Check GAD antibody positivity with the Diamyd anti-GAD RIA plate*

forewordResearch Scientists throughout the world are currently facedwith a time-old challenge – to define and understand themechanisms leading to development of autoimmune diseases,and then to determine and develop efficient means oftreating or preventing them. While these might sound like two distinctchallenges, the definition of the molecules targeted in an autoimmune diseaseprocess also provides the candidates for therapeutic targeting.To date, there is no vaccine for any of the many autoimmune diseases thataffect millions of people throughout the world. A number of candidate moleculeshave been identified and targeted, such as insulin in Type 1 diabetesand myelin basic protein in Multiple Sclerosis, but clinical vaccination trialsusing these molecules have not yet been successful. Given the recent advancesin gene technology and knowledge of the genetic codes comprising bothMan and experimental animals, the potential for discovery of new candidatemolecules is great. However, one candidate molecule identified manyyears ago still stands the test of time.Glutamic acid decarboxylase 65 (GAD65) is a candidate autoantigenimplicated in development of the autoimmune disease Type 1 diabetes. Inautumn 1994, ‘The Story of GAD’ by Robert Dinsmoor appeared in theCountdown magazine, a publication of the Juvenile Diabetes ResearchFoundation International. This article described the series of research discoveriesencompassing the period 1982-1993 that led to definition ofGAD65 as a prime candidate autoantigen in Type 1 diabetes. The articleended with two pertinent statements – that access to recombinant GADin industrial-sized quantities would be required for testing of GAD therapiesin humans, and that GAD-based therapies should provide a meansof preventing diabetes.During the last 10 years, intensive efforts have led to development ofhigh quality recombinant GAD65 proteins, which have been tested inPhase I and Phase II clinical trials.Additionally, recent research efforts havealso identified GAD as an important elementin Parkinson’s and other neurologicaldiseases, indicating that the candidate moleculehas not only remained a potential keyto prevention of diabetes, but also importantin treatment of other diseases as well. Thereis obviously still more to learn about thisfascinating protein.This issue is dedicated to GAD65 with anumber of contributions made by several prominent GAD researchers. Aswell as personal reflections of their own past experiences of GAD, thecurrent status and proposals for future applications of GAD are presented.We hope you find these articles interesting.Assoc. Prof. Robert A. HarrisintroductionIntroduction byLars KlareskogIn 1999 diamyd medical initiated a unique cooperationwith the Karolinska Hospital whensetting up its Competence Center forAutoimmune Diabetes (DMCCAD) at theCenter for Molecular Medicine (CMM) atKarolinska. Since its inception, with Robert Harris atthe helm, DMCCAD has become an interesting pointfor cross-fertilization of ideas related to various autoimmunediseases including Type 1 diabetes, MS andRheumatoid Arthritis.It is generally agreed that Glutamic AcidDecarboxylase 65 (GAD) is an important antigen inType 1 diabetes. As in other autoimmune diseases,recent research is focused on induction of tolerance todisease specific antigens. This may protect self structuresunder attack without interfering with the immunesystem’s capabilities in other areas, such as combattingviral infections or cancer.While it is already generally accepted that GADantibodies (GADA) is a major diagnostic marker forType 1 diabetes and predicting the course of developmentof Type 2 diabetes (LADA-patients), toleranceinduction to GAD may become a way to prevent thedevelopment of insulin dependency in LADA patients.I have had the pleasure of being able to follow thedevelopment of Diamyd Medical’s GAD-based vaccinefor LADA patients. The outcome of the Phase IIstudy in LADA patients is important, not only becauseit may lead to a new drug for diabetes therapy, butits results may also be of importance when designingfuture clinical trials in other diseases such as MS.page 4 dmccad june 2003

Lars Klareskog, MD, Ph.D., is Professor of Rheumatology and Head ofthe Rheumatology Research group at the Karolinska Hospital inStockholm and formerly Professor of Medical Immunology and Headof Clinical Immunology at the Uppsala Teaching Hospital. Klareskog'sresearch is specifically aimed at the causes and treatment of autoimmunedisorders. Klareskog is a member of the Nobel Foundation.Comments on Diamyd Diabetes Vaccineby Hans WigzellThat the human immune system can react against selfstructures is well known. Most such reactions do notcause disease – but some do.In the efforts to fight autoimmune disease such as Type1 diabetes, one therapeutic approach is to modify the immune reactionand induce functional tolerance to potentially relevant moleculessuch as GAD by administration of the autologous antigen itself.Similar approaches have been successful before. For exampleadministration of autoantigens have alleviated autoimmune diseasein animal models. Hyposensitization (where increasing doses of allergenare used to treat allergies) is another example.How should GAD be used to induce functional tolerance?Using our knowledge from conventional as well as from autologousvaccines it seems logical to use doses from a few micrograms to 500Hans Wigzell, MD, D.Sc, is Professor of Immunology, Dean of theKarolinska Institute and Chief Scientific Advisor to the SwedishGovernment. Wigzell is one of Sweden´s most prominent and internationallyrenowned scientists in the field of Immunology. Wigzellwas Director General of the National Bacteriological Laboratory1988-1993; Director General of the Swedish Institute for InfectiousDiseases 1993-1994; Wigzell is since 1990 Chairman of the ECConcerted Research Programme: European Vaccine against AIDS(EVA). Wigzell was Chairman of the Nobel Committee 1990-1992and is Chairman of the Nobel Foundation.micrograms.To use alum as adjuvant seems highly recommendable. It is conventionaland it is biased to the humoral rather than the cellularimmune response which is logical when the cellular response is to bereduced.Subcutaneous administration is recommendable and conventional.Intravenous administration is much more problematic and intramuscularis less efficient in terms of immunogenicity.The GAD-vaccine differs from a conventional vaccine in that theadministered antigen is already present naturally in the body. Thisprobably means that the vaccine needs to be injected a couple oftimes. Therefore a prime and boost strategy seems logical.In summary, the diamyd approach to induce functional tolerance toGAD seems logical.dmccad june 2003page 5

The Story ofGADRobert Dinsmoor1975Researchersdiscover isletcell surfaceantibodies inchildren withdiabetes1982Scientistsfind 64Kantibodiesin diabeticBB rats1986Researchshows that64K antibodiesdevelopyears beforediabetes1988Scientistsfind antibodiesto GAD inpatients withStiff ManSyndromeResearchersdiscover Islet CellAntibodies (ICAs)in peoplewith newlydiagnoseddiabetes1978Scientistsfind antibodiesto 64Kprotein indiabeticchildren1984Molecularbiologistssequencegene formaking GAD1987Researchersfind 64Kantibodies indiabeticNOD miceThis article’s original form appeared in Countdown magazine, a publication of the Juvenile Diabetes Research Foundation International, Autumn 1994.At one time, scientists thought that diabetes wascaused by a virus or toxic environmentalagent that destroyed theinsulin-producing beta cells of thepancreas. But decades ago, scientistsdiscovered that people with newlydiagnosed diabetes had inflammationof the islet cells of the pancreas, or insulitis.The islets were filled with white blood cells, whichthe body’s immune system uses to fightinfection.This led researchers to speculatethat diabetes might be an autoimmunedisease, caused by a misdirectedattack by the body’s own immunesystem.page 6 dmccad june 2003

More support for the autoimmunetheory came in the early1970’s when Dr. Bottazzo andDr. Doniach and colleagues atthe Middlesex Hospital inLondon first detected antibodiesto pancreatic islet cells in people with newly diagnoseddiabetes. The antibodies, which react to the cytoplasminside all islet cells, came to be called cytoplasmicislet cell antibodies or ICAs.Antibodies are a part of the immune system responsecalled the humoral response. Made by white blood cellscalled B lymphocytes, specific antibodies are designed toattack specific foreign proteins or antigens. It was unclear,however, whether the beta cells were actually destroyedby these antibodies or by components of the cellularimmune response, such as the killer T lymphocytes or“T cells”.In the mid-1970’s, Dr. Lernmark, who had developed atechnique for isolating beta cells from isolated islets of1990In a majorbreakthrough,scientistsidentify the64K proteinas GAD1991Scientistsdiscover that 64Kprotein isGAD65, thesmaller formof GADStudy showsthat T cellstargetGAD651993Research teamsprevent diabetesin NOD mice byinactivating GAD-Reactive T cellsResearcherssuggest 64Kantibodies maybe earlypredictor ofdiabetesMolecularbiologistssequence twoforms of GAD– GAD65 andGAD67 – fromislet cellsStudies showthat someICAs only reactto GAD65Two studiesestablish GADas first knowntarget of T cellsin diabeticNOD micethe Langerhans, was researching proteins on the surface ofbeta cells at the University of Chicago. While involved inthis research, he attended a lecture in ICAs in Chicago byDr. Doniach. During the lecture, he realized that it mustbe the proteins on the surface of the beta cells that provokedantibodies in diabetes, and he turned his research inthat direction. He and other researchers at the Universityof Chicago treated pancreatic islets cells from rats withblood serum from children who had diabetes for less thanfive years. They discovered that some of these childrenhad antibodies to proteins on the surface of the islet cells,called islet cell surface antibodies or ICAs. However, theydid not know which specific proteins had attracted theantibodies.The 64,000 Mystery ProteinIn the late 1970’s, Dr. Lernmark became Director ofResearch at the Hagedorn Research Laboratory inDenmark, where he continued to work with ICAs. Heenlisted the aid of Dr. Baekkeskov, who had developeddmccad june 2003page 7

Åke Lernmark, MD, Ph.D., and his colleagues atHagedorn Research Laboratory in Denmark, firstreported 64K proteins in humans in 1982.Steinunn Baekkeskov, Ph.D., played a primary role indetecting the 64K protein in humans and later worked incollaboration with Yale researchers to identify the proteinas GAD.Mark Atkinsson, Ph.D., and colleagues at the Universityof Florida in Gainesville, concluded in 1990 that 64Kautoantibodies might be the earliest and best marker fordiabetes yet identified.techniques for detecting membrane proteins while studyingthe role of antigens in African sleeping sickness. Usingnewly developed and very sensitive tests such as the“immunoprecipitation technique”, Drs. Baekkeskov,Lernmark and colleagues at the Hagedorn found 80-90percent of blood serum samples from children withnewly diagnosed diabetes had antibodies to an unidentifiedICSA with molecular weight of approximately 64,000daltons-or 64 kilodaltons. They dubbed the mystery protein64K, and reported their findings in 1982 in Nature.In 1984, they reported that 64K antibodies were oftenfound in diabetic BB rats (a commonly used animalmodel for human diabetes). One intriguing aspect of thisfinding was that the antibodies appeared 40 to 70 daysbefore the onset of diabetes. This suggested that the 64Kprotein was an early and major target of the immunesystem. But did this same predictive effect hold true forhumans as well?They next looked at blood serum samples of 14 peoplewho later developed diabetes. Eleven of them had antibodiesto the 64K protein - for up to seven years before theybegan to show symptoms of diabetes. This suggested that64K antibodies might be an early warning sign for theeventual development of diabetes.In the summer of 1986 Dr. Atkinson who had beenstudying environmental influences on diabetes in rats atthe University of Florida in Gainesville, traveled toHagedorn as a visiting scientist. There Baekkeskov taughthim the immunoprecipitation technique for detecting64K antibodies. Armed with this new skill, he returned tothe University of Florida, where he and Dr. NoelMaclaren, detected the 64K antibodies in nonobese diabetic(NOD) mice, a newly bred type of animal model fordiabetes. Finding the 64K antibodies in animals helpedconfirm that the 64K antigen was an important feature indiabetes.Drs. Atkinson, Maclaren, and colleagues also studiedfirst-degree relatives of people with diabetes, who were athigher risk for developing diabetes themselves. In 1990, inthe British Journal The Lancet, they reported finding 64Kautoantibodies in 23 of 28 people who developed diabetesup to seven years later. Based on these findings and thoseof the Hagedorn researchers they concluded that 64Kmight be the earliest and best marker for diabetes yetidentified and might be especially useful for predictingdiabetes.As diabetes researchers became increasingly aware ofhow important this protein was, they set out to analyzeits chemical characteristics to compare with known proteins.In 1988, Drs. Solimena, Camilli, and their colleagues atYale University reported in The New England Journal ofMedicine that they had found the target antigen in patientswith an autoimmune disorder called Stiff-ManSyndrome (“SMS”). SMS is a rare disorder of the centralnervous system in which a person’s muscles become moreand more rigid, and painful spasms occur. Patients withthe disorder were found to have antibodies to an enzymecalled Glutamic acid Decarboxylase, or GAD.GAD is responsible for converting the amino acid glutamateinto a protein called GABA, which the brain cells useto communicate. It turned out that many patients withpage 8 dmccad june 2003

GAD antibodies had diabetes – even the ones who neverdeveloped SMS. Moreover, they found that antibodies fromthe serum of SMS patients targeted the beta cells, and thesewere not any of the previously known antibodies.This led to an historic collaboration between Dr.Baekkeskov, now at the University of California, SanFrancisco, and the Yale research team. Dr. Camilli sentBaekkeskov blood serum samples from SMS patients (containingGAD antibodies), and she sent him blood serumsamples from diabetes patients (containing 64K antibodies).Then they used various methods to demonstrate thatthe proteins were completely identical. The end result ofthis meeting of the minds was their landmark paper inNature identifying the 64K protein as the enzyme GAD.The discovery set off a flurry of interest in GAD by diabetesresearchers.Send in the ClonesDr. Kaufman, who had been studying the role of GAD inneurological diseases such as epilepsy, became interested indiabetes one day in 1990 when his car broke down. Whilewaiting for it to be fixed, he went to the library, wherehe stumbled on the article by Dr. Atkinson and colleaguesin The Lancet describing the 64K antibodies as predictorsof diabetes.According to Kaufman, the landmark paper identifying64K as GAD had yet to appear in Nature, but he was ableto put the two together for himself: He knew about the roleof GAD in SMS and the fact that it was found in beta cells,and he also knew that that molecular weight of GAD wasapproximately 64kD. He joined forces with the researchteam at the University of Florida, as well as with his formermentor at the UCLA Dr. Tobin, who was researching therole of GAD in Huntington’s disease and epilepsy. Therewere known to be slightly different molecules of GADfound in the brain – one weighing 67kD (GAD67) and asmaller one weighing 65kD (GAD65). As a graduate studentworking with Tobin, Kaufman had worked out the geneticsequence for GAD67, and another graduate student Dr.Erlander, had done the same for GAD65. Thanks in part tothis work, Dr. Tobin’s laboratory was now able to makeboth forms of GAD in quantity using genetically alteredbacteria.Spurred by the news that 64K was GAD, other researchgroups cloned GAD. In 1988, Dr. Lernmark had moved tothe University of Washington in Seattle with some membersof his former Hagedorn research team. One of theseDr. Allan Karlsen, cloned the gene for human isletGAD65. One of Lernmark’s former graduate student’s, Dr.Birgitte Michelsen, cloned GAD67. According to Dr.Lernmark, cloning GAD enabled researchers to makerecombinant GAD in quantities they had never dreamedAllan Tobin, Ph.D., and Daniel L. Kaufman, Ph.D., at the University of California at Los Angeles published one of two1993 papers in “Nature” confirming that GAD is the first known target of the T cell attack in diabetes.of. Cloning GAD65 also enabled researchers to determinewhich molecule was the alter ego of 64K.The researchers in Tobin’s lab sought help from aUCLA diabetes expert, Dr. Michael Clare-Salzler, tostudy GAD’s role in diabetes. Using blood serum samplesfrom diabetic patients with newly diagnosed diabetes providedby Dr. Clare-Salzler, the researchers showed that96% of patients with newly diagnosed diabetes had antibodiesto one or both forms of GAD. Furthermore, itappeared that 64K was GAD65, the smaller of the twomolecules that had been identified.Dr. Erlander also used a computer search that uncoveredstructural similarities between both forms of GADand proteins in the Coxsackie virus, a virus known toinfect many people who eventually developed diabetes.“That moment was probably the most exciting moment inmy whole scientific life!” Dr. Tobin recalled. This findingsupported a theory known as “molecular mimcry”, whichsuggested that infection by the Coxsackie virus might setoff the immune reaction in diabetes. According to the“true identity”pageAs diabetes researchers became increasinglyaware of how important thismystery protein was, they were eager to discover itsdmccad june 20039

theory, the immune system mistakenlyattacks GAD in the beta cells aswell.GAD Provokes an Immune AttackIn 1992, a number of studies showedthat some of the ICAs, the antibodiesdescribed back in 1974 that werevery predictive for diabetes, targetedGAD. One study from Bottazzo’sresearch group in London found thatICAs from the blood serum samplesof pre-diabetic individuals seemed toattack GAD. Likewise Dr. GeorgeEisenbarth and Dr. Robert Gianiniand colleagues at Joslin DiabetesCenter in Boston studied ICAs fromHugh McDevitt, at Stanford University suggested in 1993that diabetes could be prevented in its earliest stages by relatives of people with diabetes.preventing the T cell attack against GAD.They found that GAD was one targetof ICAs, but there might be otherantigens as well.These reports showed an attack on GAD by the humoralimmune system – that is antibodies. But these antibodiesare probably not what destroys the beta cells. Morelikely, they are destroyed by the cellular immune responsein the form of T cells, white blood cells that infiltrate thepancreatic islets.The same year, Drs. Atkinson and Maclaren, working inconjunction with the UCLA research team, took T cellsfrom people with newly diagnosed diabetes, relatives ofpeople with diabetes, and nondiabetic individuals, andexposed them all to GAD65. The T cells that reacted andmultiplied in the presence of GAD65 tended to be thosefrom people with diabetes and their relatives who hadICAs and would later develop diabetes. This showed thatGAD provokes a T cell attack, which may be whatdestroys the beta cell in diabetes.In 1993, two studies published in Nature helped confirmGAD as the first known target of the T cells andsuggested that diabetes could be prevented in its earlieststages by preventing the T cell attack against GAD. Onestudy was from Kaufman and colleagues at UCLA in conjunctionwith Dr. Atkinson, and the other was from Dr.Hugh McDevitt, Dr. Roland Tisch and their colleagues atStanford University. In each case, the researchers testedNOD mice for reactivity to some known antigens in diabetes,including the two forms of GAD. They found thatthe attack on GAD by T cells coincided with the developmentof insulitis – infiltration of the beta cells with whiteblood cells. Only later did the T cells attack other knownantigens in the beta cells.This suggested that GAD is the first antigen to beattacked by T cells, and that this attack then diversifies toinclude other antigens, and that these were recognizedand attacked by the immune system. Was this actually thecause, or was there some other antigen, as yet unknown,that provoked the first attack?To help answer these questions, both groups of researcherstried deactivating the T cell response in their NODmice. The UCLA group injected NOD mice with GAD atthree weeks of age, a treatment already shown to inactivateT lymphocytes against GAD. Most of the mice treatedwith GAD showed no T cell response against GAD, indicatingcomplete immune system tolerance to GAD. TheGAD tolerant mice showed no reaction to the other betacell antigens, and never developed any degree of insulitisor diabetes. This suggested that GAD – and not any ofthe other known antigens – provoked diabetes. On theother hand, mice treated with other antigens did developa T cell attack to the other beta cell to antigens, did developinsulitis, and went on to develop diabetes.The Stanford researchers likewise tried inactivatingGAD-reactive T cells – but in this case by injecting NODmice with GAD65 into their thymus (a major site ofimmune system programming). The 70 percent of NODmice who actually became tolerant of GAD had markedlyreduced T lymphocyte responses to the other beta cellantigens, reduced insulitis, and no development of diabetes.“I think this is an important step toward understandingthe disease process,” Dr. Tisch explained. “It appears thatthe disease occurs in specific stages, and now we may beable to categorize the antigens with regard to reactivitywithin those stages.”The Prediction and Prevention Pay-OffNow that GAD has been identified, cloned, and shown tobe important in the development of diabetes, it can playan important role in diabetes treatment, to predict whowill get diabetes – probably years before they develop anysymptoms of the disease. Now we have a very specificassay to show whether a healthy person has GAD antibody.But the ultimate implications of GAD go far beyondpredicting who will get diabetes. The recent studies inNOD mice show that GAD, too, might play a role in preventingdiabetes.Unfortunately, testing oral GAD in humans won’t befeasible until researchers have access to recombinantGAD in industrial-sized quantities and at an affordablecost. “We’ve actually discovered a direction for potentiallypreventing diabetes in human beings. Using treatmentssuch as with GAD we should be able to prevent diabetesin all individuals who are at risk of developing it,” concludedDr. Clare-Salzler.page 10 dmccad june 2003

Diamyd, Inc., is pleased to provide a comprehensive portfolio ofimmunoassays for In Vitro Diagnostic UseRECOMBINANT PROTEINS10-65702-13-01 T cell GAD 1mg/vial10-65702-01 rhGAD65 1mg/vial10-IA2-01Human IA-2 0.5 mg/vialAUTOANTIBODY DETECTION KITS10-1121-01 IAA ELISA10-1123-01 Diamyd anti GAD65 RIADIABETES HUMAN IMMUNOASSAYS10-1113-01 Insulin ELISA10-1132-01 Ultra Sensitive Insulin ELISA10-1128-01 Iso-Insulin ELISA10-1136-01 C-Peptide Specific ELISA10-1141-01 Ultra Sensitive C-Peptide ELISA10-1118-01 Proinsulin ELISADIABETES MAMMALIAN IMMUNOASSAYS10-1124-01 Rat Insulin ELISA10-1137-01 Ultra Sensitive Rat Insulin ELISA10-1145-01 High Range Rat Insulin ELISA10-1149-01 Mouse Insulin ELISA10-1150-01 Ultra Sensitive Mouse Insulin ELISA10-1129-01 Porcine Insulin ELISA10-1130-01 Sheep Insulin ELISA10-1131-01 Bovine Insulin ELISACONTROL SERUM10-1134-01 Diabetes antigen control, human L and H 2X 0.5 ml10-1135-01 Insulin control, Mammalian L and H 2X 0.5 mlCARDIOVASCULAR DISEASE10-1103-01 Apo(a) ELISA10-1106-01 Apo(a) RIA10-1143-01 Oxidized LDL ELISAT cell AND B-CELL SEPARATION KITS20-7200-01 Human T cell Kit20-7100-01 Mouse T cell Kit20-6900-01 Rat T cell Kit20-6900-01 B-Cell Accessory Kit (Human, Mouse, Rat)NEUROLOGICAL DISEASE10-1198-01 Diamyd S-100β Concentration Measurement Kit10-1199-01 Diamyd S-100-β Antibody ELISAwww.diamyd.comproducts@diamyd.com

GAD Back to the Future…Åke LernmarkÅke Lernmark, MD, Ph.D., is R.H.Williams Professor of Medicine atthe University of Washington inSeattle. Lernmark focused hisattention on diabetes and theantigen that later proved to beGAD at an early stage. Lernmarkwas first to demonstrate the presenceof antibodies against GADin patients with insulin-dependentdiabetes, and to use molecularmethods to define HLA genes thatare necessary, but not sufficient todevelop the disorder.The beta cell specific loss associated withthe clinical onset of Type 1 diabetesmellitus (T1DM) is a remarkable phenomenon.The ablation of the insulinproducingcells does not seem to affectneighboring cells producing glucagon,somatostatin or pancreas polypeptide. The eradication of thebeta cells is a testimony to the laser knife-like precision ofthe immune system to recognize and attack antigen. Theimmune mediated loss of beta cells is well documented inpancreas specimens from T1DM patients of short diseaseduration. The degree and character of the inflammatory cellinfiltration vary from patient to patient and would not predictclinical onset. The loss of beta cells appears more predictiveand the major question that we and others try to answeris whether the immune attack on the beta cells can bereversed by controlling the autoimmune reaction againstglutamic acid decarboxylase (GAD65) either before or afteronset. While autoantibodies to insulin and IA-2 are predictiveof T1DM primarily in the young, the predictive value ofGAD65 autoantibodies appears less age-dependent. It is thereforeoften argued that GAD65 is the major autoantigen inT1DM, which is consistent with GAD65 autoantibodiesbeing most prevalent (more than 80% of new onset patients)at onset. Although recent follow up studies suggestthat GAD65 autoantibodies are not necessarily the firstautoantibody to appear before the clinical diagnosis is made,appearance of all three autoantibodies to GAD65, IA-2 andinsulin is the best predictor of disease. Recognition ofGAD65 by helper CD4 and killer CD8 positive cells are likelyto precede the autoantibody appearance. At present,there is a lack of reliable and standardized tests that detectT cell reactivity against islet autoantigens (1). This is in contrastto GAD65 and IA-2 autoantibodies, which can be reliablymeasured against a WHO standard (2). At least therewill be one reliable way to monitor whether SpecificImmune Therapy (SIT) with GAD65 might alter the courseof T1DM disease associated autoimmunity.Following the molecular cloning of the human (3) androdent GAD proteins (4) it was possible to carry out experimentswith recombinant GAD65 to demonstrate that spontaneousdiabetes in the NOD mouse (5, 6) was preventable.Of equal importance was the observation that diabetes wasnot inducible by GAD65 in rats (7) and mice (8). T1D thereforedid not immediately seem to mimic other autoimmunediseases, which have animal models that are based on theactual immunization of an autoantigen. About 10 yearsafter the demonstration that NOD mouse diabetes was preventable,Phase I (toxicity) and Phase II (safety) clinical trialsof GAD65 have been completed. In the absence of adverseevents among the GAD65 autoantibody positive Type 2 diabetespatients (so-called LADA patients or LatentAutoimmune Diabetes in the Adult) it may now be possibleto move into Phase III clinical trials to test whetherGAD65 SIT is efficacious.The future approach to determine whether GAD65 SIT1994Researchers showthat transplantedislets survivelonger if they arepretreated withGAD1998Researcherstransfer diabetesbetween animalswith GAD-reactiveT cellsStart ofPhase Iclinical trialDiamyd Phase Iclinical trial in 24healthyindividualscompletedDiamyd Medicalinitiatesdevelopment of aGAD baseddiabetes vaccine1996Researchers showthat progressionto T1DM isassociated with achange in GADAisotypes1999Researchers showthat ß-cellexpression of GADis required fordevelopment ofT1DM in the NODmouse2000Start of Phase IIclinical trial

is truly inducing immune tolerance, deviation, or both, is tocarefully move from adult patients who already have diabetessuch as LADA patients to new onset adult T1DM patientsand then possibly to triple autoantibody positive firstdegree relatives since our ability to predict disease in suchsubjects is now well documented thanks to the DPT-1study with parenteral insulin administration (9). Theattempts to approach prevention in the fashion modeled inanimals will by necessity require a carefully staked outpathway. Novel observations are particularly tantalizing asto the possibility to use GAD65 SIT and cause no harm inhealthy subjects.The beta cell expression of GAD65 in human beta cells iswell established. There is a lack of qualitative and quantitativestudies on GAD65 in the non-insulin producing endocrineislet cells. Also, we still do not fully understand the roleof GABA for normal beta cell function. Recent observationssuggest that increased body mass index (BMI) is associatedwith GAD65 autoantibodies in subjects not developing diabetes(10, 11). It will be critical to find out if subclinicalGAD65 autoimmunity is related to obesity phenotypes arisingbecause of altered beta cell function, insulin resistance,or both. Future studies will also require detailed studies ofGAD65 gene expression in CNS areas controlling food intakesince GAD65 gene polymorphisms affecting GABA productionmay be related to morbid obesity (P. Boutin and P.Frougel, personal communication). The role of GAD65 genepolymorphism in food intake regulation, obesity developmentand beta cell function may therefore need to be furtherstudied before we can embark on large scale GAD65SIT.The relationship between Stiff Man Syndrome and GADautoimmunity was critical to the demonstration that T1DMsera immunoprecipitated 64K protein with GAD enzymeactivity (12). We have now learned that the immune responseto GAD65 in SMS is qualitatively and quantitatively diffferentfrom T1DM and the approach to prevent either SMS,T1DM, or both may require differentGAD65 SIT. One approach may be tobase the SIT on the HLA associationwith the disease. In the InsulinAutoimmune Syndrome the autoantibodyformation is strongly associatedwith HLA DRB1*0401. This haplotypeis insufficient for the association withinsulin autoantibodies in T1DM, whichis rather associated with HLADQA1*0501-B1*0201. Hence insulin aswell as GAD65 is the autoantigen ofmore than one disorder of autoimmunecharacter. Future studies will needto take these similarities and differencesinto account to design interventiontrials that may lead to the identificationof novel approaches to prevent orreverse autoantigen-specific autoimmunity.Don’t sit around, GAD back tothe future…References1. Peakman M, et al,Characterization of preparations of GAD65, proinsulin, and the islet tyrosinephosphatase IA-2 for use in detection of autoreactive T cells in Type 1 diabetes:report of phase II of the Second International Immunology of Diabetes SocietyWorkshop for Standardization of T cell assays in Type 1 diabetes.Diabetes 50:1749-1754, 2001.2. Mire-Sluis AR, et al,The World Health Organization International Collaborative Study for Islet CellAntibodies.Diabetologia 43:1282-1292., 2003. Karlsen AE, et al,Cloning and primary structure of a human islet isoform of glutamic acid decarboxylasefrom chromosome 10.Proc Natl Acad Sci US 88:8337-8341, 1991.4. Kaufman DJ, et al,Autoimmunity to two forms of glutamate decarboxylase in Insulin-dependentdiabetes mellitus.J. Clin. Invest. 89:283-292, 19925. Kaufman DL, et al,Spontaneous loss of T cell tolerance to glutamic acid decarboxylase in murineinsulin-dependent diabetes.Nature 366:69-72, 1993.6. Tisch R, et al,Immune response to glutamic acid decarboxylase correlates with insulitis in nonobesediabetic mice.Nature 366:72-75, 1993.7. Bieg S, et al,GAD65 and insulin B chain peptide (9-23) are not primary autoantigens in theType 1 diabetes syndrome of the BB rat.Autoimmunity 31:15-24, 1999.8. Plesner A, et al,Immunization of diabetes-prone or non-diabetes-prone mice with GAD65 doesnot induce diabetes or islet cell pathology.J Autoimmun 11:335-341, 1998.9. DPT-1: Effects of insulin in relatives of patients with Type 1 diabetes mellitus. NEngl J Med 346:1685-1691, 2002.10. Rolandsson O, et al,Levels of glutamate decarboxylase (GAD65) and tyrosine phosphatase-like protein(IA-2) autoantibodies in the general population are related to glucose intoleranceand body mass index.Diabetologia 42:555-559, 1999.11. Weets I, et al,Male-to-female excess in diabetes diagnosed in early adulthood is not specific forthe immune-mediated form nor is it HLA-DQ restricted: possible relation toincreased body mass index.Diabetologia 44:40-47, 2001.12. Baekkeskov S, et al,Identification of the 64K autoantigen in insulin-dependent diabetes as the GABAsynthesizingenzyme glutamic acid decarboxylase.Nature 347:151-156, 1990.Researchers preventT1DM in theNOD mouse byvaccination withGAD-plasmid DNAScientists prevent T1DM inthe NOD mouse with GADgenemodified Vaccinia virusResearchers report thatGAD gene polymorphismmay be associated withobesity200120022003ADA says in its Clinical PracticeRecommendations for 2002 that“autoantibody tests may be of value toidentify which newly diagnosedpatients have immune-mediated Type 1-diabetes in circumstances where it is notobvious".Resarchers successfullyuse GAD gene therapy fortreatment of Parkinson’sdisease in an animalmodelDiamyd Phase IIclinical trial in47 LADApatientscompleted

100GAD in GraphsGAD65 DNA vaccination preventsdiabetes in NOD micediabetic mice (in percent)90ppInsulin (n=25)80 ICA69+GAD65+ppIns (n=14)control (untreated) (n=14)70 control (vector) (n=12)GAD65 (n=14)6050403020DNA vaccination1000 10 20 30age (in weeks)Karges et al, Diabetes 51:3237; 2002Proliferation (cpm)• 4-6 week old female NOD• Different DNA vaccinations• Insulin construct does not protectfrom diabetes• GAD65 construct preventsdiabetes• Combining constructs abolishesGAD65-induced protectionBaculovirus-encoded recombinantGAD outperforms bacterial andyeast GAD preparations in T cellproliferation assays100000100001000baculovirusyeast100bacterial1 10Peakman et al (2002) Characterization of preparationsof GAD65, proinsulin and IA-2 for use indetection of autoreactive T-celsl in Type 1 diabetes.Diabetes 50:1749-1754• T cell recall immune responseswere studied by proliferation assaysusing defined T cell lines and clonesand different preparations of GAD65• Two independently producedGAD65 preparations expressed bybaculovirus gave higher proliferationindices than either yeast- orbacterial-produced GAD65Untangling the GADsAllan TobinMy interest in GABA and GAD came from my interest in Huntington’s disease – adegenerative neurological disorder that had killed the mother of two of my friends.I decided in the early 1980s that GAD was a reasonable “candidate gene” forHuntington’s disease, and I set out to isolate it.The time was ripe for gene isolationsince new molecular techniques were emerging every month – both new ways ofmaking collections of genes and new methods for screening them for genes of interest.Just at that time, Dan Kaufman, whohad worked in my laboratory as aUCLA undergraduate and thenmoved to Berkeley to do graduatework, decided to move back to LosAngeles, to rejoin my laboratory, andto work on the brain.Never suspecting that I would everend up contributing to that field, I counseled Danto work on a more tractable problem – the isolationof the GAD gene. At the time, few neuroscientistswere using molecular techniques, and findingthat gene would undoubtedly make a significantcontribution to the field. Even ifHuntington’s disease is not caused by a mutationin GAD, I reasoned, having the gene in handwould accelerate progress in understanding theinhibitory circuits that are important in normalbrain function and that fail in Huntington’s disease.Dan quickly succeeded in finding the GAD gene(which we now call GAD67), using a combinationof immunological, molecular, and biochemical techniques.Pure luck – the ability of bacteria to makea functional GAD – allowed us to short circuitmost of the slogging that we thought would benecessary to prove we had the right gene. But therewere some anomalies that later caused another graduatestudent, Mark Erlander, to challenge our initialview that “there is only one GAD”. Using a newset of techniques, in the early days of PCR (thepolymerase chain reaction), Mark found anotherGAD gene, which we now call GAD65.In 1990, Dan Kaufman, then a postdoctoral felllowat the Salk Institute, again returned to thelab, again interested in autoimmune disease. Onthe basis of a paper by Pietro Di Camilli on StiffMan Syndrome – a rare complication of diabetesand of other conditions, Dan suspected thatGAD65 or GAD67 might be identical to theunnamed 64 kD antigen first identified by ÅkeLernmark in 1982 as the first target of Type 1(T1D) antibodies. About the same time, SteinunnBaekkeskov, Pietro Di Camilli, and their colleaguescame to a similar conclusion though they didn’tknow about GAD65, which we had onlyrecently discovered. In collaboration with NoelMaclaren and Mark Atkinson at the Universityof Florida, we examined GAD autoimmunity inT1D patients, using GAD65 and GAD67 that wecould produce from recombinant DNA.Autoantibodies to GAD indeed provided anpage 14dmccad june 2003

“our real excitement came from a thoroughly unexpected(and still unexplained) finding, that part of GAD65 showed astriking similarity to a piece of Coxsackievirus, a virus”that isepidemiologically associated with Type 1 diabeteseffective diagnostic or predictive test for T1D.While autoantibodies are hallmarks of the autoimmuneprocess and are valuable prediagnosticmarkers, autoreactive T cells, rather than autoantibodies,are likely to be the underlying cause ofT1D. In further collaboration with Maclaren andAtkinson, we showed that newly diagnosed T1Dpatients, in contrast to healthy controls, have significantT cell responses to purified recombinanthuman GAD65.We knew that these discoveries would be usefuland important, but our real excitement came froma thoroughly unexpected finding – that part ofGAD65 showed a striking similarity to a piece ofCoxsackie virus a virus that is epidemiologicallyassociated with T1D. This sequence match sugggesteda possible mechanism for the initiation ofan immune response. We thought that T1D mightinvolve “molecular mimicry,” in which theCoxsackie viral antigen triggered an immuneresponse that cross-reacted with GAD. With thishighly speculative hypothesis of molecular mimicryin mind, we set about to look for evidencethat GAD autoimmunity might actually be pathogenicrather than merely diagnostic in T1D.To test the hypothesis that an autoimmuneresponse toGAD65 initiatesautoimmunity,DanKaufman andour otherUCLA colleaguesturned tothe non-obesediabetic mouse(NOD mice).Allan Tobin, Ph.D., is aProfessor and Directorof the Brain ResearchInstitute at UCLA, LosAngeles. Tobin is alsoScientific Director of theHereditary DiseaseFoundation, which participatedin the identification of the gene thatcauses Huntington´s disease. Tobin has specializedin the use of molecular methods for synthesis,function and breaking down of GABA.GABA provides a significant inhibitory signal inthe brain and the pancreas.Early treatment with GAD65 induced tolerancenot only to GAD65 itself, as expected, but alsoprevented the development of autoimmunity toother ß cell antigens as well.Still more amazingly, mice made tolerant toGAD65 never developed insulitis or diabetes. Thiswas our first strong indication that GAD-specificautoimmunity may be important in the pathogenesisof T1D. This finding, together with subsequentwork by Dan Kaufman and others, was thebasis for the clinical trial now being conductedby diamyd medical.dmccad june 2003page 15

In Nature, Anything that CanHappen Does HappenDaniel Kaufman, Ph.D.,is a Professor, activewithin the departmentof Molecular andMedical Pharmacologyat the UCLA School ofMedicine in Los Angeles.In a paper in November1993, Kaufman demonstrated that the administrationof GAD to mice that would otherwisedevelop insulin-dependent diabetes preventedthe outbreak of this disorder.Kaufman was also a member of the groupassociated with Allan Tobin, which was thefirst to submit a patent application for thefull cDNA code for GAD, the application thatDiamyd Medical licenses. Kaufman’s sphereof interest is focused on GAD and its relationto diabetes.IDan Kaufmandid not want to clone GAD. It was themid-1980’s, and I was first year graduatestudent of Allan Tobin at UCLA.New cloning technologies had just beendeveloped that could allow neurobiologiststo clone rare brain mRNAs. Allanand I thought that some of the most importantgenes in brain development and disease would beneurotransmitter receptors and neurotransmittersynthesizing proteins. I wanted to use antibodiesto screen brain cDNA expression libraries forGABA receptors, or use serum from multiple sclerosispatients to identify the protein targets oftheir autoimmune response. Allan, however, knewthat there was a good antibody available againstGAD (developed by Oertel) and instructed me togo after GAD.I constructed and screened a brain cDNAexpression library and quickly isolated one immunoreactiveclone. In those days, there was greatcontention about the molecular weight of GAD,so I tested the putative GAD clone for its GADenzymatic activity. Luckily, the one immunoreactiveclone I had isolated encoded an enzymaticallyactive protein when expressed in E. coli. Ourreport describing the cloning of GAD (Science,1986) showed that expression libraries could bescreened for functional activity, led to a greater ofunderstanding GAD biochemistry, and providedneuroscientists with a cDNA probe for studyingGAD and GABA in brain development and disease.My discussions with a new graduate student inthe Tobin lab, Mark Erlander, led us to suspect thatanother GAD gene existed, despite our failure todetect another gene by low stringency screening.Using degenerate PCR techniques, Mark was thefirst to isolate the second GAD gene, which encodeda 65 kD protein (GAD65). The clone that I had isolatedwas renamed GAD67, based on its molecularweight.Using recombinantly produced GAD67 protein,I next generated a GAD67-specific antibody. Thisantibody has been widely used by neuroscientiststo study GABAergic neurons. Using this antibodytogether with a GAD65-specific antibody, I showedfor the first time that while GAD67 is distributedthroughout the neuron, GAD65 is localizedin the nerve axon terminals. While most ofGAD67 is saturated with its cofactor, less thanhalf of GAD65 exists as active holoenzyme. Wesuggested that the modulation of GAD65apo/holoenzyme levels in nerve terminals couldcouple GABA production to neuronal activity (J.Neurochem, 1991).My change in course to study Type I diabetes(T1D) resulted from my car breaking down. I wasthen a postdoc at the Salk Institute in San Diego.After working in the lab and finding that my carwouldn’t start, I decided to pass time in the library.I picked up the latest issue of Lancet, and it fellopen to a report entitled “64,000 Mr autoantibodiesas predictors of insulin-dependent diabetes” byMark Atkinson and colleagues. A then littleknown factlet came to mind that besides the brain,GAD was also expressed in the cells that madeinsulin. From my work in the Tobin lab, I knewthat there were two forms of GAD, both of whichcould be candidates for this 64K autoantigen.I learned that the 64K autoantigen was a longsought after ß-cell antigen described by SteinunnBaekkeskov and Åke Lernmark. In addition,Michele Solimena and Pietro De Camilli had justreported an association between GAD autoimmunityand Stiffman syndrome. However, they alsoreported that they did not detect GAD autoimmmunityin T1D patients, which in retrospect, wasmissed only because the GAD antibodies in T1Dpatients could not recognize the denatured GADepitopes in their assay system. Despite this reportedlack of association, I decided that it wasworth a try to test whether T1D sera could recognizerecombinant GAD67 or GAD65. This wouldbe an easy experiment using the recombinantGADs available in the Tobin lab. I was again fortunate,since at this time, the existence of twoGADs was only known, and their cDNAs onlyavailable, in the Tobin lab.I got permission from Allan to do an immunoprecipitationexperiment in his lab over the nextweekend. I got T1D sera from Michael Clare-Salzler, and later, from Mark Atkinson and NoelMaclaren. Using the recombinant GADs, I quicklypage 16 dmccad june 2003

found that sera from T1D patients contained GADautoantibodies that were primarily directed againstGAD65. Using recombinantly produced fragmentsof GAD65, I was able to map some of the autoantibodyepitope recognition patterns. These were thefirst T1D diagnostics based on using recombinantGADs. I had also read about an epidemiologicalassociation between Coxsackie virus and T1D.When we compared Mark Erlander’s GAD65 DNAsequence with other sequences in the DNA sequencedata banks, we found a large sequence similaritywith a Coxsackie virus. This was a very provocativefinding, which could explain how ß-cell autoimmunitywas started, or augmented.We submitted a manuscript with our findingsto Science essentially at the same time asSteinunn Baekkeskov and colleagues submittedtheir findings to Nature. Unfortunately, a reviewerinsisted that we provide further proof ofGAD-Coxsackie virus molecular mimicry beforeit could be published. Consequently, we were notable to publish our work until after that ofBaekkeskov and colleagues. While very disappointingat the time, this may have been in a way fortuitous– I had just become a new AssistantProfessor at UCLA, and it prompted me to goafter the next major question – what was the roleof GAD in T1D?As T1D is thought to be T cell, and not autoantibodymediated, it was crucial to understand Tcell responses to ß-cell antigens. I was again extremelyfortunate to have Jide Tian join my newlab, who was highly knowledgeable and skilled incellular immunology. Using the NOD mousemodel of T1D, Jide found that a Th1-type responsefirst arose to GAD65, and then spread intraandinter-molecularly to other ß-cell antigens. Jidetreated very young NOD mice with GAD65 orother ß-cell antigens in a way that inactivatedreactive T cells. The mice that were tolerized toGAD65 had no autoimmune responses or insulitis,while those tolerized to other antigens did displayß-cell autoreactivity (Nature, 1993). Thus,the early inactivation of GAD65 reactive T cellscould circumvent the development of autoimmunity,qualifying GAD65 as a key antigen in theinduction of murine T1D.We next examined what could be done to inhibitan autoimmune process after it had alreadystarted. Jide showed that injecting GAD65 in anadjuvant that induced Th2 responses to GAD65could greatly inhibit the progression of the autoimmuneprocess in NOD mice that already had“These and subsequent experiments suggested that the greaterprotective effect of GAD65 was due to its greater abilityto induce Th2 responses. It is these experiments that”led to the GAD65vaccine that Diamyd is testing in clinical trialsan established autoimmune process. Moreover,GAD65 vaccination could also protect transplantedislets in diabetic mice, significantly betterthan the other ß-cell autoantigens (Nature-Medicine, 1996). These and subsequent experimentssuggested that the greater protective effectof GAD65 was due to its greater ability to induceTh2 responses. It is these experiments that led tothe GAD65 vaccine that diamyd is testing in clinicaltrails.We also began to examine the possibility of Tcell cross-reactivity between GAD and Coxsackievirus. Jide showed that T cell cross-reactivity didoccur, both at the peptide level and at the wholeprotein level between GAD65 and the Coxsackievirus protein – but only in mice with the diabetessusceptibly MHC II allele, and not in mice withother MHC II alleles (J. Exp. Med, 1994). Thejury is still out as to whether the epidemiologicalassociation of Coxsackie virus with T1D is due tothe ability of Coxsackie virus to infect islet cells,or due to T cell cross-reactivity with GAD. As itis generally found that in nature, anything thatcan happen does happen, it would not be surprisingto find that both mechanisms can occur, andin some cases, initiate or augment ß-cell autoimmmunity.It has been most gratifying to see our work atthe bench contribute to new diagnostics andpotential treatments for diabetes, as well as neurologicaldiseases. Our work has been highly dependenton the support of Allan Tobin, as well ascontributions by Jide Tian, Michael Clare-Salzler,Mark Atkinson and Paul Lehmann – who havealso had the patience of saints in teaching meimmunology.dmccad june 2003page 17

References1. Baekkeskov, S., et al,Autoantibodies in newly diagnosed diabetic childrenimmunoprecipitate specific human pancreatic islet cellproteins.Nature 298, 167-169 (1982).2. Baekkeskov, S., et al,Antibodies to a Mr 64,000 human islet cell antigen precedethe clinical onset of insulin-dependent diabetes. J.Clin. Invest. 79, 926-934 (1987).3. Baekkeskov, S., et al,Revelation of specificity of 64k autoantibodies in IDDMserums by high-resolution 2D-gel electrophoresis.Unambiguous identification of 64k target antigen.Diabetes 38, 1133-1141 (1989).4. Christgau, S., et al,Pancreatic ß-cells express two autoantigenic forms of glutamicacid decarboxylase, a 65kDa hydrophilic form anda 64kDa amphiphilic form which can be both membrane-boundand soluble.J. Biol. Chem. 266, 21257-21264 (1991).5. Baekkeskov, S., et al,Identification of the 64k autoantigen in insulin-dependentdiabetes as the GABA-synthesizing enzyme glutamicacid decarboxylase.Nature 347, 151-156 (1990).6. Kim, J., et al,Higher autoantibody levels and recognition of a linearN-terminal epitope in the autoantigen GAD65 distinguishstiff-man syndrome from insulin dependent diabetes.J. Exp. Med. 180, 595-606 (1994).7. Kanaani, J., et al,A combination of three distinct trafficking signals mediatesaxonal targeting and presynaptic clustering of GAD65.J. Cell Biol. 158, 1229-1238. (2002).8. 8. Kash, S. F., et al,Increased anxiety and altered responses to anxiolytics inmice deficient in the 65 kDa isoform of glutamic aciddecarboxylase.Proc. Natl. Acad. Sci. 96, 1698-1703 (1999).9. Shi, Y., et al,Increased expression of glutamic acid decarboxylase andGABA in pancreatic b-cells impairs glucose-stimulatedinsulin secretion at a step proximal to membrane depolarization.Am. J. Physiol. Endocrinol. Metab. 279, 684-694, (2000).10. Schwartz, H., et al,High resolution epitope mapping and structural modelingof human glutamic acid decarboxylase 65.J. Mol. Biol. 287, 983-999 (1999).11. Jaume, J. C., et al,Suppressive effect of GAD65-specific autoimmune Blymphocytes on processing of T cell determinants locatedwithin the antibody epitope.J. Immunology 169, 665-672, (2002).Contributions tothe GAD65 FieldISteinunn Baekkeskovn the 1960’’s and 1970’s several investigatorsdiscovered evidence to show thatinsulin dependent diabetes mellitus(Type 1 diabetes or T1D) is an autoimmmunedisease characterized by circulatingautoantibodies to the β cell in humanpancreatic islets and infiltration of lymphocytesinto the islets. However, the target antigen of thisautoimmunity was completely unknown, when Ijoined the Hagedorn Laboratory in Copenhagen inlate 1979 to work with Åke Lernmark. Usingexpertise I had gained in studying target antigensin trypanosomes, I developed a method that alloweda very sensitive detection of human β cellproteins that bound to circulating autoantibodiesin the blood of patients. To our surprise, we discoveredthat autoantibodies in the blood from 8 outof 10 diabetic children bound to the same protein,a 64kD antigen from human islets while none of10 healthy children recognized this protein. Thiswas the very first indication that there was a specifictarget antigen in the autoimmune response inT1D (1). We then showed that the autoantibodiesto the 64kD protein can be present many yearsbefore clinical onset of T1D (2), suggesting thatthese autoantibodies can be used as a sensitivemarker to identify individuals at risk of developingT1D. Also, in all autoimmune diseases the targetantigen is of critical importance because of itspotential to be used to prevent the autoimmunedisease. Having established my own laboratoryfirst at the Hagedorn Research Laboratory andlater at the University of California, San Francisco,we developed amethod for partiallypurifyingthe 64kD protein(2-3). Wecharacterizedimportantbiochemicaland biophysicalparameters ofthe antigen (2-4) in a painstakingeffort thatculminated inits identificationas theSteinunn Baekkeskov,Ph.D., is a Professor atthe University ofCalifornia. Baekkeskov´scurrent research projectsfocus on four mainareas: i) structure,function, and cell biologyof GAD65, IA-2 and glima 38; ii) characterizationof disease specific B-cell epitopesin GAD65, and IA2, and the temporalpattern of their recognition during early andlate phases of §-cell destruction; iii) characterizationof autoimmune T cell epitopes inGAD65 and IA2, and how they can be used toinduce apoptosis in autoimmune T cells duringthe preclinical phase to prevent disease; iv)and mechanisms of transplantation tolerance,and development of methods to preventallo- as well as autoimmune destruction ofislet cell transplants.smaller isoform of the GABA-synthesizing enzymeglutamic acid decarboxylase or GAD65 (5). Theidentification of a component of the 64kD autoantigenas GAD65 transformed the field ofImmunology of Diabetes, because it became possibleto use recombinant protein for developmentof autoantibody assays and for studies of autoimmunemechanisms in the NOD mouse. A fewyears later, a second component of the 64kD autoantigen,left behind by our purification method,was identified as the protein IA-2 by Michaelpage 18 dmccad june 2003

GAD65 moleculeChristie, my first postdoctoral fellow, in his ownlab. Together, GAD65 and IA-2, which are expresssedin both ß cells and neurons, are recognized by80-90% of T1D patients. In neurons, GAD65 isalso a target of autoimmunity in a rare neurologicaldisorder Stiff-man syndrome (6).Following the discovery of GAD65 as a major targetautoantigen in T1D (5), my lab has focused onunderstanding the structure, cell biology, and functionof this molecule, and how it is recognized bythe immune system. The subcellular trafficking ofGAD65 is the parameter that most clearly distinguishesit from the highly homologous GAD isoform,GAD67, and is dependent on unique trafficking signalsin the N-terminal region (7 and refs therein).By knocking-out GAD65 in the mouse, we showedthat GABA generated by GAD65 is involved infine tuning of inhibitory neurotransmission inresponse to a variety of environmental stimuli (8and refs therein), while overexpression of the proteinin β cells revealed a role of GABA in negativeregulation of first phase insulin secretion (9). Finemapping of the autoimmune epitopes recognizedby GAD65 autoantibodies in human patients revealedthat they target almost the entire surface of themolecule (10 and refs therein) and led to the firstThe figure is printed with permission from Academic Press (Journal of Molecular Biology, Vol.287,No 5, April 16, 1999, pp. 983-999, Baekkeskov S, Schwartz HL).3D model of the GAD65 dimer (10). This structuralinformation enabled us to show how the epitopespecificity of autoimmune B cells, a critical playerin GAD65 presentation to T cells and developmentof T1D, influences the autoimmune T cellepitope repertoire in the protein (11).dmccad june 2003 page 19

GAD in GraphsIncidence of diabetes (%)10080604020In Type 1 diabetes patients, GAD65-reactive T cells are primarily memoryT cellsAnti–B7-1[ 3 H] Thymidine604020(%) Inhibition0IFN-γ6040200IL-136040200GAD65 (µg/ml)Viglietta et al (2002) GAD65-reactive T cells areactivated in patients with autoimmune Type 1adiabetes. J Clin Invest 109:895-903.• T cell recall immune responseswere studied by proliferation andcytokine assays using GAD65 andinhibition of the costimulatorymolecule B7.1• T cells from Type 1 diabetics werenot inhibited by anti-B7.1, whilethose from controls were. This indicatesthat the Type 1 diabetic T cellsare of a memory T cell phenotype.Treatment of young NOD mice withrrVV-GAD65 prevents diabetesdevelopmentrVV-GAD65rVV-MJ601rVV-NS1Uninjected010 20 30age (in weeks)Sun et al. (2002). Prevention of autoimmunediabetes by immunogene therapy using recombinantvaccinia virus expressing glutamic aciddecarboxylase.Diabetologia 45:668-676• 3 week old female NOD mice wereinjected with vaccina constructscoding for GAD or control antigens• Mice receiving the GAD65construct were protected fromdiabetes40The GAD65 Stanford PerspectiveHugh McDevittInitial work with GAD65 identified it as one of the first to elicit a spontaneousimmune response in 3-4 week old NOD mice. Next, we used GAD65 to suppress ordelay the diabetic process. Treatment with 4 doses of GAD65 intravenously at 12weeks of age resulted in a large decrease in diabetes incidence at 30 weeks. Otherislet cell proteins, or foreign proteins had no effect.Tcell hybridomas were used toidentify immunodominantpeptide epitopes of GAD 65in NOD mice: GAD65 aminoacids 206-220, 221-235, and286-300.Transgenic mice with T Cell Receptors (TCRs)for the 206 and 286 peptides, and a hybrid TCRtransgenic with the 221α chain and the 286βchain, revealed that in all 3 TCR transgenic NODlines neither insulitis nor Type 1 diabetes developed.T cells in all 3 lines are reactive to cognatepeptide and produce IL-2, IFNγ, IL-10, and someTNFα. The T cells are positively selected in thethymus, but only escape negative selection via theexpression of a second α chain. The 206 and 286TCR transgenic lines on the Cα -/- backgroundhad near complete deletion of T cells. MHC tetrameranalysis of 286 TCR affinity of the 286 TCRshowed that it was of extremely low affinity.T cells were partially negatively selected in thethymus, and underwent further negative selectionin the periphery. Only 10-20% of peripheral CD4 +T cells are 286 tetramer positive, while 30-40% ofCD8 + T cells are tetramer positive. Cell transferexperiments showed that these mice have T cellscapable of partially suppressing, and definitelydelaying the onset of Type 1 diabetes in a standardtransfer system.To date, none of these mice have developed diabetesor insulitis. Crosses with a RIP-HuGAD65high expressing transgenic NOD line has not resultedin diabetes or insulitis in either the 206 or 286lines. Other studies are currently underway.However, it appears that in all 3 TCR transgenicmouse lines,GAD65 specificT cells areprotective.A review ofthe literatureshows thatmost GAD65-specific T cellclones eitherdo not causediabetes orHugh McDevitt, MD, isProfessor of theDepartment ofMicrobiology andImmunology atStanford UniversitySchool of Medicine,with a joint appointmentin the Department of Medicine.Following his discovery of linkage of the majorhistocompatibility complex with specificimmune responses, McDevitt has pursued themechanisms of this association over the pastseveral decades, and most recently has beenstudying the immune response of NOD miceto GAD65.insulitis, or in some cases are protective. (Oneclone, reported by Robert Sherwin, did cause Type1 diabetes.) Further, von Boehmer (J. Exp. Med., inpress) has shown that induction of recessive toleranceto GAD65 with a transgene insuring highlevel GAD65 expression in endosomal compartmentsdoes not change Type 1 diabetes incidence.These results suggest that GAD65 is primarilya protective antigen in Type 1 diabetes in NODmice. Whether it is also a protective antigen inpatients is an open question. Nonetheless, experimentsfrom our and other laboratories suggestthat administration of GAD65 in non-inflammatoryvehicles result in a delay in onset, or adecrease in incidence of NOD diabetes. This raisesthe question of whether GAD65 is a protectiveislet cell antigen because of its unusual locationin ß cells (in vesicles similar to neuronalvesicles), with this presentation resulting in apreferential elicitation of a TH2 response to thisislet cell protein. This increase in TH2 T cells inthe islets may then be a factor in delaying onsetof diabetes and/or decreasing its incidence.page 20 dmccad june 2003

Islet Cell Antibody Dependent Progression to Insulin Treatment in LADA Patientsnon-insulin requiring diabetes (%)age4510090807060504030201000 1 2 3 4(years)5 6The Search for Islet AntigensTDavid Leslie and Marco Londeihe search for islet antigens associatedwith the immuneresponse which precedes andpredicts the clinical onset ofType 1 diabetes received a substantialboost with the identificationof GAD as one the major antigens involvedin the aberrant immune response. We used a largecohort of identical twins which had been followedprospectively for several years to show for the firsttime that in these twins GAD antibodies were highlypredictive of progression to clinical diabetes.Further studies found that the isotypes of theseGAD antibodies were largely restricted to a singleisotype IgG1. In contrast, patients with a rare nervedisease Stiff Man Syndrome, who also have GADantibodies, and in whom about half the cases alsohave diabetes, were found to have a broad isotypeprofile involving antibody isotypes other than IgG1.These observations suggested that the antibodyimmune response to GAD in Type 1 diabetes showedmaturation at an early stage and was distinctfrom that found in Stiff Man Syndrome. We showedthat the T cell responses to GAD also differedin Type 1 diabetes as compared with normal subjectsand patients with Stiff man syndrome.Specifically, the dominant T cell response in diabeteswas to peptide epitopes (fragments) at the carboxy-“These studies, takentogether, suggest adistinct mature immune response toGAD involving cellular and humoralimmunity in Type 1 diabetes”terminal end of the GAD molecule. These studies,taken together, suggest a distinct mature immuneresponse to GAD involving cellular and humoralimmunity in Type 1 diabetes and raised the possibilitythat modulation of that response might modifythe destructive immune process associated with progressionto insulin dependence.The potential to modify that GAD immuneresponse and thereby modify the disease coursewas illustrated by others in animal models and isnow being tested in patients with autoimmunediabetes. Should these studies prove successful thenext stage would be to consider intervention in atrisksubjects since GAD antibodies, allied to otherdisease markers, are highly predictive of diseaseprogression.David Leslie, MD,FRCP., is Professor ofDiabetes and Autoimmunityat the RoyalLondon and St.Bartholomew´s Schoolof Medicine,University of London.Leslie has been involved in diabetes researchand clinical studies since 1975. Lesliehas been Director of the British DiabeticTwin Study since 1982, the world’s largesttwin study of its type. By studying twinsProfessor Leslie has been able to show thepossibilities for predicting and preventingdiabetes.Marco Londei, MD, isa Professor at theImperial CollegeSchool of Medicine inLondon, is one ofGreat Britain's mostinternationally renownedscientists in thefield of autoimmunity. Londei´s research hasbeen concentrated on T cells in autoimmunity.dmccad june 2003page 21

References1. Quinn A, et al,MHC class I-restricted determinants on the glutamicacid decarboxylase 65 molecule induce spontaneousCTL activity.J Immunol. 2001;167:1748-57.2. Bowie L, et al,Generation and maintenance of autoantigen-specificCD8+ T cell clones isolated from NOD mice.J. Immunol. Methods. 1999;228:87-95.3. Gauvrit A, et al,Identification of a relevant GAD65 H-2Kd-restrictedepitope recognized by CD8+ T cells in NOD mice.2003. Unpublished data.4. Bach J-M, et al,High affinity presentation of an autoantigenic peptidein Type I diabetes by an HLA class II protein encodedin a haplotype protecting from disease.J. Autoimmun. 19975. Bach JM, et al,Identification of mimicry peptides based on structuralmotifs of epitopes derived from 65-kDa glutamic aciddecarboxylase.Eur. J. Immunol. 1998;28:1902-1910.6. Endl J, et al,Identification of naturally processed T cell epitopesfrom glutamic acid decarboxylase presented in the contextof HLA-Dr. alleles by T lymphocytes of recentonset IDDM patients.J. Clin. Invest. 1997;99:2405-2415.J.-M. Bach, Ph.D., D.V.M., is AssistantProfessor in Endocrinology and Director ofthe ImmunoendocrinologyUnit of Nantes, France. Doctor Bach hasspecialized in the study of the cellularimmune response to GAD in Type 1 diabetes(mouse, dog and human).GAD Peptide Vaccines forT1DM:Not Just a Blueprint?TJ.-M. Bachype 1 diabetes is an autoimmunedisease resulting in thedestruction of pancreatic islet ß-cells, which secrete insulin, byautoreactive T lymphocytes.One of the major challenges ofthe new century will be the treatment of organspecificautoimmune disease. For Type 1 diabetes, itmeans disease prevention before the completedestruction of the ß-cell mass, or replacement ofinsulin-secretive cells after their disappearance.Pancreas and isletcell transplantation are effectiveß-cell replacement therapy for diabetes, but theshortage of human donor pancreata has led tosearch for potential alternative sources of isleT cells(pig islets, in vitro generation of ß-cells from pancreaticduct cells or from stem cells). Beside replacementapproaches, diabetes prevention implicatesthe development of immunotherapies, especiallyspecific immunomodulations of auto-antigensinvolved in Type 1 diabetes pathogenesis.Immunotherapies are also necessary for extensive ß-cell replacement approaches, to block the autoimmmuneprocess and to avoid strong immunosuppressortreatments. To be attractive, immunotherapiesof Type 1 diabetes have to produce a long-lastingprotective response specific for self-antigens.On one hand, specific modulations and tolerancecould be obtained in rodent models withwhole self-proteins or DNA encoding auto-antigens.On a second hand, synthetic peptide-basedvaccines could be a very attractive approach forspecific immunomodulations, permitting to targetclonal elimination or unresponsiveness of aggressiveT cells. Peptides can be produced in largequantities with high purity at low cost, and aresafer than recombinant proteins. To allow thedesign of antigen-specific immune therapies targetingpathogenic autoreactive T cells, we need tocharacterize peptide specificities of autoreactive Tcells, which should also provide fabulous markersof isleT cell damage, useful for therapeutic trialsand diagnosis of individuals with increased riskfor disease, and could help to unravel the pathophysiologyof autoimmune diabetes.Several self-antigens have been identified in Type1 diabetes. Among them, GAD (Glutamic AcidDecarboxylase) is a crucial early target involved indiabetes in human and non-obese diabetic (NOD)mouse. Its role is considered presently as decisive inautoimmunity initiation as well as in progression toovert disease. A majority of studies of diabetes autoimmunityprocess and prevention are focused onthis autoantigen. In rodent models, the role of apeculiar set of T lymphocytes, expressing the CD8marker, in ß-cell aggression is not clearly understood.CD8 + T cells could be implicated not only in diabetesinduction, but also in progression to destructiveinsulitis and overt diabetes. Teams of E. Sercarz (1)and A. Cooke (2) have identified two peptides derivedfrom GAD and recognized by CD8 + T lymphocytesin NOD mouse (peptide 546 and 515,respectively). CD8 + T cells specific for these peptidesare detected early in young NOD mouse. Oursquad have very recently specific another GADderivedpeptide (peptide 90), which may be crucialin diabetes progression in mice [3]. For the firsttime, we showed that CD8 + T cells specific for aGAD-derived peptide are implicated in diabetesaggravation. Specific cytotoxic CD8 + T cells wereactivated and could play an important role in progressionof insulitis to overt disease. We presentlytest immunoprevention of spontaneous diabetes inNOD mice using CD8 + T cell-inducing peptides.And in humans? In humans, while autoantibodyresponses have been extensively characterized,very little is known of the natural history of Tcell autoreactivities, neither for CD8 + T cells (probablythe first T lymphocytes to colonize pancreaticislets), nor for T lymphocytes expressing CD4antigen. We and others have already identifiedGAD-derived peptides presented to CD4 + T lymphocytesof recent-onset diabetic patients[4-6]. Inhumans, we need also to identify relevant CD8 + Tcell-inducing peptides of GAD. In the future, discoveryof GAD-peptides recognized by self-reactiveaggressive CD8 + T lymphocytes and elucidationof key immunological events leading to diabetesinitiation or aggravation in humans may permit todefine efficient peptide-vaccine strategies.page 22 dmccad june 2003

No GAD No Type 1 Diabetes?TJi-Won Yoon, Hee-Sook Jun and Julia McFarlanehe prevention of Type 1 diabeteshas long been a primaryarea of research for our laboratory.Several ß cell autoantigenshave been implicated in thetriggering of ß cell-specificautoimmunity, and GAD65 is a strong candidatein both humans and the diabetes-prone nonobesediabetic (NOD) mouse, which is one of the bestanimal models for human autoimmune diabetes.In the NOD mouse, GAD, as compared withother ß cell autoantigens examined, provokes theearliest T cell proliferative response.We hypothesized that GAD expression in the ßcells may be important for disease initiation. Toaddress this, we took a transgenic approach and selectivelysuppressed GAD expression in the ß cells ofNOD mice. We found that complete suppression ofGAD expression in the ß cells of anti-sense GADtransgenic mice ([SJL x C57BL/6] F2 mice) backcrosssedwith NOD mice resulted in the prevention ofautoimmune Type 1 diabetes. These results supportthe hypothesis that GAD may play an important rolein the development of T cell-mediated autoimmunediabetes. However, we cannot exclude the possibilitythat diabetes-resistance genes from the founder micemay have been transmitted to the anti-sense GADtransgenic mice. To address this issue, we are presentlyexamining the development of insulitis and diabetesin ß cell-specific GAD knock-out NOD mice to confirmthe role of GAD in the initiation of Type 1autoimmune diabetes.Several different approaches for immune therapyusing GAD have been tried for the prevention ofdiabetes in animal models. We used a recombinantvaccinia virus (rVV) expressing GAD as a vaccine,as rVVs can induce humoral and cell-mediatedimmune responses to target proteins and the inducedimmune responses can be long lived. We wereable to show that administration of GAD-expresssingrVV effectively prevented autoimmune diabetesin an age- and dose-dependent manner throughactive suppression of effector T cells in NOD mice.Although the therapeutic effect of GAD-basedimmunotherapy is different depending on theroute of administration, experimental conditions,and quality of antigen, treatment using GAD maybe of importance with regard to future strategiesfor the prevention of Type 1 diabetes in humans.What is theGenetic Basis of T1DM?Michael Clare-SalzlerDr. Ji-Won Yoon holds aCanada Research Chairin Diabetes and is thedirector of the Laboratoryof Viral andImmunopathogenesis ofDiabetes in the Facultyof Medicine, Universityof Calgary. Previously, Yoon spent 10 years as asenior investigator at the National Institutesof Health. Over the past 5 years, Yoon hasbeen involved in studies on the role of glutamicacid decarboxylase (GAD) in the pathogenesisof diabetes and the development of methodsto prevent autoimmune diabetes byimmunogene therapy using recombinant vaccciniavirus vectors expressing GAD.The goal of my research is to establishthe cellular, molecular, andgenetic basis for the immunopathogenesisof Type 1 diabetes(T1D) and to develop methodsto prevent this disease. I haveconcentrated on identifying autoantigens and determiningthe role of antigen presenting cells, e.g. dendriticcells (DC), in T1D. Our publications wereamong the first defining glutamate decarboxylase(GAD) as an autoantigen. Our subsequent studiesdemonstrated GAD administration prevented diseasein diabetes-prone NOD mice. Recently, the NIHbaseddiabetes prevention study, TrialNet, developedplans for a trial to test whether GAD administrationprevents T1D in humans.Michael Clare-Salzler, MD. Professor ofPathology, Immunology and LaboratoryMedicine, University of Florida, Director ofResearch and Academic affairs, research; part ofthe UCLA team that characterized GAD as anautoantigen in NOD mice, demonstrated GADresponses in NOD mice, study of the role antigenpresenting cells in Type 1 diabetes, first todemonstrate a tolerogenic role for dendritic cellsin the autoimmunity of the NOD mouse, first labto demonstrate abnormal regulation Cox-2regulation and prostaglandin metabolism in Type1 diabetes.10090807060Percent Diabetic5040GAD in GraphsInduction of GAD65-specific regulatoryT cells modulates diabetes inNOD mice30alone20 + OVA T cells+ GAD T cells10010 20 30 40 50 60Days Post TransferTisch et al. (1998) Induction of GAD65-specificregulatory T cells inhibits ongoing autoimmunediabetes in nonobese diabetic miceDiabetes 47:894-899Insulitis severity• Irradiated NOD mice developdiabetes when diabetogenicsplenic cells are transferred alone orcotransfered with OVA-specific Tcells.• Cotransfer of GAD-specificregulatory T cells prevents diabetesdevelopmentImmunization with GAD65 does notinduce murine diabetesA2.01.51.00.50.0C2.01.51.00.5BALB/cB2.01.51.00.50.0D2.01.51.00.5C57B10.00.0NMRINODPlesner et al. (1998) Immunization of diabetesproneor non-diabetes-prone mice with GAD65does not induce diabetes or islet cell pathologyJ Autoimmunity 11:335-341• 11 wk old NOD mice were injectedtwice sc with 75mg GAD65 or BSA.• Non-diabetes prone mice (Balb/c,C57/Bl and NMRI) did not developdiabetes.• NOD mice injected with GAD65developed less insulitisdmccad june 2003page 23

References1. Tisch, R., et al,Induction of GAD65-specific regulatory T cells inhibitsongoing autoimmune diabetes in nonobese diabeticmice.Diabetes, 47:894, 1998.2. Tisch, R., et al,Induction of GAD65-specific Th2 cells and preventionof autoimmune diabetes at late stages of disease developmentis epitope dependent.J. Immunol. 163:1178, 1999.3. Tisch, R., et al,Antigen-specific mediated suppression of ß cell autoimmunityby plasmid DNA vaccination.J. Immunol. 166:2122, 2001.4. Seifarth, C., et al,. More stringent conditions of plasmidDNA vaccination are required to protect graftedversus endogenous islets in nonobese diabetic mice.J. Immunol. (in press). 2003.Roland Tisch, Professor in Microbiology andImmunology at the University of NorthCarolina. Currently, the laboratory is studyingType I diabetes, an autoimmune diseasecharacterized by the T cell mediateddestruction of insulin produ-cing beta cells.The non-obese diabetic (NOD) mouse, aspontaneous animal model for Type I diabetesis being utilized to eluci-date the keymolecular and cellular events involved in thediabetogenic response.GAD65 and the Immunoregulationof Autoimmune DiabetesORoland Tischur group has used GAD65as a model ß cell autoantigenfor the purpose ofdeveloping strategies ofantigen-specific basedimmunotherapy, and definingevents involved in the breakdown of selftolerancewithin the T cell compartment in nonobesediabetic (NOD) mice. As in Man, GAD65-specific reactivity can be detected in NOD miceearly in the diabetogenic response, suggesting akey role for this ß cell autoantigen in both murineand human Type 1 diabetes (T1D).The goal of antigen-specific based immunotherapyis to selectively prevent and/or suppress pathologicalautoimmunity without hindering the“normal” function of the immune system. In anumber of models of autoimmunity includingT1D, prevention and/or treatment of disease istypically accomplished through induction of socalledimmunoregulatory T cells, of which thereappear to be a number of “types” exhibitingvarious physical and functional properties. Weand others have found that administration ofGAD65 protein or peptides via different routes toyoung NOD mice can prevent initiation of thediabetogenic response. More importantly, administrationof GAD65 can prevent the onset ofovert diabetes even at late stages of preclinicalT1D in NOD mice (1-4). This latter scenario isreflective of a clinical setting in which prediabeticindividuals exhibiting ongoing ß cell autoimmunitywould be considered for some form of intervention.Furthermore, we have recently foundthat GAD65 vaccination can also be effectivelyapplied to induce islet graft tolerance. Specifically,islets transplanted into diabetic NOD mice areprotected long-term via administration of geneticvaccines encoding a portion of GAD65 and cytokinesknown to promote T cells with immunoregulatoryfunction. One key characteristic of theprotection mediated by GAD65 administration isthat infiltration of T cells and other immune cellsinto the islets is efficiently blocked regardless ofthe stage of disease progression. This propertyappears to be unique to GAD65 since protectionmediated by administration of other ß cell autoantigensis usually marked by continued infiltrationof the islets, especially when the diabetogenicresponse is ongoing at the time of treatment.Currently, there is a need to further define theproperties of GAD65-specific immunoregulatoryT cells. For example, different “types” of T cellsexhibiting distinct as well as overlapping immunoregulatoryfunctions appear to be induced byadministration of GAD65 under the appropriateconditions. Therefore, the general immunotherapeuticefficacy of GAD65 vaccination may reflectthe relative number and “variety” of immunoregulatoryT cells elicited by treatment.Indeed, GAD65-specific T cells may “normally”regulate the progression of T1D. We have establishedNOD mice in which a significant frequencyof developing T cells express a “transgenic” T cellreceptor (TCR) specific for a peptide derivedfrom GAD65. T cells expressing the transgenicTCR are efficiently deleted through mechanismsthat prevent the development of autoreactive Tcells and maintain tolerance to self-proteins suchas GAD65. Interestingly, T1D is exacerbated inthese transgenic NOD mice. This finding suggeststhat deletion of T cells expressing the transgenicTCR results in the loss of GAD65-specific T cellswhich regulate the diabetogenic response. In thecontext of immunotherapy, GAD65 vaccinationmay in part lead to the expansion of this pool ofimmunoregulatory T cells. The task at hand is todetermine whether findings regarding the role ofGAD65-specific immunoregulatory T cells inNOD mice can be extrapolated to diabetic individualsand/or those at high risk and exploited forthe purpose of immunotherapy.page 24 dmccad june 2003

Using DNA Vaccines Expressing IsletAntigens to Prevent Type 1 DiabetesOMatthias von Herrathur laboratory is interested indevising novel strategies toprevent Type 1 diabetes byutilizing beta cell derivedautoantigens for immunization.The underlying rationaleis that self-antigens might frequently induceimmune responses that are regulatory in nature.Indeed, during the past 8 years, we and othergroups have identified autoreactive regulatory lymphocytesthat can be induced by administering betacell antigens, such as GAD. For example, oral administrationof these antigens can induce immunedeviation in the sense that islet-specific responsesare skewed away from TH1 more towards a TH2or TH3-like effector phenotype. In some cases thesecells are dependent on TGF-beta, in our hands theyfunctionally rely on secretion of interleukin-4 (IL-4).In our experimental systems, insulin-B induced IL-4producing CD4 + lymphocytes can act as bystandersuppressors by reducing numbers of autoaggressiveCD8 T cells in the pancreatic draining lymph nodeand the islets. At least in mouse models oral autoantigensrequire quite large dosages to be effective.Therefore, we have developed DNA vaccines thatexpress the beta cell antigens insB and GAD. Bothvaccines can prevent diabetes in the RIP-LCMV orNOD mouse models. However, in order to avoid augmentationof autoaggression (which one might fearas the most undesirable outcome when immunizingwith autoantigens), response modifiers in the form ofcytokines and possibly anti-CD3 should be given.This strategy does not only increase the safety marginbut also enhances efficacy.We would like to propose that immunizationwith autoantigens such as GAD can be an effectivemeans to prevent Type 1 diabetes, if augmentationof autoaggression is prevented by suitable responsemodifiers. It is noteworthy that in certain experimentalsituations, diabetes was prevented withoutresponse modifiers. This allows for the speculationthat some islet antigen specific immune responsesmight be more prone to a regulatory rather thanaggressive phenotype.Peptide derivedfrom self-antigenBystander suppression by Ag-specific TregsMHC class I or IImoleculeT cell receptorAPCSolublemediatorsCell-cellcontactAPCCytokine/chemokineAPCTregEffectorT cell of differentAg specificitythan TregNaive or antigen-experienced autoaggressors inthe affected organ or draining lymphoid nodesare, as a consequence, not activated or amplifiedGeneration of moreTregsAfter interaction withTregs, APC expressesless co-stimulatormolecules and IL-12,and more antiinflammatoryIL-10Matthias von Herrath,MD, Associate Memberwith Tenure, La JollaInstitute for Allergyand Immunology. Thefocus of von Herrath’sresearch is in devisingnovel strategies toprevent Type 1 diabetes by inducing autoreactiveregulatory T cells. Their modes ofaction via modulation of antigen presentingcells and combinatorial application withother immune interventions (aCD3, aCD40L)are of particular interest. Positive and negativeassociations between viral infections andautoimmune disease are being investigated.dmccad june 2003page 25

Vaccination with GAD PlasmidSuppresses Diabetes in the NOD MouseEven After Development of InsulitisNora Sarvetnick is Professor of Immunologyat The Scripps Research Institute (TSRI) in LaJolla. Sarvetnick is focusing on new therapiesfor the prevention of autoimmune diabetes.TNora SarvetnickThe NOD mouse developsspontaneous T cell dependentautoimmune diabetes and isused as an experimental modelto explore many features sharedwith Type 1 diabetes patients.Much research has focused on pancreaticautoantigens and GAD65 is one of the bestknown targets of autoreactive T cells during theearly phase of disease development. GAD expresssionin the pancreatic islets increases with age inthe NOD mouse and when its expression is suppressed,diabetogenic T cells are not generated,resulting in protection from diabetes. The importanceof GAD in Type 1 diabetes has also beenconfirmed by adoptive transfer experiments, thatis, a GAD-specific CD4 T cell clone from a NODmice that normally develops diabetes transferreddiabetes into NOD-scid/scid mice that normallydo not develops diabetes.Autoimmune diabetes has been successfullysuppressed at the prediabetic stage via Th2 immunedeviation. For example, young NOD mice injectedwith GAD protein (intravenously or intraperitoneally)or its peptides (intranasally) respondedwith a reduction of T cell proliferation to GAD65and their insulitis and diabetes decreased. Likewise,induction of a GAD-reactive Th2 response prolongedthe survival of syngeneic islets grafted indiabetic NOD mice and inhibited disease progresssionin animals with early signs of Type 1 diabetes.Additionally, the transgenic expression of IL-4 inthe pancreas of NOD mice completely protectedthem from the disease due to the inhibition ofdiabetogenic Th1 lymphocytes by islet-reactiveTh2 cells.Recently it was shown that vaccination of micewith plasmids encoding foreign (viral) antigensinduced effective T cell responses and circumventedthe requirement for conventional immunizationwith proteins in complete Freunds adjuvant.We therefore hypothezised that genetic vaccinationwith plasmids encoding GAD may interferewith development of autoimmune diabetes.Indeed, GAD-plasmid vaccination effectively preventeddiabetes in NOD female mice when theywere treated at 4 to 5 weeks of age (before insulitis)and even at 10 to 11 weeks of age (after insulitis).This protection, however, did not stem frominduction of the classical Th2 shift as reported inseveral other GAD-vaccination protocols. On thecontrary, the mechanism of GAD-plasmid mediateddiabetes protection was due to insufficientcostimulation of GAD-specific T cells at the siteof antigen presentation. In fact when costimulationwas provided in vivo at the time of GAD-plasmidvaccination, the protective effect was abrogated.With these results our lab was first to report(Clinical Immunology, May, 2001) that GAD65-plasmid vaccination is effective in suppressing diabetesin the NOD mouse even after developmentof insulitis and that that this protection is notdependent on a systemic or pancreatic Th2 immuneresponse.page 26 dmccad june 2003

GAD in GraphsImmune deviation of T cell responsesto GAD65 inhibits disease progressionand protects islet transplantsin NOD mice100GAD Peptides and T1DM-Associated HLA MoleculesMay Hold the Key to DiseasePrediction and PreventionC.B. TSanjeevi1DM is an autoimmune diseasewhich occurs predominantlyin children.Autoantibodies to T1DMautoantigens (GAD65 orGlutamic acid decarboxylaseisoform 65, IA-2 or Tyrosine phosphatase andinsulin) are used to identify ongoing autoimmuneprocess both in newly diagnosed T1DM childrenand prediabetic individuals. The autoreactive Tcells are implicated in the destruction of the insulinproducing beta cells leading to the disease andthe autoantibodies are considered only markersfor the disease.We have studied HLA-DQ that is associatedwith T1DM in my laboratory. These studies havehelped us to identify naturally processed and presentedpeptides from susceptible and protectiveHLA-DQ, sequence these peptides and show thebinding motifs (from peptide sequencing andmolecular modelling).We have screened the diabetes autoantigenGAD65 and identified several 15 amino acid longpeptides that carry the right binding motifs forHLA-DQ associated with T1DM.We believe that T cells reacting to the DQrestricted peptides from T1DM autoantigen(GAD65) are in the peripheral blood of newlyCarani B. Sanjeevi isan Associate Professorin the division ofDiabetes andEndocrinology at theDepartment ofMolecular Medicine atthe KarolinskaInstitute in Stockholm. Sanjeevi is also headat the ‘Molecular Immunogenetics’ researchgroup at Karolinska Hospital. Sanjeevi isassociated with his contribution of MHC inautoimmune Diabetes and is currently workingon the role of peptides from diabetesautoantigens (GAD65) and viruses that carrymotifs for diabetes associated HLA-DQ, indisease prediction.diagnosedT1DM childrenand prediabeticchildren andtheir detectionis predictive ofthe disease.An assay toidentify the Tcells reacting tothe DQ restrictedpeptides hasbeen standardizedand is being tested in newly diagnosed Type 1diabetes children and their first degree relatives. Wehope that this assay will help us to both diagnosethe disease and also to predict the disease in firstdegree relatives.Our subsequent aim would be to use these peptidesin some form of peptide-based vaccine approach.Future potential for GAD65:GAD65 has a very good potential in the diagnostic,predictive as well as preventive approach. The keyis to find the form in which GAD65 could beused to achieve the goals. It could be wholeGAD65 or modified GAD65 or peptides fromGAD65.Diabetes Incidence (%)% Diabetes Recurrence1008060402080604020§-GAL.Insulin B-chainHSP277GAD6500 15 30 45 60 75 90 105Post-transplantation Time (days)Tian et al. (1996) Modulating autoimmuneresponses to GAD inhibits disease progressionand prolongs islet graft survival in diabetespronemiceNature Medicine 2:1348-1353• Pancreatic islets transplantedinto diabetic NOD mice that hadbeen pretreated with GAD65survived longer than in those thathad been pretreated with eitherß-galactosidase, insulin B chainor hsp65• GAD65 treatment increases islettransplant efficacyNasal administration of GAD65 preventsdiabetes in NOD miceGAD65 peptidesControl peptide010 15 20 25 30 35 40 45 50 55 60Time (weeks)Tian et al. (1996) Nasal administration of glutamatedecarboxylase (GAD65) peptides inducesTh2 responses and prevents murine insulindependentdiabetesJ Exp Med 183:1561-1567• Intranasal administration ofGAD65 peptide into 2-3wk old NODmice induced IgG1 anti-GAD antibodies,reduced IFN-γ and increasedIL-5responses to GAD65 (i.e. diversion ofthe GAD response from Th1-Th2)• Nasal administration ofGAD65 prevents diabetesdmccad june 2003page 27

GADGAD in GraphsGAD ELISPOT outperforms proliferationassays for detection of GADreactiveT cellsStimulation IndexIL-5 (pg/ml)121086420300250200150100500Control Type 2 Type 1*(n=23) (n=12) (n=31)Kotani et al (2002) Detection of GAD65 reactiveT cells in Type 1 diabetes by immunoglobulinfreeELISPOT assays. Diabetes Care 25:1390-1397• Cellular immune responses toß-cell autoantigens were studiedby proliferation and ELISPOT inPMBCs of Type 1, Type 2 diabetespatients and healthy controls• GAD65-specific proliferation wasdetected in 32% Type 1 patients,while IFN-γ ELISPOT GAD65-specificactivity was detected in 66% patientsImmunization with a GAD65/IL-4plasmid DNA vaccine prevents diabetesin NOD miceCMedium*untreatedHELGADGAD/IL-4GAD65 HSP60 CPHTisch et al. (2001) Antigen-specific mediatedsuppression of cell autoimmunity by plasmidDNA vaccination.J Immunol 166:2122-2132p=.003• Enhanced secretion of IL-5 anddecreased secretion of IFN-γ by Tcells from NOD mice immunized at12 wks old with pDNA encodingGAD65-IgFc and IL-4 and restimulatedwith GAD65.• Vaccination caused a shift in cytokineresponse to GAD65 stimulation*Human T Cells RecognizingGAD65 in Type 1a DiabetesODavid Hafler, Jack Sadie, David Breakstone and Sally Kentur laboratory has beeninterested in the functionof autoreactive T cells inhumans with autoimmunedisease and in particular,Multiple Sclerosis (MS) formany years. The T cell response is regulated in partthrough two signaling events. The first signal isthrough the T cell receptor by peptide processedfrom an antigen in the context MajorHistocompatibility Complex (MHC) proteins onan antigen-presenting cell (APC) and the second isthrough costimulation proteins, CD28 and CTLA-4by B7-1 and B7-2 proteins on APCs. We and othershad previously found that myelin basic protein(MBP) reactive T cells in patients with MS werecostimulation-independent (second signal) as comparedto T cells from normal individuals. This indicatesthat the autoreactive T cells behaved morelike T cells in a memory response and that the patienthas had T cells reactive to MBP for some time.This also suggests a method for differentiating patientT cell responses from those of controls andways of intervening in the autodestructive immuneresponse by the T cells.These data prompted us to examine this issuein Type 1 diabetes. Insulin-dependent Type 1a diabetesis an autoimmune disease mediated by Tlymphocytes recognizing pancreatic islet cell antigens.Glutamic acid decarboxylase 65 (GAD65)appears to be an important autoantigen in the disease.We found that in patients with new-onsetType 1a diabetes, GAD65-reactive T cells werestrikingly less dependent on CD28 and B7-1 costimulationto enter into cell cycle and proliferatethan were equivalent cells derived from healthycontrols. B7-2 appears to be the primary costimulatorymolecule engaging CD28 in T cell activationof GAD65-reactive T cells, and its engagementwith CTLA-4 appears to deliver a negativesignal. We hypothesize that these autoreactive Tcells have been activated in vivo and have differentiatedinto memory cells, suggesting a pathogenicrole in Type 1 diabetes. These findings stronglyindicate that the activation state of antigen-specificcells plays a role in the autoimmune processand selected costimulatory molecules may representthe target of future therapies.We are currently utilizing GAD65 for other studiesin Type 1a diabetes. These include studies examiningthe quantitation and phenotype of GAD65reactive T cells from controls and Type 1 diabeticswith a GAD peptide-loaded tetramer (HLADR*0401 loaded with GAD p555-567) and monitoringpotentially destructive GAD65 T cell autoreactiveresponses in long-term Type 1a diabetics receivingislet transplants. We are enthusiastic to continueto utilize GAD65 as a means of examining autoreactiveT cellresponses inhuman Type 1adiabetes.David Hafler, MD. is Professor of Neurology(Neuroscience) at Brigham and Women’sHospital and Harvard Medical School inBoston, MA., and the head of the MolecularImmunology Laboratory. Hafler’s main clinicaland research interests are in human autoimmunediseases: Multiple Sclerosis, Type 1diabetes and rheumatoid arthritis. His goalsare to understand the nature of self-recognitionby T cells, to understand how that immuneresponse leads to autoimmune disease andhow one can alter this response to developnovel therapiesSally Kent, Ph.D., is an Instructor in Neurologyand Associate Immunologist at Brigham andWomen’s Hospital and Harvard MedicalSchool in Boston, MA. Kent has specialized inNKT cell function in Type 1a diabetics by examiningperipheral blood and pancreatic draininglymph nodes. In Type 1a diabetes, Kenthas focused on GAD65 T cell reactivity as ameasure of autoreactivity and memory T cellresponses in controls and patients and inpatients undergoing islet cell transplantation.page 28 dmccad june 2003

GAD in GraphsGAD65 Specific RegulatoryT Cells May ProvideProtection from DiabetesNAnthony Quinnumerous studies haveshown that T cells of theCD8 + and CD4 + subsetsare both involved in theimmunopathogenesis ofType 1 diabetes (T1D) inhumans. Likewise, both these cell types are alsonecessary for T1D in NOD mice, the murine modelfor spontaneous autoimmune diabetes. We havebeen studying the cellular immune responses toself-antigens in NOD mice to determine significancein the initiation of islet-specific damage. Thefocus has been on two very relevant issues regardingthe initiation of autoimmune disease: the identificationof host-derived factors that influence susceptibilityto autoimmune disease and the role ofinfectious agents in the induction and persistenceof autoimmunity. Although both issues are broadgeneralized topics – particularly the study of susceptibilityfactors – there are very specific issuesthat can be readily addressed and that provide valuableinformation for our understanding of the pathogenicprocesses in autoimmune disease.First, does the quality of the autoimmuneresponse to self-antigens influence the susceptibilityto autoimmune disease? Recently, we have beenseeking to determine if CD8 + T cell responses toglutamic acid decarboxylase (GAD65) are clinicalllyrelevant in the NOD mouse model of T1D.GAD65 is one of the first beta cell antigens toprime autoimmune responses that are detectablein the spleens of naive prediabetic NOD mice.Our preliminary data demonstrates that GAD65-reactive CD8 + T cells can be found in prediabeticNOD mice, while others have shown that suchcells are present in the peripheral blood of humanpatients recently diagnosed with diabetes.Currently, we are investigating the nature ofGAD65-specific regulatory T cells which may provideprotection from diabetes in individuals whoare considered to be at-risk but remain diabetesfree.Such mechanisms may provide protection tosiblings/relatives of diabetic individuals and maybe represented in male NOD mice and F1 mice,both of which display a reduced susceptibility todiabetes compared to female NOD mice.Secondly, can islet proteins, or fragments fromthem, be used to prevent diabetes in young NODmice or humans? Autoimmune diseases such as diabetesare likely the result of immunological dysregulationsuch that the balance between the antagonisticforces of regulation and pathogenesis tilts in favor ofthe pathogenic mechanisms. One of our objectives isto initiate GAD65-specific mechanisms that are capableof restoring the immunological balance in diabetes-pronemice. Hopefully, the clues gained in themurine models can also provide insights into applicationsto humandisease.Tony Quinn, Ph.D., isAssociate Professor inBiological Sciences atthe University ofToledo, and AdjunctProfessor inMicrobiology andImmunology at theMedical College of Ohio, Toledo, Ohio. Quinnis an immunologist with a long-standinginterest in the regulation of immunity, as itrelates to the control of autoimmune disease,and the enhancement of targeted immuneresponses. Quinn is currently studying thecellular immune responses to self-antigens inautoimmune diabetes and their roles in theinitiation of pancreas-specific damage.Relative amount of GAD65immunoprecipitated (% of PAS)(%) Diabetes10080604020Clinical onset of IDDM is associatedwith a decreased Th2 GAD response2520151050IgMGAD65-GMCSFPre. IDDM -IDDM -IDDM(High)Petersen et al (1999). Progression to Type 1 diabetesis associated with a change in the immunoglobulinisotype profile of autoantibodies toglutamic acid decarboxylase (GAD65). ChildhoodDiabetes in Finland Study Group. Clin Immunol90:2 276-81• Isotype-specific IgE and IgMGAD65 autoantibodies in firstdegree relatives before clinical onset(Pre), IDDM at onset (IDDM), lowrisk IDDM relatives (-IDDM) andhigh risk relatives (-IDDM High).• Individuals with a low risk forType 1 diabetes have a higher Th2response to GAD65 than do individualsat high risk or already withdiseaseAdoptive transfer of GAD-reactiveCD4 + Th1 cells induces diabetes inNOD/SCID mice5A injeced miceOVA3E3 injected miceSALINE injected mice00 2 4 6 8 10 12Time (wk)Zekzer et al. (1998) GAD-reactive CD4 + Th1 cellsinduce diabetes in NOD/SCID miceJ Clin Invest 101:68-73• Adoptive transfer of the GADspecificCD4 + T cell line 5A inducesdiabetes In NOD/SCID mice, buttransfer of an OVA-specific controlT cell line (OVA3E3) does not• GAD-specific T cells cause diabetes**dmccad june 2003page 29

GAD in GraphsA1004 wks of age% Incidence of diabetes% Incidence of diabetes80604020B10001480604018 22 2610 wks of age(n=12)PlasmidsGAD65GAD65-GMCSFcontrol(n=8)(n=13)Vaccination with GAD plasmid DNAprevents diabetes in NOD mice(n=11)PlasmidsGAD65GAD65-GMCSFcontrol20(n=8)(n=7)016 14 20 24 28 32 36Balasa et al, 2001. Vaccination with GAD plasmidDNA protects mice from spontaneousautoimmune diabetes and B7/CD28 costimulationcircumvents that protectionClin Immunol 99:241-252• Mice vaccinated at both 4-5 (A)and 10-11 weeks of age (B) wereprotected from developing diabetesIs GAD All There Is?TBart Roephe discovery of GAD65 as oneof the first islet autoantigens inthe pathogenesis of type 1 diabeteswas a breakthrough thatenabled study of specific autoimmuneresponses. I worked asa postdoc in the laboratory of Steinunn Baekkeskovin San Francisco at the time that the identity ofthe 64,000 kDa target of autoantibodies was unraveledas GAD65. My aim as a T cell immunologistwas to determine T cell autoreactivity to this proteinin humans, and to study whether immunizationof genetically predisposed mice with GAD65led to the induction of insulitis or diabetes. Theresults in these pioneer studies to check for T cellautoimmunity were unexpected since I foundvery few differences between patients and controlsubjects, although in both groups significantresponses were detectable. With few exceptions,this finding turned out to be reproducible bymany colleagues all over the world during the folllowingdecade. My second aim to induce diabetesby immunization with GAD65 was in vain:although very strong immune responses could beinduced in various strains of mice, none of thesedeveloped insulitis or diabetes. The latter observationproved useful for the application of usingGAD65 as an immune modulator rather than asan inducer of pathogenesis. Currently there is consensusthat, immunization with GAD65 delays orprevents diabetes in mice, rather than inducing oraccelerating the disease.For the next step as T cell immunologists, weintroduced GAD65 as a candidate target autoantigenin our immune monitoring of Type 1 diabetespatients receiving islet allografts. Indeed, chronicrecurrent loss of islet allograft function wasaccompanied with increment in T cell autoreactivityto GAD65, even in the absence of increases inantibody titers, while this reactivity was neverobserved in patients successfully transplanted withislets. This clinical trial proved that longitudinalstudies on cellular immune responses to GAD65were informative to determine the clinical fate ofbeta-cells in Type 1 diabetes patients. It proved oftremendous importance to test GAD65 of highpurity to avoid reactivity to contaminants in thepreparation of recombinant GAD65. This was theprime result from the first international workshopon T cell reactivity to islet autoantigens. Eversince, I have used GAD65 produced and purifiedby diamyd as golden standard, since this materialwas of reproducibly high quality and void of toxicor mitogenic contaminants. A second satisfactoryencounter with GAD65 came from studies in arare autoimmune disorder called Stiff-ManSyndrome (SMS) that shares GAD65 with Type 1diabetes as major target autoantigen. One-third ofSMS patients develop Type 1 diabetes. We studieda case of SMS without Type 1 diabetes to understandwhy SMS patients often do not developType 1 diabetes despite phenomenal autoimmunityto this neuroendocrine autoantigen GAD65, inorder to develop immunotherapy for Type 1 diabetesin patients that lost tolerance to this protein.Indeed, the GAD65 reactivity measured in thispage 30 dmccad june 2003

“Indeed, chronic recurrent loss of islet allograftfunction was accompanied with incrementin T cell autoreactivity to GAD65, even”in theabsence of increases in antibody titersnon-diabetic SMS patient was characterized as nonoreven anti-inflammatory by its IL-10 producingnature. Unexpectedly, this SMS patient did not elicitprimary proliferative responses to GAD65.However, when she developed Type 1 diabetesfour years later, strong primary responses weredetectable, while the only cytokine produced bythese T cells was the proinflammatory cytokineinterferon-γ. At that time, all symptoms of SMShad resolved. This observation demonstrated thatthe nature of cellular autoimmunity to GAD65 isan important parameter to distinguish healthy subjectsfrom Type 1 diabetics from SMS patients.Moreover, this was a clinical demonstration thatan anti-inflammatory cytokine profile was favorablefor diabetes-prone subjects against developmentof Type 1 diabetes.Thirteen years after the discovery of GAD65 astarget autoantigen in Type 1 diabetes mellitusthrough its recognition by autoantibodies, we arestill left with many open issues. The disease specificityof immune responses to GAD65 is unclear.Autoantibodies against GAD65 are rarely found innon-diabetic subjects, but the majority of subjectswho do have such antibodies will remain healthy,while others suffer from different autoimmunediseases. With regard to T cell responses toGAD65, the consensus is that there are only minimaldifferences in T cell proliferation to GAD65in Type 1 diabetes patients versus controls. Thedifference most likely lies in the quality of theimmune response, and how GAD65 autoimmunityis regulated. The big question now, however, iswhether islet autoantigens such as GAD65 can beused as an immunotherapeutic to divert autoimmmunityfrom destruction to regulation.It should also be considered to assess whethercombinations of immunotherapies includingGAD65 as an immune modifying agent are effectiveto suppress disease activity. In preliminary invitro experiments we have observed that stimulationof GAD65 specific autoreactive T cell clonesisolated from a prediabetic subject is affected bythe new generation of humanized monoclonalantibodies against CD3, and only upon stimulationwith GAD65.Bart Roep is AssociateProfessor in Medicine,in particular theimmunology andimmunogenetics ofType 1 diabetes, at theLeiden UniversityMedical Center, Leiden,The Netherlands. Roep pioneered studies onthe role of T cells in the pathogenesis ofType 1 diabetes, demonstrating that autoreactiveT cells play a key role in beta celldestruction in humans. Roep is currentlyinvolved in the design and evaluation of newimmunotherapies to prevent beta-cell autoimmunity.dmccad june 2003page 31

Mark Atkinson, Ph.D.,is an AmericanDiabetes AssociationProfessor for DiabetesResearch, and theDirector of the Centerfor Immunology andTransplantation at theUniversity of Florida. Atkinson was amongstthe first groups of researchers to identifythe value of measuring immune responsesagainst GAD, and to describe the whiteblood cell response against this protein inpersons with the disease. Atkinson holdspositions on a number of scientific advisoryboards/research panels including theJuvenile Diabetes Foundation Inter-national(JDFI), the American Diabetes Associationand the National Institutes of Health (NIH).While Atkinson’s current research extends tounderstanding the molecular immunologicaland genetic mechanisms underlying the formationof diabetes, his primary researchgoal lies in the development of an effectivemethod for preventing insulin-dependentdiabetes. Professors Atkinson and NoelMaclaren were first to file a US patent forGAD and diabetes therapy which is exclusivelylicensed by Diamyd Medical.More Effective Methods forpreventative Interventionβ cell MassLower Accuracy inDisease PredictionHigher Accuracy inDisease PredictionLess Effective Methodsfor PreventativeInterventionOnset of OvertType 1 DiabetesAbility to Predict Type 1DiabetesAbility to Predict Type 1DiabetesType 1 Diabetes: a Dilemmafor Clinical TreatmentTMark Atkinsonhroughout much of the lastdecade, guarded hope existedthat an agent capable of preventingor reversing Type 1diabetes would be uncovered.As of today, such an agentdoes not unequivocally exist. As a result, manyhave addressed the question ‘why?’ The answers tothis question are many; some of which are readilyaddressable, others are by their nature more inherentlydifficult. Among the latter obstacles facingthe diabetes prevention field is a situation that hasbeen referred to as the ‘treatment dilemma’. Awide body of evidence, both in animal models ofType 1 diabetes as well as in persons with or – atincreased risk for – the disease, supports thenotion that the most effective interventions willbe those that are begun early in the autoimmunedisease process. In contrast, the process of diseaseprediction [that is, using immunologic (e.g. GADautoantibodies), genetic (e.g. HLA types), andmetabolic (e.g. glucose tolerance tests) markers ofthe disease to identify risk for eventual diseasedevelopment] is most accurate in the period closeto the onset of overt diabetes. As a result, a conflict(both ethical and clinical) exists wherein themost effective forms of therapy may involve theearly treatment of subjects in a period in whichdisease prediction is less accurate; a situation thathas positioned the need for a safe and benignform of therapy against treating persons who maynever develop Type 1 diabetes. The identificationof such an idealized agent has thus far provenextremely difficult to uncover.Yet another challenge relates to properlyaddressing, in combination, the questions of “whodo we treat” and “what agent will we use”? In theory,attempts to prevent Type 1 diabetes will mostlikely address two distinct populations. The firstwould involve therapy of high-risk individuals (e.g.GAD/islet autoantibody positive relatives of aproband with Type 1 diabetes) or those who alreadyhave one form of the disease (e.g. LADA subjects).The second would be that of a generalpopulation approach such as is common practicefor vaccinations against infectious disease. Bothmodels have inherent strengths and weaknesses interms of therapeutic intervention. In the lattermodel, a safe and benign therapy capable of interrruptingadverse immune events/environmentalagents (e.g. a vaccine) or alterations in lifestyle providingavoidance of disease risk factors (e.g. diabetogenicdietary components) would ideally beimplemented while the costs associated with screeninggeneral populations forms a barrier. Indeed,one could speculate that in designing preventativemeasures within the general population, the diseasefrequency and unpredictable time of onsetform major obstacles that screening would be eliminatedand vaccination would become universal.Performing clinical trials in increased-risk populationsor those already diagnosed with the diseasemay prove more cost effective (in terms of a trial)and efficient, yet in terms of humanitarian benefit,it could be argued that the general populationapproach may ultimately be more importantas approximately 85% of newly-diagnosedpatients have no family history of the disease.A final barrier for this discussion is the lack ofobvious candidates for the next round of largeprevention trials. As a result, current interest isdirected at studies involving recent-onset Type 1diabetes and LADA patients for the purpose ofidentifying new and perhaps more promisingagents such as diamyd . If diamyd is shown to beclinically effective, then prospective, randomizedcontrolled studies with appropriate statisticalpower and objective endpoints can be designedand prevention strategies in different populationgroups at different stages of the disease processcan be undertaken. Not only will these studiesascertain potential efficacy and safety, but shouldalso lead to greater insight into disease pathogenesis.page 32 dmccad june 2003

Diamyd S-100ß ConcentrationMeasurement ELISADetermination of S-100β protein concentrations in serum is used in neurologyto assess the extent of brain damage in stroke, in head injuries, during extracorporealcirculation and during circular arrest.It is also used for follow-up and prognosis of malignant melanoma.Diamyd, Inc has developed prototype ELISA kits for detecting the presence ofS-100ß protein in human serum/plasma.Diamyd S-100ß-antibody ELISAS-100β autoantibodies have recently been shown to be a possiblemarker for autoimmune diabetes. An article (Nature 2003*) shows evidencethat islet cell death is related to autoreactive T- and β-cellresponses to neighbouring peri-islet Schwann cells, which express S-100β protein. Diamyd, Inc has developed prototype ELISA kits fordetecting the presence of autoantibodies to S-100ß protein in humanserum/plasma. The ELISA uses visual detection (requires a visibleplate reader that measures absorbance at 492 nm).page 32www.diamyd.comproducts@diamyd.compage 21dmccad june 2003

GAD in GraphsIL-4 (pg/ml)250100755025Immunization with a GAD65/IL-4plasmid DNA vaccine preventsdiabetes in NOD miceB *p=.002untreatedHELGADGAD/IL-40Medium GAD65 HSP60 CPHTisch et al. (2001) Antigen-specific mediatedsuppression of cell autoimmunity by plasmidDNA vaccination.J Immunol 166:2122-2132• Enhanced secretion of IL-4 anddecreased secretion of IFN-γ by Tcells from NOD mice immunized at12 wks old with pDNA encodingGAD65-IgFc and IL-4 and restimulatedwith GAD65.• Vaccination caused a shift in cytokineresponse to GAD65 stimulationOur Story of GAD– Serendipity in SciencePaul Zimmet and Ian MackayOur involvement with the GAD story is one of serendipity. One of us (PZ) was attendinga major international pharmaceutical company advisory board meeting inRome in 1991 and heard the presentation of one of their lead scientists, Dr. BillKnowles. He presented work that he was involved in on islet cell antibodies.This wasjust around the time that GAD had been identified by Baekkeskov as the 64kD antigenthat she had discovered in the 1980’s – a putative key autoantigen in Type 1 diabetes.I told Bill that I was very keen to have access to a method to measure anti-GAD. Over a quiet glass of Chianti, Bill told me that he had purified GAD from pigbrain and was attempting to develop an assay for antibodies but his company was% Islets100β cell-specific expression of GAD isrequired for development of autoimmunediabetes806040200H-AS-GADTg(-)Islets Status4 retracted3 >50%2 25-50%1 >25%0 NormalYoon et al. (1999) Control of autoimmunediabetes in NOD mice by GAD expression orsuppression in β cells Science 284:1183-1190• Islet grafts from transgenic miceexpressing high levels of antisenseGAD65/67 (H-AS-GAD) survivewhen grafted into acutely diabeticNOD mice, while grafts expressingGAD (Tg-) are infiltrated anddestroyed• GAD is the target of autoimmuneattack in diabetesnot all that interested as they were more interested in Type 2 diabetes. I asked himwhether I could have some GAD and he agreed. Knowles became a valued advisorSand collaborator in the next phase of our work.o the GAD came to Melbourneand within a few weeks, the nimblefingers of Dr. Merrill Rowleyin our laboratory resulted in thedevelopment of the first radioimmmunoassay(RIA) for anti-GAD,based on radioiodine labelling of the Knowles’GAD preparation. I was involved with the developmentof a Type 1 Diabetes Register in ourisland state of Tasmania so I was able to quicklyfind samples to test and we demonstrated highlevels in newly diagnosed and long-standing casesof Type 1 diabetes. The results were publishedin Diabetes as the first RIA for anti-GAD.Then again, serendipity came into play. Wewere very interested in the fact that a numberof adults were identified with diabetes that presentedclinically as Type 2 yet their naturalhistory over a period of a year or more was indicativeof insulin dependency. We were awarethat a close friend and colleague, Leif Groop,then in Finland, had undertaken a study someyears previously on a group of these patients andhad demonstrated a positive test for islet cellantibodies. Fortuitously, one of his outstandingyoung researchers, Dr. Tiinamaija Tuomi, hadcome to our laboratory as a visiting researcher.She obtained the serum samples from Groop’sstudy and we were quickly able to measure anti-GAD in these subjects: the frequency of anti-GAD in these adult subjects was very high andappeared to be a better predictor of insulindependency than ICA. Indeed, when vigorouslychallenged by a leading authority on ICA at anAmerican Diabetes Association presentation asto our ability to perform the measurement ofpage 34 dmccad june 2003

ICA properly, Dr. Tuomi was able to point outthat the ICA assays had been performed in hislaboratory!We then had a vigorous debate on what weshould call this slow-onset Type 1 diabetes. LeifGroop favoured the term AIDA (autoimmunediabetes in adults) and PZ suggested SODA(slow onset diabetes of adults), but the wisdomand status of IM prevailed and we adopted theterm LADA (latent immune diabetes in adults),not to be confused with the Mexican telephonecompany or the Czech motor car! Tuomi quicklywrote up the results and submitted the paperto Diabetes – it was accepted within 5 days.With a number of other studies with internationalcollaborators we were able to confirm ourfindings in several different countries and ethnicgroups. The biggest challenge still lay ahead. Wewere battling against the ICA “mafia” who stillbelieved that this test was the gold standard.Would it be feasible to use the anti-GAD test topredict future diabetes?We then came again through serendipity to astudy we called “Back to the Future”.Fortuitously, my colleague in Finland, ProfessorJaakko Tuomilehto, brought to our attentionthat the Finnish Women’s Register had takenblood samples from every woman during pregnancysince 1984 and these samples were storedaway at the National Finnish Public HealthInstitute. By linking these samples to the FinnishType 1 Diabetes Register, and testing them allfor anti-GAD, we were able to show that up to10 years prior to a woman developing Type 1diabetes, antibodies to GAD were present. Thiswas a critical study that confirmed the utility ofthe anti-GAD test as a powerful weapon for theprediction of Type 1 diabetes, and establishedthat the “latent period” for an autoimmune diseasecould be very long indeed.Later, in a landmark collaboration with thelate Professor Robert Turner on the UKPDScohort, we were able to demonstrate that 10% ofthe “pedigree” Type 2 diabetics of this cohortactually had LADA. It is now well demonstratedthat this 10% figure applies in many countriesaround the world in terms of the number ofsubjects “misclassified” as having Type 2 diabetes.Of course, early prediction of Type 1 diabetesin its long preclinical phase is not that much of“It is now welldemonstrated that this 10%figure applies in many countriesaround the world interms of the number of subjects“misclassified” as”having Type 2 diabetesa blessing in the absence of an effective preventiveregimen. At the time of our studies onLADA referred to above, there was a peaking ofinterest in administration of autoantigenic preparationsmucosally (oral tolerance) to abrogateautoimmune disorders diabetes included.Administration of GAD to NOD mice to retardonset of diabetes, in our studies and those ofothers, have given at best “promising” results.While earlier enthusiasm for oral tolerance maybe diminishing, we certainly look forward toseeing results of human trials of the preventativeuse of GAD in LADA.Paul Zimmet isDirector of theInternationalDiabetes Instituteand Professor ofDiabetes, MonashUniversity inMelbourne, Australia.His current research includes Type 1 diabetesetiology and the molecular mechanismsof Type 2 diabetes, insulin resistance andobesity and the effects of life-style changeleading to diabetes, obesity, coronary heartdisease and hypertension in developingcountries in the Asia-Pacific region.Intraperitoneal injection of GAD65prevents diabetes in NOD mice100 NaClMBPGAD-MBP8060402000 10 20 30GADtreatment Age (weeks)Pleau et al. (1995). Prevention of autoimmunediabetes in nonobese diabetic female mice bytreatment with recombinant glutamic aciddecarboxylase (GAD65). Clin Immunol Pathol76:90-95Diabetic animals %Stimulation index log100101GAD in Graphs• 4 wk old NOD mice which hadbeen injected ip 3 times with GAD65were protected from diabetes developmentcompared to mice injectedwith fusion protein (MBP) or saline(NaCl)• ip administration of GAD65prevents diabetesGAD65 is a key target antigen in theinduction of murine IDDMGADhsp65cbphinsulin3 4 6 8 12 15Age of mice (weeks)Kaufman et al. (1993) Spontaneous loss of T celltolerance to glutamic acid in murine insulindependentdiabetes. Nature 366:69-72• Spontaneous T cell proliferativeresponses in NOD micefirst develop to GAD65, then tohsp65, carboxypeptidase H andto insulin in a defined order.• GAD65 is a key diabetogenicautoantigen in murine IDDMdmccad june 2003page 35

References1. Kobayashi T, et al,IsleT cell antibodies in Insulin-dependent and non-insulindependent diabetics in Japan: their prevalence andclinical significance. In clinico-genetic genesis of diabetesmellitus. Mimura G, Baba S, Goto Y, Kobberling J,Eds. Amsterdam,Excerpta Med 150-160 (ICS No. 597), 1982.2. Kobayashi T, et al,Time course of islet cell antibodies and beta cell functionin non-insulin-dependent stage of Type I diabetes.Diabetes 36: 510-7, 1987.3. Kobayashi T, et al,Maleness as risk factor for slowly progressive IDDM.Diabetes Care 12: 7-11, 1989.4. Kobayashi TSubtype of insulin-dependent diabetes mellitus (IDDM)in Japan: slowly progressive IDDM-the clinical characteristicsand pathogenesis of the syndrome.Diabetes Res Clin Pract 24 Suppl: S95-9. Review, 1994.5. Kobayashi T, et al,GAD antibodies seldom disappear in slowly progressiveIDDM.Diabetes Care 19: 1031, 1996.6. Kobayashi T, et al,Immunogenetic and clinical characterization of slowlyprogressive IDDM.Diabetes Care 16: 780-8, 1993.7. Nakanishi K, et al,Predictive value of insulin autoantibodies for furtherprogression of beta cell dysfunction in non-insulindependentdiabetics.Diabetes Res 9: 105-109, 1988.8. Nakanishi K, et al,Exocrine pancreatic ductograms in insulin-dependentdiabetes mellitus.Am J Gastroenterol 89: 762-6,1994.9. Nakanishi K, et al,Relationships among residual beta cells, exocrine pancreas,and islet cell antibodies in insulin-dependent diabetesmellitus.Metabolism. 142: 196-203, 1993.10. Groop LC, et al,Islet cell antibodies identify latent Type I diabetes inpatients aged 35-75 years at diagnosis.Diabetes 35: 237-41, 1986.11. Zimmet PZ, et al,Latent autoimmune diabetes mellitus in adults(LADA): the role of antibodies to glutamic acid decarboxylasein diagnosis and prediction of insulin dependency.Diabet Med 11: 299-303, 1994.12. Thai AC, et al,Anti-GAD antibodies in Chinese patients with youthand adult-onset IDDM and NIDDM. Diabetologia 40:1425-1430, 1997.13. Sutanegara D, et al,The epidemiology and management of diabetes mellitusin Indonesia.Diabetes Res Clin Pract 50 Suppl 2: S9-S16, 2000.14. Brooks-Worrell BM, et al,Cellular immune responses to human islet proteins inantibody-positive Type 2 diabetic patients. Diabetes48: 983-8, 1999.15. Kobayashi T, et al,Small doses of subcutaneous insulin as a strategy forpreventing slowly progressive beta cell failure in isletcell antibody-positive patients with clinical features ofNIDDM.Diabetes 45: 622-6, 1996.16. Nakanishi K, et al,Residual beta cell function and HLA-A24 in IDDM.Markers of glycemic control and subsequent developmentof diabetic retinopathy.Diabetes 44: 1334-9, 1995.17. Kobayashi T, et al,Insulin Intervention to Preserve ß-cells in SlowlyProgressive Insulin-Dependent (Type 1) DiabetesMellitus.Ann NY Acad Sci 958: 117-30, 2002.Tetsuro Kobayashi, MD, Ph.D., KanazawaUniversity School of Medicine, Kanazawa,Professor and Chairman, Third Departmentof Internal Medicine, School of Medicine,University of Yamanashi, Tamaho City,Yamanashi, 409-3898 Japan 1968-1974.Chief, Department of Endocnology andMetabolism, Toranomon Hospital andOkinaka Memorial Institute of MedicalResearch, Tokyo. Main Works: Discovery ofthe presence of SPIDDM, Insulin interventionof SPIDDM.Intervention to Preserve ß-Cells in SPIDDMTetsuro Kobayashi , Shoichiro Tanakaa, Kaoru Aidaa and Taro MaruyamabSince the late 1970’s our laboratoryhas made rigorous efforts to examinethe clinical significance of the presenceof islet cell antibodies (ICA) inpatients with non-insulin-dependentdiabetes mellitus (NIDDM) (1-6). Wehave found that the clinical features of ICA-positiveNIDDM patients are largely different fromICA-negative NIDDM patients because ß-cell dysfunctionis progressive and most of them lapse intoan insulin dependent state indistinguishable fromthat of insulin-dependent diabetes mellitus(IDDM) (2-6). In 1982, we described ICA-positiveNIDDM as slowly progressive insulin-dependentdiabetes mellitus (SPIDDM) based on the characteristicclinical courses (1). The clinical characteristicsof SPIDDM include; 1) Late age onset with an initialclinical phenotype of NIDDM with progressiveß-cell failure and subsequent features of IDDM (1-4); 2) Persistent pancreatic humoral autoimmunemarkers including glutamic acid decarboxylaseautoantibody (GADAb), ICA, ICA512/IA-2 autoantibodies(IA-2Ab) and insulin autoantibodies (IAA)(2, 5, 7); 3) Male predominance (3, 6); 4)Involvement of exocrine as well as endocrine pancreas(8); 5) Less marked insulitis with preserved ß-cell mass (9); 6) Association with specific HLA-DQA1*0302-DQB1*0401 haplotype (6). After developmentof a convenient GADAb assay, an increasingnumber of studies have shown that NIDDM patientswho have GADAb as well as ICA are confirmedto be distinct from GADAb-negative NIDDMpatients and are called latent Type 1 diabetes (10)or latent autoimmune diabetes in adults (LADA)(11).The clinical importance for intervention to maintainß-cell function or to prevent ß-cell failure inSPIDDM is based on the following points: 1)SPIDDM is more prevalent than classical IDDMand the prevalence is as high as 10% amongNIDDM patients in some ethnic groups includingCaucasian (10, 11), Japanese (5), Chinese (12),Indonesian (13), and Thai (4) populations. 2) the Tcell response to islet antigens in SPIDDM is weakerthan that in classical IDDM (14). 3) A pilot studydemonstrated that small doses of insulin prevent ß-cell failure in SPIDDM patients (15). 4) Insulinsecreted from preserved ß-cells in SPIDDM contributesto stable glycemic control and subsequentlyprevents late diabetic complications (16). In 1996,we organized a multicenter randomized clinical trial(The Tokyo Study) to examine the effect of earlytreatment with insulin in SPIDDM. At seven hospitalsin the Tokyo area, about 4000 NIDDM patientswere screened for autoantibodies against GAD.Patients were randomly assigned to one of twogroups. One group received subcutaneous insulininjection (Insulin group) and the other received oralsulfonylurea (SU group). The primary outcomemeasures for the study were serum C-peptideresponse and level of blood glucose during 75gOGTT.During the trial C-peptide responses to OGTT(Sigma C-peptide) decreased progressively in the SUgroup and became significant at 24 and 36 months(17). Seven patients lapsed into an insulin-dependentstage when their sigma C-peptide reached lessthan 4 ng/ml. In contrast, the sigma C-peptidevalue remained unchanged in the patients in theInsulin group and the value was significantly differentfrom that of the SU group at 36 months (17).Our multicenter randomized study demonstratedthat insulin intervention is effective and safe for gradualß-cell failure in SPIDDM, specifically in the patientswith preserved ß-cell function and high titer ofGADAb at the initiation of insulin.In a recent study, we have found GADAbs to aunique epitope in the N-terminal region of theGAD65 molecule. This region includes anchoringdomains of GAD65 molecules, which potentiallycan be accessed by GADAb during the exocytosisof GABA from the ß-cell (unpublished data). Theseresults open the door to the prevention of ß-cell failureby vaccination of GAD in SPIDDM patients,because GADAb may have a causative role in ß-celldysfunction and vaccination with GAD may modifythe pathological process of ß-cell failure in thissyndrome.page 36 dmccad june 2003

GAD Antibodies and LatentAutoimmune Diabetes of theAdult (LADA)TA.G. Unnikrishnan and S. K. Singhhe prevalence of Type 2 diabetesis rapidly increasing in theIndian subcontinent.While themajority of subjects with Type2 diabetes in developed countriesare obese, those fromIndia are mostly non-obese, and many of them arelean (1). The proposed hypotheses to explain thisinclude coexistent malnutrition, autoimmunityand metabolic abnormalities.Recent research from North India shows thatone-fourth of the recently diagnosed Type 2 diabeticswho are lean (i.e with a body mass index ofless than 18.5 kg/m 2 ) have positivity to GADantibodies. These adult subjects with autoimmunitywere significantly younger, had a lower waisthip ratio and beta cell function (HOMA) as comparedto lean Type 2 diabetics without antibodypositivity. Hence they had a clinical profile consistentwith latent autoimmune diabetes of theadult (LADA) . They also had lower insulinresistance (HOMA), showing that reduced betacell function is the predominant metabolic abnormalityin these subjects (2).The above study included only adult-onsetType 2 diabetes, which is seen commonly in India.There is also evidence that 23% of emaciated, veryyoung subjects with a ketosis-resistant form ofdiabetes, a phenotype which is rare yet peculiar toIndia, are positive for GAD antibodies (3). Thisform of diabetes has also been termed malnutritionmodulated diabetes mellitus (MMDM).Exciting work is in progress to unravel the geneticbasis of Indian diabetics with GAD65 antibodypositivity. It has been reported that MHC-relatedgenes can discriminate between acute-onset andslow-onset Type 1 diabetes in India, and can alsodistinguish the MMDM phenotype (4).Clearly, it is important to avoid the misclassificationof these lean diabetic subjects, as some ofthem could have LADA. The detection of GADantibodies could predict the early onset of insulindependency, prompting more aggressive glucoseloweringtherapy. This could prevent unneccessaryexposure to hyperglycemia. Further Indian studiesare needed to assess the prevalence of eventualbeta cell failure in these subjects, and the speed ofprogression.References1. Mohan V, et al,Clinical Profile of lean NIDDM in South India.Diabetes Res Clin Pract 38(2): 101-8, 1997.2. Unnikrishnan AG, et al,Clinical and immunological profile of underweightType 2 diabetes in north India.( Abstract number: A-03-255-EASD). Accepted for presentationat the 18th International Diabetes FederationCongress, 24 - 29 August 2003, Paris, France.3. Singh AK, et al,Role of islet autoimmunity in the aetiology of differentclinical subtypes of diabetes mellitus in young northIndians.Diabet Med Apr;17(4):275-80, 2000.4. Sanjeevi CB, et al,MHC class I chain-related gene a alleles distinguishmalnutrition-modulated diabetes, insulin-dependent diabetes,and non-insulin- dependent diabetes mellitus patientsfrom eastern India.Ann N Y Acad Sci Apr 958:341-4, 2002.A.G. Unnikrishnan 1 , S. K. Singh 21 Asst. Professor, Department of Diabetesand Endocrinology, Amrita Institute ofMedical Sciences, Kochi, India, 2 Reader,Department of Endocrinology, Institute ofMedical Sciences, Banaras Hindu University,Varanasi, India.dmccad june 2003page 37

References1. A.Falorni, et al,Radioimmunoassays for glutamic acid decarboxylase(GAD65) and GAD65 autoantibodies using 35S or 3Hrecombinant human ligands.J.Immunol.Methods 186:89-99,19952. C.L.Vandewalle, et al,Belgian Diabetes Registry: High diagnostic sensitivityof glutamate decarboxylase autoantibodies in IDDMwith clinical onset between age 20 and 40 years.J.Clin.Endocrinol. & Metab. 80:846-851,19953. A.Falorni, et al,Diagnostic sensitivity of immunodominant epitopes ofglutamic acid decarboxylase (GAD65) autoantibodiesin childhood IDDM.Diabetologia 39:1091-98,19964. A.Falorni, et al,Autoantibody recognition of COOH-terminal epitopesof GAD65 marks the risk for insulin requirement inadult-onset diabetes mellitus. J.Clin.Endocrinol.Metab85:309-316,2000.5. L.Avesani, et al,Improved in planta expression of the human isletautoantigen glutamic acid decarboxylase (GAD65).Transgenic Research 12:203-212,20036. Marcovina et al,Evaluation of a novel radioimmunoassay using 125 I-labelled human recombinant GAD65 for the determinationof glutamic acid decarboxylase (GAD65) autoantibodiesInt J Lab Res (2000) 30: 21-26Alberto Falorni, MD, PhD is AssociateProfessor in Internal Medicine at theUniversity of Perugia, Italy. Falorni has specializedin the techniques for the study ofhumoral autoimmunity and genetics of Type1 diabetes mellitus and other endocrine autoimmunediseases. Falorni’s clinical activity isfocused on the diagnosis and treatment ofendocrine diseases. Falorni’s main scientificinterests are pathogenesis and prevention ofType 1 diabetes mellitus and of autoimmuneprimary adrenal insufficiency.LADA Diagnostics and GADTransgenic PlantsMAlberto Falorniy main research interest ispathogenesis and preventionof Type 1 diabetesmellitus and other endocrineautoimmune diseases. Ihave initially focused myattention on the role of glutamic acid decarboxylase(GAD65) as a target molecule of autoantibodies in Type1 diabetes. The development of a semi-automated procedurefor the radioimmunological determination ofGAD65 autoantibodies in human serum (1) has made itpossible to test the diagnostic sensitivity (frequency ofType 1 diabetic patients positive) and specificity (frequencyof non-diabetic subjects negative) of this immunemarker for the disease. It was shown that GAD65 autoantibodiescan be detected in over 80% of recently diagnosedType 1 diabetic patients and occur more frequentlyamong diabetic females than males. Interestingly,GAD65Ab have emerged as the immune marker athighest diagnostic sensitivity for adult-onset Type 1 diabetes(2) which has paved the way to the diagnosticuse of this marker in routine clinical practice to discriminateautoimmune from non-autoimmune cases.In the attempt of both identifying novel markersat highest diagnostic accuracy for Type 1 diabetesand elucidating the molecular mechanisms of autoantibodyformation, I have constructed chimeric molecules,generated by substitution of regions of humanGAD65 with homologous regions of GAD67 (aLeanType II diabetesLow risk forinsulin requirement-Clinical diagnosis of adult-onset diabetes mellitus28ObeseType II diabetesScreening for otherautoimmune diseasesGAD isoenzyme which is not a major diabetes-relatedautoantigen), to define the epitope regions of theautoantigen recognized by human autoantibodies (3).It was shown that human GAD65 autoantibodies areprimarily directed against epitopes located in themiddle and COOH-terminal regions of the enzymeand that levels of GAD65 antibodies specific for theCOOH-terminal end of the autoantigen discriminateType 1 diabetic children from antibody-positive childrenwho do not progress towards clinical diabetes.In addition, I have constructed a mutant form ofhuman GAD65, generated by site-directed mutagenesisof the active site of the enzyme, which has provenenzymatically inactive but immunologically indistinguishablefrom “wild-type” human GAD65.More recently, I have focused my interest on thediagnosis and clinical characteristics of the so-calledlatent autoimmune diabetes in adults (LADA). Ihave demonstrated that GAD65 antibodies can bedetected in approximately 10% of a hospital-basedpopulation of patients diagnosed with Type 2 diabeteson clinical grounds in Italy (4). AmongGAD65Ab-positive individuals, high antibody levelsand presence of antibodies directed to the COOHterminalend of the autoantigen predicted an extremelyhigh risk of progression towards insulin-dependencyand of associated organ-specific autoimmunediseases, such as thyroid autoimmune diseases orautoimmune Addison’s disease. In contrast, the presenceof low levels of GAD65 antibodies directedonly against the middle region of the enzyme discriminateda sub-population of LADA patients withclinical characteristics very similar to that of antibody-negativeType 2 patients.At present, I am testing the hypothesis that theislet autoimmune process may be modulated and theappearance of clinical signs of the disease delayed orprevented by the oral administration of recombinanthuman GAD65. To make this strategy potentiallyapplicable to clinical studies of primary prevention,we have constructed transgenic plants expressinghuman GAD65 (5), that can potentially allow us toadminister the human autoantigen without the needfor costly and time-consuming procedures of proteinpurification. We have demonstrated that, in transgenicplants, GAD65 accumulates in chloroplast tylacoidsand mitochondria and that the targeting ofhuman GAD65 to the plant cell cytosol (by substitutionof the NH 2 -terminal end of the protein with ahomologous region of GAD67) is associated with a 5-fold increase of expression levels. Ongoing studies aretesting the effect of the oral administration of transgenicplant material containing human GAD65 inanimal models of spontaneous autoimmune diabetes.page 38 dmccad june 2003

ReferencesT1DM and LADA Differ inGADA Epitope SpecificityDChristiane Hampe, Rattan Juneja, Åke Lernmark, Jerry Palmeriabetes Mellitus is classifiedinto two major forms, Type1 and Type 2 diabetes. Type1 diabetes is characterizedby an autoimmune-mediateddestruction of betacells,leading to insulin deficiency. The autoimmunereaction involves both T cells and antibodies directedagainst islet cell autoantigens that can be detectedin the majority of Type 1 diabetes patients (1, 2).The main autoantigens identified are insulin (3), theMr 65,000 isoform of glutamic acid decarboxylase(GAD65) (4), and the tyrosine phosphatase-like IA-2antigen (5). These autoantibodies are often detectedlong before the clinical onset of Type 1 diabetes andare useful to predict disease risk (5, 6). GAD65 andIA-2 autoantibodies (Ab) are readily detected bynow standardized (7), precise and reproducible radioimmunoassays(4, 7, 8) suitable for large scale analysisand population screening (6, 9). In contrast, classicalType 2 diabetes patients do not show evidence ofautoimmune beta cell destruction. Patients withType 1 diabetes usually require insulin treatment atthe time of diagnosis whereas Type 2 patients can besuccessfully treated by diet and oral agents for manyyears. These patients do not show evidence of autoimmunebeta cell destruction. A third group of patientsis referred to as latent autoimmune diabetes inadults (LADA) (10), Type 1.5 diabetes (11), or slowlyprogressive insulin dependent diabetes mellitus(SPIDDM) (12). These patients lose beta cell function,fail oral agents early and require insulin treatment(13, 14). Evidence for an underlying autoimmmunepathogenesis is provided by the observationthat many of these patients have islet cell antibodies(ICA), autoantibodies to GAD65 (GAD65Ab) (10,15, 16), or both. The presence of GAD65Ab alone isa sufficient marker for future insulin requirementin younger patients (44 years or younger) while inolder patients positivity for both ICA andGAD65Ab is a stronger predictor of insulin requirement(17). The question has been raised whetherType 1.5 diabetes represents a separate clinical diseaseor is a slowly progressive form of Type 1 diabetes(18, 19). Epitope mapping of GAD65Ab can assist inthe classification of the underlying autoimmunity.Using both GAD65/67 fusion proteins (20, 21) andGAD65-specific recombinant Fab we and otherswere able to identify phenotype-specific GAD65Abepitopes. GAD65Ab in newly diagnosed young Type1 diabetes patients recognize restricted epitopes primarilylocated at the combined middle-carboxyterminalconformational epitope of GAD65, while bindingto GAD67 or the N-terminus of GAD65 isdetected only at a low level. In contrast, GAD65Abpositive Type 1.5 diabetes patients exhibit aGAD65Ab epitope pattern that is characterized bybinding to both the N-terminus of GAD65 and to atentative conformational epitope formed of themiddle and carboxyterminal part of GAD65 (22).This GAD65Ab epitope profile clearly differs fromthat found in Type 1 diabetes patients and moreresembles the broader GAD65Ab epitope specificityfound in GAD65Ab-positive healthy individuals andfirst-degree relatives (20). This difference in the bindingpattern of GAD65Ab of Type 1.5 diabetes patientscompared to that of Type 1 diabetes patientssupports the notion that the disease process maydiffer between these two types of patients. We thereforesuggest that Type 1.5 diabetes might be a subtypeof Type 1 diabetes characterized by separateimmunologic features. GAD65Ab epitope patternsmay be useful to identify Type 1.5.Christiane Hampe, Ph.D., has a position asjunior faculty at the University of Washingtonin Seattle. Her research interests are the disssectionof the role of GAD65 and its autoantibodiesin the pathogenesis of Type 1 diabetes.While these autoantibodies are widely acceptedas markers for the disease, preliminarydata indicate that disease-specific GAD65Abmodulate T cell responses. Hampe’s currentresearch goals are to understand the effect ofGAD65Ab on processing and presentation ofGAD65.1. Bonifacio E, et al,Islet autoantibody markers in IDDM: risk assessmentstrategies yielding high sensitivity.Diabetologia 38:816-822, 19952. Landin-Olsson M, et al,Islet cell and other organ-specific autoantibodies in allchildren developing Type 1 (insulin-independent) diabetesmellitus in Sweden during one year and in matchedcontrols.Diabetologia 32:387-395, 19893. Palmer JP, et al,Insulin antibodies in insulin-dependent diabetics beforeinsulin treatment. Science 222:1337-1339, 19834. Grubin CE, et al,A novel radioligand binding assay to determine diagnosticaccuracy of isoform-specific glutamic acid decarboxylaseantibodies in childhood IDDM.Diabetologia 37:344-350, 19945. Verge CF, et al,Prediction of Type I diabetes in first-degree relativesusing a combination6. Bingley PJ, et al,Prediction of IDDM in the general population:Strategies based on combinations of autoantibodymarkers.Diabetes 46:1701-1710, 19977. Mire-Sluis AR, et al,The development of a World Health Organisationinternational standard for islet cell antibodies: the aimsand design of an international collaborative study.Diabetes Metab Res Rev 15:72-77, 19998. Verge CF, et al,Combined use of autoantibodies (IA-2) autoantibody,GAD autoantibody, insulin autoantibody, cytoplasmicislet cell antibodies) in Type 1 diabetes: CombinatorialIslet Autoantibody Workshop.Diabetes 47:1857-1866, 19989. Rolandsson O, et al,Glutamate decarboxylase (GAD65) and tyrosine phosphatase-likeprotein (IA-2) autoantibodies index in aregional population is related to glucose intoleranceand body mass index.Diabetologia 42:555-559, 199910. Tuomi T, et al,Antibodies to glutamic acid decarboxylase reveallatent autoimmune diabetes mellitus in adults with anon-insulin-dependent onset of disease.Diabetes 42:359-362, 199311. Harris MI, et al,Classification of diabetes mellitus and other catagoriesof glucose intolerance. In: Keen H, DeFronzo R,Alberti K, Zimmet P, ed.The international textbook of diabetes mellitus.London: Wiley, 1992, 3-18.12. Ludvigsson J, et al,HLA-DR3 is associated with amore slowly progressiveform of Type 1 (insulin-dependent) diabetes.Diabetologia 29:207-210, 198613. Temple RC, et al,Insulin deficiency in non-insulin-dependent diabetes.Lancet 1:293-295, 198914. Gjessing HJ, et al,Fasting plasma c-peptide, glucagon stimulated plasmac-peptide, and urinary c-peptide in relation to clinicaltype of diabetes.Diabetologia 32:305-311, 198915. Groop LC, et al,Islet cell antibodies identify latent Type 1 diabetes inpatients aged 35-75 years at diagnosis.Diabetes 35:237-241, 198616. Rowley MJ, et al,Antibodies to glutamic acid decarboxylase discriminatemajor types of diabetes mellitus.Diabetes 41:548-551, 199217. Turner R, et al,UKPDS 25: autoantibodies to isle T cell cytoplasm andglutamic acid decarboxylase for prediction of insulinrequirement in Type 2 diabetes.UK Prospective Diabetes Study Group [published erratumappears in Lancet 1998 Jan 31;351 (9099): 376].Lancet 350:1288-1293, 199718. Juneja R, et al,Autoimmunity 29:65-83, 199919. Tuomi T, et al,Clinical and genetic characteristics of Type 2 diabeteswith and without GAD antibodies.Diabetes 48:150-157, 199920. Hampe CS, et al,Recognition of Glutamic Acid Decarboxylase (GAD)by Autoantibodies from Different GAD Antibody-Positive Phenotypes.J Clin Endocrinol Metab 85:4671-4679, 200021. Falorni A, et al,Diagnostic sensitivity of immunodominant epitopes ofglutamic acid decarboxylase (GAD65) autoantibodiesepitopes in childhood IDDM.Diabetologia 39:1091-1098, 199622. Hampe CS, et al,GAD65 antibody epitope patterns of Type 1.5 diabeticpatients are consistent with slow-onset autoimmunediabetes.Diabetes Care 25:1481-1482., 2002dmccad june 2003page 39

References1. Chattopadhyay, S., et al,An autoantibody inhibitory to glutamic acid decarboxylasein the neurodegeneartive disorder Batten disease.Hum. Mol. Genet. 11, 1421-1431, 2002.2. International Batten Disease Consortium. Isolation ofa novel gene underlying Batten disease.Cell 82, 949-957, 1995.3. Mitchison, H. M., et al,Targeted disruption of the Cln3 gene provides a mousemodel for Batten disease.Neurobiol. Dis. 6, 321-334, 1999.4. Chattopadhyay, S., et al,An autoantibody to GAD65 in sera of patients withjuvenile neuronal ceroid lipofuscinoses.Neurology. 59, 1816-1817, 2002.David Pearce is Assistant Professor ofBiochemistry and Biophysics in the Centerfor Aging and Developmental Biology at theUniversity of Rochester School of Medicineand Dentistry, Rochester, NY. Pearce is researchingthe molecular basis of juvenileneuronal ceroid lipofuscinosis (JNCL), orBatten disease. Pearce’s group discovered thepresence of autoantibodies to GAD65 inindividuals with Batten disease as well as ina mouse model for the human disease.GAD and Batten’s DiseaseTDavid Pearcehe discovery that individualswith juvenile neuronal ceroidlipofuscinosis, or Batten diseasehave circulating autoantibodiesto GAD65 is perhapsthe most recent chapter onGAD65 autoantibodies and disease, and certainlyone that requires further exploration. We firstpublished that individuals with Batten disease, anda mouse model for the disease had circulatingautoantibodies to GAD65 in Human MolecularGenetics in 2002 (1). The circumstances that ledto this report do not necessarily follow conventionalscientific reasoning, as much as to intuition.Batten disease is a pediatric onset devastatingfatal neurodegenerative disease. Individuals withBatten disease have inherited mutations in bothcopies of a still to be functionally characterizedgene product, designated CLN3 (2). We were studyinga mouse model that lacks a CLN3-gene productconstructed by others (3). Gene expressionstudies indicated that in the brain of cln3-knockoutmice, a shift in the expression of enzymesassociated to the synthesis and utilization of theneurotransmitter glutamate was altered. We nextconfirmed that endogenous levels ofglutamate/glutamic acid were elevated in thebrains of cln3-knockout mice. There was no reasonto suspect, and still isn’t, that the CLN3-gene productis directly involved in glutamate metabolism.I convinced Subrata Chattopadhyay the postdoctoralfellow working with me on this to use serumdrawn from the cln3-knockout mice as the primaryantibody in a western blot against a brain proteinextract. The basis for this was that a block inGAD activity, which converts glutamate to GABAcould be mediated by the presence of an autoantibodyto GAD, thus causing the elevation in glutamatethat we observed. It turned out that thecln3-knockout mice sera had circulating autoantibodiesreactive to a number of brain proteins.Most notably, there was a predominant protein ofaround 65kD. We knew that this would beGAD65, we just had to prove it. Thanks to thevast amount of research on GAD65 in diabetesand stiff persons syndrome we were able to benefitfrom reagents available and a plethora ofpublished techniques. Within the space of a fewmonths Subrata confirmed that the cln3-knockoutmice in addition to having elevated glutamate inthe brain, also had a decrease in enzymatic activityof GAD, most likely due to a demonstrated associationof autoantibody in the brain. His in vitrostudies showed that serum drawn from the cln3-knockout mice also contained an element thatcould inhibit the activity of GAD, and that wecould block the reactivity of the autoantibody toGAD65 by pre-incubation with recombinantGAD. At this point I felt we had biochemicallymet the criteria of showing that the autoantibodycould block an enzyme, GAD, resulting in accumulationof it’s substrate, glutamate, and that associationto the brain resulted in a decrease in theactivity of the enzyme. The final piece of this initialstudy was to show a possible link between theautoantibody and the disease. I might add that thisin many ways is an ongoing task.Nevertheless, this is where Jim Powers, a neuropathologistcame to the rescue. With the aid of hispostdoctoral fellow, Masumi Ito, he confirmed adecrease in GAD65 positive neurons in Batten diseasepost-mortem brain. I might add that Jim did ahuge amount of other work that aided our understandingof this phenomenon, and most importantly,he believed what initially was a strange idea,namely that GAD65 autoantibodies may be presentin the disease. Of course we have gone on toshow that individuals with Batten disease haveautoantibodies to GAD65, in fact to date everychild we have tested has come up positive.Furthermore, we have shown that other pediatricneurodegenerative diseases that fall in a similarcategory to Batten disease do not have circulatingautoantibodies to GAD65 (4). Our focus right nowis to deduce whether GAD65 autoantibodies arereally pathological, or simply epi-phenomenal inBatten disease. This of course is a mammoth biologicaltask, however, thanks again to the resourcesalready put forth in researching GAD65 autoantibodiesin other diseases, and some excellent collaborationswith researchers in this field, I believe wewill again benefit in expediting answering thisquestion. Importantly, establishing the potentialcontribution of the autoantibody to Batten disease,will reveal whether or not this is an element of thedisease process that should be targeted.page 40 dmccad june 2003

GAD and Parkinson’s DiseaseOHelen Fitzsimons and Matthew Duringur primary interest inGAD has been in the developmentof gene therapybased treatments for neurologicaldisorders that involveabnormalities in neuronalactivity. As neurotransmission in the brain canbe modulated by manipulation of GABA levels,we are using adeno-associated viral vector (AAV)-mediated gene transfer of GAD into targetedbrain areas to augment inhibition. The expressionof GAD in neurons, which contain high intracelllularlevels of glutamate and co-factors, leads tobiosynthesis and release of GABA, which actingon GABAA receptors facilitates transport of chlorideions into the cell, hyperpolarization and relevantsilencing of cell activity. AAV is our preferrredgene delivery vector as it provides a high levelof stable gene expression with minimal inflammatoryor immune response (Xu et al, 2001;Mastakov et al, 2002) and in addition, direct injectionof AAV allows transduction of specificneuronal pathways without affecting the rest ofthe brain (Xu et al, 2001; During et al, 2003).Our main focus to date has been on the developmentof an AAVGAD gene therapy based treatmentfor Parkinson’s disease (PD). The motorabnormalities of PD are caused by alterations inbasal ganglia network activity. Briefly, the deathof dopaminergic neurons in the substantia nigrapars compacta and the associated depletion ofdopamine in the striatum causes disinhibition ofthe subthalamic nucleus (STN). This in turn causesoveractivation of the output nuclei of thebasal ganglia, the substantia nigra pars reticulata(SNr) and the internal segment of the globus palllidus,leading to impaired motor function.We rationalized that an increase in GABAergictransmission from the STN to the SNr wouldlead to suppression of firing activity of SNrneurons. AAVGAD65 or AAVGAD67 gene transferto the rat STN via stereotactic surgery resultedin robust expression of recombinant GADprotein that was restricted to STN neurons, and afour-fold increase in GABA release from the SNrfollowing stimulation of the STN. Single unitrecording from the SNr following stimulation ofthe STN showed rare (6%) inhibitory responses incontrol rats, which was increased to 78% inGAD65-treated rats (P

Diamyd’s Commercial Development of a GAD VaccineJohn RobertsonDiamyd Medical has been pioneering both the evaluation of clinical safetyand clinical efficacy of GAD – with a view to its potential use as a vaccine toprevent autoimmune diabetes. Both these achievements were made possibleby Diamyd’s early definition of a manufacturing process – capable ofproviding the quantity and quality of GAD required for different stages ofcommercial drug development.Our manufacturing process relies on the expression ofrecombinant human GAD65 in an insect cell line -grown under special conditions – after infection withan insect specific baculovirus containing the GADcDNA (our “baculoGAD” recombinant clone). This is referred to asthe baculovirus/insect cell expression system (or BVES). Both thequantity and quality of GAD manufactured by our BVES processhave proven appropriate up to the current stage of development – andseem likely to meet our future requirements up to market introduction.While currently at the 50 litre scale (with each 50L batch providingsufficient GAD for thousands of vaccinations) we have alreadyfound that our process can readily be scaled-up to 500 litres (providingtens of thousands of vaccinations per batch). So, pending thesuccessful outcome of our Phase II, our manufacturing process seemscapable of producing sufficient GAD for further clinical developmentand market introduction.Apart from its suitability for manufacturing active GAD, theBVES also has inherent safety advantages – as a non-mammalianexpression system modified to avoid contact with any mammaliancomponents. This implies the low risk of contamination of our vaccineby harmful viruses. Moreover, because the vast majority of humanviruses are not able to grow in the insect cells used, the likelihood ofthese inadvertently being propagated during manufacture and contaminatingour vaccine is considerably reduced. Similarly, because the“baculoGAD” recombinant clone is an insect-specific virus (that cannot infect mammalian cells) the risk imposed by possible residual tracesof residual virus in our vaccine is greatly reduced.A final attribute of our GAD therapeutic strategy has recentlySf9InsectcellsDiamyd - cGMP manufacturerhGAD65 Bulk Drug50LfermenterExtract/PurifyrhGA D65Bulk DrugDiamydGADrecombinantbaculovirusFormulationDiamydFormulation/FillAlumDiamyd Vaccinebecome apparent. Despite only one (or a few) small regions (ca. 12amino acids) of GAD being thought as likely to be active, we madethe decision at the outset (now 9 years ago) that we would not riskpage 42 dmccad june 2003

No Impact on Experimental Autoimmune Encephalomyelitis DevelopmentClinical Score3.532.521.510.50-70 7 8 9 10 11 12 13 14 15 16 17 18DayPBSALUM-GAD 2 µgALUM-GAD 20 µgALUM-GAD 200 µgCTRL-ALUMMean cpm immunoprecipitatedwith serum dilution 1:1000Immunogeneticity of GAD is Increased with Use of Adjuvant100009000800070006000500040003000200010000Mean GAD bufferMean ALUM-GADMean IFA-GADAssoc. Prof. Robert Harris CMM, Karolinska Institute, Sweden, 2000• Clinical Course of MBP 63-88 –EAE in Lewis ratsProf. Åke Lenmark, University of Washington, Seattle, USA, 2000• Total rat serum lgG anti-GAD radioimmunoassaychoosing the “wrong” portion - and our vaccine would contain thefull-length GAD protein (consisting of 585 amino acids). So, despitethe obvious difficulties imposed in manufacturing a recombinant proteinof this size (which would be avoided by synthesising a peptideinstead) we decided that our vaccine would contain full-length, naturalGAD protein - leaving antigen processing and presentation of theappropriate region (“determinant”) to that individual patients immunesystem. This decision not to pursue GAD peptide therapy seems tohave paid off in view of recent scientific reports that repeat injectionsof protein fragments (“peptides”) can confuse the immune system andresult in immune over-reaction (anaphylaxis) – that can be life threateningin experimental animals. This response is not shown by therespective (intact) proteins. Our use of the full-length GAD65 proteinin our vaccine avoids this.Our initial manufacturing was conducted to the principles of GLP(“Good Laboratory Practice”) and was used successfully for all preclinicalsafety studies and clinical Phase I (in healthy volunteers).These ca. 25 different pre-clinical safety studies used various quantitiesof either GAD alone, or the GAD vaccine (alum-GAD), in severaldifferent species, and administered by different routes. All thesestudies followed international regulatory requirements to establish thepre-clinical safety after GAD administration, and thereby providedthe basis for Phase II development. In contrast to the pre-clinical andPhase I stages, however, the GAD vaccineused for Phase II is manufactured to the highestquality standard available for manufactureof clinical therapeutics. This is “cGMP”(or “current Good Manufacturing Practice”)– that will also be required for all future clinicaldevelopment.Diamyd’s development of the GAD vaccinehas now culminated with completion of ourPhase II clinical trial in 47 “Type 2” diabetespatients with GAD-antibodies (LADA). Thisstudy is truly pioneering – in that this is thefirst time patients have received the GADvaccine.Prior to un-blinding, I am highly optimisticregarding the outcome of this PhaseII. I think this study will be a resoundingsuccess if the study outcome includes thefollowing:1. there are no safety concerns of alum-GAD vaccination2. there is evidence for responses that areconsistent with a positive therapeutic effect.John Robertson, Ph.D., has beenDirector of Research Development inDiamyd Medical since 1994.Robertson has experience in biotechnologyand toxicology, both fromindustry (Inveresk ResearchInternational, U.K. and ScheringAgrochemicals, U.K.) and academicinstitutes (Karolinska Institute,Stockholm; The Pasteur Institute,Paris; and the NIH in Washington).dmccad june 2003page 43

T cell GAD65For use of GAD in immunological assaysbulk rhGAD65For use of GAD in enzymatic assaysDiamyd, Inc.,Research Triangle Park6213-D Angus DriveRaleigh, NC. 27617USAwww.diamyd.comproducts@diamyd.comDiamyd, Medical AB (publ)Djurgårdsbrunnsvägen 54SE-115 25 StockholmSweden