the full-length assembled GFPmut3. 

(c) Results of assembling three fluorescent 

proteins using the longer oligonucleotides in 

OLS pool 2 and a PCR assembly protocol 

that did not require gel isolation. (d) Results 

of assembling 42 variable regions of singlechain 

antibody fragments that contained 

challenging GC-rich linkers. Of the 

42 assemblies, all but two (7 and 24) 

resulted in strong bands of the correct size. 

We gel isolated and re-amplified these two, 

850 bp 

650 bp 

200 bp 

100 bp 

resulting in bands of the correct size (Supplementary Fig. 10b). The antibody that corresponds to each number is given in Supplementary Table 3 

and the amino acid sequence of the linker region used is given above each gel with differing amino acids in red. 

We gel isolated, digested and then cloned the assembly products 

into an expression vector (Fig. 2b and Supplementary Fig. 4). We 

used flow cytometry tests, manual colony counts and sequencing 

of individual clones to measure the error rates (Supplementary 

Fig. 5a,b). All three of the assays correlated well, and the error rates 

determined through sequencing were 1/1,500 bp, 1/1,130 bp and 

1/1,350 bp for the GFP43, GFP35 and GFP20 synthesis reactions, 

respectively (Fig. 3 and Supplementary Table 2). 

These results illustrate a number of important points. First, our 

subpool assembly primers were sufficiently well-designed to provide 

stringent subpool amplification of as few as 5 oligonucleotides 

a 

Fluorescent (percent) 

b 

Bp/error 

100 

90 

80 

70 

60 

50 

40 

30 

20 

10 

0 

2,000 

1,800 

1,600 

1,400 

1,200 

1,000 

800 

600 

400 

200 

0 

GFP20 

GFP43 

GFP35 

GFP43 ErrASE 

GFP35 ErrASE 

mTFP1 

mCitrine 

mApple 

a 

d 

650 bp 

650 bp 

mTFP1 ErrASE 

GFP43 

MW 

GFP35 

* 

out of a 12,995 oligonucleotide background. Second, the relative 

quantities of the oligonucleotides in the assembly subpools were sufficient 

to allow PCR assembly. Third, the error rates from 130-mer 

OLS pools were sufficiently low to construct gene-sized fragments 

(717 bp) such that >50% of the sequences were perfect. In fact, the 

error rates from both the GFP43 and GFP35 assemblies were indistinguishable 

from the column-synthesized GFP20 assemblies. Fourth, 

our data show that the level of fluorescence of our gene assemblies 

correlated with the number of constructs with perfect sequence, providing 

a useful screen to test fluorescent gene assemblies in OLS pool 2 

(Supplementary Fig. 6). Finally, although PCR assembly was able 

to generate full-length product, many smaller misassembled products 

were also formed, requiring the use of difficult-to-automate gel 

isolation steps. 

In OLS pool 2, we designed 836 assembly subpools split into 

11 plate subpools, encoding 2,456,706 bases of oligonucleotides that 

could potentially result in 869,125 bp of final assembled sequence. 

We first constructed three fluorescent proteins to test assembly protocols 

in OLS pool 2: mTFP1, mCitrine and mApple. We found that 

the PCR assembly protocols developed for ssDNA subpools in OLS 

pool 1 only produced short (

l e t t e r s 


isolation (Fig. 2c and Supplementary Fig. 7a,b). Cloning followed 

by flow cytometry screening showed that 6.8%, 7.5% and 6.8% of the 

cells were fluorescent for mTFP1, mCitrine and mApple assemblies, 

respectively (Fig. 3a). 

Assuming 6% correct sequence per construct and no selection 

against errors in the assembly process, the error rate was ~1/250 bp 

for 200-mer OLS pool 2, significantly above that of the estimates for 

130-mer OLS pool 1 (~1/1,000 bp) and the sequenced 55,000 150- 

mer OLS pool (~1/500 bp). This is not completely unexpected, as the 

amount of depurination is dependent upon the number of deprotection 

steps during synthesis and thus the oligonucleotide length. 

Despite the higher error rate, there were several advantages to the 

200-mer OLS pool 2. First, the extensive overlaps designed in OLS 

pool 1 caused spurious processing of the primers from the assembly 

subpools. The use of type IIs restriction endonucleases to process 

primers to form dsDNA resulted in more robust processing. Second, 

the use of two amplification steps conserves chip-eluted DNA to 

allow for future scaling of the gene synthesis process (Supplementary 

Note). Third, the assemblies of OLS pool 1 produced many smaller 

bands and required lower-throughput gel isolation procedures. This 

could be due to mispriming during PCR assembly because of the long 

overlap lengths used in the design process. The assemblies in OLS 

pool 2 used much shorter overlap lengths and resulted in no smaller 

molecular weight misassembled products. 

To improve the error rates of the genes assembled from OLS pool 2, 

we used ErrASE, a commercially available enzyme cocktail that 

detects and corrects mismatched base pairs, to remove errors in the 

assembled fluorescent proteins. For each gene, we applied ErrASE 

at six different stringencies, reamplified the constructs, cloned the 

PCR products and rescreened the cloned genes using flow cytometry. 

Improvement of the level of fluorescence progressively increased 

with greater ErrASE stringency. At the highest levels of error correction, 

the fluorescence levels were 31%, 49% and 26% for mTFP1, 

mCitrine and mApple respectively (Fig. 3a and Supplementary 

Fig. 8). We also performed the ErrASE procedure on our GFP43 

and GFP35 pools from OLS pool 1, resulting in fluorescence levels 

of 89% and 92%, respectively (Fig. 3a and Supplementary Fig. 5c). 

We sequenced clones of GFP43 and GFP35 and found three errors in 

21,510 (1/7,170 bp) and four errors in 20,076 (1/5,019 bp) sequenced 

bases, respectively. 

As a more challenging test for our DNA synthesis technology, 

we designed and synthesized oligonucleotides in OLS pool 2 for 

42 genes encoding the variable regions of single-chain antibody 

fragments (scFv) regions corresponding to a number of well-known 

antibodies. We have previously had trouble synthesizing these 

genes using commercial gene synthesis companies. This might be 

partly due to the prototype (Gly 4 Ser) 3 linker, which is designed 

to maximize flexibility and allow the heavy and light V regions to 

assemble 22 . The repetitive nature and high GC content of the linkerencoding 

sequences often represents a challenge for accurate DNA 

synthesis. We therefore tested three different linker sequences that 

varied in GC content and repetitive character of the linker encoding 

sequence. In addition, the presence of high sequence homology in 

the antibody backbones and linkers represented a potential source of 

cross-hybridization that could interfere with assembly (61% average 

sequence identity). 

As expected, the antibody sequences did not assemble as robustly 

as the fluorescent proteins, and thus we further optimized the conditions 

during pre- and post-assembly (Supplementary Figs. 7c, 9 

and 10a). Under the best protocol, 40 of the 42 constructs assembled 

to the correct size (Fig. 2d and Supplementary Table 3). The two 

misassembled genes displayed faint bands at the correct size, which 

were gel isolated and reamplified to produce strong bands of the correct 

size. We sequenced 15 antibodies including representatives from 

all three linker types. We performed enzymatic error correction using 

ErrASE, gel isolated the product and finally cloned the constructs 

into an expression vector. One of the 15 antibodies did not clone, 

and another had a deleted linker region in all 21 sequenced clones. 

Both of these antibodies were encoded with the highest GC content 

linker. The average error rate of the 14 antibodies that did clone was 

1/315 bp (Fig. 3b and Supplementary Table 2); this was considerably 

higher than the GFP assemblies, but still sufficient for construction 

of genes of this size (~10% of clones should be perfect, on average). 

In addition, the high levels of sequence similarity between the antibodies, 

combined with the successful assembly and sequencing (which 

showed no instances of cross-contamination) further validates that 

the selective amplification is at least stringent enough to make highly 

related protein sequences. 

Our results show the assembly of gene-sized DNA fragments 

totaling ~35,000 bp from oligonucleotide pools of more than 

50 kilobases. A number of key features are important to make the 

process work, including the use of low-error starting material, wellchosen 

orthogonal primers, subpool amplification of individual 

assemblies, optimized assembly methods and enzymatic error 

correction. Together, these features enabled gene assembly from 

oligonucleotide pools containing at least 50 times more sequences 

than previously reported (Supplementary Note). We describe two 

separate OLS pool lengths and assembly methods, which have their 

own advantages and disadvantages (Supplementary Fig. 1). The 

shorter, 130-mer OLS pool 1 assemblies have lower error rates, but 

because there are no plate amplifications, will be harder to scale as 

we begin to utilize larger OLS pools. The longer 200-mer OLS pool 2 

is easier to scale, but contained higher error rates. The costs of 

oligonucleotides in both processes are

l e t t e r s 








1. Carr, P.A. & Church, G.M. Genome engineering. Nat. Biotechnol. 27, 1151–1162 

(2009). 

2. Tian, J., Ma, K. & Saaem, I. Advancing high-throughput gene synthesis technology. 

Mol. Biosyst. 5, 714–722 (2009). 

3. Tian, J. et al. Accurate multiplex gene synthesis from programmable DNA microchips. 

Nature 432, 1050–1054 (2004). 

4. Richmond, K.E. et al. Amplification and assembly of chip-eluted DNA (AACED): a 

method for high-throughput gene synthesis. Nucleic Acids Res. 32, 5011–5018 

(2004). 

5. Zhou, X. et al. Microfluidic PicoArray synthesis of oligodeoxynucleotides and 

simultaneous assembling of multiple DNA sequences. Nucleic Acids Res. 32, 

5409–5417 (2004). 

6. Nirenberg, M.W. & Matthaei, J.H. The dependence of cell-free protein synthesis in 

E. coli upon naturally occurring or synthetic polyribonucleotides. Proc. Natl. Acad. 

Sci. USA 47, 1588–1602 (1961). 

7. Söll, D. et al. Studies on polynucleotides, XLIX. Stimulation of the binding of 

aminoacyl-sRNA’s to ribosomes by ribotrinucleotides and a survey of codon 

assignments for 20 amino acids. Proc. Natl. Acad. Sci. USA 54, 1378–1385 

(1965). 

8. Gibson, D.G. et al. Creation of a bacterial cell controlled by a chemically synthesized 

genome. Science 329, 52–56 (2010). 

9. Gibson, D.G. Synthesis of DNA fragments in yeast by one-step assembly of 

overlapping oligonucleotides. Nucleic Acids Res. 37, 6984–6990 (2009). 

10. Li, M.Z. & Elledge, S.J. Harnessing homologous recombination in vitro to generate 

recombinant DNA via SLIC. Nat. Methods 4, 251–256 (2007). 

11. Bang, D. & Church, G.M. Gene synthesis by circular assembly amplification. 

Nat. Methods 5, 37–39 (2008). 

12. Shao, Z., Zhao, H. & Zhao, H. DNA assembler, an in vivo genetic method for rapid 

construction of biochemical pathways. Nucleic Acids Res. 37, e16 (2009). 

13. Lee, C.-C., Snyder, T.M. & Quake, S.R. A microfluidic oligonucleotide synthesizer. 

Nucleic Acids Res. 38, 2514–2521 (2010). 

14. Kim, C. et al. Progress in gene assembly from a MAS-driven DNA microarray. 

Microelectron. Eng. 83, 1613–1616 (2006). 

15. LeProust, E.M. et al. Synthesis of high-quality libraries of long (150mer) 

oligonucleotides by a novel depurination controlled process. Nucleic Acids Res. 38, 

2522–2540 (2010). 

16. Patwardhan, R.P. et al. High-resolution analysis of DNA regulatory elements by 

synthetic saturation mutagenesis. Nat. Biotechnol. 27, 1173–1175 (2009). 

17. Schlabach, M.R. et al. Synthetic design of strong promoters. Proc. Natl. Acad. Sci. 

USA 107, 2538–2543 (2010). 

18. Li, J.B. et al. Multiplex padlock targeted sequencing reveals human hypermutable 

CpG variations. Genome Res. 19, 1606–1615 (2009). 

19. Li, J.B. et al. Genome-wide identification of human RNA editing sites by parallel 

DNA capturing and sequencing. Science 324, 1210–1213 (2009). 

20. Porreca, G.J. et al. Multiplex amplification of large sets of human exons. 

Nat. Methods 4, 931–936 (2007). 

21. Xu, Q. et al. Design of 240,000 orthogonal 25mer DNA barcode probes. Proc. Natl. 

Acad. Sci. USA 106, 2289–2294 (2009). 

22. Huston, J.S. et al. Medical applications of single-chain antibodies. Int. Rev. 

Immunol. 10, 195–217 (1993). 

23. Carr, P.A. et al. Protein-mediated error correction for de novo DNA synthesis. Nucleic 

Acids Res. 32, e162 (2004). 




Reanalysis of OLS pool error rates. We reanalyzed a previously published 

data set for determining sequencing error rates 18 . Briefly, the data set was 

derived from high-throughput sequencing using the Illumina Genome 

Analyzer platform of a 53,777 150-mer OLS pool. Two sequencing runs were 

performed, the first before any amplification, and the second after two rounds 

of ten cycles of PCR (20 cycles total). As our previous analyses were mostly 

looking for distribution effects, we reanalyzed these existing data to get an estimate 

of error rates before and after PCR amplification. We realigned the data 

set using Exonerate to allow for gapped alignments and analysis of indels 24 . 

Specifically, we used an affine local alignment model that is equivalent to the 

classic Smith-Waterman-Gotoh alignment, a gap extension penalty of −5, 

and used the full refine option to allow for dynamic programming–based 

optimization of the alignment. These reads were solely mapped on base calls 

by the Illumina platform. We used these alignments to count mismatches, 

deletions and insertions as compared to the designed sequences. However, 

as base-calling can be more error prone on next-generation platforms than 

traditional Sanger-based approaches, we filtered the results based only on 

high-quality base-calls (Phred scores of ≥30 or >99.9% accuracy). This was 

accomplished by converting Illumina quality scores to Phred values using 

the Maq utility sol2sanger 25 and only using statistics from base calls of Phred 

30 or higher. All error rate analysis scripts were implemented in Python and 

are available upon request. Although this method provides an estimate for 

error rates, unmapped reads may have higher error rates, thus underestimating 

the total average error rate. In addition, base-calling errors might still overestimate 

the error rate. Finally, using only high-quality base calls, which usually 

occur only in the first ten bases of a read, might only reflect error rates on the 

5′ end of the synthesized oligonucleotide. 

Design and synthesis of OLS pools. The 13,000 oligos in the first OLS library 

(OLS pool 1) were broken up into 12 separately amplifiable subpools (assembly 

subpools). Each assembly subpool was defined by unique 20 bp priming 

sites that flanked each of the oligos in the pool. The priming sites were 

designed to minimize amplification of oligos not in the particular assembly 

subpool. This was done by designing set of orthogonal 20-mers (assemblyspecific 

primers) using a set of 240,000 orthogonal 25-mers 21 as a seed. From 

these sequences we selected 20-mers with 3′ sequence ending in thymidine 

or GATC for the forward and reverse primers, respectively. We screened for 

melting temperatures of 62–64 °C and low primer secondary structure. After 

the additional filtering, 12 pairs of forward and reverse primers were chosen 

to be the assembly-specific primers. The 13,000 oligos in the second OLS 

library (OLS pool 2) were broken up into 11 subpools corresponding to 11 

sets of up to 96 assemblies (plate subpools), which were further divided into a 

total of 836 assembly subpools. A new set of orthogonal primers were designed 

similarly to the previous set (without the GATC and thymidine constraints) 

but further filtered to remove type IIS restriction sites, secondary structure, 

primer dimers and self-dimers. The final set of primer pairs was distributed 

among the plate-specific primers, assembly-specific primers and construction 

primers. See Supplementary Methods for more detailed design information 

and primer sequences. 

OLS pools were synthesized by Agilent Technologies and are available upon 

signing a Collaborative Technology Development agreement with Agilent. 

Costs of OLS pools are a function of the number of unique oligos synthesized 

and of the length of the oligos (

alemtuzumab, ranibizumab, cetuximab, efungumab, pertuzumab, tadocizumab 

and trastuzumab; see Fig. 2d and Supplementary Table 3). Using the same 

methods as with the first set of cloned antibodies, this second set was errorcorrected, 

gel-isolated, cloned and sequenced. See the Supplementary 

Methods for more detailed protocols. 

24. Slater, G.S. & Birney, E. Automated generation of heuristics for biological sequence 

comparison. BMC Bioinformatics 6, 31 (2005). 

25. Li, H. Maq: mapping and assembly with qualities. Welcome Trust Sanger Institute 

(2010). Available at: . 

26. Cormack, B.P., Valdivia, R.H. & Falkow, S. FACS-optimized mutants of the green 

fluorescent protein (GFP). Gene 173, 33–38 (1996). 


doi:10.1038/nbt.1716 


l e t t e r s 

Rapid translocation of nanoparticles from the lung 

airspaces to the body 

Hak Soo Choi 1 , Yoshitomo Ashitate 1 , Jeong Heon Lee 1 , Soon Hee Kim 1 , Aya Matsui 1 , Numpon Insin 2 , 

Moungi G Bawendi 2 , Manuela Semmler-Behnke 3 , John V Frangioni 1,4,6 & Akira Tsuda 5,6 


Nano-size particles show promise for pulmonary drug 

delivery, yet their behavior after deposition in the lung 

remains poorly understood. In this study, a series 

of near-infrared (NIR) fluorescent nanoparticles were 

systematically varied in chemical composition, shape, size and 

surface charge, and their biodistribution and elimination were 

quantified in rat models after lung instillation. We demonstrate 

that nanoparticles with hydrodynamic diameter (HD) less than 

≈34 nm and a noncationic surface charge translocate rapidly 

from the lung to mediastinal lymph nodes. Nanoparticles of 

HD < 6 nm can traffic rapidly from the lungs to lymph nodes 

and the bloodstream, and then be subsequently cleared by the 

kidneys. We discuss the importance of these findings for drug 

delivery, air pollution and carcinogenesis. 

Nanoparticles have been proposed as diagnostic, therapeutic and theranostic 

agents for a wide variety of human diseases 1–3 . Delivery of nanoparticles 

through the lung is receiving increased attention owing to the 

large lung surface area available and the minimal anatomical barriers 

limiting access to the body 4 . The behavior of nanoparticles in the lung 

is also important for understanding the health effects of air pollution. 

Recent toxicological studies have confirmed that nano-size particles 

reach deep into the alveolar region of the lungs 5,6 and can cause severe 

inflammatory reactions because of their large surface area–to-mass 

ratio 6 . Inhalation of nanoparticles is increasingly recognized as a major 

cause of adverse health effects and has an especially strong influence 

on the cardiovascular system and hemostasis, leading to increased 

cardiovascular morbidity and mortality 6–8 . Inhaled nanoparticles can 

pass from the lungs into the bloodstream and extrapulmonary organs 9 . 

Alveolar macrophage-mediated translocation from the lung surface 

to the regional tracheobronchial lymph nodes has been shown for 

micrometer-sized particles 10 . However, this process takes a relatively 

long time, from several hours to weeks 10–12 . 

Here we study the behavior of nanoparticles in the first hour 

after administration to the rat lung and attempt to define the key 

parameters that control their translocation to draining lymph nodes, 

the bloodstream and the rest of body, and subsequent elimination 

(that is, clearance). The standard approach for studying the translocation 

of inhaled nanoparticles and ultrafine air pollutants from the 

lungs to extrapulmonary compartments in animals is to perform postmortem 

analysis of tissues after inhalation of carbon-based particles 13 , 

radiotracers 9 or neutron-activated metal particles 11,14,15 . Compared to 

these methods, real-time NIR fluorescence imaging provides several 

advantages for quantifying the process of nanoparticle trafficking, 

including a low autofluorescence background, real-time imaging, 

real-time overlay with anatomy and highly sensitive detection of targets 

several millimeters deep in tissue 16 . We systematically varied the 

chemical composition, size, shape and surface charge of NIR fluorescent 

nanoparticles to determine how their physicochemical properties 

affect trafficking across the alveolar surface barrier into tissue, 

and once in the tissue, their biodistribution and clearance. Enabling 

this study is our intraoperative fluorescence-assisted resection and 

exploration (FLARE) imaging system, which permits two independent 

channels (700 nm and 800 nm) of NIR fluorescence images to be 

acquired simultaneously with color video images in real time 17,18 . 

Two distinct families of nanoparticles were engineered with varying 

chemical composition, shape, size and surface charge (Table 1 

and Fig. 1a). Inorganic/organic hybrid nanoparticles (INPs) emitted 

at 800 nm. Quantum dots with various core(shell) and organic coating 

ligands were used to produce INPs of various sizes and charges 

(INP1–INP5; Supplementary Fig. 1). Gel-filtration chromatography 

was used to measure the final HD, which ranged from 5 to 23 nm in 

PBS (Supplementary Fig. 2). It is important to note that the surfacecharged 

INPs with either carboxylic (INP3 and INP5) or amino 

(INP4) groups showed high protein adsorption, which contributed 

to a significant increase in final HD when exposed to 100% serum 19 . 

To study the effect of surface charge in detail, we engineered INP1 

and INP2 to have zwitterionic and polar coatings, respectively, which 

eliminated serum protein binding and resulted in a constant HD in 

PBS and serum 19,20 . To engineer INPs with HD > 50 nm, we used 

silica nanospheres doped with quantum dots and conjugated NIR 

fluorophore CW800 on the silica surface (INP6–INP9). These silica 

nanoparticles did not associate with serum proteins; thus, the original 

size was maintained after 1 h serum incubation at 37 °C. Organic 

1 Division of Hematology/Oncology, Department of Medicine, Beth Israel Deaconess Medical Center, Boston, Massachusetts, USA. 2 Department of Chemistry, 

Massachusetts Institute of Technology, Cambridge, Massachusetts, USA. 3 Institute of Lung Biology and Disease, Helmholtz Center München—German Research 

Center for Environmental Health, Neuherberg/Munich, Germany. 4 Department of Radiology, Beth Israel Deaconess Medical Center, Boston, Massachusetts, USA. 

5 Molecular and Integrative Physiological Sciences, Harvard School of Public Health, Boston, Massachusetts, USA. 6 These authors contributed equally to this work. 

Correspondence should be addressed to J.V.F. (jfrangio@bidmc.harvard.edu) or A.T. (atsuda@hsph.harvard.edu). 

Received 16 April; accepted 5 October; published online 7 November 2010; doi:10.1038/nbt.1696 


l e t t e r s 


Table 1 Chemical and physical properties of inorganic/organic hybrid nanoparticles (INPs) and organic nanoparticles (ONPs) 

HD a (nm) 

Translocation at 1 h 

Inorganic/organic 

hybrid nanoparticles 

(INPs) 

Organic nanoparticles 

(ONPs) 

Nanoparticle 

Core(shell) Organic coating In PBS In serum Surface charge 

nanoparticles (ONPs) were designed to emit 700 nm NIR fluorescence, 

which permitted comparison of the effect of nanoparticle chemical 

composition with 800 nm–emitting INPs. 

To investigate the role of HD in the translocation of nanoparticles from 

the lung to extrapulmonary compartments of the body, we systematically 

increased the size of fluorescent nanoparticles from 5 to 300 nm. 

Translocation of INP5 (38 nm HD) and larger particles was severely 

restricted at 30 min after administration (Supplementary Fig. 3), 

which indicates that the size threshold for rapid translocation of nanoparticles 

from lungs to lymph nodes corresponds to an HD

l e t t e r s 

a 

SBR (normalized) 

25 

15 

5 

0 

Translocation of INP1 

(HD = 5 nm, Zwitterionic) 

LN 

Blood 

Urine 

10 20 30 40 50 60 

Time (min) 

b 

Lymph node Lung 

Mu 

LN+ LN– 

Mu 

LN+ LN– 

c 

%ID/g 

0.4 Inset 

0.3 

0.2 

0.1 

0 

0 10 20 30 40 50 60 

Time (min) 

– 

TcO 4 

Tc-INP1 

Tc-INP3 

Tc-INP4 

0.3 

0.2 

0.1 

Inset 

0 

0 5 10 

Lag 

SBR (normalized) 

25 

15 

5 

0 

Translocation of INP3 

(HD = 27 nm, Anionic) 

LN 

Blood 

Urine 

10 20 30 40 50 60 

Time (min) 

Liver Kidney 

%ID/g 

2.5 

2.0 

1.5 

1.0 

0.5 

0 

– 

TcO 4 

Tc-INP1 

Tc-INP3 

Tc-INP4 

TcO 4 

– Tc-INP1 Tc-INP3 Tc-INP4 

%ID 

100 

75 

50 

25 

0 

Urine 

Lung 

Body 

Total 


Figure 2 Biodistribution, clearance and histological analysis of INPs in Sprague-Dawley rats. (a) Translocation into lymph node, blood and urinary 

excretion of INP1 (left) and INP3 (right) using real-time NIR fluorescence imaging. Each point represents the mean ± s.d. of n = 3 animals. SBR, 

signal-to-background ratio. (b) Frozen sections obtained from resected organs of INP1-administered Sprague-Dawley rats at 1 h after instillation. 

From top to bottom are representative images of lung, lymph node, kidney (arrow: cortex; arrowhead: calyces), and liver. Mu, muscle; LN+, posterior 

mediastinal lymph node; LN−, negative para-aortic lymph node. Scale bars, 5 mm. Shown are color video and NIR fluorescence of intact specimens 

(left two panels, respectively) along with representative histological images from the same organ/tissue (H&E, NIR, right two panels, respectively). Green 

dotted circle, bronchiole; blue dotted circle, glomerular basement membrane; red dotted circle, portal area (portal vein, hepatic artery and bile duct). 

Scale bars, 200 μm. All NIR fluorescence images (λ Exc = 760 ± 20 nm and λ Em = 795 nm longpass) have identical exposure times and normalizations. 

(c) Quantitative biodistribution and clearance using 99m Tc-conjugated INPs administered intratracheally into Sprague-Dawley rats. The small molecule 

TcO 4 

− was used as a control. Translocation from lung to blood over time (top). Translocation from lung to regional lymph nodes (bottom left) 1 h after 

injection. Recovery of injected dose in urine, lung, body (without lungs) and total (bottom right). Each data point represents the mean ± s.d. of n = 3 

animals. All values in blood curves are statistically different (ANOVA) from each other at 1 h. 

translocation of nanoparticles into the mediastinal lymph nodes. 

A previous study has shown the importance of surface properties 

of nanoparticles in hemostasis-inducing effects, such as platelet 

activation, which may lead to thrombosis 13 . On the other hand, the 

behavior of highly net-charged nanoparticles (charge-to-volume 

ratio >±1) was more complicated. As both anionic and cationic 

charged molecules quickly adsorb endogenous proteins in the lungs, 

similar migration results were expected among the batches 19,21 . The 

translocation behavior of nanoparticles in the alveolar lining fluid 

was, however, very different. Carboxylic group–modified INP3 

was found in the lymph nodes within 30 min after administration, 

whereas no significant translocation was observed for the cationic 

INP4 (Fig. 1c, bottom), even though they were engineered from the 

same batch of CdTe(CdSe) nanocrystals. As it is well known that 

cationic-charged materials can be taken up by cells 22 , sometimes 

causing acute cytotoxicity, it is likely that cationic particles (INP4) 

were rapidly taken up by pulmonary macrophages and/or epithelial 

cells shortly after administration, making them unavailable for 

translocation to lymph nodes. As cationic nanoparticles remain in 

pulmonary cells for a long time, they may cause severe lung injury 23 . 

Our results are also consistent with recent reports on pulmonary 

drug delivery systems that suggest that surface charge affects translocation 

efficiency 24 . 

The majority of the nanoparticles administered in this study 

remained in the lungs, with small amounts rapidly appearing in the 

mediastinal lymph nodes within 30 min after administration. Even 

more rapid translocation (in minutes) from the site of deposition to 

lymph nodes was only observed for smaller nanoparticles. In particular, 

for the smallest nanoparticles (INP1, 5 nm), migration to the mediastinal 

lymph nodes was very fast (within 3 min), and there was measurable 

nanoparticle accumulation in the kidneys and excretion into urine 

at 30 min after administration (Fig. 2a and Supplementary Video 1). 

On the other hand, INP3 (27 nm) showed slower accumulation into 

lymph nodes (≈10 min) and blood (≈20 min). Most of our ultra-small 

nanoparticles were found in the kidneys, with the smallest ones also 

found in the urine (Fig. 2b and Supplementary Fig. 4). Because of the 

zwitterionic surface coating and small HD 19,25 , there were virtually no 

nanoparticles trapped in the hepatic clearance route, such as liver, bile 

and intestine (Fig. 2b and Supplementary Video 2). 

To confirm the qualitative observations made using NIR fluorescence, 

we labeled INPs with 99m Tc-MAS 3 (MAS 3 is S-acetylmercaptoacetyltriserine) 

and repeated experiments to better quantify 

nanoparticle translocation from the lung airspaces to lymph node, 

blood and eventually urine (Fig. 2c and Supplementary Figs. 5 and 6). 

Blood clearance was measured by intermittent sampling of the tail 

vein. The small-molecule sodium pertechnetate (TcO 4 − ) was used as 

a control and showed rapid uptake into blood, consistent with direct 

translocation through the alveolar air-blood barrier 9 , excretion into 

urine and moderate uptake in lymph nodes at 1 h post-instillation. On 

the other hand, INP1 and INP3 appeared in the blood after a short lag 

and at concentrations inversely proportional to HD. Because lymph 

node uptake of nanoparticles was more pronounced than TcO 4 − , 

this suggests that there could be another pathway for translocation 

of nanoparticles from lung airspaces into blood, one involving transient 

passage through lymph nodes. 

In this study, we have systematically engineered various nanoparticles 

to investigate the effect of particle characteristics on the ability 

of nanoparticles to translocate from the lungs into the lymph nodes 

and bloodstream, and on their fate in vivo. Although nanoparticles are 

stable and show a narrow size distribution in PBS, this study suggests 

that the final HDs as measured in serum, along with surface charge, 

are critical in determining the ultimate migration, biodistribution and 

clearance of nanoparticles deposited in the lungs. The key findings 

of our study are: (i) a size threshold of ≈34 nm determines whether 

there is rapid transepithelial translocation of nanoparticles from the 

alveolar luminal surface into the septal interstitium, followed by quick 


l e t t e r s 


translocation to the regional draining lymph nodes where further 

translocation into the bloodstream could occur; (ii) below this size 

threshold of ≈34 nm, surface charge is a major factor that influences 

translocation, with zwitterionic, anionic and polar surfaces being 

permissive and cationic surfaces being restrictive; and (iii) when 

HD is 100 nm) cationic liposomes provide an improved therapeutic 

window to treat lung diseases. That is, systemic absorption is very 

low, presumably because these nanoparticles exceed the size and 

charge thresholds defined above (with whatever systemic absorption 

is observed probably due to liposome instability). Second, and more 

importantly, engineering nanoparticle-based drugs to be zwitterionic 

and 34 nm) and cationic could potentially minimize 

airspace-to-body translocation and provide time for mucociliaryand/or 

macrophage-mediated clearance. Once in the body and lodged 

in lymph nodes (that is 6 nm ≤ HD ≤ 34 nm), noncationic nanoparticles 

could cause inflammation and/or contribute to carcinogenesis. Smaller 

nanoparticles, such as those ≈5 nm in HD, are of even more concern 

for carcinogenesis and distal inflammation because they are capable 

of traveling from the lung to the bloodstream, and once in the bloodstream, 

can potentially reach every tissue and organ in the body 26 . 

Methods 

Methods and any associated references are available in the online 

version of the paper at http://www.nature.com/naturebiotechnology/. 



We thank R. Oketokoun and S. Gioux for help with developing the FLARE imaging 

system and E.P. Lunsford of the Longwood Small Animal Imaging Facility for 

assistance with SPECT/CT imaging. The Biophysical Instrumentation Facility for 

the Study of Complex Macromolecular Systems (National Science Foundation- 

0070319 and US National Institutes of Health (NIH) GM68762) is gratefully 

acknowledged. We thank W. Liu and B.I. Ipe for providing quantum dots, L. Moffitt 

for editing, and L. Keys and E. Trabucchi for administrative support. This work 

was supported in part by NIH grant HL054885 (A.T.), HL070542 (A.T.), HL074022 

(A.T.) and R01-CA-115296 (J.V.F.). 


H.S.C., Y.A., J.H.L., S.H.K., A.M., N.I. and A.T. performed the experiments. H.S.C., 

M.G.B., M.S.-B., A.T. and J.V.F. reviewed, analyzed and interpreted the data. H.S.C., 

A.T. and J.V.F. wrote the paper. All authors discussed the results and commented on 

the manuscript. 







1. Janib, S.M., Moses, A.S. & Mackay, J.A. Imaging and drug delivery using theranostic 

nanoparticles. Adv. Drug Deliv. Rev. 3, 1–2 (2010). 

2. Zrazhevskiy, P., Sena, M. & Gao, X. Designing multifunctional quantum dots for 

bioimaging, detection, and drug delivery. Chem. Soc. Rev. 3, 1–2 (2010). 

3. Choi, H.S. & Frangioni, J.V. Nanoparticles for biomedical imaging: Fundamentals 

of clinical translation. Mol. Imaging 9, 291–310 (2010). 

4. Courrier, H.M., Butz, N. & Vandamme, T.F. Pulmonary drug delivery systems: recent 

developments and prospects. Crit. Rev. Ther. Drug Carrier Syst. 19, 425–498 

(2002). 

5. Tsuda, A., Rogers, R.A., Hydon, P.E. & Butler, J.P. Chaotic mixing deep in the lung. 

Proc. Natl. Acad. Sci. USA 99, 10173–10178 (2002). 

6. Oberdorster, G. Pulmonary effects of inhaled ultrafine particles. Int. Arch. Occup. 

Environ. Health 74, 1–8 (2001). 

7. Mills, N.L. et al. Adverse cardiovascular effects of air pollution. Nat. Clin. Pract. 

Cardiovasc. Med. 6, 36–44 (2009). 

8. Wichmann, H.E. et al. Daily mortality and fine and ultrafine particles in Erfurt, 

Germany part I: role of particle number and particle mass. Res. Rep. Health Eff. 

Inst. November, 5–86, discussion 87–94 (2000). 

9. Moller, W. et al. Deposition, retention, and translocation of ultrafine particles from 

the central airways and lung periphery. Am. J. Respir. Crit. Care Med. 177, 426–432 

(2008). 

10. Harmsen, A.G., Muggenburg, B.A., Snipes, M.B. & Bice, D.E. The role of 

macrophages in particle translocation from lungs to lymph nodes. Science 230, 

1277–1280 (1985). 

11. Semmler-Behnke, M. et al. Biodistribution of 1.4- and 18-nm gold particles in rats. 

Small 4, 2108–2111 (2008). 

12. Geiser, M. et al. The role of macrophages in the clearance of inhaled ultrafine 

titanium dioxide particles. Am. J. Respir. Cell Mol. Biol. 38, 371–376 

(2008). 

13. Nemmar, A. et al. Ultrafine particles affect experimental thrombosis in an in vivo 

hamster model. Am. J. Respir. Crit. Care Med. 166, 998–1004 (2002). 

14. Kreyling, W.G. et al. Translocation of ultrafine insoluble iridium particles from lung 

epithelium to extrapulmonary organs is size dependent but very low. J. Toxicol. 

Environ. Health A 65, 1513–1530 (2002). 

15. Semmler, M. et al. Long-term clearance kinetics of inhaled ultrafine insoluble 

iridium particles from the rat lung, including transient translocation into secondary 

organs. Inhal. Toxicol. 16, 453–459 (2004). 

16. Frangioni, J.V. In vivo near-infrared fluorescence imaging. Curr. Opin. Chem. Biol. 

7, 626–634 (2003). 

17. Gioux, S. et al. High-power, computer-controlled, light-emitting diode-based light 

sources for fluorescence imaging and image-guided surgery. Mol. Imaging 8, 

156–165 (2009). 

18. Troyan, S.L. et al. The FLARE intraoperative near-infrared fluorescence imaging 

system: a first-in-human clinical trial in breast cancer sentinel lymph node mapping. 

Ann. Surg. Oncol. 16, 2943–2952 (2009). 

19. Choi, H.S. et al. Renal clearance of quantum dots. Nat. Biotechnol. 25, 1165–1170 

(2007). 

20. Choi, H.S. et al. Tissue- and organ-selective biodistribution of NIR fluorescent 

quantum dots. Nano Lett. 9, 2354–2359 (2009). 

21. Burns, A.A. et al. Fluorescent silica nanoparticles with efficient urinary excretion 

for nanomedicine. Nano Lett. 9, 442–448 (2009). 

22. Zhang, L.W. & Monteiro-Riviere, N.A. Mechanisms of quantum dot nanoparticle 

cellular uptake. Toxicol. Sci. 110, 138–155 (2009). 

23. Brown, D.M., Wilson, M.R., MacNee, W., Stone, V. & Donaldson, K. Size-dependent 

proinflammatory effects of ultrafine polystyrene particles: a role for surface area 

and oxidative stress in the enhanced activity of ultrafines. Toxicol. Appl. Pharmacol. 

175, 191–199 (2001). 

24. Duncan, R. Polymer conjugates as anticancer nanomedicines. Nat. Rev. Cancer 6, 

688–701 (2006). 

25. Liu, W. et al. Compact cysteine-coated CdSe(ZnCdS) quantum dots for in vivo 

applications. J. Am. Chem. Soc. 129, 14530–14531 (2007). 

26. Choi, H.S. et al. Design considerations for tumour-targeted nanoparticles. 

Nat. Nanotechnol. 5, 42–47 (2010). 




Synthesis of inorganic/organic nanoparticles (INPs). CdSe(ZnCdS) nanocrystals 

were used to produce INP1 and INP2 with cysteine and dihydrolipoic 

acid (DHLA)-conjugated polyethylene glycol spacer (DHLA-PEG, 

MW 1 kDa) coating, respectively, via modification of previously reported 

procedures (Supplementary Methods) 19,25 . 800 nm NIR-emitting fluorophore 

CW800 (LI-COR) was conjugated on the quantum dot surface in PBS, 

pH 7.8 (labeling ratio ≈1.4). CdTe(ZnS) core(shell) NIR quantum dots were 

purchased from Invitrogen (QDot 800 ITK) with carboxylated PEG (INP3) 

or amino PEG (INP4) ligand coating 27 . To engineer INP5, α-amino-ωcarboxylate 

PEG 3.4 kDa (Nektar) was conjugated through conventional EDC 

chemistry. INP6 to INP9 silica nanoparticles were prepared by incorporating 

CdSe(ZnS) quantum dots in the silica shells and conjugating CW800 fluorophores 

on the surface of nanospheres to engineer larger sizes of nanoparticles 

(Supplementary Methods) 28 . 

Synthesis of organic nanoparticles (ONPs). ONP1 was prepared by a reaction 

of Cy5.5-NHS (Amersham Biosciences) with human serum albumin (HSA; 

American Red Cross) in PBS at pH 7.8, followed by purification by gel-filtration 

chromatography (GFC) using an Econo-Pac P6 cartridge (Bio-Rad). The labeling 

ratio was 2.6, estimated using the extinction coefficients of HSA (ε 280nm = 

32,900 M −1 cm −1 ) and Cy5.5 (ε 675nm = 19,000 M −1 cm −1 ). ONP2 was obtained 

by simply mixing two equivalents of Cy5.5-NHS with amino mPEG 20 kDa 

(Nektar) in PBS at pH 7.8, followed by GFC purification (labeling ratio = 1). 

A size series of nanospheres based on polystyrene and polyacrylate block 

copolymers (ONP3–ONP7) were purchased from Phosphorex and conjugated 

covalently to Cy5.5 (5–10 fluorophores per mole of polyacrylate) to produce 

700 nm NIR fluorescence emission (Supplementary Methods). 

Double-lumen balloon catheter for nanoparticle administration into lungs. 

Administration of nanoparticles into the lung without backflow entering the 

trachea and obscuring mediastinal lymph nodes imaging was a major challenge. 

To solve this problem, we developed a custom-built, double-lumen 

balloon catheter that blocked mucociliary flow into the trachea while animals 

were being ventilated (Supplementary Fig. 3). A custom-made 20 cm 

double-lumen balloon catheter was developed by Pelham Plastics with the 

following specifications: OD = 1.5 mm; a thru lumen = 0.71 mm for ventilation; 

a balloon lumen = 0.5 mm; inflation balloon OD = 2.5 mm. The catheter 

was inserted through a trachea cannula (OD = 2.1 mm, ID = 1.8 mm) into 

the right mainstem bronchus and the outer balloon was inflated. The inflated 

balloon remained in position for 1 h to block backflow of deposited nanoparticles. 

The thru lumen was used for administration of nanoparticle solutions 

deep into the lung followed by a puff of air during an inspiration phase to 

ensure delivery of a nanoparticle bolus deep into the lung. Both side lungs 

were ventilated for 60 min. 

Animal models. All animals were used under the supervision of an approved 

institutional protocol. 500 to 550 g Sprague-Dawley (SD) male rats from Charles 

River Laboratories were anesthetized with 65 mg/kg intraperitoneal pentobarbital 

and were mechanically ventilated by tracheostomy using an MRI-1 

ventilator (CWE) at 80 breaths per min using room air. A specially designed 

double-lumen balloon catheter was placed in the right mainstem bronchus as 

described above, and a minimal volume (50–100 μl) of nanoparticle solution 

(a mixture of 800 nm INPs and/or 700 nm ONPs) in PBS was then administered 

through the thru lumen into lungs at a dose of 10 pmol/g of animal weight. After 

each study, animals were euthanized by intraperitoneal injection of 200 mg/kg 

pentobarbital, a method consistent with the recommendations of the Panel on 

Euthanasia of the American Veterinary Medical Association. 

Real-time intraoperative NIR fluorescence imaging. The dual-channel intraoperative 

NIR fluorescence imaging system (FLARE) optimized for animal 

surgery has been described in detail previously 17,18 . Excitation fluence rate for 

white light, 700 nm NIR excitation light and 800 nm NIR excitation light were 

20,000 lux, 3.5 mW/cm 2 and 10 mW/cm 2 , respectively. The translocation of 

nanoparticles into lymph nodes and bladder was measured using the following 

method: the fluorescence (Fl) and background (BG) intensity of a region 

of interest over each organ was quantified using custom FLARE software at 

the following time points: 0, 1, 2, 5, 10, 20, 30, 40, 50 and 60 min. Signal-tobackground 

ratio (SBR), defined as SBR = (Fl–BG)/BG, was then plotted as 

a function of time. For measuring nanoparticle concentration in blood, ~20 

μl of blood was collected from the tail vein using glass capillary tubes, and 

the SBR was measured over time. At least three animals were analyzed at each 

time point. Statistical analysis was carried out using the unpaired Student’s 

t-test or one-way analysis of variance (ANOVA). Results were presented 

as mean ± s.d. and curve fitting was performed using Prism version 4.0a 

software (GraphPad). 

99m Tc-labeling of nanoparticles and radioscintigraphic imaging. 99m Tc-conjugated 

nanoparticles were prepared using high-specific-activity N-hydroxysuccinimide 

(NHS) ester of 99m Tc-MAS 3 , as described previously 19,20,26 . 

50–100 μl of 99m Tc-conjugated nanoparticles (100–300 μCi) were administered 

intratracheally into Sprague-Dawley rats. Sodium pertechnetate (TcO 4 − ) was 

used as a control. Measurement of blood clearance was performed by intermittent 

sampling of the tail vein. Rats were euthanized at 1 h after administration. 

To measure total urinary excretion, the bladder was removed en masse 

and combined with excreted urine before measurement of radioactivity in 

a dose calibrator. The radioactivity of each resected major organ was measured 

on a Wallac Wizard (PerkinElmer) 10-detector gamma counter. Gamma 

radioscintigraphy was performed with a Research Digital Camera (Isocam 

Technologies) equipped with a 1/2″ NaI crystal, 86 photomultiplier tubes and 

high-resolution (1 mm) lead collimator. 

Histological analysis of tissues. Tissues were resected 1 h after administration, 

fixed in 10% formalin for 3 h, molded with Tissue-Tek OCT compound (Fisher 

Scientific) and frozen in liquid nitrogen. Frozen sections were cut to a 10 μm 

thickness and stained with hematoxylin and eosin (Histotec Laboratories). 

Lung, liver, spleen, kidney, pancreas, intestine, muscle and lymph nodes 

were examined by a 4-channel fluorescence microscope. The excitation and 

emission filters used were 650 ± 22 nm and 710 ± 25 nm for NIR 700 nm, and 

760 ± 20 nm and 790 nm longpass for NIR 800 nm, respectively. 

27. Kim, S. et al. Near-infrared fluorescent type II quantum dots for sentinel lymph 

node mapping. Nat. Biotechnol. 22, 93–97 (2004). 

28. Insin, N. et al. Incorporation of iron oxide nanoparticles and quantum dots into 

silica microspheres. ACS Nano 2, 197–202 (2008). 


doi:10.1038/nbt.1696

l e t t e r s 

Suppressing resistance to Bt cotton with sterile 

insect releases 

Bruce E Tabashnik 1 , Mark S Sisterson 2 , Peter C Ellsworth 3 , Timothy J Dennehy 1,4 , Larry Antilla 5 , 

Leighton Liesner 5 , Mike Whitlow 5 , Robert T Staten 6 , Jeffrey A Fabrick 7 , Gopalan C Unnithan 1 , Alex J Yelich 1 , 

Christa Ellers-Kirk 1 , Virginia S Harpold 1 , Xianchun Li 1 & Yves Carrière 1 


Genetically engineered crops that produce insecticidal toxins 

from Bacillus thuringiensis (Bt) are grown widely for pest 

control 1 . However, insect adaptation can reduce the toxins’ 

efficacy 2–5 . The predominant strategy for delaying pest 

resistance to Bt crops requires refuges of non-Bt host plants to 

provide susceptible insects to mate with resistant insects 2–7 . 

Variable farmer compliance is one of the limitations of this 

approach. Here we report the benefits of an alternative strategy 

where sterile insects are released to mate with resistant insects 

and refuges are scarce or absent. Computer simulations 

show that this approach works in principle against pests with 

recessive or dominant inheritance of resistance. During a largescale, 

four-year field deployment of this strategy in Arizona, 

resistance of pink bollworm (Pectinophora gossypiella) to Bt 

cotton did not increase. A multitactic eradication program that 

included the release of sterile moths reduced pink bollworm 

abundance by >99%, while eliminating insecticide sprays 

against this key invasive pest. 

Transgenic cotton and corn that produce proteins from Bacillus 

thuringiensis (Bt) for insect control have been planted on a cumulative 

total of >200 million ha worldwide since their commercial introduction 

in 1996 (ref. 1). Although Bt crops remain effective against most 

pest populations, several pests have evolved resistance 2–5 . The main 

strategy for delaying pest resistance to Bt crops promotes survival of 

susceptible insects by providing host plants that do not produce Bt 

toxins 6,7 . These are commonly referred to as ‘refuges’. Ideally, most of 

the rare, resistant insects emerging from Bt crops will mate with the 

relatively abundant susceptible insects from nearby refuges. If resistance 

is inherited as a recessive trait, the Bt crops will kill the hybrid 

progeny produced by such matings and evolution of resistance will 

be substantially slowed 6,7 . 

Retrospective evaluations of global resistance monitoring data 

suggest that refuges have delayed pest resistance to Bt crops 3,7 . In 

particular, theoretical and empirical analyses imply that refuges 

have delayed resistance in pink bollworm (Pectinophora gossypiella), 

one of the world’s most destructive pests of cotton 7–10 . This invasive 

insect, which was first detected in the United States in 1917, feeds 

almost exclusively on cotton in some parts of the southwestern United 

States, including Arizona 9,10 . Field and greenhouse data show that 

transgenic cotton that produces Bt toxin Cry1Ac (Bt cotton) kills 

essentially 100% of susceptible pink bollworm larvae 11–14 . However, 

laboratory selection with Cry1Ac quickly produced several resistant 

strains of pink bollworm from Arizona that could survive on Bt cotton 

plants 11,12,15 . Furthermore, pink bollworm resistance to Bt cotton 

has been reported in the field in India, where farmer compliance with 

the refuge strategy has been low 5,16 . In contrast, compliance with the 

refuge strategy is considered a primary reason why pink bollworm 

susceptibility to Bt cotton did not decrease in the field in Arizona 

from 1997 to 2005 (refs. 8,17). 

As part of a coordinated, multitactic effort to eradicate pink bollworm 

from the southwestern United States and northern Mexico, a 

new strategy that replaced refuges with season-long releases of sterile 

pink bollworm moths was initiated in Arizona in 2006 (refs. 18,19) 

(Online Methods). Under this new strategy, Arizona cotton growers 

were permitted to plant up to 100% transgenic cotton that produces 

either one Bt toxin (Cry1Ac) or two Bt toxins (Cry1Ac and 

Cry2Ab) 18,19 . The concept underlying this alternative approach is that 

if enough sterile moths are released, resistant moths will mate mainly 

with sterile moths, rather than with fertile, wild moths that are either 

resistant or susceptible to Bt. In principle, this approach has several 

advantages over the refuge strategy. First, farmers could greatly reduce 

or eliminate planting of refuges and thus avoid associated complications 

and yield losses. Moreover, because matings with sterile insects 

do not produce fertile progeny, this approach could delay resistance 

that is based on either recessive or dominant inheritance. Unlike the 

refuge strategy, this approach does not require maintenance of pest 

populations. It is thus compatible with eradication efforts. To test 

the idea of delaying resistance with sterile insect releases, we conducted 

computer simulations and analyzed more than a decade of 

field data from before and after deployment of this strategy statewide 

in Arizona. 

In the computer simulations, sterile moth releases suppressed 

resistance to Bt cotton by decreasing the pest’s population size and 

1 Department of Entomology, University of Arizona, Tucson, Arizona, USA. 2 USDA-ARS, San Joaquin Valley Agricultural Sciences Center, Parlier, California, USA. 

3 Department of Entomology, University of Arizona, Maricopa Agricultural Center, Maricopa, Arizona, USA. 4 Monsanto Company, St. Louis, Missouri, USA. 5 Arizona 

Cotton Research & Protection Council, Phoenix, Arizona, USA. 6 USDA-APHIS, retired. 7 USDA-ARS, US Arid Land Agricultural Research Center, Maricopa, Arizona, 

USA. Correspondence should be addressed to B.E.T. (brucet@cals.arizona.edu). 

Received 17 August; accepted 8 October; published online 7 November 2010; doi:10.1038/nbt.1704 


l e t t e r s 


Years to resistance 

>20 

15 

10 

5 

* 

0 

0 

5 

Steriles 

Low 

Very low 

None 

15 20 

reducing the probability of mating between resistant moths (Fig. 1, 

Supplementary Fig. 1 and Supplementary Methods). In the simulations, 

when sufficient numbers of sterile moths were released, pest 

populations did not persist and resistance did not occur over a 20-year 

period (Fig. 1 and Supplementary Fig. 1). Based on experimental 

evidence of pink bollworm responses to Bt cotton 11–13,15,17 , we first 

modeled recessive inheritance of resistance with a fitness cost and 

incomplete resistance (Fig. 1 and Supplementary Table 1). With no 

refuges, resistance evolved in 3 years without releases of sterile moths, 

but populations did not persist and resistance did not occur with 

weekly ‘low’ releases of 0.62 sterile moths per ha of Bt cotton (Fig. 1). 

With refuges accounting for 2 to 20% of the total area planted to 

cotton, resistance evolved more slowly with increases in the number 

of sterile moths released, increases in the refuge percentage, or both 

(Fig. 1). With 20% of cotton planted to non-Bt cotton refuges, resistance 

did not occur in 20 years, even without sterile releases (Fig. 1). 

Because of fitness costs associated with pink bollworm resistance to 

Bt cotton, higher refuge percentages not only reduced the proportion of 

the population exposed to selection for resistance but also increased 

selection against resistance 8,20 . In a hypothetical worst-case scenario 

with dominant inheritance of resistance and no refuges, resistance 

evolved in 1 year with no sterile moths, but populations did not persist 

and resistance did not occur with weekly releases of 78 sterile moths 

per ha of Bt cotton (Supplementary Fig. 1). 

The mean release rate of sterile pink bollworm moths achieved from 

2006 to 2009 in Arizona was >600 times higher than the simulated rate 

that suppressed recessive resistance to Bt cotton for >20 years without 

refuges (Fig. 1). For each year from 2006 to 2009, the mean number 

of sterile moths released per ha per week was 380 (range = 170–830) 

for Bt cotton and 4,400 (range = 3,900–5,200) for non-Bt cotton, with 

10 

Refuge (%) 

Figure 1 Computer simulations of effects of sterile moth releases on 

evolution of resistance to Bt cotton. We used a stochastic, spatially 

explicit model with a region of 400 cotton fields of 15 ha each 

(Supplementary Methods). Two alleles (‘r’, resistant; ‘s’, susceptible) 

at a single locus controlled larval survival, which was 0% for ss and rs 

(recessive inheritance), and 15% for rr on Bt cotton; 20.8% for ss and 

rs, and 17.7% for rr on non-Bt cotton. The initial r allele frequency was 

0.018. The criteria for resistance were an r allele frequency >0.50 and a 

population size >10% of the initial population size. Each point represents 

the median of ten simulations. The simulated sterile moth release rate 

was 10 times higher in non-Bt cotton fields than in Bt cotton fields to 

mimic field releases. We simulated three sterile release rates: none, very 

low and low. Sterile release rates (in moths per ha per week) were 0.16 in 

Bt cotton and 1.6 in non-Bt cotton for the very low release rate, and 0.62 

in Bt cotton and 6.2 in non-Bt cotton for the low release rate. The actual 

release rates in the field were >600 times higher than the low release 

rate (see text). With no refuges and the low release rate, the regional 

population size decreased to 0 after 2 to 4 years (indicated by asterisk). 

In the cases with refuges where resistance did not evolve in 20 years, the 

regional population persisted but the r allele frequency remained 25% in all 

years, with a mean of 37.4% for 1997 to 2005 (Supplementary Fig. 2). 

With the onset of sterile releases, the statewide refuge percentage 

declined to 15.4% in 2006, 8.4% in 2007, 2.3% in 2008 and 3.1% in 

2009 (Supplementary Fig. 2). 

Consistent with the simulation results (Fig. 1), monitoring of pink 

bollworm field populations showed no net decrease in susceptibility 

to Cry1Ac from 1997 to 2005, when the refuge percentage was 

>25% every year, or from 2006 to 2009, when sterile insects were 

released and the mean refuge percentage was 7.3% (Fig. 2 and 

Supplementary Fig. 2). DNA screening for the three mutations in 

the cadherin gene that are linked with pink bollworm resistance to 

Cry1Ac did not identify any resistant alleles during 2006 to 2009 

(n = 2,499) (Online Methods). Based on larval survival on diet treated 

with Cry1Ac, bioassays detected a single resistant individual during 2006 

(n = 3,822), but no resistant individuals were found during 2007 or 2008 

(n = 3,602) (Fig. 2) (Online Methods). Bioassays also detected no larvae 

resistant to Cry2Ab during 2007 or 2008 (n = 2,572). As detailed 

below, in 2009, this pest was so scarce in Arizona that we could not 

collect enough individuals to conduct bioassays. 

Since the eradication program began in 2006, pink bollworm populations 

have declined dramatically (Fig. 3). In 2009, only two pink 

bollworm larvae were found in 16,600 bolls of non-Bt cotton screened 

statewide. This yields an infestation rate of 0.012%, which represents 

a 99.9% decline from the 15.3% infestation rate in 2005 (Fig. 3A). 

Likewise, the number of wild male pink bollworm moths caught per 

trap per week dropped from 26.7 in 2005 to 0.0054 in 2009, a 99.98% 

decrease (Fig. 3b). The decrease in pink bollworm populations during 

the eradication program was steeper than the decline observed 

with the planting of Bt cotton before the eradication program began 

in Arizona (Fig. 3) 21 and the declines in other target pests associated 

with planting of Bt crops in other regions 22–26 . 

Along with declines in pink bollworm populations, insecticide 

sprays against this pest fell to historic lows (Fig. 4). The mean 

number of sprays per ha per year targeting pink bollworm in Arizona 

was 2.7 from 1990 to 1995, which dropped to 0.64 from 1996 to 2005 

with use of Bt cotton, before the eradication program (Fig. 4) 10,27 . 

Under the eradication program, this mean decreased to 0.14 in 2006, 

0.013 in 2007, 0.0029 in 2008 and 0 in 2009 (Fig. 4). The mean 

yearly cost of pink bollworm to Arizona cotton growers, including 

Resistance allele frequency 

0.30 

0.25 

0.20 

0.15 

0.10 

0.05 

0 

Before eradication program 

Eradication 

program 

1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 

Year 

Figure 2 Pink bollworm resistance allele frequency (with 95% confidence 

intervals) in Arizona from 1997 to 2008, as estimated from laboratory 

bioassays with Cry1Ac (Online Methods). Data from 1997 to 2004 were 

reported previously 8 . 


l e t t e r s 


a 

Infested non-Bt bolls (%) 

(log scale) 

b 

Moths/trap/week 

(log scale) 

100 

10 

1 

0.1 


0.01 

1997 1998 1999 2000 2001 2002 2003 

100 

10 

1 

0.1 

0.01 

Year 


Eradication 

program 

2004 2005 2006 2007 2008 2009 

Eradication 

program 

0.001 

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 

Year 

yield losses and insecticide sprays, was $18 million for 1990 to 1995, 

$5.4 million from 1996 to 2005 and only $172,000 for 2006 to 2009 

(ref. 28). By using Bt cotton as one component of a comprehensive 

integrated pest management program, Arizona growers also greatly 

reduced insecticide use against all cotton pests, including those not 

killed by Bt cotton, saving a cumulative total of $200 million from 

1996 to 2009 (refs. 10,27). 

The estimated yearly cost of pink bollworm to US cotton growers 

before the eradication program was $32 million per year 19 , which is 

$2 million more than the mean annual cost of the eradication program 

in the United States and northern Mexico from 2006 to 2009 (ref. 29). 

These estimates do not include the value of the indirect advantages 

of reduced insecticide use associated with the eradication program, 

such as conservation of natural enemies that control pests other 

than pink bollworm, and benefits to the environment and human 

health 27 . Although increasing economic gains are expected if pink 

bollworm remains scarce and program costs decline, eradication of 

pink bollworm from the United States and northern Mexico remains 

challenging because this invasive pest is widespread and resilient 9,10 . 

Nevertheless, our results show that pink bollworm resistance to 

Figure 3 Pink bollworm abundance in Arizona before and during the 

eradication program. (a) Larval infestation of non-Bt cotton bolls from 

1997 to 2009. Analysis of covariance (Online Methods) shows that 

infestation (log [% infested non-Bt cotton bolls]) was significantly 

affected by year, treatment (before versus during the eradication 

program), and a year-by-treatment interaction (P ≤ 0.0001 for each 

factor and their interaction, r 2 = 0.97). Linear regression shows that 

the slope, which indicates the decrease in infestation per year, was 18 

times steeper from 2006 to 2009 (−0.81, r 2 = 0.97, P = 0.012) than 

from 1997 to 2005 (−0.044, r 2 = 0.42, P = 0.059). Data from 1997 

to 2005 were reported previously 14 . (b) Wild male pink bollworm moths 

trapped in Bt cotton fields from 1998 to 2009. Analysis of covariance 

shows that number of moths caught per trap per week (log transformed) 

was significantly affected by year, treatment (before versus during the 

eradication program), and a year-by-treatment interaction (P < 0.0001 

for each factor and their interaction, r 2 = 0.95). Linear regression shows 

that the slope, which indicates the change in moths trapped per year, 

was significantly negative from 2006 to 2009 (−1.0, r 2 = 0.92, 

P = 0.04), but did not differ significantly from 0 from 1998 to 2005 

(0.017, r 2 = 0.071, P = 0.52). 

Bt cotton in Arizona did not increase from 2006 to 2009, despite the 

low abundance of non-Bt cotton refuges. Although simulation results 

suggest that sterile releases alone can delay resistance to Bt crops 

(Fig. 1 and Supplementary Fig. 1), the dramatic decline in Arizona’s 

pink bollworm population probably reflects the combined effects of 

the sterile releases, high adoption and sustained efficacy of transgenic 

cotton producing either one or two Bt toxins and other control tactics 

used in the eradication program 18,19 . 

Although the sterile insect technique has been known for decades 

and used successfully in some cases 30,31 , the program described here 

is, to our knowledge, the first large-scale effort using this approach 

to suppress pest resistance to a transgenic crop. This program has 

benefitted from a strong grower commitment, public investment in 

sterile insect technology, a well-developed infrastructure for monitoring 

pink bollworm resistance and population density, virtually 100% 

efficacy of Bt cotton against pink bollworm, and this pest’s nearly 

exclusive dependence on cotton in Arizona 9,10,18,19 . We do not know 

whether the success of this program can be replicated with other pests 

or even with pink bollworm in other parts of the world. Analyses 

of mathematical models imply that refinements of the sterile insect 

technique, such as release of transgenic insects carrying a dominant 

lethal gene, could be more widely applicable for suppression of pests 

that harm crops or transmit pathogens 31–33 . Our results suggest that 

further exploration of such tactics could help to enhance the sustainability 

of Bt crops. The results reported here also illustrate the idea 

that Bt crops are likely to be most useful when combined with other 

tactics for integrated control of pests. 

Figure 4 Mean number of insecticide sprays per ha per year targeting 

pink bollworm on cotton in Arizona from 1996 to 2009. The asterisk 

indicates 0 sprays in 2009. Analysis of covariance (Online Methods) 

shows that insecticide use (log [sprays + 0.0001]) was significantly 

affected by year, treatment (before versus during the eradication 

program), and a year-by-treatment interaction (P < 0.0001 for each 

factor and their interaction, r 2 = 0.96). Linear regression shows that the 

slope, which indicates the change in sprays per year, was significantly 

negative from 2006 to 2009 (−1.0, r 2 = 0.98, P = 0.01), but did not 

differ significantly from 0 from 1996 to 2005 (−0.035, r 2 = 0.17, 

P = 0.17). The regression lines are not plotted because the 0 sprays 

for 2009 cannot be represented on the log scale used here and the 

regressions were calculated on a different scale (log [sprays + 0.0001]). 

Data from 1996 to 2005 were reported previously 27 . 

Sprays per year 

(log scale) 

10 

1 

0.1 

0.01 

0.001 

0.0001 

1996 

1997 


1998 

1999 

2000 

2001 

2002 

2003 

2004 

2005 

Eradication 

program 

2006 

2007 

2008 

* 

2009 

Year 


l e t t e r s 


Methods 





We thank N. Alphey, D. Crowder, F. Gould, W. Hutchison, S. Naranjo, 

G. Rosenthal, R.L. Smith and K.M. Wu for their thoughtful comments and 

assistance; and E. Miller, D. Parker, M. Krueger, N. Manhardt and Arizona Cotton 

Research and Protection Council (ACRPC) staff members for their contributions 

to the eradication program. This work was supported by funding from the USDA- 

National Institute of Food and Agriculture program, ACRPC, Arizona Cotton Growers 

Association, Cotton Foundation, Cotton Inc., National Cotton Council, Western IPM 

Center, Arizona Pest Management Center, and Monsanto and Dow AgroSciences. 


M.S.S. conducted computer simulations; P.C.E. collected and summarized 

insecticide use data; L.A., L.L., M.W. and R.T.S. directed the eradication program 

and contributed to its design; J.A.F., G.C.U., A.J.Y., C.E.-K., and V.S.H. collected data; 

B.E.T., M.S.S., P.C.E., T. J. D., L.A., R.T.S., J.A.F., G.C.U., X. L. and Y.C. contributed 

to research design; Y.C. analyzed data. B.E.T. wrote the paper. All authors discussed 

the results and commented on the manuscript. 







1. James, C. Global status of commercialized biotech/GM crops: 2009. ISAAA Briefs 

41 (ISAAA, Ithaca, New York, USA, 2009). 

2. Kruger, M.J., Van Rensburg, J.B.J. & Van den Berg, J. Perspective on the development 

of stem borer resistance to Bt maize and refuge compliance at the Vaalharts irrigation 

scheme in South Africa. Crop Prot. 28, 684–689 (2009). 

3. Tabashnik, B.E., Van Rensburg, J.B.J. & Carrière, Y. Field-evolved insect resistance to 

Bt crops: definition, theory, and data. J. Econ. Entomol. 102, 2011–2025 (2009). 

4. Storer, N.P. et al. Discovery and characterization of field resistance to Bt maize: 

Spodoptera frugiperda (Lepidoptera: Noctuidae) in Puerto Rico. J. Econ. Entomol. 

103, 1031–1038 (2010). 

5. Bagla, P. Hardy cotton-munching pests are latest blow to GM crops. Science 327, 

1439 (2010). 

6. Gould, F. Sustainability of transgenic insecticidal cultivars: integrating pest genetics 

and ecology. Annu. Rev. Entomol. 43, 701–726 (1998). 

7. Tabashnik, B.E., Gassmann, A.J., Crowder, D.W. & Carrière, Y. Insect resistance to 

Bt crops: evidence versus theory. Nat. Biotechnol. 26, 199–202 (2008). 

8. Tabashnik, B.E., Dennehy, T.J. & Carrière, Y. Delayed resistance to transgenic cotton 

in pink bollworm. Proc. Natl. Acad. Sci. USA 102, 15,389–15,393 (2005). 

9. Henneberry, T.J. & Naranjo, S.E. Integrated management approaches for pink 

bollworm in the southwestern United States. Integrated Pest Management Rev. 3, 

31–52 (1998). 

10. Naranjo, S.E. & Ellsworth, P.C. Fourteen years of Bt cotton advances IPM in Arizona. 

Southwest. Entomol. 35, 439–446 (2010). 

11. Tabashnik, B.E. et al. Frequency of resistance to Bacillus thuringiensis in field populations 

of pink bollworm. Proc. Natl. Acad. Sci. USA 97, 12,980–12,984 (2000). 

12. Tabashnik, B.E. et al. Association between resistance to Bt cotton and cadherin 

genotype in pink bollworm. J. Econ. Entomol. 98, 635–644 (2005). 

13. Carrière, Y. et al. Cadherin-based resistance to Bt cotton in hybrid strains of 

pink bollworm: fitness costs and incomplete resistance. J. Econ. Entomol. 99, 

1925–1935 (2006). 

14. Dennehy, T.J. et al. Susceptibility of Southwestern Pink Bollworm to Bt toxins 

Cry1Ac and Cry2Ab2 in 2005. in Arizona Cotton Report (P-151) (Cooperative 

Extension, Agricultural Experiment Station, The University of Arizona, Tucson, 

USDA, 2007) . 

15. Liu, Y.B., Tabashnik, B.E., Dennehy, T.J., Patin, A.L. & Bartlett, A.C. Development 

time and resistance to Bt crops. Nature 400, 519 (1999). 

16. Stone, G.D. Biotechnology and the political ecology of information in India. Hum. 

Organ. 63, 127–140 (2004). 

17. Carrière, Y. et al. Long-term evaluation of compliance with refuge requirements for 

Bt cotton. Pest Manag. Sci. 61, 327–330 (2005). 

18. Antilla, L. & Liesner, L. Program advances in the eradication of pink bollworm, 

Pectinophora gossypiella, in Arizona cotton. in Proceedings of the 2008 Beltwide 

Cotton Conferences, Nashville Tennessee, January 8–11, 2008, 1162–1168 

(National Cotton Council of America, Memphis, Tennessee, USA, 2008). 

19. Grefenstette, B., El-Lissy, O. & Staten, R.T. Pink Bollworm Eradication Plan in 

the US. (USDA, 2009). . 

20. Gassmann, A.J., Carrière, Y. & Tabashnik, B.E. Fitness costs of insect resistance 

to Bacillus thuringiensis. Annu. Rev. Entomol. 54, 147–163 (2009). 

21. Carrière, Y. et al. Long-term regional suppression of pink bollworm by 

Bacillus thuringiensis cotton. Proc. Natl. Acad. Sci. USA 100, 1519–1523 

(2003). 

22. Adamczyk, J.J. Jr. & Hubbard, D. Changes in populations of Heliothis virescens (F.) 

(Lepidoptera: Noctuidae) and Helicoverpa zea (Boddie) (Lepidoptera: Noctuidae) in 

the Mississippi Delta from 1986 to 2005 as indicated by adult male pheromone 

traps. J. Cotton Sci. 10, 155–160 (2006). 

23. Micinski, S., Blouin, D.C., Waltman, W.F. & Cookson, C. Abundance of Helicoverpa zea 

and Heliothis virescens in pheromone traps during the past twenty years in 

northwestern Louisiana. Southwest. Entomol. 33, 139–149 (2008). 

24. Storer, N.P., Dively, G.P. & Herman, R.A. Landscape effects of insect-resistant 

genetically modified crops. in Integration of Insect-Resistant Genetically Modified 

Crops within IPM Programs, (eds. Romeis, J., Shelton, A.M. & Kennedy, G.G.) 

273–302 (Springer, The Netherlands, 2008). 

25. Wu, K.M., Lu, Y.H., Feng, H.Q., Jiang, Y.Y. & Zhao, J.Z. Suppression of cotton 

bollworm in multiple crops in China in areas with Bt toxin-containing cotton. Science 

321, 1676–1678 (2008). 

26. Hutchison, W.D. et al. Areawide suppression of European corn borer with Bt maize 

reaps savings to non-Bt maize growers. Science 330, 222–225 (2010). 

27. Naranjo, S.E. & Ellsworth, P.C. Fifty years of the integrated control concept: moving 

the model and implementation forward in Arizona. Pest Manag. Sci. 65, 1267–1286 

(2009). 

28. Ellsworth, P.C., Fournier, A. & Smith, T.D. Cotton Insect Losses (The University 

of Arizona, Cooperative Extension, 2010). . 

29. Anonymous. Pink Bollworm Eradication Current Status. (US Department of 

Agriculture, February, 2009). < http://www.aphis.usda.gov/plant_health/plant_pest_ 

info/cotton_pests/downloads/PBW_prog_update2009.pdf>. 

30. Krafsur, E.S. Sterile insect technique for suppressing and eradicating insect 

populations: 55 years and counting. J. Agric. Entomol. 15, 303–317 (1998). 

31. Gould, F. & Schliekelman, P. Population genetics of autocidal control and strain 

replacement. Annu. Rev. Entomol. 49, 193–217 (2004). 

32. Thomas, D.D., Donnelly, C.A., Wood, R.J. & Alphey, L.S. Insect population control 

using a dominant, repressible, lethal genetic system. Science 287, 2474–2476 

(2000). 

33. Alphey, N., Bonsall, M.B. & Alphey, L. Combining pest control and resistance 

management: synergy of engineered insects with Bt crops. J. Econ. Entomol. 102, 

717–732 (2009). 




Pink bollworm eradication program. The goal of this ongoing program is to 

eradicate pink bollworm from the United States and northern Mexico 18,19 . The 

program includes: (i) mapping cotton field locations, sizes and types (Bt or 

non-Bt); (ii) measuring pink bollworm abundance by checking cotton bolls for 

larval infestation and by trapping adult males; (iii) monitoring pink bollworm 

resistance to Bt toxins Cry1Ac and Cry2Ab; and (iv) controlling pink bollworm 

using Bt cotton, sterile moth releases, cultural practices, mating disruption 

with pheromone in non-Bt cotton and minimal insecticide applications. The 

grower-sponsored Arizona Cotton Research and Protection Council (ACRPC) 

proposed a plan to allow Arizona cotton growers to plant up to 100% Bt cotton 

that produces either one or two Bt toxins, with sterile insect releases instead 

of non-Bt cotton refuges for delaying resistance 18 . The US Environmental 

Protection Agency (EPA) convened a scientific advisory panel to review the 

plan 34 and the EPA subsequently approved the plan. 

The eradication program began in Texas, New Mexico and Mexico in 2001 

and has expanded each year since 19 . It was initiated in phases in Arizona, 

starting in 2006 with eastern and central Arizona (Cochise, Graham, Greenlee, 

Maricopa, Pima and Pinal Counties), which accounted for 83% of Arizona’s 

cotton acreage that year. The program extended to northwestern Arizona 

(La Paz and Mohave Counties) and southern California in 2007, and to southwestern 

Arizona (Yuma County) and Baja California in 2008. Funding for the 

binational program is provided by growers (80%) and the USDA-Animal Plant 

Health Inspection Service (APHIS) (20%) 29 . 

Computer simulations. To assess the potential effects of sterile moth releases 

on evolution of resistance to Bt cotton, we used a previously described 

stochastic, spatially explicit model of pink bollworm resistance to Bt cotton 35 

with some modifications. We modeled scenarios where resistance would evolve 

with limited refuges and no sterile releases, as reported for pink bollworm 

resistance to Bt cotton producing Cry1Ac in India 5 and for some cases with 

other pests 3 . As detailed in the Supplementary Methods, we based assumptions 

primarily on empirical data for pink bollworm. However, to conservatively 

test the potential for sterile moth releases to delay resistance, we also 

used some assumptions that overestimate the rate of resistance evolution. 

Supplementary Table 1 summarizes the parameter values that we examined. 

Sterile moth releases. The sterile moths released were from the APHIS strain 

of pink bollworm 36 maintained at the USDA-APHIS Pink Bollworm Rearing 

Facility in Phoenix, Arizona. This strain originated from Arizona and was 

infused yearly with wild insects until 2000. Moths were marked internally by 

rearing larvae on artificial diet containing a fat soluble dye 37 (oil red dye 

no. 2114, Passaic Color and Chemical Co.). Moths were irradiated with 200 Gy 

in a Shepherd 484R Cobalt-60 irradiator and stored in containers in groups 

of 2 million at 4 °C for 1–2 d until release. Moths were released from small 

airplanes such as Cessna model 206. Each plane had a tube underneath for 

releasing moths and a device that controlled the release rate. Moths were usually 

released from dawn to 11 a.m. at an altitude of roughly 150 m and a speed 

of ~180 km per hour. Throughout each cotton-growing season from 2006 to 

2009, 1.7–2.1 billion sterile pink bollworm moths were released over Arizona 

cotton fields. From May to October, each cotton field received releases two or 

three times per week. For each year from 2006 to 2009, the mean number of 

sterile moths released per ha per week was 380 for Bt cotton (range = 170–830) 

and 4,400 (range = 3900–5200) for non-Bt cotton. The release rate was higher 

for non-Bt cotton than for Bt cotton because larval survival and emergence of 

wild moths was expected to be higher in non-Bt cotton. 

Whereas spatial separation between refuges and Bt crops could limit the efficacy 

of the refuge strategy, sterile insect releases were made directly into Bt and 

non-Bt cotton fields. Also, although temporal asynchrony in moth emergence 

between refuges and Bt cotton fields could reduce the effectiveness of the refuge 

strategy 15 , sterile releases were made frequently throughout the season, so that 

sterile moths were available consistently for mating with wild moths. 

Refuge percentage. For 1998 to 2009, we determined the total area of cotton 

(Gossypium hirsutum (upland cotton) and G. barbadense (Pima cotton)) 

planted and the area planted to non-Bt cotton in Arizona using methods 

similar to those described previously 17 . We calculated the refuge percentage 

as the area of non-Bt cotton divided by the total area of cotton, multiplied by 

100%. Thus, our estimate of the refuge percentage includes non-Bt cotton 

planted by growers who planted no Bt cotton. In each year, an experienced 

field crew trained by the ACRPC mapped the position of cotton fields and 

collected information from producers on cotton type (Bt or non-Bt) throughout 

Arizona. In each year, data collected by field crews were mapped with 

Geographic Information System (GIS) software and validated by comparing 

cotton field locations between field-generated paper maps and computergenerated 

maps. Enzyme-linked immunosorbent assays for Cry1Ac from 

a randomly chosen subset of fields showed that all of the fields tested had 

been correctly identified on the GIS map 17 . We analyzed the GIS maps using 

ArcView software to calculate the area planted to each type of cotton in each 

year. The statewide Bt cotton percentage for 1997 was reported previously 38 . 

For 1997 to 2009, Arizona’s yearly mean total area planted to cotton was 

98,000 ha (range = 58,000–123,000 ha). In addition to cotton, pink bollworm 

larvae in Arizona feed on okra (Abelmoschus esculentus), which typically grew 

on less than 150 ha per year in Arizona, approximately 1/500th (0.2%) or less 

of the area planted to cotton. We did not include okra in our calculations of 

non-Bt cotton refuges, but sterile moths were released on okra at the same 

rate used on non-Bt cotton. 

Resistance monitoring. We monitored pink bollworm resistance to Bt toxins 

Cry1Ac and Cry2Ab using bioassays 8,11 . We also used DNA screening to monitor 

resistance to Cry1Ac 8,12,13,39–41 . The bioassays can detect resistance caused 

by any mechanism. Because pink bollworm resistance to the diagnostic concentrations 

of Cry1Ac and Cry2Ab used in bioassays is recessive 11–13,15,17,42 , 

however, the bioassays do not distinguish between homozygous susceptible 

larvae and heterozygous larvae carrying only one copy of a resistance gene. The 

DNA screening detects any of the three mutations in a cadherin gene that are 

tightly linked with resistance to Cry1Ac in several laboratory-selected strains 

of pink bollworm from Arizona that survive on Bt cotton plants 12,13,39–41 . 

Although the DNA screening can identify single resistance alleles in heterozygous 

insects, it detects only the three known cadherin resistance alleles. 

Bioassays. To monitor pink bollworm resistance to Cry1Ac and Cry2Ab, we 

used previously described methods for field sampling, laboratory bioassays 

and data analysis 8,11,42 . To monitor resistance to Cry1Ac, each year from 1997 

to 2008, we tested an average of 2,730 larvae from an average of 11.6 cotton 

fields in Arizona. The progeny of field-collected pink bollworm from each site 

were reared and tested separately. Neonates were tested individually for 21 d 

on artificial diet without toxin (control) or on diet with 10 μg Cry1Ac per ml 

diet, which kills susceptible homozygotes and heterozygotes but not resistant 

homozygotes 11 . Based on recessive inheritance of resistance, the Cry1Ac 

resistance allele frequency for each site was estimated as the square root of the 

frequency of survivors after adjustment for control mortality 11 . We calculated 

the 95% confidence interval for each yearly statewide mean resistance allele 

frequency using the bootstrap method with 10,000 repetitions 8 . Bioassay data 

from 1997 to 2004 were reported previously 8 . To monitor resistance to Cry2Ab, 

we used methods similar to those described above for Cry1Ac. Neonates were 

tested individually for 21 d on artificial diet without toxin (control) or on diet 

with a diagnostic concentration of 10 μg Cry2Ab per ml diet 42 . The numbers of 

larvae tested at the diagnostic concentration of Cry2Ab were 2,052 larvae from 

nine cotton fields in 2007 and 520 larvae from two cotton fields in 2008. The 

sample size was smaller in 2008 because the scarcity of pink bollworm in that 

year made it difficult to collect enough live individuals to conduct bioassays. 

DNA screening. We used previously described field sampling procedures and 

allele-specific PCR methods to screen for three cadherin mutations linked 

with pink bollworm resistance to Cry1Ac 8,12,13,39–41 . Details of the methods 

and results of DNA screening from 2001 to 2005 were reported previously 41 . 

DNA screening was completed here for the following numbers of wild pink 

bollworm individuals collected from the field from Arizona: 1,033 in 2006, 

884 in 2007, 364 in 2008 and 218 in 2009 (total = 2,499). 

Pink bollworm abundance. Pink bollworm abundance was measured with 

two complementary methods: checking bolls of non-Bt cotton for larvae and 

capturing male moths in Bt cotton fields with pheromone-baited traps. 


doi:10.1038/nbt.1704


Bolls. Pink bollworm abundance in bolls of non-Bt cotton plants in Arizona 

was determined from 1997 to 2009 by sampling bolls from commercial cotton 

fields during August to November and cutting them open to check for larvae as 

described previously 14 . The mean sample size per year was 18,200 bolls (range 

= 2,900–54,300) from 7 to 44 cotton fields. 

Pheromone traps. Male pink bollworm moths were captured in Bt cotton 

fields from 1998 to 2009 using delta traps baited with septa impregnated with 

4 mg of the pink bollworm female sex pheromone gossyplure (1:1 mixture 

of the Z,E-7,11 and Z,Z-7,11 isomers of hexadecadienyl acetate, Shin-Etsu 

Corporation) 43 . Traps were near field edges, usually at 0.8 m high (range = 

0.5–1.5 m). Traps and lures were changed every week. Traps were brought into 

the laboratory where moths were identified under magnification. During the 

eradication program when sterile moths were released, sterile males in traps 

were identified visually by the red dye used in their larval diet. When wild pink 

bollworm males were rare in traps, particularly during 2008 and 2009, pink 

bollworm males that did not readily show dye were subjected to additional 

testing as follows: each male was ground individually with a glass rod in a 

glass shell vial containing several milliliters of acetone. A strip of Whatman 

no. 4 filter paper trimmed to a point at the top was put in the glass vial. After 

the acetone rose to the top of the filter paper and evaporated, the male was 

deemed sterile if red dye appeared at the top of the filter paper and wild if no 

red dye was visible. 

We analyzed data on wild males caught in traps that were collected from 

the field each year from 15 April to 15 June because data for this period were 

available for all years from 1998 to 2009. The mean number of traps per year 

was 10,400 (range = 1,498–29,928) with a seasonal total of up to nine traps 

(one per week) at each monitoring site. More than 150 Bt cotton fields were 

monitored with traps each year. 

A previous analysis based on data from 1992 to 2001 (before the eradication 

program) showed significant declines in pink bollworm males captured 

in pheromone traps in regions of Arizona where abundance of Bt cotton was 

high but not in regions of Arizona where abundance of Bt cotton was low 21 . 

Compared with the data reported and analyzed here (Fig. 3b), the previous 

analysis differs in terms of the years studied (1992 to 2001 before; 1998 to 2009 

here), the criteria for standardizing the time period examined within years 

(accumulation of degree-days before; calendar dates here); spatial scale (15 

regions of Arizona analyzed separately before; statewide data pooled here), 

and the distribution of pheromone traps (traps near all cotton fields before; 

only traps in Bt cotton fields here). The analysis here shows a steep statewide 

decline in males captured in pheromone traps in Bt cotton fields during the 

eradication program (2006 to 2009) but not before the eradication program 

(1998 to 2005) (Fig. 3b). 

Insecticide sprays. Insecticide sprays targeting pink bollworm in Arizona 

cotton from 2006 to 2009 were calculated by adding the number of sprays 

made by growers and by the ACRPC as part of the eradication program. The 

ACRPC sprayed when larval infestation reached or exceeded 5% of bolls. 

Growers sprayed based on their own criteria. The number of sprays made 

by growers against pink bollworm was estimated as described previously 28 . 

The mean number of sprays per ha per year made by growers against 

pink bollworm was 0.068 in 2006, 0.0095 in 2007, 0 in 2008 and 0 in 2009 

(ref. 10). The mean number of sprays per ha per year made by the ACRPC 

against pink bollworm was 0.070 in 2006, 0.0035 in 2007, 0.00029 in 2008 and 

0 in 2009. Data for 1996 to 2005, which reflect only sprays made by growers, 

were reported previously 27 . 

Analysis of data on pink bollworm abundance and insecticide sprays. We 

calculated the percentage decrease in abundance as: 100% − [(final abundance/ 

initial abundance) × 100%]. For example, infestation of non-Bt cotton bolls 

was 0.012% in 2009 (final abundance) and 15.3% in 2005 (initial abundance). 

The percentage decrease in abundance was 100% − [(0.012%/15.3%) × 100%], 

which equals 99.9%. 

We tested for effects of the eradication program on three response variables: 

(i) infestation of non-Bt cotton bolls, (ii) wild males caught per trap per 

week in Bt cotton fields, and (iii) insecticide sprays per ha per year targeting 

pink bollworm. To test for effects of the eradication program on infestation 

of non-Bt cotton bolls, we compared trends for two time periods: years with 

Bt cotton before the eradication program (1997–2005) and years during the 

eradication program (2006–2009). For each period, simple linear regression 

was used to evaluate the association between the percentage of infested bolls 

(log transformed) and year. We also used covariance analysis to evaluate the 

effects of year, treatment (before versus during the eradication program), and 

their interaction on the percentage of infested bolls (log transformed). In 

this analysis a significant interaction term indicates that the slope before the 

eradication program differs from the slope during the eradication program. 

For the covariance analysis of the boll infestation data, years before the eradication 

program were coded as 1 (1997) to 9 (2005) and during the eradication 

program as 1 (2006) to 4 (2009). As with the boll infestation data, we used 

simple linear regression and covariance analysis to evaluate the effects of the 

eradication program on the number of wild males caught per trap per week in 

Bt cotton fields (log transformed). For this analysis, years with available trap 

data before the eradication program were 1998–2005 and were coded as 1–8. 

We used the same approach to analyze the effects of the eradication program 

on the number of insecticide sprays targeting pink bollworm. Because the 

spray data included a zero (for 2009), we added 0.0001 to the number of sprays 

before performing the log transformation 44 . For sprays, years analyzed before 

the eradication program were 1996 to 2005 and were coded as 1–10. Analyses 

were performed in JMP 45 . 

34. U.S. Environmental Protection Agency. Scientific Advisory Panel, October 24–26, 

2006: Evaluation of the resistance risks from using 100% Bollgard and Bollgard II 

cotton as part of a pink bollworm eradication program in the state of Arizona 

. 

35. Sisterson, M.S., Antilla, L., Carrière, Y., Ellers-Kirk, C. & Tabashnik, B.E. Effects 

of insect population size on evolution of resistance to transgenic crops. J. Econ. 

Entomol. 97, 1413–1424 (2004). 

36. Liu, Y.B. et al. Effects of Bt cotton and Cry1Ac toxin on survival and development 

of pink bollworm (Lepidoptera: Gelechiidae). J. Econ. Entomol. 94, 1237–1242 

(2001). 

37. Graham, H.M. & Mangum, C.L. Larval diets containing dye for tagging pink bollworm 

moths internally. J. Econ. Entomol. 64, 376–379 (1971). 

38. Sims, M.A. et al. Arizona’s multi-agency resistance management program for Bt 

cotton: sustaining the susceptibility of pink bollworm. in Proceedings of the 2001 

Beltwide Cotton Conferences, January 9–13, 2001, Anaheim, California, 2, 

1175–1179 (National Cotton Council of America, Memphis, Tennessee, USA, 

2001). 

39. Morin, S. et al. Three cadherin alleles associated with resistance to Bacillus 

thuringiensis in pink bollworm. Proc. Natl. Acad. Sci. USA 100, 5004–5009 

(2003). 

40. Morin, S. et al. DNA-based detection of Bt resistance alleles in pink bollworm. 

Insect Biochem. Mol. Biol. 34, 1225–1233 (2004). 

41. Tabashnik, B.E. et al. DNA screening reveals pink bollworm resistance to Bt cotton 

remains rare after a decade of exposure. J. Econ. Entomol. 99, 1525–1530 

(2006). 

42. Tabashnik, B.E. et al. Asymmetrical cross-resistance between Bacillus thuringiensis 

toxins Cry1Ac and Cry2Ab in pink bollworm. Proc. Natl. Acad. Sci. USA 105, 

11,889–11,894 (2009). 

43. Tabashnik, B.E. et al. Dispersal of pink bollworm (Lepidoptera: Gelechiidae) males 

in transgenic cotton that produces a Bacillus thuringiensis toxin. J. Econ. Entomol. 

92, 772–780 (1999). 

44. Sokal, R.R. & Rohlf, F.J. Biometry (W.H. Freeman and Company, San Francisco, 

1969). 

45. SAS Institute. JMP 8.0, Cary, North Carolina, USA (2008). 

doi:10.1038/nbt.1704 


errata and corrigenda 

Erratum: GM alfalfa—who wins? 

Jeffrey L Fox 

Nat. Biotechnol. 28, 770 (2010); published online 9 August 2010; corrected after print 7 December 2010 

In the version of this article initially published, Geertson Seed Farms was spelled “Geerston.” The error has been corrected in the HTML and PDF 

versions of the article. 

Erratum: South-South entrepreneurial collaboration in health biotech 

Halla Thorsteinsdóttir, Christina C Melon, Monali Ray, Sharon Chakkalackal, Michelle Li, Jan E Cooper, Jennifer Chadder, 

Tirso W Saenz, Maria Carlota de Souza Paula, Wen Ke, Lexuan Li, Magdy A Madkour, Sahar Aly, Nefertiti El-Nikhely, Sachin Chaturvedi, 

Victor Konde, Abdallah S Daar & Peter A Singer 

Nat. Biotechnol. 28, 407–416 (2010); published online 7 May 2010; corrected after print 7 December 2010 

In the version of this article initially published, a line was missing connecting India and China in Figure 3. The error has been corrected in the 

HTML and PDF versions of the article. 


Corrigendum: The BioPAX community standard for pathway data sharing 

Emek Demir, Michael P Cary, Suzanne Paley, Ken Fukuda, Christian Lemer, Imre Vastrik, Guanming Wu, Peter D’Eustachio, 

Carl Schaefer, Joanne Luciano, Frank Schacherer, Irma Martinez-Flores, Zhenjun Hu, Veronica Jimenez-Jacinto, Geeta Joshi-Tope, 

Kumaran Kandasamy, Alejandra C Lopez-Fuentes, Huaiyu Mi, Elgar Pichler, Igor Rodchenkov, Andrea Splendiani, Sasha Tkachev, 

Jeremy Zucker, Gopal Gopinath, Harsha Rajasimha, Ranjani Ramakrishnan, Imran Shah, Mustafa Syed, Nadia Anwar, Özgün Babur, 

Michael Blinov, Erik Brauner, Dan Corwin, Sylva Donaldson, Frank Gibbons, Robert Goldberg, Peter Hornbeck, Augustin Luna, 

Peter Murray-Rust, Eric Neumann, Oliver Reubenacker, Matthias Samwald, Martijn van Iersel, Sarala Wimalaratne, Keith Allen, 

Burk Braun, Michelle Whirl-Carrillo, Kei-Hoi Cheung, Kam Dahlquist, Andrew Finney, Marc Gillespie, Elizabeth Glass, Li Gong, 

Robin Haw, Michael Honig, Olivier Hubaut, David Kane, Shiva Krupa, Martina Kutmon, Julie Leonard, Debbie Marks, David Merberg, 

Victoria Petri, Alex Pico, Dean Ravenscroft, Liya Ren, Nigam Shah, Margot Sunshine, Rebecca Tang, Ryan Whaley, Stan Letovksy, 

Kenneth H Buetow, Andrey Rzhetsky, Vincent Schachter, Bruno S Sobral, Ugur Dogrusoz, Shannon McWeeney, Mirit Aladjem, 

Ewan Birney, Julio Collado-Vides, Susumu Goto, Michael Hucka, Nicolas Le Novère, Natalia Maltsev, Akhilesh Pandey, Paul Thomas, 

Edgar Wingender, Peter D Karp, Chris Sander & Gary D Bader. 

Nat. Biotechnol. 28, 935–942 (2010); published online 09 September 2010; corrected after print 7 December 2010 

In the version of this article initially published, the affiliation for Ken Fukuda was incorrect. The correct affiliation is Computational Biology 

Research Center, National Institute of Advanced Industrial Science and Technology, Tokyo, Japan. The error has been corrected in the HTML and 

PDF versions of the article. 

Corrigendum: Analysis of a genome-wide set of gene deletions in the fission 

yeast Schizosaccharomyces pombe 

Dong-Uk Kim, Jacqueline Hayles, Dongsup Kim, Valerie Wood, Han-Oh Park, Misun Won, Hyang-Sook Yoo, Trevor Duhig, 

Miyoung Nam, Georgia Palmer, Sangjo Han, Linda Jeffery, Seung-Tae Baek, Hyemi Lee, Young Sam Shim, Minho Lee, Lila Kim, 

Kyung-Sun Heo, Eun Joo Noh, Ah-Reum Lee, Young-Joo Jang, Kyung-Sook Chung, Shin-Jung Choi, Jo-Young Park, Youngwoo Park, 

Hwan Mook Kim, Song-Kyu Park, Hae-Joon Park, Eun-Jung Kang, Hyong Bai Kim, Hyun-Sam Kang, Hee-Moon Park, Kyunghoon Kim, 

Kiwon Song, Kyung Bin Song, Paul Nurse & Kwang-Lae Hoe 

Nat. Biotechnol. 28, 617–623 (2010); published online 16 May 2010; corrected after print 7 December 2010 

In the version of this article initially published, the address of one of the authors, Young-Joo Jang, was incorrect. The correct address is Laboratory 

of Cell Cycle & Signal Transduction, WCU Department of NanoBioMedical Science, Institute of Tissue Regeneration Engineering, Dankook 

University, Cheonan, Korea. The error has been corrected in the HTML and PDF versions of the article. 



Catching the wave in China 

Connie Johnson Hambley 

China is presenting the pharmaceutical world with unprecedented opportunity. Here is what is happening there and 

how you can be a part of it. 


Catching a trend at its earliest stage and 

reaping its rewards as it advances is easy 

only in hindsight. It takes more than being 

able to identify a trend to ensure job security: 

it takes understanding the factors behind the 

trend to gauge its strength, longevity, direction 

and sustainability, as well as your place 

in its potential opportunities. Armed with 

this information, you can best assess your 

career options. 

Seeds of a boom 

In hindsight, it is easy to see the factors that 

created the biotech boom in the United 

States. The confluence of world-class academic 

institutions, human talent, innovative 

ideas, abundant investment capital, economic 

incentives and enjoyment of profits 

combined to create a magnet for the world’s 

brightest who wanted to be educated and 

work in the United States. Sporadic layoffs in 

big pharma were not inherently destabilizing 

because biotech companies were being created 

fast enough to absorb the talent. Many 

say there was no better place to be during the 

boom years than in Cambridge, San Diego or 

Research Triangle Park. 

During the boom, having a PhD or handson 

experience was a guarantee to lifelong 

employment. It was easy to manage a career 

when you were in demand and had a pick 

of jobs. But now, those who can find jobs 

are often relocating families, taking on less 

desirable or part-time positions or taking 

pay cuts. This comes as a surprise to many 

who feel that they have finally arrived at 

the station only to find that their train has 

already left. 

In addition, venture capital investment 

in biotech startups has declined, with some 

Connie Johnson Hambley is at Steele Executive 

Search, Boston, Massachusetts, USA. 

e-mail: connie@steelesearch.com 

venture capitalists citing ‘investor fatigue’— 

characterized as a reluctance to ‘re-up’ 

financing of laboratories and talent as a drug 

moves closer to the clinic. As a result, companies 

cut research staff and run virtually, 

meaning they consist of management and 

project managers who coordinate with contract 

research organizations (CROs). Drug 

development is increasingly moving offshore 

through outsourcing and big pharmas such 

as GlaxoSmithKline and Merck are creating 

R&D outposts in China, which are taking the 

jobs with them. In the new reality, a global 

industry must now include Asia, not just the 

United States and Europe. 

China is poised for a 

prolonged period of 

sustainable growth 

and that can translate 

into job longevity and 

career advancement, 

says Connie Johnson 

Hambley. 

As a result, opportunities for pharmaceutical 

professionals are exploding in places like 

India and China. However, it is China that is 

poised to enjoy the greatest growth. The ZRG 

Partners Global Life Science Hiring Index 1 

states that in the third quarter of 2010 the Asia 

Pacific region enjoyed a 20% increase in hiring 

over the second quarter, whereas the Americas 

saw a 2% decline in hiring in the same period. 

Even before the recent decline, the seeds for 

the boom in China were being sown. 

From ‘Made in China’ to ‘Created in China’ 

“A tidal wave of change is coming to China, 

and the goal for many is to ride the wave 

rather than be buried by it,” says Jason Mann, 

who advises biopharmaceutical firms on 

emerging market strategy with a focus on 

China. “As Americans we often view the 

Chinese through the viewpoint of our own 

lifetime, as the sudden rise of a new power, 

but many Chinese view themselves through 

a historical lens of thousands of years.” 

According to Mann, we are now seeing the 

impact of reforms in China over the last 30 

years. He notes that some observe an unspoken 

contract at work between the people and 

the government in post-Mao China: “Give 

the people an ever-rising standard of living 

through economic reform and they will focus 

less on political reform.” This emphasis on 

economic reform combined with China’s 

scale means that the government can mobilize 

tremendous resources behind favored 

sectors, with an increasing focus on life science 

investment and growth. 

The Chinese government created a 

complex series of reforms and initiatives 

designed to move China from an economy 

based on manual labor and manufacturing 

to one of innovation and creation, as articulated 

in their most recent five-year plan. 

Biotechnology is a key focus, and the biotech 

industry in China has grown to nearly $9 

billion from $3 billion barely five years ago, 

with much of that growth heavily supported 

with government investment. An estimated 

additional $1.5 billion investment into drug 

development has been announced. 

Barriers to entrepreneurial growth have 

been removed. Intellectual property laws 

were revised and high-profile enforcement 

of those laws has given investors and scientists 

greater confidence. And with the reemergence 

of the stock exchange barely 20 

years ago, the longstanding Chinese tradition 

of private enterprise has been enhanced by 

strengthening the mechanisms for distributing 

ownership over a larger number of people. 

Chinese companies are now mastering 

that truly entrepreneurial holy grail—the initial 

public offering. All of these factors show 

that China is poised for a prolonged period 




of sustainable growth and that can translate 

into job longevity and career advancement. 

The human factor 

Restructuring laws and funneling money into 

infrastructure are helpful supports, but it is the 

human talent that gets the work done. In drug 

discovery and development, Western-trained 

pharmaceutical professionals have an edge. 

Wang Huiyao, visiting fellow at the 

Brookings Institution and director general 

of the Center for China and Globalization, 

states that China has had 1.62 million students 

go overseas to get their education since 

1978. Only 497,000 have returned to China 

and only 8% of science and engineering PhD 

graduates have returned. As China builds its 

industry, the lack of seasoned professionals 

is strongly felt. 

A hallmark of Western R&D culture is to 

question what is being presented and not to 

merely accept information. Scientific rigor 

is the intellectual process of innovation 

and creativity where peers and superiors 

engage in a dynamic dialog of questions 

and critiques requiring scientists to push 

back with rebuttals to support their findings. 

Weaknesses are identified, unusual 

relationships are discovered and creative 

problem-solving results. This is in contrast 

to traditional Chinese education, which has 

a more top-down approach, where challenging 

an authority can be uncomfortable. This 

cultural ethic produces high-quality skills 

uniquely suited to drug discovery. 

In 2008, China launched its ‘Thousand 

Talents’ program designed to change China’s 

economy from an investment-driven to a 

more human resources–driven one. With its 

large population viewed as a valuable natural 

resource, the Thousand Talents program is 

designed to invest in and cultivate human 

capital. Business entrepreneurs, technical 

professionals and highly skilled workers 

make up three of the six targeted categories 

and each is essential for creating and 

sustaining a biotech and life science boom. 

The plan lures highly trained professionals 

and researchers to its shores with financial 

incentives and then uses those professionals 

to further train local talent. 

China is also pouring money and resources 

into revising and deepening their educational 

systems to address the needs of their population 

as a whole and to supply the human 

capital that their targeted industries require. 

The Thousand Talents program acknowledges 

that it is not only book training that 

young minds need, but experienced professionals 

to guide them and impart knowledge. 

And these professionals need not just be ‘sea 

turtles’, or Chinese returnees—the plan specifically 

allows for the recruitment of foreign 

nationals. Western-trained pharmaceutical 

talent receives the most interest. 

“The young scientists in China are hungry 

for knowledge. They want to learn and 

they want to learn from the best,” says Yun 

He, chief scientific officer of BioDuro, the 

Beijing subsidiary of US-based CRO giant 

PPD. “The most important factor for success 

is to have the right talent. The people part is 

hard. [The] money part is easier.” Focusing 

on cultivating talent in-house, BioDuro 

launched the BioDuro Learning Institute in 

2009 using professional instructors and their 

own senior management to provide instruction 

on everything from management skills 

to good laboratory practices to English. 

Limited time of opportunity 

Language is not a barrier to working in 

China. “Most scientists are surprised when 

they walk into a lab in Beijing and sit down 

and interact with a team,” states John Oyler, 

president and CEO of BeiGene, a startup 

cancer research company in Beijing. English 

is the predominant language in most laboratories 

because senior and middle managers 

received essential scientific training in the 

United States. But Oyler cautions there could 

be “a short window of time” for a non-Chinese 

speaking employee to reap all of the rewards 

of the current climate. 

Both He and Oyler roughly estimate the true 

unequalled opportunity horizon at five years. 

“The quality that already exists in China makes 

it almost too late for some synthetic chemistry 

and perhaps structural biology professionals,” 

continues Oyler, citing the more systematic 

and formulaic disciplines more suited to current 

Chinese educational standards. “Medicinal 

chemistry scientists have a shorter window. 

Folks in development or other more specialized 

areas of biology may have ten years and translational 

research may be even longer given the 

sheer number of patients and the importance 

of proximity to the clinic.” These time horizon 

estimates are supported by both the five-year 

plan and the Thousand Talents program goals. 

Certainly stellar talent knows no geographical 

bounds. If you are preeminent in your field, you 

will have opportunities regardless of country. 

So what does this mean for you? 

In China, Western-trained scientists have 

an edge most are not aware of, and it comes 

in the very essence of their training. The 

pharmaceutical industry has always shifted 

workers around the globe to meet its needs 

and China is the new frontier. Its growth is 

not a passing trend but a major force in its 

early stages. Individual employees may be 

powerless to avoid being laid off, but they 

are not powerless to take advantage China’s 

opportunities. 


The author declares competing financial 

interests: details accompany the full-text HTML 

version of the paper at http://www.nature.com/ 

naturebiotechnology/. 

ACKNOWLEDGMENTS 

C.J.H. would like to thank Michael Russo at the 

Bruckner Group for his help in the preparation of 

this article. 

1. ZRG Partners. Global Life Science Hiring Index, 

http://www.zrgpartners.com/Documents/lifescience 

indexq32010.pdf 


people 


PolyTherics (London) has announced the appointment of John Burt 

(left) as chief business officer. He was most recently CEO and cofounder 

of Thiakis. His previous posts include positions at Imperial 

Innovations, GlaxoSmithKline and Vernalis. 

“We are delighted that John has joined us and are sure that his 

experience, notably with Thiakis and GSK, will be invaluable,” 

says PolyTherics CEO Keith Powell. “We now have a management 

team which, together with our chairman, Ken Cunningham, has 

the breadth of experience with both biotech and large pharma 

companies to take PolyTherics to the next stage in its evolution.” 

Vical (San Diego) has announced the appointment 

of Igor P. Bilinsky as senior vice president, 

corporate development. Bilinsky was 

previously at Halozyme Therapeutics, where he 

served most recently as vice president, business 

development and special operations. 

Aastrom Biosciences (Ann Arbor, MI, USA) has 

named Ronald Cresswell to its board of directors. 

Cresswell is the former senior vice president 

and CSO of Warner-Lambert and has served on 

the board of directors of Esperion Therapeutics, 

Allergan, Curagen and Vasogen. 

Milind S. Deshpande has been promoted to 

president of R&D and retains his position as 

CSO, and Mary Kay Fenton has been promoted 

to senior vice president and retains her 

position as CFO at Achillion Pharmaceuticals 

(New Haven, CT, USA). Deshpande previously 

served as the company’s executive vice 

president of R&D and Ms. Fenton served as 

vice president. 

Jeffrey D. Edelson has been appointed chief 

medical officer of Palatin Technologies 

(Cranbury, NJ, USA). Edelson has more than 

15 years of clinical development experience 

in the pharma and biotech industries, most 

recently serving as executive vice president 

of R&D and chief medical officer of Ikano 

Therapeutics. Prior to that, he was vice president 

and therapeutic area head of novel therapeutics 

for Johnson & Johnson Pharmaceutical 

Research and Development. In addition, Palatin 

announced the resignation of Trevor Hallam 

as executive vice president of R&D, effective 

December 31. His departure results from the 

company’s decision last quarter to eliminate 

research and discovery activities, and focus on 

advancing its phase 2 clinical candidates. 

Ambit Biosciences (San Diego) has named 

biotech industry veteran Faheem Hasnain as 

chairman of the board and Alan Fuhrman 

as CFO. From December 2008 until its 

acquisition by Abbott Laboratories in April, 

Hasnain was president, CEO and a director 

of Facet Biotech. Previously he was with PDL 

BioPharma, Biogen Idec, Bristol-Myers Squibb 

and GlaxoSmithKline. Fuhrman joins Ambit 

from Naviscan, where he served as vice president 

and CFO. He also held positions at Sonus 

Pharmaceuticals and Integrex. 

James Patrick Kelly has been appointed senior 

vice president, CFO, treasurer and secretary of 

Vanda Pharmaceuticals (Rockville, MD, USA). 

Kelly has more than 18 years of financial and 

operating experience in the healthcare industry, 

most recently as vice president, controller 

at MedImmune. 

The board of directors of Athersys (Cleveland, 

OH, USA) has elected Ismail Kola as a director 

of the company. Kola currently serves as executive 

vice president of Belgium-based pharma 

company UCB and as president of UCB New 

Medicines, UCB’s discovery research through 

proof-of-concept organization since November 

2009. He currently serves on the board of 

directors of Astex Pharmaceuticals, Synosia 

and Ondek, and has previously served on the 

board of directors for companies such as Eragen 

Biotech, Promega and Ingene. 

Regeneron Pharmaceuticals (Tarrytown, NY, 

USA) has elected Christine A. Poon to fill 

a new seat on its expanded board of directors. 

Poon currently serves as dean of Ohio 

State University’s Fisher College of Business. 

She was formerly vice chairman of the board 

of directors of Johnson & Johnson and 

worldwide chairman of the Johnson & Johnson 

Pharmaceuticals Group. 

Vestaron (Kalamazoo, MI, USA) has announced 

the election of John Sorenson to the position of 

chairman and Elin Miller to the position of vice 

chairman of the company’s board of directors. 

Both were previously members of the board. 

Former Chairman Eli Thomssen remains on 

the board. Sorenson was formerly president of 

Syngenta Biotechnology and before that president 

of Syngenta Seeds. Miller previously held 

senior management positions with Dow, Dow 

AgroSciences and ArystaLifeScience. 

Amarin (Dublin, Ireland, and Mystic, CT, 

USA) has announced the appointment of two 

replacements for president, CEO and director 

Colin Stewart, who resigned in November to 

address personal matters. Joseph S. Zakrzewski, 

who has served as Amarin’s chairman of the 

board since January 2010, has been named 

CEO while continuing to serve as chairman. 

He most recently served as CEO of Xcellerex. 

Additionally, John F. Thero has been appointed 

president of Amarin. Since November 2009, 

Thero had been the company’s CFO. 

Gentris (Morrisville, NC, USA) has appointed 

Sam C. Tetlow executive chairman of the 

company’s board of directors. He currently 

serves as vice president of business development 

for Calvert Research and as a director 

of Immunologix. In addition, Gentris 

has named Howard McLeod as the company’s 

first chief scientific advisor. McLeod 

is the Fred N. Eshelman Distinguished 

Professor and director of the Institute for 

Pharmacogenomics and Individualized 

Therapy at the University of North Carolina 

at Chapel Hill. He serves as principal investigator 

for the CREATE Pharmacogenomics 

Research Network and is a member of the 

FDA Committee on Clinical Pharmacology. 

Timothy P. Walbert has been elected to the 

board of directors of antibody therapeutics 

developer XOMA (Berkeley, CA, USA). A 

seasoned executive with experience launching 

therapeutic products including Abbott 

Laboratories’ Humira (adalimumab), Walbert 

currently serves as chairman, president and 

CEO of privately held Horizon Pharma.

Untitled

Create successful ePaper yourself

Delete template?

Save as template?