Membranes and mammalian glycolipid transferring ... - Åbo Akademi
Membranes and mammalian glycolipid transferring ... - Åbo Akademi
Membranes and mammalian glycolipid transferring ... - Åbo Akademi
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Accepted Manuscript<br />
Title: <strong>Membranes</strong> <strong>and</strong> <strong>mammalian</strong> <strong>glycolipid</strong> <strong>transferring</strong><br />
proteins<br />
Author: Jessica Tuuf Peter Mattjus<br />
PII:<br />
S0009-3084(13)00151-5<br />
DOI:<br />
http://dx.doi.org/doi:10.1016/j.chemphyslip.2013.10.013<br />
Reference: CPL 4241<br />
To appear in:<br />
Chemistry <strong>and</strong> Physics of Lipids<br />
Received date: 28-8-2013<br />
Revised date: 29-10-2013<br />
Accepted date: 30-10-2013<br />
Please cite this article as: Tuuf, J., Mattjus, P.,<strong>Membranes</strong> <strong>and</strong> <strong>mammalian</strong><br />
<strong>glycolipid</strong> <strong>transferring</strong> proteins, Chemistry <strong>and</strong> Physics of Lipids (2013),<br />
http://dx.doi.org/10.1016/j.chemphyslip.2013.10.013<br />
This is a PDF file of an unedited manuscript that has been accepted for publication.<br />
As a service to our customers we are providing this early version of the manuscript.<br />
The manuscript will undergo copyediting, typesetting, <strong>and</strong> review of the resulting proof<br />
before it is published in its final form. Please note that during the production process<br />
errors may be discovered which could affect the content, <strong>and</strong> all legal disclaimers that<br />
apply to the journal pertain.
<strong>Membranes</strong> <strong>and</strong> <strong>mammalian</strong> <strong>glycolipid</strong> <strong>transferring</strong> proteins<br />
Jessica Tuuf <strong>and</strong> Peter Mattjus<br />
Biochemistry, Department of Biosciences, <strong>Åbo</strong> <strong>Akademi</strong> University, Turku, Finl<strong>and</strong>.<br />
Keywords: GLTP, FAPP2, glycosphingolipid, membrane binding<br />
Abbreviations:<br />
ARF, ADP-ribosylation factor; DPPC, 1,2-dipalmitoyl-sn-glycero-3-phosphocholine; ER,<br />
endoplasmic reticulum; FAPP1/2, phosphoinositol 4-phosphate adaptor protein-1/2; FFAT,<br />
two phenylalanines (FF) in an acidic tract; GalCer, galactosylceramide; GCS,<br />
glucosylceramide synthase; GlcCer, glucosylceramide; GLTP, <strong>glycolipid</strong> transfer protein;<br />
GSL, glycosphingolipid; LacCer, lactosylceramide; OSBP, oxysterol-binding protein; PC,<br />
Accepted Manuscript<br />
phosphatidylcholine; PE, phosphatidylethanolamine; PH, pleckstrin homology;<br />
phosphatidylinositol; PI3P, phosphatidylinositol-3-phosphate; PI4P, phosphatidylinositol-4-<br />
phosphate; PM, plasma membrane; POPC, 1-palmitoyl-2-oleyl-sn-glycero-3-phosphocholine;<br />
PS, phosphatidylserine; SM, sphingomyelin; START, steroidogenic acute regulatory protein<br />
(StAR)-related lipid transfer; TGN, trans-Golgi-network; VAP, vesicle-associated membrane<br />
protein-associated protein<br />
PI,<br />
1<br />
Page 1 of 34
Abstract<br />
Glycolipids are synthesized in <strong>and</strong> on various organelles throughout the cell. Their<br />
trafficking inside the cell is complex <strong>and</strong> involves both vesicular <strong>and</strong> protein-mediated<br />
machineries. Most important for the bulk lipid transport is the vesicular system, however,<br />
lipids moved by transfer proteins is also becoming more characterized. Here we review<br />
the latest advances in the <strong>glycolipid</strong> transfer protein (GLTP) <strong>and</strong> the phosphoinositol 4-<br />
phosphate adaptor protein-2 (FAPP2) field, from a membrane point of view.<br />
Accepted Manuscript<br />
Page 2 of 34
1. Introduction<br />
The hydrophobic nature of the lipids requires different transport <strong>and</strong> trafficking mechanisms<br />
compared to water soluble biomolecules. Lipids are synthesized on <strong>and</strong> in different<br />
organelles, <strong>and</strong> cells constantly need to adjust <strong>and</strong> respond to changes in the lipid<br />
requirements. To do this efficiently they need different transport machineries, connected to<br />
the synthesis <strong>and</strong> degradation pathways. Several mechanisms in the cells are used for lipid<br />
distribution. Most important for the bulk lipid transport is the vesicular system, however, lipid<br />
movement mediated by transfer proteins is also widely characterized. In addition, lipid<br />
transfer proteins have also often been given the role as sensors, responsible for coordinating<br />
the transport routes within the synthesis <strong>and</strong> degradation machineries. There are several<br />
<strong>glycolipid</strong>-binding proteins in mammals, proteins that specifically recognize <strong>glycolipid</strong>s. The<br />
sphingolipid activator proteins (saposins) bind to different glycosphingolipids (GSL) <strong>and</strong> help<br />
in the degradation process of GSLs, with short oligosaccharide chains, in the lysosomes<br />
(Schulze et al., 2009). Mutations in the saposins can cause severe lysosomal storage disorders,<br />
like Gaucher disease (Schnabel et al., 1991). There is also evidence that non-specific lipid<br />
transfer proteins, isolated from beef liver, can recognize GSLs <strong>and</strong> transfer them between<br />
membranes (Bloj <strong>and</strong> Zilversmit, 1981). Another emerging group are the <strong>glycolipid</strong> transfer<br />
proteins. In this review we will focus mostly on <strong>glycolipid</strong> transfer protein (GLTP), but also<br />
on phosphoinositol 4-phosphate adaptor protein-2 (FAPP2), two proteins that belong to the<br />
GLTP superfamily.<br />
2. Glycolipid transfer protein (GLTP)<br />
GLTP is a small soluble, 24 kDa, protein that has been identified in many organisms. GLTP<br />
Accepted Manuscript<br />
was first discovered in the cytosolic fraction from bovine spleen (Metz <strong>and</strong> Radin, 1980;Metz<br />
<strong>and</strong> Radin, 1982) <strong>and</strong> has since then been found in e.g. liver <strong>and</strong> brain from various<br />
<strong>mammalian</strong> sources (Abe et al., 1982;Yamada <strong>and</strong> Sasaki, 1982a;Yamada <strong>and</strong> Sasaki,<br />
1982b;Wong et al., 1984). In these studies GLTP was always purified from the cytosolic<br />
fraction <strong>and</strong>, consequently, the protein was postulated to be cytosolic. It has later been<br />
confirmed that GLTP indeed remains in the cytoplasm <strong>and</strong> not within other major cellular<br />
organelles (Tuuf <strong>and</strong> Mattjus, 2007). Homologs to <strong>mammalian</strong> GLTPs have furthermore been<br />
discovered in both plants (West et al., 2008) <strong>and</strong> yeast (Saupe et al., 1994). The expression<br />
2<br />
Page 3 of 34
levels of GLTP in bovine tissue have been analyzed (Lin et al., 2000). The results show that<br />
the cerebrum <strong>and</strong> kidney display the highest GLTP mRNA levels, which coincides with<br />
normally quite high amount of glycosphingolipids. GLTP can accelerate the transfer of<br />
<strong>glycolipid</strong>s, both diacylglycerol- <strong>and</strong> sphingoid-based, between two lipid membranes.<br />
Substrates that can be used by GLTPs include glucosylceramide (GlcCer), lactosylceramide<br />
(LacCer), galactosylceramide (GalCer), sulfatide, GM 1 , <strong>and</strong> GM 3 (Yamada et al., 1985;Brown<br />
et al., 1985). Furthermore, the initial sugar residue, linked to the ceramide or diacylglycerol<br />
backbone, needs to be in β-configuration for GLTP to recognize it as a substrate (Yamada et<br />
al., 1986). On the other h<strong>and</strong>, GLTP cannot transfer phosphatidylcholine (PC),<br />
phosphatidylethanolamine (PE), sphingomyelin (SM), phosphatidylinositol (PI), cholesterol<br />
<strong>and</strong> cholesterol oleate (Yamada et al., 1985). The activity of GLTP is sensitive to changes in<br />
the surrounding pH (West et al., 2006). Results show that GLTP has the highest transfer<br />
activity at pH 7.0, 66% transfer activity at pH 10 <strong>and</strong> only 11% activity at pH 4.0. This could<br />
give an important clue on GLTP subcellular localization <strong>and</strong> cellular activity, since GLTP<br />
will encounter different pH conditions within the cell.<br />
The amino acid sequence of GLTPs from different mammals is highly conserved. The protein<br />
sequence of GLTP from pig brain was the first to be determined by Edman degradation (Abe,<br />
1990). Later, it was demonstrated that GLTP contains 209 amino acids <strong>and</strong> that the sequence<br />
between porcine <strong>and</strong> bovine is 100% identical (Lin et al., 2000). Human GLTP has almost as<br />
high identity <strong>and</strong> shows 98% identity compared to porcine <strong>and</strong> bovine amino acid sequences<br />
(Li et al., 2004). Furthermore, the human GLTP amino acid sequence is 100% identical to the<br />
amino acid sequences of chimpanzee <strong>and</strong> macaque, 99% to the nonprimate mammal Canis<br />
familaris <strong>and</strong> 98% identical to the Bos taurus sequence (Zou et al., 2008). Two single-copy<br />
GLTP genes have been found in human cells, on chromosome 11 <strong>and</strong> 12. The<br />
transcriptionally active GLTP is on chromosome 12 <strong>and</strong> contains five exons <strong>and</strong> four introns,<br />
Accepted Manuscript<br />
while the inactive pseudogene is located on chromosome 11 (Zou et al., 2008). The<br />
expression of GLTP mRNA is regulated by two transcription factors, Sp1 <strong>and</strong> Sp3. These<br />
factors bind to multiple sites in the GC-boxes localized to the promoter of GLTP (Zou et al.,<br />
2011). In the same study, the ability of different sphingolipids to enhance the GLTP<br />
transcription of GLTP was analyzed. Interestingly, only ceramide (but not for example<br />
GlcCer, GM 1 or sulfatide) could enhance the transcription levels of GLTP via the Sp1 <strong>and</strong><br />
Sp3 transcription factors. The structure of both bovine <strong>and</strong> human GLTP has been<br />
determined, both in its apo-form <strong>and</strong> in complex with different <strong>glycolipid</strong>s (Malinina et al.,<br />
3<br />
Page 4 of 34
2004;West et al., 2004;Airenne et al., 2006;Malinina et al., 2006). The GLTP structure<br />
contains eight α-helices, that are arranged in a two-layer fold, a structure that is completely<br />
novel for lipid <strong>transferring</strong> proteins. The carbohydrate group will bind to a recognition center<br />
on the surface of GLTP, through hydrophobic contacts <strong>and</strong> hydrogen bonding, <strong>and</strong> the<br />
hydrocarbon acyl chains will be accommodated into a hydrophobic pocket within the protein<br />
(Malinina et al., 2004). The binding of <strong>glycolipid</strong>s to GLTP will be discussed extensively in<br />
chapter 4.<br />
The formation of GLTP dimers has been discussed in early publications. There are three<br />
cysteine residues in the GLTP sequence, two of them reside inside the protein <strong>and</strong> the third is<br />
on the surface of GLTP. Abe <strong>and</strong> coworker suggested that the two internal cysteines might<br />
form a intramolecular disulfide bond, while the cysteine on the outside could form a disulfide<br />
bond with another GLTP molecule (Abe <strong>and</strong> Sasaki, 1989a;Abe <strong>and</strong> Sasaki, 1989b). The<br />
disulfide bridge forming ability of GLTP was suggested to function as a means for regulating<br />
GLTP transfer activity, since it was demonstrated that the dimer had almost no transfer<br />
activity in contrast to GLTP with an intermolecular disulfide bridge, which had a high transfer<br />
activity. However, results from crystallization studies show no intra- or intermolecular<br />
disulfide bonds in the GLTP structure (Malinina et al., 2004;Airenne et al., 2006), but based<br />
on structural requirements it was suggested that it might be possible that an internal disulfide<br />
bridge would help in regulating GLTP activity (Airenne et al., 2006). It is still unknown<br />
whether disulfide bond-formation actually occurs in vivo as a regulatory mechanism for<br />
GLTP function. Recently a new transfer protein was discovered that moves ceramide-1-<br />
phosphate, <strong>and</strong> was named CPTP, ceramide-1-phosphate transfer protein (Simanshu et al.,<br />
2013). Structurally, CPTP has a very similar fold compared to the GLTP fold, however, CPTP<br />
associates with the trans-Golgi network (TGN), nucleus <strong>and</strong> the plasma membrane (PM). A<br />
CPTP-mediated ceramide-1-phosphate decrease in plasma membranes <strong>and</strong> increase in the<br />
Accepted Manuscript<br />
Golgi complex stimulates cPLA 2 α release of arachidonic acid, triggering pro-inflammatory<br />
eicosanoid generation.<br />
3. Phosphoinositol 4–phosphate adaptor protein-2 (FAPP2)<br />
There are two closely related homologs of four-phosphate adaptor proteins, type 1 <strong>and</strong> 2<br />
(FAPP1 <strong>and</strong> FAPP2). FAPP1 was originally discovered in a study, when trying to identify<br />
4<br />
Page 5 of 34
novel proteins that could interact with various phosphatidylinositol-phosphates (Dowler et al.,<br />
2000). The FAPP1 gene is located on chromosome 2 <strong>and</strong> encodes a protein with 300 amino<br />
acids. FAPP1 has in its amino-terminal a pleckstrin homology (PH) domain <strong>and</strong> in the C-<br />
terminus resides a proline rich-motif (Godi et al., 2004). FAPP2, on the other h<strong>and</strong>, is a<br />
protein with 519 amino acids <strong>and</strong> is encoded by a single-copy gene on chromosome 7. In<br />
addition to a PH domain in the N-terminus, FAPP2 has a GLTP domain in the carboxyterminal,<br />
which makes it a member of the GLTP superfamily (Godi et al., 2004;Kamlekar et<br />
al., 2013) (Fig. 1A). The PH domain of FAPP1 <strong>and</strong> FAPP2 is 80% identical (Godi et al.,<br />
2004). FAPP2 is able to transfer GlcCer, both fluorescently labeled <strong>and</strong> natural, but not SM,<br />
PC or ceramide (D'Angelo et al., 2007). Recently it was demonstrated that in addition to<br />
GlcCer, the GLTP domain of FAPP2 can specifically transfer GalCer <strong>and</strong> LacCer between<br />
membranes, but not the ganglioside GM 1 or the negatively charged sulfatide (Kamlekar et al.,<br />
2013). This substrate specificity makes the GLTP domain of FAPP2 more similar to HET-C2,<br />
a fungal homolog of GLTP, than to human GLTP itself.<br />
The FAPPs appear to be ubiquitously expressed <strong>and</strong> have a cytoplasmic distribution, though,<br />
with a strong preference towards the TGN via their targeting signals (Godi et al., 2004). For<br />
example, in mouse tissue FAPP2 is highly expressed in kidney, but is also found in testis, the<br />
small intestine, spleen <strong>and</strong> cerebrum (D'Angelo et al., 2013). Homologs of FAPP2 have been<br />
found in vertebrates <strong>and</strong> echinoderms (D'Angelo et al., 2008) <strong>and</strong> sequence analysis of<br />
various FAPP2 homologs reveals that both the PH domain as well as the GLTP domain are<br />
highly conserved (D'Angelo et al., 2012).<br />
By analyzing FAPP2 by small-angle x-ray scattering <strong>and</strong> analytical ultracentrifugation, it<br />
appears that FAPP2 is dimeric in solution (Cao et al., 2009). Whether this is the case in<br />
biological systems remains to be demonstrated. To date, it has been proven to be difficult to<br />
Accepted Manuscript<br />
crystallize FAPP2, probably because of its large size <strong>and</strong> flexibility. However, a lowresolution<br />
structure of dimeric FAPP2 by small-angle x-ray scattering has been obtained,<br />
showing a molecular shape that is well extended <strong>and</strong> curved, with a length of 30 nm (Cao et<br />
al., 2009). The GLTP domain has been analyzed using homology modeling. These models<br />
show a conserved folding structure against human GLTP, with the key residues, important for<br />
the <strong>glycolipid</strong> <strong>transferring</strong> properties retained (D'Angelo et al., 2007). A truncated FAPP2<br />
form, with a 212 amino acids C-terminus (the GLTP domain), was used to study the<br />
resemblance to GLTP (Kamlekar et al., 2013). This protein showed a structure with high<br />
5<br />
Page 6 of 34
content of alpha-helices <strong>and</strong> results from experiments using fluorescence emission of the<br />
intrinsic tryptophan residues suggested that the GLTP domain actually adopts a GLTP fold.<br />
However, there are some differences in FAPP2 compared to GLTP, e.g. FAPP2 has a more<br />
selective substrate specificity which is due to a restricted <strong>glycolipid</strong> head group recognition<br />
center (Kamlekar et al., 2013). The FAPP1 PH domain has been analyzed using nuclear<br />
magnetic resonance-based solution structures in a free state, together with micelles or<br />
phosphatidylinositol-4-phosphate (PI4P) (Lenoir et al., 2010). Shortly after, the crystal<br />
structure of FAPP1 PH domain was determined (He et al., 2011). The PH domain was<br />
reported to consist of a β-barrel capped by an α-helix at one edge. Three additional loops<br />
connect the str<strong>and</strong>s; a long β1- β2 loop, a short β6- β7 loop <strong>and</strong> a β3-β4 loop that is partially<br />
hidden in the structure. Between the β4 <strong>and</strong> β5 str<strong>and</strong> one can also find a one-turn α-helix.<br />
Hydrogen-deuterium exchange mass spectrometry has been used on FAPP2 to look at<br />
conformational changes upon GlcCer binding (D'Angelo et al., 2013). The GLTP <strong>and</strong> the PH<br />
domain appear to be separated by a very flexible linker region <strong>and</strong> results show that when<br />
FAPP2 binds GlcCer a conformational change will occur in the whole structure, not only in<br />
the <strong>glycolipid</strong>-binding domain, but also the linker region <strong>and</strong> the PH domain will be affected.<br />
4. Glycolipid binding<br />
The broad specificity of GLTP for various glycosphingolipids, for example <strong>glycolipid</strong>s with<br />
large variations in their head group structure, is due to the hydrogen bonding adaptable head<br />
group recognition site localized at the entrance to the hydrophobic tunnel of the protein. This<br />
enables GLTP to selectively bind the GSL head <strong>and</strong> the proximal ceramide amide group with<br />
multiple hydrogen bonds.<br />
Accepted Manuscript<br />
GLTP from different species all have the characteristic feature of an adaptive hydrophobic<br />
cavity. The physical function of GLTP is to protect the lipid substrate hydrocarbon chains<br />
from the aqueous environment once the <strong>glycolipid</strong> is extracted from the membrane. Due to the<br />
large variation in the chain lengths <strong>and</strong> saturation of natural GSLs, an adaptable cavity seems<br />
to be the evolutionary structural optimized form. This is in contrast to the precursor synthases<br />
responsible for ceramide syntheses, that seem to have developed into six different ceramide<br />
synthases for various chain lengths of ceramides (Tidhar <strong>and</strong> Futerman, 2013). Part of the<br />
hydrophobic tunnel of GLTP is partly collapsed in the apo-state. The 3D structure of GLTP,<br />
6<br />
Page 7 of 34
as determined by crystallization (West et al., 2004), reveals a s<strong>and</strong>wiched all-alpha-helical<br />
conformation (Malinina et al., 2004;Airenne et al., 2006) (Fig. 2A). Several amino acid<br />
residues have been identified based on the crystal structure studies, that are required for<br />
<strong>glycolipid</strong> binding (Malinina et al., 2004;Malakhova et al., 2005;Airenne et al., 2006;Malinina<br />
et al., 2006). Tryptophan at position 96 (helix 4) is in a key role in the forming of a platform<br />
of hydrogen bonds between the hydroxyl group of the first glucose or galactose. For FAPP2<br />
the corresponding tryptophan required for <strong>glycolipid</strong> transfer activity is at position 407.<br />
Further hydrogen bonds are formed between His-140 <strong>and</strong> the ceramide amide region of the<br />
<strong>glycolipid</strong> backbone (Fig. 2B). By comparison of the apo-GLTP <strong>and</strong> GLTP with a lipid<br />
bound, it is evident that there is a shifting of the alpha-2 <strong>and</strong> alpha-6 helices <strong>and</strong> by changing<br />
the side chain conformations of residues Phe-42, Ile-45, Phe-148 <strong>and</strong> Leu-152 when a lipid is<br />
bound. Presumable, when the <strong>glycolipid</strong> acyl chains come in contact with these hydrophobic<br />
amino acid side chains inside the hydrophobic tunnel, they change their conformation, <strong>and</strong><br />
consequently, a shift in helix-2 <strong>and</strong> helix-6 occurs. We believe that this acts as the process by<br />
which GLTP desorbs from the membrane into the surrounding aqueous solution. Neither<br />
GLTP nor FAPP2 contain the “KRTIQK“ amino acid motif described by Fantini <strong>and</strong><br />
coworkers as a <strong>glycolipid</strong>-binding algorithm (Fantini et al., 2006). They identified this domain<br />
in several amino acid sequences from Helicobacter pylori bacterial adhesion protein adhesin<br />
A (HpaA). HpsA is known to interact with membranes containing <strong>glycolipid</strong>s. HpaA, to our<br />
knowledge, appears not to be able to transfer <strong>glycolipid</strong>s.<br />
4.1 Sphingosine in versus sphingosine out<br />
The cavity that accommodates the hydrophobic parts of the GSL is adaptable <strong>and</strong> aligned with<br />
nonpolar side chains effectively excluding water (Malinina et al., 2004;Airenne et al., 2006).<br />
Thorough examination of different crystal structures of GLTP with different bound GSLs<br />
Accepted Manuscript<br />
with varying acyl chain composition reveals that either one or both acyl chains are<br />
accommodated inside the hydrophobic tunnel. In the sphingosine-in form, both chains are<br />
inside, whereas in the sphingosine-out conformation, the amide-linked chain occupies more of<br />
the hydrophobic cavity preventing the sphingosine chain to enter (Fig. 3A). Most of the<br />
crystallized forms of GLTP are in the sphingosine-out conformation. However, it is still not<br />
clear if GLTP has its substrate bound in the sphingosine-out mode in biological systems. It is<br />
neither known how different diacylglycerol-based <strong>glycolipid</strong>s are bound by GLTP, if they<br />
also adopt both a sphingosine-in <strong>and</strong> sphingosine-out binding mode. Encapsulation of both<br />
7<br />
Page 8 of 34
chains has only been observed for wildtype GLTP monomers <strong>and</strong> with either GlcCer or<br />
LacCer containing an amide-linked oleoyl (18:1 Δ 9 ) acyl chain (Malinina et al.,<br />
2004;Samygina et al., 2011). A proposed mechanism for the accommodation of the<br />
sphingosine chain inside GLTP involves His-140 hydrogen bonding to the ceramide amide<br />
group <strong>and</strong> movement of two phenylalanines (Phe-42 & Phe-148), creating additional space for<br />
chain encapsulation. A mutation of the residue His-140 leads to a complete inactivation of<br />
GLTP (Malinina et al., 2004). Another mutant GLTP (D48V) crystallizes with the 24:1 Δ 9<br />
sulfatide in a sphingosine-in form. The mutant has a structural resemblance very close to that<br />
of the wildtype GLTP with sphingosine-in with bound 18:1 GlcCer. The D48V mutation as<br />
well as another double mutant A47D/D48D are interesting proteins in that they do not bind<br />
neutral GSL, but can be engineered to become transfer proteins for 24:1 Δ 9 sulfatide<br />
(Samygina et al., 2011).<br />
What determines the sphingosine-in conformation, <strong>and</strong> is this simply a crystallization<br />
phenomenon? There are very small conformational differences between the sphingosine-in<br />
<strong>and</strong> sphingosine-out GLTP-lipid complexes. At a first glance, it appears that increasing the N-<br />
linked acyl chain length, <strong>and</strong> due to the adaptable hydrophobic cavity of GLTP, sphingosine<br />
appears to be simply forced out by a long chain. However, co-crystallization with short chains<br />
N-linked chains, such as C8 & C10, did not allow for a sphingosine-in alignment. Most likely<br />
the shorter acyl chain never reaches all the way to the bottom of the hydrophobic tunnel <strong>and</strong>,<br />
consequently, the tunnel remains narrow preventing the sphingosine from entering. Another<br />
peculiarity that has been complicating the GLTP crystal structure studies has been the<br />
presence of short extraneous hydrocarbon chains not originating from the bound GSL, found<br />
inside the hydrophobic cavity (Malinina et al., 2004;West et al., 2004;Airenne et al., 2006).<br />
These most likely originate from the bacterial expression of the protein used for<br />
crystallization. Finding hydrocarbons <strong>and</strong> detergents inside the crystal structure of<br />
Accepted Manuscript<br />
hydrophobic proteins is seen frequently, <strong>and</strong> is sometimes crucial for obtaining protein<br />
crystals (Haapalainen et al., 2001;Guan et al., 2001;Wright et al., 2003). The extraneous<br />
hydrocarbon seen in the apo-GLTP crystal structures is displaced from the hydrophobic<br />
tunnel when GLTP forms a complex with the <strong>glycolipid</strong> (Malinina et al., 2004;Airenne et al.,<br />
2006). It could be possible that in vivo, GLTP could also carry a hydrophobic molecule when<br />
no <strong>glycolipid</strong> is bound. It is not clear if the molecule enters from the <strong>glycolipid</strong> binding site or<br />
enters into the hydrophobic cavity from the opposite end, through the openings at the end of<br />
the tunnel, formed by helix-1, helix-7 <strong>and</strong> helix-8. The opening is structurally wide enough to<br />
8<br />
Page 9 of 34
let a hydrocarbon chain pass. The far end opening of the hydrophobic cavity can be seen in<br />
Fig. 3A, black arrow.<br />
4.2 Crystal-Related Dimerization in GSL-GLTP Complexes<br />
In a recent paper, the dimerization states of GLTP were analyzed (Samygina et al., 2013).<br />
Previously, homodimers of GLTP has been found in various crystal structures when GLTP<br />
has been complexed with different glycosphingolipids. It appears that when GLTP binds the<br />
<strong>glycolipid</strong> in the sphingosine–out mode, a dimer will form (Malinina et al., 2006). This event<br />
is suggested to be due to interaction of two different GLTP molecules as well as the adjacent<br />
sphingosine hydrocarbon chains (Samygina et al., 2013). The authors proposed that the<br />
function of GLTP might be regulated by forming reversible homodimers. This is further<br />
supported by the finding that when wildtype GLTP <strong>and</strong> a mutant GLTP (D48V) bind 24:1<br />
sulfatide in the absence of lipid membranes, a homodimer of GLTP will form in solution<br />
(Samygina et al., 2011). In contrast to these results it was shown, based on analytical<br />
ultracentrifugation experiments, size-exclusion chromatography <strong>and</strong> affinity-tag<br />
immunoadsorbtion experiments, that both apo-GLTP <strong>and</strong> GLTP complexed with GalCer<br />
(24:1) are monomeric in solution (Malinina et al., 2006). However, the authors still proposed,<br />
based on optimal docking area (ODA) energy contact calculations, that upon binding of<br />
GalCer in the sphingosine-out mode, the sphingosine chain would interact with a sphingosine<br />
molecule on a partner GLTP molecule, thereby forming crystal-related cross dimers.<br />
The crystal-related dimerization of GLTP allows the two sphingosine chains to interact with<br />
each other presumably in a more energetically favorable conformation than being inside the<br />
tunnel. Is the sphingosine-out form, <strong>and</strong> the subsequent dimerization of GLTP the biologically<br />
active carrier of GSLs in vivo? More than a dozen crystal structures of GLTP have the bound<br />
Accepted Manuscript<br />
<strong>glycolipid</strong> in the sphingosine-out conformation. Could this simple be a form generated during<br />
high protein concentrations <strong>and</strong> under limited hydration? The sphingosine chain might be<br />
jutted outwards into a low hydration environment, hydrophobic amino acid residues in close<br />
proximity to the tunnel entrance (Fig. 3B, red patch indicated with a black arrow) <strong>and</strong><br />
extraneous hydrocarbons occupying the far end of the tunnel might drive a sphingosine-out<br />
conformation. It is of vital importance to keep in mind that all dimers form when no<br />
membrane surfaces are present. Analysis of GLTP in solution over a wide range of<br />
concentrations using light scattering <strong>and</strong> gel-filtration all indicate that in the presence of<br />
9<br />
Page 10 of 34
membranes GLTP is a monomer (Zhai et al., 2009).<br />
4.3 GLTP-like proteins<br />
The Arabidopsis thaliana GLTP-like protein (Brodersen et al., 2002), ACD11 has also<br />
conserved residues, the GLTP amino acids Asp-48 <strong>and</strong> His-140 participate in the hydrogen<br />
binding to the sphingosine base <strong>and</strong> are conserved in ACD11 as Asp-60 <strong>and</strong> His-143<br />
respectively (Petersen et al., 2008). ACD11 appears to be able to adopt a GLTP-like structure<br />
(Petersen et al., 2008). ACD11 appears not to be capable of <strong>transferring</strong> <strong>and</strong> (binding?)<br />
glycosphingolipids in vitro, only the single chain sphingosine (Brodersen et al., 2002).<br />
Replacing the sphingosine hydrogen bond forming amino acids in ACD11 (Asp-60 & His-<br />
143) causes a complete loss of transfer activity. However, if tryptophan is introduced in<br />
position 103, to make ACD11 resemble GLTP, it does not improve its <strong>glycolipid</strong> transfer<br />
activity.<br />
5. Membrane composition consequences on protein activity<br />
The composition of the membrane has a great impact on protein function <strong>and</strong> activity.<br />
Processes like binding to membranes <strong>and</strong> <strong>glycolipid</strong> extraction by the GLTPs have been<br />
studied quite intensively. We will here give the latest insights in such events.<br />
5.1 Membrane binding<br />
Accepted Manuscript<br />
The rate-limiting step in the <strong>glycolipid</strong> transfer process is thought to be the GLTP-<strong>glycolipid</strong><br />
complex formation, or the dissociation of the protein-lipid complex from the membrane into<br />
the surrounding aqueous environment (Rao et al., 2005). The initial GLTP binding to the<br />
membrane, however, is not the limiting process <strong>and</strong> appears to have a minimal contribution to<br />
the overall changes to the enthalpy <strong>and</strong> entropy for the overall transfer process. Once docked<br />
to the membrane surface, GLTP is searching for the <strong>glycolipid</strong> substrate in the membrane.<br />
This process is probably a combination of both a lateral flow of both GLTP <strong>and</strong> lipids. GLTP<br />
transfer activity is different depending on the membrane composition <strong>and</strong> the lateral<br />
10<br />
Page 11 of 34
miscibility of the lipids in the membrane. If the <strong>glycolipid</strong> is in a membrane in a fluid phase<br />
the <strong>glycolipid</strong> removal is favored. GLTP binds more strongly to sphingomyelin containing<br />
membranes than to phosphatidylcholine membranes. This is true regardless of acyl chain<br />
composition, i.e. saturated or unsaturated (Ohvo-Rekilä <strong>and</strong> Mattjus, 2011).<br />
5.2 Lipid extraction<br />
Membrane composition. It has been shown that the rate of transfer by GLTP is sensitive to the<br />
membrane composition surrounding the <strong>glycolipid</strong>. The positively charged areas of GLTP<br />
affect how well GSLs are moved from donor to acceptor membranes. GalCer in negatively<br />
charged vesicles (5 or 10 mol% negative lipids) is transferred significantly slower than from<br />
the donor vesicles composed of neutral membranes (Mattjus et al., 2000). The same amount<br />
of negative charge in acceptor vesicles does not impede the transfer rate as effectively.<br />
Positively charged lipids in the donor vesicle membrane neither affect the GSL transfer<br />
(Mattjus et al., 2000). GLTP is positively charged at neutral pH (pI=9.0) <strong>and</strong> will be attracted<br />
to the negatively charged donor membrane through electrostatic interactions between the<br />
protein <strong>and</strong> the membrane surface, resulting in a slow transfer. It is speculated that the GLTP<br />
off-rate from the donor surface becomes slow <strong>and</strong> consequently the transfer rate is slow.<br />
GLTP cannot transfer GalCer <strong>and</strong> GlcCer from vesicles made of saturated SMs, like<br />
palmitoyl-SM, regardless of curvature of the donor vesicle membranes (Mattjus et al.,<br />
2002;Nylund <strong>and</strong> Mattjus, 2005). Addition of cholesterol has only minimal effect on the<br />
transfer rate. If donor vesicles are made of different SM analogues, that more resemble a PC,<br />
transfer is possible. If the 3-hydroxy group of SM is replaced by a hydrogen, <strong>and</strong> the trans-4,5<br />
double bond of sphingosine is reduced, transfer will occur (Nylund et al., 2006). GLTP is also<br />
able to transfer <strong>glycolipid</strong>s from membranes composed of unsaturated SMs, however much<br />
Accepted Manuscript<br />
slower than from a chain-matched PC (Mattjus et al., 2002;Nylund <strong>and</strong> Mattjus, 2005).<br />
Membrane curvature. The transfer rates for GLTP are more than 5 times greater for small<br />
highly curved fluid 1-palmitoyl-2-oleyl-sn-glycero-3-phosphocholine (POPC) membranes<br />
compared to that for large membranes, that are comparable to a planar bilayer (Rao et al.,<br />
2005;Nylund et al., 2007). The transfer rate of fluorescently labeled GalCer from fluid POPC<br />
vesicles decreases in a linear fashion as the vesicles become larger, to an almost undetectable<br />
rate for vesicles above 800 nm in diameter. From planar monolayers no <strong>glycolipid</strong> transfer<br />
11<br />
Page 12 of 34
occurs (Nylund et al., 2007). If the membrane is composed of 1,2-dipalmitoyl-sn-glycero-3-<br />
phosphocholine (DPPC) there is no difference in the anthrylvinyl-GalCer transfer rates<br />
regardless of the donor vesicle size except for the smallest vesicles (50 nm in diameter) that<br />
are significantly higher. It is noteworthy that the transfer is very small from gel-like<br />
membranes in general. It is speculated that GLTP can bind clustered <strong>glycolipid</strong>s in a fluid<br />
environment such as POPC, but with limited accessibility to <strong>glycolipid</strong>s well dispersed into a<br />
more tightly packed environment such as that in DPPC <strong>and</strong> palmitoyl-SM (Nylund <strong>and</strong><br />
Mattjus, 2005;Nylund et al., 2006;Brown <strong>and</strong> Mattjus, 2007;Nylund et al., 2007;Mattjus,<br />
2009;Ohvo-Rekilä <strong>and</strong> Mattjus, 2011). Recently Brown <strong>and</strong> colleagues found that GLTP was<br />
able to extract BODIPY-GlcCer from a planar monolayer composed of POPC <strong>and</strong> low<br />
amounts (physiological) of either PE or phospatidic acid (PA) (15 <strong>and</strong> 5 mol% respectively)<br />
at bilayer comparable surface pressures (30-35 mN/m) (Zhai et al., 2013). It is unclear what is<br />
happening with phosphatidylserine (POPS) in these experiments, it appears not to stimulate<br />
GlcCer transfer to the same extent as PE <strong>and</strong> PA (Zhai et al., 2013), but rather penetrate the<br />
monolayer. Previously Nylund <strong>and</strong> co-workers showed that the exclusion pressure for GLTP<br />
i.e. the pressure where GLTP no longer can penetrate the monolayer, was clearly the highest<br />
with DPPS (30.7 mN/m) of all lipids analyzed, even negatively charged sulfatide (Nylund et<br />
al., 2007). GLTP is positively charged at neutral pH <strong>and</strong> penetrates negatively charged<br />
monolayer lipid films at higher surface pressures than those of neutral lipid films. However,<br />
DPPS is also more easily penetrated than the <strong>glycolipid</strong> sulfatide. The relatively larger<br />
penetration of GLTP into the lipid monolayers at low surface pressures does not necessarily<br />
indicate a direct interaction of the protein with the lipid molecules, as GLTP alone is surfaceactive<br />
<strong>and</strong> can spontaneously migrate from the subphase to a lipid free interface. This shows<br />
that the extraction of GLTP can be enhanced by alterations in the matrix lipid, independent of<br />
any additional proteins. The hydration around the GSL is likely to be the driving force for the<br />
effects seen by PE <strong>and</strong> PA. Saturated-chain PEs (12:0, 14:0 & 16:0) are not miscible with<br />
Accepted Manuscript<br />
chain-mixed GlcCer either above or below T m of the phospholipid. However, the unsaturated<br />
POPE, OPPE <strong>and</strong> SOPE are all miscible with GlcCer at physiological temperatures (Quinn,<br />
2011;Quinn, 2012). It is likely that GlcCer <strong>and</strong> GalCer interact more favorable with POPE<br />
than with POPC in monolayers because of the smaller <strong>and</strong> less hydration of the PE head. The<br />
hydration of GlcCer <strong>and</strong> GalCer more resembles PE than PC. In the monolayer system one<br />
leaflet is missing, this inevitably allows the lipids to move more freely out from the<br />
membrane plane, less hydrophobic forces keep the lipid stronger into the membrane.<br />
12<br />
Page 13 of 34
5.3 GLTP membrane binding, desorption mechanism <strong>and</strong> lipid release<br />
Trp-142 has been confirmed to be essential in the GLTP membrane binding process. In<br />
FAPP2 <strong>and</strong> HET-C2 the conserved amino acids are Trp-447 <strong>and</strong> Phe-149 respectively.<br />
Several other hydrophobic amino acids take part in the docking of GLTP onto the membrane<br />
surface (Malinina et al., 2004;Rao et al., 2005;Airenne et al., 2006;West et al., 2006;Neumann<br />
et al., 2008;Zhai et al., 2009;Kamlekar et al., 2010;Kenoth et al., 2010). A coloring of the<br />
amino acids according to their hydrophobic scale is presented in Fig. 3B. The most<br />
hydrophobic amino acids are shown in red, <strong>and</strong> amino acids with lower hydrophobicity in<br />
lighter red shades, white being fully hydrophilic in nature (Eisenberg et al., 1984). Examining<br />
the changes to the crystal structure of GLTP after the <strong>glycolipid</strong> is bound to the hydrophobic<br />
tunnel reveals several movements in two interhelical loops <strong>and</strong> two different helices. Alphahelix<br />
2 moves along its axis <strong>and</strong> outwards <strong>and</strong> alpha-helix 6 moves outwards. The net result of<br />
these movements is an expansion of a hydrophobic tunnel to accommodate the lipid, either<br />
with one or two chains inside. A putative membrane binding domain with high<br />
hydrophobicity is indicated with a black arrow in Fig. 3B. There are only small changes in<br />
this domain between the apo-GLTP <strong>and</strong> the lig<strong>and</strong>-bound form. The domain is formed by a<br />
group of tryptophan, tyrosine, isoleucine <strong>and</strong> nonpolar residues on the protein surface, in<br />
close proximity to the <strong>glycolipid</strong>-binding site. The small change in the hydrophobic domain<br />
suggests that after GSL binding also other parts of the protein are responsible for the release<br />
of GLTP from the membrane.<br />
The mechanism how the GSL is lifted from the membrane into the GLTP cavity is unclear.<br />
Perhaps the ability of the sphingosine chain to bind to the outer grow of GLTP will allow the<br />
amide-linked chain to enter the hydrophobic tunnel, specifically recognized <strong>and</strong> bound by the<br />
amino acids of the sugar binding pocket of GLTP. We speculate that the mechanism could be<br />
Accepted Manuscript<br />
that the membrane-binding domain of GLTP generates a disturbance in the hydrogen-bond<br />
network between the water molecules <strong>and</strong> the membrane lipid head groups, allowing an acyl<br />
chain from the lipids in the membrane to move upward towards the hydrophobic amino acids,<br />
out from the hydrophobic interior, onto the GLTP surface. If the acyl chain belongs to a lipid<br />
that is also recognized by the sugar-binding pocket of GLTP, a lipid binding takes place. If<br />
the chain belongs to a lipid that is not recognized, it simply moves back into the membrane<br />
core, <strong>and</strong> the next lipid is allowed to be tested by GLTP. Once a <strong>glycolipid</strong> binding has<br />
occurred <strong>and</strong> the <strong>glycolipid</strong> has been captured by GLTP, the second chain would then be<br />
13<br />
Page 14 of 34
pulled along <strong>and</strong> would enter the hydrophobic tunnel. Once GLTP lifts of the interface, as a<br />
consequence of the helical movements <strong>and</strong> structural changes in the membrane-binding<br />
domain, the sphingosine chain would be forced away form the aqueous environment into a<br />
more energetically favorable milieu inside GLTP. Very little is known about the release of the<br />
bound <strong>glycolipid</strong> from GLTP, once it encounters new membranes. For instance, the binding<br />
of FAPP2 carrying glucosylceramide is stronger than an empty apo-FAPP2 (D'Angelo et al.,<br />
2013), if this is the case also for GLTP is at present unknown.<br />
6. Subcellular Targeting of GLTP <strong>and</strong> FAPP2<br />
There are numerous sequences, motifs <strong>and</strong> domains within proteins that help them find their<br />
right destinations in the cell. Both protein-protein interactions as well as protein-lipid<br />
interactions are important in guiding cellular components to various subcellular<br />
compartments. The C1 <strong>and</strong>/or C2 domains of for example protein kinase C, diacylglycerol<br />
kinases <strong>and</strong> some Ca 2+ -dependent phospholipases, bind most commonly to phospholipids in<br />
cell membranes (Cho, 2001). Another domain, the steroidogenic acute regulatory protein<br />
(StAR)-related lipid transfer (START) domain, is important in lipid trafficking <strong>and</strong><br />
metabolism <strong>and</strong> the different START domains identify several lipids, including cholesterol<br />
<strong>and</strong> ceramide (Alpy <strong>and</strong> Tomasetto, 2005). The Zn 2+ -binding FYVE domains (about 70<br />
residues) are, on the other h<strong>and</strong>, very specific for phosphatidylinositol-3-phosphates (PI3P)<br />
(Gaullier et al., 1998;Patki et al., 1998) <strong>and</strong> are responsible for endosomal targeting<br />
(Stenmark et al., 1996). Another group of phosphoinositides-binding domains are the PX<br />
domains, found in a diverse set of proteins (Simonsen <strong>and</strong> Stenmark, 2001). Many of the PX<br />
domain-containing proteins also preferentially recognize PI3P <strong>and</strong> localize to the endosomes<br />
(Tessier <strong>and</strong> Woodgett, 2006). The PH domains are very divergent in their sequences, but<br />
Accepted Manuscript<br />
mediate several protein-protein interactions as well as recognizing a variety of phosphorylated<br />
derivatives of phosphatidylinositol (Scheffzek <strong>and</strong> Welti, 2012). The specific binding of PH<br />
domains to distinct phosphoinositides, located at different sites in the cell, are especially<br />
important amongst lipid-binding proteins, since this recruitment decide the protein subcellular<br />
localization <strong>and</strong> may regulate protein activity (Levine <strong>and</strong> Munro, 2002). Although there are a<br />
large variety of domains regulating protein trafficking <strong>and</strong> activity, more targeting signals will<br />
for sure be found in the future. So this raise the important question, how can the cytoplasmic<br />
proteins GLTP <strong>and</strong> FAPP2 find their specific sites of action?<br />
14<br />
Page 15 of 34
6.1 GLTP<br />
GLTP is a cytoplasmic protein <strong>and</strong> results obtained from different experimental methods<br />
suggest that GLTP does not reside within the endoplasmic reticulum (ER), the nucleus, the<br />
mitochondria or the Golgi (Tuuf <strong>and</strong> Mattjus, 2007). However, analysis of the amino acid<br />
sequence of GLTP from different organisms, indicate that there still might exist motifs or<br />
sequences within GLTP that would enable targeting of the protein to different compartments<br />
in the cell. In 2009 a motif in GLTP was characterized, a motif the authors referred to as the<br />
FFAT-like motif (Tuuf et al., 2009). Previously it has been shown that several lipid-binding<br />
proteins possess a conserved motif designated FFAT (two phenylalanines (FF) in an acidic<br />
tract) (Loewen et al., 2003), a targeting signal which directs them to the cytosolic surface of<br />
the ER for association with the vesicle-associated membrane protein-associated proteins<br />
(VAMP-associated proteins or VAPs) (Wyles et al., 2002;Loewen et al., 2003;Kawano et al.,<br />
2006). The VAPs have been imlicated in several important processes in the cell, like vesicle<br />
transport (Skehel et al., 1995), the unfolded protein response (Kanekura et al., 2006) <strong>and</strong> in<br />
the neurodegenerative disorder Amyothrophic lateral sclerosis, type 8 (Nishimura et al.,<br />
2004). Two of the isoforms of the VAPs, VAP-A <strong>and</strong> VAP-B, are mainly located in the ER<br />
(Weir et al., 2001;Nishimura et al., 2004), <strong>and</strong> it is proposed that these integral membrane<br />
proteins localize to specific ER contact sites, which could act as meeting points for several<br />
lipid transfer proteins (Levine <strong>and</strong> Loewen, 2006). The consensus sequence of the FFAT<br />
motif, almost exclusively found in lipid-binding proteins, is reported to contain the residues<br />
EFFDAxE appearing in an acidic environment (Loewen et al., 2003). In the same study it was<br />
demonstrated that minor substitutions in some of the amino acids are allowed, while other<br />
residues are essential for an interaction to occur with the VAPs. This conclusion is in<br />
agreement with more recent structural <strong>and</strong> experimental studies, where it is demonstrated that<br />
Accepted Manuscript<br />
many amino acid substitutions in the FFAT motif can indeed be tolerated (Kaiser et al.,<br />
2005;Mikitova <strong>and</strong> Levine, 2012). The FFAT-like motif identified in GLTP contains the<br />
amino acids 32 PFFDCLG 38 , a region which is also responsible for VAP-A-binding in vitro<br />
(Tuuf et al., 2009) (Fig. 1B). This FFAT-like motif is closely related to EFFDAxE <strong>and</strong><br />
contains the two phenylalanines (F) <strong>and</strong> the aspartate (D) residue, together with proline,<br />
cysteine, leucine <strong>and</strong> glycine. The fact that the FFAT-like motif is exposed on the surface of<br />
the protein (Airenne et al., 2006) also indicates that it is structurally available for a interaction<br />
with the VAPs (Fig. 2A). However, in a recent analysis on proteins with FFAT-like motifs the<br />
15<br />
Page 16 of 34
authors proposed that the FFAT-like motif in GLTP might not be optimal situated for<br />
interaction with VAP (Mikitova <strong>and</strong> Levine, 2012). This is due to 32 PFFDCLG 38 in GLTP is<br />
residing in a α-helix <strong>and</strong> not in an β-str<strong>and</strong>-like conformation, a structure that has previously<br />
been found in FFAT-containing proteins in VAP-FFAT-binding studies (Kaiser et al.,<br />
2005;Furuita et al., 2010). However, <strong>and</strong> interestingly, another FFAT-like-motif in the GLTP<br />
amino acid sequence has been identified (Mikitova <strong>and</strong> Levine, 2012). This sequence is weak,<br />
contains the residues 22 PFLEAVS 28 <strong>and</strong> lies upstream the region 32 PFFDCLG 38 previously<br />
found in GLTP (Tuuf et al., 2009) (Fig. 1B). It was suggested that both the FFAT-like motifs<br />
might act in concert, <strong>and</strong> consequently, a stronger VAP interaction would be obtained. Other<br />
proteins, with FFAT-like motifs even further from the consensus sequence, have been<br />
reported to interact with the VAPs under physiological conditions (Saita et al.,<br />
2009;Saravanan et al., 2009), which could imply that GLTP also might associate with the<br />
VAPs in vivo <strong>and</strong> not just in vitro. However, whether GLTP actually interacts with the VAPs<br />
in biological systems still remains unclear.<br />
Although GLTP is suggested to have a cytosolic distribution (Tuuf <strong>and</strong> Mattjus, 2007), it<br />
should not be excluded that GLTP might interact with various organelles during different<br />
stages in the cell cycle, upon certain cellular signals or that the association of GLTP with<br />
organelle membranes might be weak. In a large phosphoproteomic study on phosphorylation<br />
sites using mass spectrometric methods, there was a site identified in the GLTP sequence that<br />
was phosphorylated, threonine 69 (Olsen et al., 2010). This might raise the possibility that the<br />
subcellular localization of GLTP <strong>and</strong>/or interactions with other proteins might be regulated by<br />
phosphorylation/dephosphorylation events. The phosphorylation status of ceramide transfer<br />
protein <strong>and</strong> oxysterol-binding protein (OSBP), which also interact with the VAPs, can affect<br />
the targeting of the proteins to different organelles (Kumagai et al., 2007;Fugmann et al.,<br />
Accepted Manuscript<br />
2007;Saito et al., 2008;Tomishige et al., 2009;Nhek et al., 2010). Other possible posttranslational<br />
modifications of GLTP that could act as regulatory mechanisms could involve<br />
ubiquitylation or perhaps attachment of various lipids. Interestingly, a recent study shows that<br />
GLTP can be ubiquitylated (Wagner et al., 2011).<br />
6.2 FAPP2<br />
The FAPPs possess a PH domain, which is important for targeting of the protein. This domain<br />
specifically recognizes PI4P <strong>and</strong> the GTPase ADP-ribosylation factor (ARF) on the TGN<br />
16<br />
Page 17 of 34
(Godi et al., 2004) (Fig. 1A). The same mechanism is utilized by OSBP in yeast, that uses its<br />
PH domain to associate with both ARF1 <strong>and</strong> PI4P in the Golgi complex (Levine <strong>and</strong> Munro,<br />
2002) . In vitro studies on the GlcCer transport ability of FAPP2 showed that if ARF1 <strong>and</strong><br />
PI4P were added to acceptor membranes, the transport of GlcCer mediated by FAPP2<br />
between lipid donors to acceptors was enhanced (D'Angelo et al., 2007). It is the PI4P that<br />
dictates the targeting <strong>and</strong> upon GlcCer binding, FAPP2 translocates to the PI4P binding site<br />
on TGN (D'Angelo et al., 2013). However, the phospatidylinositol-4-OH kinase-β is recruited<br />
to the Golgi by the ARFs, which consequently links PI4P production <strong>and</strong> ARF localization<br />
together (Godi et al., 1999). The binding of the FAPP1 PH domain to ARF1 <strong>and</strong> PI4P has<br />
been analyzed (He et al., 2011). It appears that PI4P is absolutely necessary for the PH<br />
domain to be inserted into the membrane, <strong>and</strong> that the PI4P binding is pH-dependent, showing<br />
a stronger binding at pH 6.5 than 7.4. Furthermore, in the same study it was demonstrated that<br />
although the binding sites of ARF1 <strong>and</strong> PI4P are in close proximity, they can bind<br />
independently of each other <strong>and</strong> on the same time. These features of the PH domain<br />
contribute to the specific targeting of the FAPPs to the Golgi complex. FAPP2 with a bound<br />
GlcCer is targeted to the TGN via its PH domain <strong>and</strong> is capable to bind to PI4P <strong>and</strong> small<br />
GTPase ARF1 proteins (Godi et al., 2004). The mechanism how the bound GlcCer would<br />
increase the affinity of FAPP2 towards PI4P in the TGN remains unclear. Presumable, when<br />
GlcCer is bound to the GLTP domain of FAPP2, the protein adopts another conformation,<br />
different than the apo-form, allowing for an association by the PH domain to the membranes<br />
in the TGN. Once the GlcCer is released, the apo-FAPP2 leaves the membrane, <strong>and</strong> the<br />
affinity to bind PI4P by the PH domain again becomes lower. How the release of GlcCer<br />
takes place is not known (D'Angelo et al., 2013). GlcCer needs to be located on the inner<br />
leaflet of the TGN membranes in order to become further glycosylated to LacCer <strong>and</strong> further<br />
to Gb 3 by the enzymes residing in the lumen of the Golgi. It is tempting to speculate that<br />
when GlcCer exits the hydrophobic cavity of FAPP2, it would directly be pushed, with the<br />
Accepted Manuscript<br />
glucose head group first, directly to the inner leaflet. This mechanism would render a GlcCer<br />
flipping from the outer to the inner leaflet of the TGN unnecessary.<br />
FAPP2 in mammals also contain a weak FFAT-like motif, defined as TFFSTMN (Mikitova<br />
<strong>and</strong> Levine, 2012) (Fig. 1B). However, results from experiments from the same study showed<br />
that FAPP2 did not target the VAPs on ER as wildtype, but if the serine at position 4 was<br />
pseudophosphorylated, it showed a weak targeting to the VAPs. To our knowledge, it has not<br />
yet been demonstrated that FAPP2 actually interacts with the VAPs in vivo. In addition, there<br />
17<br />
Page 18 of 34
are several phosphorylation sites in FAPP2 that has been identified, however, whether these<br />
modifications are of importance for the function or subcellular localization of FAPP2 remains<br />
to be determined.<br />
7. Biological functions of GLTP <strong>and</strong> FAPP2<br />
The substrate specificity of both GLTP <strong>and</strong> FAPP2 suggests that the members of the GLTP<br />
superfamily are vital for regulating the GSL metabolism. Amongst the lipid transfer proteins,<br />
exist both genuine lipid transporters (Hanada et al., 2003) <strong>and</strong>/or sensors of various lipids<br />
(Litvak et al., 2005;Vihervaara et al., 2011). What about the proteins that are specific for<br />
glycosphingolipids? The biological function of GLTP is still not clear. However, implications<br />
have been made that GLTP might catalyze the transfer of GlcCer from the Golgi to the PM<br />
(Warnock et al., 1994;Halter et al., 2007). Other reports propose that GLTP might in addition<br />
function as an intracellular sensor of GSLs. Malakhova <strong>and</strong> coworkers mutated the lig<strong>and</strong>ing<br />
site in GLTP <strong>and</strong> looked at the ability of GLTP to acquire <strong>and</strong> release its substrate<br />
(Malakhova et al., 2005). The authors speculated that in addition to acting as a <strong>glycolipid</strong><br />
carrier, GLTP might function as a GlcCer sensor or a presenter of <strong>glycolipid</strong>s to protein<br />
receptors, since the results showed that wildtype GLTP was not keen to release the lig<strong>and</strong> to<br />
POPC membranes. Furthermore, GLTP appears to sense changes in the amount of newly<br />
synthesized GlcCer (Kjellberg <strong>and</strong> Mattjus, 2013). When GLTP is overexpressed in HeLa<br />
cells, de novo synthesis of GlcCer, but not LacCer or GalCer, is enhanced (Tuuf <strong>and</strong> Mattjus,<br />
2007). Downregulation of GLTP resulted in no changes in de novo syntheses of the<br />
sphingolipids, which made the authors suggest that GLTP is not necessarily involved in<br />
GlcCer transport, bur rather has a sensor function (Tuuf <strong>and</strong> Mattjus, 2007). Together, these<br />
studies indicate that GlcCer is the relevant lipid substrate for GLTP. This goes well with<br />
Accepted Manuscript<br />
GlcCer being the only <strong>glycolipid</strong> synthesized at the cytoplasmic side of organelle membranes<br />
(Futerman <strong>and</strong> Pagano, 1991;Jeckel et al., 1992). Another proposed function of GLTP is as a<br />
regulator of adjusting levels of cellular ceramide <strong>and</strong>, consequently, also the production of<br />
GSLs (Zou et al., 2011).<br />
FAPP2 has been implicated in several biological processes in the cell. Initially, FAPP2 was<br />
suggested to be involved in the transport of cargo between the TGN <strong>and</strong> the PM (Godi et al.,<br />
2004;Vieira et al., 2005). FAPP2 has also a tubulation activity <strong>and</strong> is suggested to be<br />
18<br />
Page 19 of 34
esponsible for tubulating the membranes in the TGN through its PH domain, which gives<br />
further evidence for a role of FAPP2 in Golgi-PM trafficking (Cao et al., 2009). The PH<br />
domain of FAPP1 is furthermore able to tubulate membranes (Lenoir et al., 2010). Another<br />
major role of FAPP2 was demonstrated by D´Angelo <strong>and</strong> coworkers, when they discovered<br />
that FAPP2 was an important factor in the glycosphingolipid synthetic pathways (D'Angelo et<br />
al., 2007). FAPP2 was responsible for transporting GlcCer from the cytosolic side of cis-<br />
Golgi to the later Golgi compartments, where the more complex GSLs are synthesized<br />
(Lannert et al., 1998). This transport was shown to be non-vesicular <strong>and</strong> require both the<br />
presence of ARF <strong>and</strong> PI4P on the later Golgi (D'Angelo et al., 2007). In a recent work it has<br />
been shown that GlcCer can undergo two fates. After GlcCer is synthesized in can either flip<br />
into the lumen of Golgi <strong>and</strong> be a precursor molecule for LacCer <strong>and</strong> GM 3 or alternatively, it<br />
can be transported by FAPP2 through the cytosol to TGN for globoside production, Gb 3<br />
(D'Angelo et al., 2013). This interesting <strong>and</strong> novel finding shows how the sorting <strong>and</strong><br />
branching of GlcCer by two different transport mechanisms, the non-vesicular (FAPP2) <strong>and</strong><br />
the vesicular, will decide which GSL end product is produced. However, FAPP2 has also<br />
been shown to mediate a GlcCer-transport in another direction, from the cis-Golgi back to the<br />
ER (Halter et al., 2007). Here it is suggested that after GlcCer reaches the ER, it would flipflop<br />
into the lumen of ER <strong>and</strong> be transported through Golgi by vesicle transport to the TGN<br />
where it could be processed to more complex GSLs. The flip-flop of GlcCer into the lumen of<br />
ER might be catalyzed by a ATP-independent phospholipid flippase (Chalat et al., 2012). A<br />
possible VAP-interaction, as previously discussed in chapter 6.2, might enable ER targeting<br />
of FAPP2 (Mikitova <strong>and</strong> Levine, 2012).<br />
8. Conclusion<br />
What is known about the depth of GLTP penetration into the membrane surface (Fig. 4)?<br />
Accepted Manuscript<br />
Experimental evidence indicated that the GLTP binding to the membrane is nonperturbing<br />
<strong>and</strong> not deep (Rao et al., 2005;Nylund et al., 2007). A theoretical orientation of proteins in<br />
membranes (OPM) analyses indicate a membrane penetration depth for GLTP of 3.4 Å with a<br />
tilt angle of 86° for the apo-GLTP <strong>and</strong> 3.9 Å <strong>and</strong> 87° for GLTP with a LacCer bound. It<br />
would be interesting to relate this predicted penetration depth with the membrane curvature,<br />
head group hydration <strong>and</strong> the depth of the GSLs in the surrounding lipid microenvironment,<br />
to mechanistically underst<strong>and</strong> how the GLTP lipid extraction from the membrane takes place.<br />
19<br />
Page 20 of 34
The <strong>glycolipid</strong> depth is dependent on the acyl chain length <strong>and</strong> saturation of the ceramide<br />
backbone, <strong>and</strong> consequently, the lipid species is directly affecting the exposure of the<br />
saccharide moieties into the plane of the bilayer. This aglycone (non-sugar part of the<br />
<strong>glycolipid</strong>) modulation of <strong>glycolipid</strong>s will strongly affect how they are recognized for<br />
instance by their transfer proteins (Lingwood, 1996;Lingwood, 2011). It is clear that all<br />
membrane GSLs are not distributed equally over the cellular membranes <strong>and</strong> their amount<br />
only controlled by their respective synthesis <strong>and</strong> breakdown machineries. It is also<br />
fundamental to know whether the GSLs are transported by vesicular means or by proteins.<br />
The function of the <strong>glycolipid</strong> <strong>transferring</strong> proteins is slowly emerging while new members<br />
are found, like the CPTP protein. This further adds complexity to the intracellular lipid<br />
trafficking picture.<br />
Acknowledgements<br />
We thank Anders Backman for help with the images.<br />
References<br />
References<br />
Abe, A. 1990. Primary structure of <strong>glycolipid</strong> transfer protein from pig brain. J. Biol. Chem.<br />
265, 9634-9637.<br />
Abe, A. <strong>and</strong> T.Sasaki. 1989a. Formation of an intramolecular disulfide bond of <strong>glycolipid</strong><br />
transfer protein. Biochim. Biophys. Acta 985, 45-50.<br />
Abe, A. <strong>and</strong> T.Sasaki. 1989b. Sulfhydryl groups in <strong>glycolipid</strong> transfer protein: formation of an<br />
intramolecular disulfide bond <strong>and</strong> oligomers by Cu2+-catalyzed oxidation. Biochim.<br />
Biophys. Acta 985, 38-44.<br />
Abe, A., K.Yamada, <strong>and</strong> T.Sasaki. 1982. A protein purified from pig brain accelerates the<br />
inter-membranous translocation of mono- <strong>and</strong> dihexosylceramides, but not the<br />
translocation of phospholipids. Biochem. Biophys. Res. Commun. 104, 1386-1393.<br />
Airenne, T.T., H.Kidron, Y.Nymalm, M.Nylund, G.West, P.Mattjus, <strong>and</strong> T.A.Salminen.<br />
2006. Structural evidence for adaptive lig<strong>and</strong> binding of <strong>glycolipid</strong> transfer protein. J.<br />
Mol. Biol. 355, 224-236.<br />
Alpy, F. <strong>and</strong> C.Tomasetto. 2005. Give lipids a START: the StAR-related lipid transfer<br />
(START) domain in mammals. J. Cell Sci. 118, 2791-2801.<br />
Bloj, B. <strong>and</strong> D.B.Zilversmit. 1981. Accelerated transfer of neutral glycosphingolipids <strong>and</strong><br />
ganglioside GM1 by a purified lipid transfer protein. J. Biol. Chem. 256, 5988-5991.<br />
Brodersen, P., M.Petersen, H.M.Pike, B.Olszak, S.Skov, N.Odum, L.B.Jorgensen,<br />
R.E.Brown, <strong>and</strong> J.Mundy. 2002. Knockout of Arabidopsis accelerated-cell-death11<br />
Accepted Manuscript<br />
20<br />
Page 21 of 34
encoding a sphingosine transfer protein causes activation of programmed cell death <strong>and</strong><br />
defense. Genes Dev. 16, 490-502.<br />
Brown, R.E. <strong>and</strong> P.Mattjus. 2007. Glycolipid transfer proteins. Biochim. Biophys. Acta 1771,<br />
746-760.<br />
Brown, R.E., F.A.Stephenson, T.Markello, Y.Barenholz, <strong>and</strong> T.E.Thompson. 1985. Properties<br />
of a specific <strong>glycolipid</strong> transfer protein from bovine brain. Chem. Phys. Lipids 38, 79-93.<br />
Cao, X., U.Coskun, M.Rossle, S.B.Buschhorn, M.Grzybek, T.R.Dafforn, M.Lenoir,<br />
M.Overduin, <strong>and</strong> K.Simons. 2009. Golgi protein FAPP2 tubulates membranes. Proc. Natl.<br />
Acad. Sci. U. S. A 106, 21121-21125.<br />
Chalat, M., I.Menon, Z.Turan, <strong>and</strong> A.K.Menon. 2012. Reconstitution of glucosylceramide<br />
flip-flop across endoplasmic reticulum: implications for mechanism of glycosphingolipid<br />
biosynthesis. J. Biol. Chem. 287, 15523-15532.<br />
Cho, W. 2001. Membrane targeting by C1 <strong>and</strong> C2 domains. J. Biol. Chem. 276, 32407-32410.<br />
D'Angelo, G., E.Polishchuk, T.G.Di, M.Santoro, C.A.Di, A.Godi, G.West, J.Bielawski,<br />
C.C.Chuang, A.C.van der Spoel, F.M.Platt, Y.A.Hannun, R.Polishchuk, P.Mattjus, <strong>and</strong><br />
M.A.De Matteis. 2007. Glycosphingolipid synthesis requires FAPP2 transfer of<br />
glucosylceramide. Nature 449, 62-67.<br />
D'Angelo, G., L.R.Rega, <strong>and</strong> M.A.De Matteis. 2012. Connecting vesicular transport with lipid<br />
synthesis: FAPP2. Biochim. Biophys. Acta 1821, 1089-1095.<br />
D'Angelo, G., T.Uemura, C.C.Chuang, E.Polishchuk, M.Santoro, H.Ohvo-Rekila, T.Sato,<br />
T.G.Di, A.Varriale, S.D'Auria, T.Daniele, F.Capuani, L.Johannes, P.Mattjus, M.Monti,<br />
P.Pucci, R.L.Williams, J.E.Burke, F.M.Platt, A.Harada, <strong>and</strong> M.A.De Matteis. 2013.<br />
Vesicular <strong>and</strong> non-vesicular transport feed distinct glycosylation pathways in the Golgi<br />
1. Nature 501, 116-120.<br />
D'Angelo, G., M.Vicinanza, <strong>and</strong> M.A.De Matteis. 2008. Lipid-transfer proteins in<br />
biosynthetic pathways. Curr. Opin. Cell Biol. 20, 360-370.<br />
Dowler, S., R.A.Currie, D.G.Campbell, M.Deak, G.Kular, C.P.Downes, <strong>and</strong> D.R.Alessi.<br />
2000. Identification of pleckstrin-homology-domain-containing proteins with novel<br />
phosphoinositide-binding specificities. Biochem. J. 351, 19-31.<br />
Eisenberg, D., E.Schwarz, M.Komaromy, <strong>and</strong> R.Wall. 1984. Analysis of membrane <strong>and</strong><br />
surface protein sequences with the hydrophobic moment plot. J. Mol. Biol. 179, 125-142.<br />
Fantini, J., N.Garmy, <strong>and</strong> N.Yahi. 2006. Prediction of <strong>glycolipid</strong>-binding domains from the<br />
amino acid sequence of lipid raft-associated proteins: application to HpaA, a protein<br />
involved in the adhesion of Helicobacter pylori to gastrointestinal cells<br />
4. Biochemistry 45, 10957-10962.<br />
Fugmann, T., A.Hausser, P.Schoffler, S.Schmid, K.Pfizenmaier, <strong>and</strong> M.A.Olayioye. 2007.<br />
Regulation of secretory transport by protein kinase D-mediated phosphorylation of the<br />
ceramide transfer protein. J. Cell Biol. 178, 15-22.<br />
Furuita, K., J.Jee, H.Fukada, M.Mishima, <strong>and</strong> C.Kojima. 2010. Electrostatic interaction<br />
between oxysterol-binding protein <strong>and</strong> VAMP-associated protein A revealed by NMR <strong>and</strong><br />
mutagenesis studies. J. Biol. Chem. 285, 12961-12970.<br />
Futerman, A.H. <strong>and</strong> R.E.Pagano. 1991. Determination of the intracellular sites <strong>and</strong> topology<br />
of glucosylceramide synthesis in rat liver. Biochem. J. 280 ( Pt 2), 295-302.<br />
Gaullier, J.M., A.Simonsen, A.D'Arrigo, B.Bremnes, H.Stenmark, <strong>and</strong> R.Aasl<strong>and</strong>. 1998.<br />
FYVE fingers bind PtdIns(3)P. Nature 394, 432-433.<br />
Godi, A., C.A.Di, A.Konstantakopoulos, T.G.Di, D.R.Alessi, G.S.Kular, T.Daniele, P.Marra,<br />
J.M.Lucocq, <strong>and</strong> M.A.De Matteis. 2004. FAPPs control Golgi-to-cell-surface membrane<br />
traffic by binding to ARF <strong>and</strong> PtdIns(4)P. Nat. Cell Biol. 6, 393-404.<br />
Accepted Manuscript<br />
21<br />
Page 22 of 34
Godi, A., P.Pertile, R.Meyers, P.Marra, T.G.Di, C.Iurisci, A.Luini, D.Corda, <strong>and</strong> M.A.De<br />
Matteis. 1999. ARF mediates recruitment of PtdIns-4-OH kinase-beta <strong>and</strong> stimulates<br />
synthesis of PtdIns(4,5)P2 on the Golgi complex. Nat. Cell Biol. 1, 280-287.<br />
Guan, R.-J., M.Wang, X.-Q.Liu, <strong>and</strong> D.-C.Wang. 2001. Optimization of soluble protein<br />
crystallization with detergents. Journal of Crystal Growth 231, 273-279.<br />
Haapalainen, A.M., D.M.van Aalten, G.Merilainen, J.E.Jalonen, P.Pirila, R.K.Wierenga,<br />
J.K.Hiltunen, <strong>and</strong> T.Glumoff. 2001. Crystal structure of the lig<strong>and</strong>ed SCP-2-like domain<br />
of human peroxisomal multifunctional enzyme type 2 at 1.75 A resolution. J. Mol. Biol.<br />
313, 1127-1138.<br />
Halter, D., S.Neumann, S.M.van Dijk, J.Wolthoorn, A.M.de Maziere, O.V.Vieira, P.Mattjus,<br />
J.Klumperman, G.van Meer, <strong>and</strong> H.Sprong. 2007. Pre- <strong>and</strong> post-Golgi translocation of<br />
glucosylceramide in glycosphingolipid synthesis. J. Cell Biol. 179, 101-115.<br />
Hanada, K., K.Kumagai, S.Yasuda, Y.Miura, M.Kawano, M.Fukasawa, <strong>and</strong> M.Nishijima.<br />
2003. Molecular machinery for non-vesicular trafficking of ceramide. Nature 426, 803-<br />
809.<br />
He, J., J.L.Scott, A.Heroux, S.Roy, M.Lenoir, M.Overduin, R.V.Stahelin, <strong>and</strong><br />
T.G.Kutateladze. 2011. Molecular basis of phosphatidylinositol 4-phosphate <strong>and</strong> ARF1<br />
GTPase recognition by the FAPP1 pleckstrin homology (PH) domain. J. Biol. Chem. 286,<br />
18650-18657.<br />
Jeckel, D., A.Karrenbauer, K.N.Burger, G.van Meer, <strong>and</strong> F.Wiel<strong>and</strong>. 1992. Glucosylceramide<br />
is synthesized at the cytosolic surface of various Golgi subfractions. J. Cell Biol. 117,<br />
259-267.<br />
Kaiser, S.E., J.H.Brickner, A.R.Reilein, T.D.Fenn, P.Walter, <strong>and</strong> A.T.Brunger. 2005.<br />
Structural basis of FFAT motif-mediated ER targeting. Structure. 13, 1035-1045.<br />
Kamlekar, R.K., Y.Gao, R.Kenoth, J.G.Molotkovsky, F.G.Prendergast, L.Malinina, D.J.Patel,<br />
W.S.Wessels, S.Y.Venyaminov, <strong>and</strong> R.E.Brown. 2010. Human GLTP: Three distinct<br />
functions for the three tryptophans in a novel peripheral amphitropic fold. Biophys. J. 99,<br />
2626-2635.<br />
Kamlekar, R.K., D.K.Simanshu, Y.G.Gao, R.Kenoth, H.M.Pike, F.G.Prendergast,<br />
L.Malinina, J.G.Molotkovsky, S.Y.Venyaminov, D.J.Patel, <strong>and</strong> R.E.Brown. 2013. The<br />
<strong>glycolipid</strong> transfer protein (GLTP) domain of phosphoinositol 4-phosphate adaptor<br />
protein-2 (FAPP2): structure drives preference for simple neutral glycosphingolipids.<br />
Biochim. Biophys. Acta 1831, 417-427.<br />
Kanekura, K., I.Nishimoto, S.Aiso, <strong>and</strong> M.Matsuoka. 2006. Characterization of amyotrophic<br />
lateral sclerosis-linked P56S mutation of vesicle-associated membrane protein-associated<br />
protein B (VAPB/ALS8). J. Biol. Chem. 281, 30223-30233.<br />
Kawano, M., K.Kumagai, M.Nishijima, <strong>and</strong> K.Hanada. 2006. Efficient trafficking of<br />
ceramide from the endoplasmic reticulum to the Golgi apparatus requires a VAMPassociated<br />
protein-interacting FFAT motif of CERT. J. Biol. Chem. 281, 30279-30288.<br />
Kenoth, R., D.K.Simanshu, R.K.Kamlekar, H.M.Pike, J.G.Molotkovsky, L.M.Benson,<br />
H.R.Bergen, III, F.G.Prendergast, L.Malinina, S.Y.Venyaminov, D.J.Patel, <strong>and</strong><br />
R.E.Brown. 2010. Structural determination <strong>and</strong> tryptophan fluorescence of heterokaryon<br />
incompatibility C2 protein (HET-C2), a fungal <strong>glycolipid</strong> transfer protein (GLTP),<br />
provide novel insights into <strong>glycolipid</strong> specificity <strong>and</strong> membrane interaction by the GLTP<br />
fold. J. Biol. Chem. 285, 13066-13078.<br />
Kjellberg, M.A. <strong>and</strong> P.Mattjus. 2013. Glycolipid transfer protein expression is affected by<br />
glycosphingolipid synthesis. PLoS. One. 8, e70283.<br />
Kumagai, K., M.Kawano, F.Shinkai-Ouchi, M.Nishijima, <strong>and</strong> K.Hanada. 2007. Interorganelle<br />
trafficking of ceramide is regulated by phosphorylation-dependent cooperativity between<br />
the PH <strong>and</strong> START domains of CERT. J. Biol. Chem. 282, 17758-17766.<br />
Accepted Manuscript<br />
22<br />
Page 23 of 34
Lannert, H., K.Gorgas, I.Meissner, F.T.Wiel<strong>and</strong>, <strong>and</strong> D.Jeckel. 1998. Functional organization<br />
of the Golgi apparatus in glycosphingolipid biosynthesis. Lactosylceramide <strong>and</strong><br />
subsequent glycosphingolipids are formed in the lumen of the late Golgi. J. Biol. Chem.<br />
273, 2939-2946.<br />
Lenoir, M., U.Coskun, M.Grzybek, X.Cao, S.B.Buschhorn, J.James, K.Simons, <strong>and</strong><br />
M.Overduin. 2010. Structural basis of wedging the Golgi membrane by FAPP pleckstrin<br />
homology domains. EMBO Rep. 11, 279-284.<br />
Levine, T. <strong>and</strong> C.Loewen. 2006. Inter-organelle membrane contact sites: through a glass,<br />
darkly. Curr. Opin. Cell Biol. 18, 371-378.<br />
Levine, T.P. <strong>and</strong> S.Munro. 2002. Targeting of Golgi-specific pleckstrin homology domains<br />
involves both PtdIns 4-kinase-dependent <strong>and</strong> -independent components. Curr. Biol. 12,<br />
695-704.<br />
Li, X.M., M.L.Malakhova, X.Lin, H.M.Pike, T.Chung, J.G.Molotkovsky, <strong>and</strong> R.E.Brown.<br />
2004. Human <strong>glycolipid</strong> transfer protein: probing conformation using fluorescence<br />
spectroscopy. Biochemistry 43, 10285-10294.<br />
Lin, X., P.Mattjus, H.M.Pike, A.J.Windebank, <strong>and</strong> R.E.Brown. 2000. Cloning <strong>and</strong> expression<br />
of <strong>glycolipid</strong> transfer protein from bovine <strong>and</strong> porcine brain. J. Biol. Chem. 275, 5104-<br />
5110.<br />
Lingwood, C.A. 1996. Aglycone modulation of <strong>glycolipid</strong> receptor function. Glycoconj. J. 13,<br />
495-503.<br />
Lingwood, C.A. 2011. Glycosphingolipid functions. Cold Spring Harb. Perspect. Biol. 3.<br />
Litvak, V., N.Dahan, S.Ramach<strong>and</strong>ran, H.Sabanay, <strong>and</strong> S.Lev. 2005. Maintenance of the<br />
diacylglycerol level in the Golgi apparatus by the Nir2 protein is critical for Golgi<br />
secretory function. Nat. Cell Biol. 7, 225-234.<br />
Loewen, C.J., A.Roy, <strong>and</strong> T.P.Levine. 2003. A conserved ER targeting motif in three families<br />
of lipid binding proteins <strong>and</strong> in Opi1p binds VAP. EMBO J. 22, 2025-2035.<br />
Malakhova, M.L., L.Malinina, H.M.Pike, A.T.Kanack, D.J.Patel, <strong>and</strong> R.E.Brown. 2005. Point<br />
mutational analysis of the lig<strong>and</strong>ing site in human <strong>glycolipid</strong> transfer protein:<br />
Functionality of the complex. J. Biol. Chem. 280, 26312-26320.<br />
Malinina, L., M.L.Malakhova, A.T.Kanack, M.Lu, R.Abagyan, R.E.Brown, <strong>and</strong> D.J.Patel.<br />
2006. The lig<strong>and</strong>ing of <strong>glycolipid</strong> transfer protein is controlled by <strong>glycolipid</strong> acyl<br />
structure. PLoS. Biol. 4, e362.<br />
Malinina, L., M.L.Malakhova, A.Teplov, R.E.Brown, <strong>and</strong> D.J.Patel. 2004. Structural basis for<br />
glycosphingolipid transfer specificity. Nature 430, 1048-1053.<br />
Mattjus, P. 2009. Glycolipid transfer proteins <strong>and</strong> membrane interaction. Biochim. Biophys.<br />
Acta 1788, 267-272.<br />
Mattjus, P., A.Kline, H.M.Pike, J.G.Molotkovsky, <strong>and</strong> R.E.Brown. 2002. Probing for<br />
preferential interactions among sphingolipids in bilayer vesicles using the <strong>glycolipid</strong><br />
transfer protein. Biochemistry 41, 266-273.<br />
Mattjus, P., H.M.Pike, J.G.Molotkovsky, <strong>and</strong> R.E.Brown. 2000. Charged membrane surfaces<br />
impede the protein-mediated transfer of glycosphingolipids between phospholipid<br />
bilayers. Biochemistry 39, 1067-1075.<br />
Metz, R.J. <strong>and</strong> N.S.Radin. 1980. Glucosylceramide uptake protein from spleen cytosol. J.<br />
Biol. Chem. 255, 4463-4467.<br />
Metz, R.J. <strong>and</strong> N.S.Radin. 1982. Purification <strong>and</strong> properties of a cerebroside transfer protein.<br />
J. Biol. Chem. 257, 12901-12907.<br />
Mikitova, V. <strong>and</strong> T.P.Levine. 2012. Analysis of the key elements of FFAT-like motifs<br />
identifies new proteins that potentially bind VAP on the ER, including two AKAPs <strong>and</strong><br />
FAPP2. PLoS. One. 7, e30455.<br />
Accepted Manuscript<br />
23<br />
Page 24 of 34
Neumann, S., M.Opacic, R.W.Wechselberger, H.Sprong, <strong>and</strong> M.R.Egmond. 2008. Glycolipid<br />
transfer protein: clear structure <strong>and</strong> activity, but enigmatic function. Adv. Enzyme Regul.<br />
48, 137-151.<br />
Nhek, S., M.Ngo, X.Yang, M.M.Ng, S.J.Field, J.M.Asara, N.D.Ridgway, <strong>and</strong> A.Toker. 2010.<br />
Regulation of oxysterol-binding protein Golgi localization through protein kinase D-<br />
mediated phosphorylation. Mol. Biol. Cell 21, 2327-2337.<br />
Nishimura, A.L., M.Mitne-Neto, H.C.Silva, A.Richieri-Costa, S.Middleton, D.Cascio, F.Kok,<br />
J.R.Oliveira, T.Gillingwater, J.Webb, P.Skehel, <strong>and</strong> M.Zatz. 2004. A mutation in the<br />
vesicle-trafficking protein VAPB causes late-onset spinal muscular atrophy <strong>and</strong><br />
amyotrophic lateral sclerosis. Am. J. Hum. Genet. 75, 822-831.<br />
Nylund, M., C.Fortelius, E.K.Palonen, J.G.Molotkovsky, <strong>and</strong> P.Mattjus. 2007. Membrane<br />
curvature effects on <strong>glycolipid</strong> transfer protein activity. Langmuir 23, 11726-11733.<br />
Nylund, M., M.A.Kjellberg, J.G.Molotkovsky, H.S.Byun, R.Bittman, <strong>and</strong> P.Mattjus. 2006.<br />
Molecular features of phospholipids that affect <strong>glycolipid</strong> transfer protein-mediated<br />
galactosylceramide transfer between vesicles. Biochim. Biophys. Acta 1758, 807-812.<br />
Nylund, M. <strong>and</strong> P.Mattjus. 2005. Protein mediated <strong>glycolipid</strong> transfer is inhibited FROM<br />
sphingomyelin membranes but enhanced TO sphingomyelin containing raft like<br />
membranes. Biochim. Biophys. Acta 1669, 87-94.<br />
Ohvo-Rekilä, H. <strong>and</strong> P.Mattjus. 2011. Monitoring <strong>glycolipid</strong> transfer protein activity <strong>and</strong><br />
membrane interaction with the surface plasmon resonance technique. Biochim. Biophys.<br />
Acta 1808, 47-54.<br />
Olsen, J.V., M.Vermeulen, A.Santamaria, C.Kumar, M.L.Miller, L.J.Jensen, F.Gnad, J.Cox,<br />
T.S.Jensen, E.A.Nigg, S.Brunak, <strong>and</strong> M.Mann. 2010. Quantitative phosphoproteomics<br />
reveals widespread full phosphorylation site occupancy during mitosis. Sci. Signal. 3, ra3.<br />
Patki, V., D.C.Lawe, S.Corvera, J.V.Virbasius, <strong>and</strong> A.Chawla. 1998. A functional PtdIns(3)Pbinding<br />
motif. Nature 394, 433-434.<br />
Petersen, N.H., L.V.McKinney, H.Pike, D.Hofius, A.Zakaria, P.Brodersen, M.Petersen,<br />
R.E.Brown, <strong>and</strong> J.Mundy. 2008. Human GLTP <strong>and</strong> mutant forms of ACD11 suppress cell<br />
death in the Arabidopsis acd11 mutant. FEBS J. 275, 4378-4388.<br />
Quinn, P.J. 2011. The structure of complexes between phosphatidylethanolamine <strong>and</strong><br />
glucosylceramide: a matrix for membrane rafts. Biochim. Biophys. Acta 1808, 2894-<br />
2904.<br />
Quinn, P.J. 2012. Lipid-lipid interactions in bilayer membranes: married couples <strong>and</strong> casual<br />
liaisons. Prog. Lipid Res. 51, 179-198.<br />
Rao, C.S., T.Chung, H.M.Pike, <strong>and</strong> R.E.Brown. 2005. Glycolipid transfer protein interaction<br />
with bilayer vesicles: modulation by changing lipid composition. Biophys. J. 89, 4017-<br />
4028.<br />
Saita, S., M.Shirane, T.Natume, S.Iemura, <strong>and</strong> K.I.Nakayama. 2009. Promotion of neurite<br />
extension by protrudin requires its interaction with vesicle-associated membrane proteinassociated<br />
protein. J. Biol. Chem. 284, 13766-13777.<br />
Saito, S., H.Matsui, M.Kawano, K.Kumagai, N.Tomishige, K.Hanada, S.Echigo, S.Tamura,<br />
<strong>and</strong> T.Kobayashi. 2008. Protein phosphatase 2Cepsilon is an endoplasmic reticulum<br />
integral membrane protein that dephosphorylates the ceramide transport protein CERT to<br />
enhance its association with organelle membranes. J. Biol. Chem. 283, 6584-6593.<br />
Samygina, V.R., B.Ochoa-Lizarralde, A.N.Popov, A.Cabo-Bilbao, F.Goni-de-Cerio,<br />
J.G.Molotkovsky, D.J.Patel, R.E.Brown, <strong>and</strong> L.Malinina. 2013. Structural insights into<br />
lipid-dependent reversible dimerization of human GLTP. Acta Crystallogr. D. Biol.<br />
Crystallogr. 69, 603-616.<br />
Accepted Manuscript<br />
24<br />
Page 25 of 34
Samygina, V.R., A.N.Popov, A.Cabo-Bilbao, B.Ochoa-Lizarralde, F.Goni-de-Cerio, X.Zhai,<br />
J.G.Molotkovsky, D.J.Patel, R.E.Brown, <strong>and</strong> L.Malinina. 2011. Enhanced selectivity for<br />
sulfatide by engineered human <strong>glycolipid</strong> transfer protein. Structure. 19, 1644-1654.<br />
Saravanan, R.S., E.Slabaugh, V.R.Singh, L.J.Lapidus, T.Haas, <strong>and</strong> F.Br<strong>and</strong>izzi. 2009. The<br />
targeting of the oxysterol-binding protein ORP3a to the endoplasmic reticulum relies on<br />
the plant VAP33 homolog PVA12. Plant J. 58, 817-830.<br />
Saupe, S., C.Descamps, B.Turcq, <strong>and</strong> J.Begueret. 1994. Inactivation of the Podospora<br />
anserina vegetative incompatibility locus het-c, whose product resembles a <strong>glycolipid</strong><br />
transfer protein, drastically impairs ascospore production. Proc. Natl. Acad. Sci. U. S. A<br />
91, 5927-5931.<br />
Scheffzek, K. <strong>and</strong> S.Welti. 2012. Pleckstrin homology (PH) like domains - versatile modules<br />
in protein-protein interaction platforms. FEBS Lett. 586, 2662-2673.<br />
Schnabel, D., M.Schroder, <strong>and</strong> K.S<strong>and</strong>hoff. 1991. Mutation in the sphingolipid activator<br />
protein 2 in a patient with a variant of Gaucher disease. FEBS Lett. 284, 57-59.<br />
Schulze, H., T.Kolter, <strong>and</strong> K.S<strong>and</strong>hoff. 2009. Principles of lysosomal membrane degradation:<br />
Cellular topology <strong>and</strong> biochemistry of lysosomal lipid degradation. Biochim. Biophys.<br />
Acta 1793, 674-683.<br />
Simanshu, D.K., R.K.Kamlekar, D.S.Wijesinghe, X.Zou, X.Zhai, S.K.Mishra,<br />
J.G.Molotkovsky, L.Malinina, E.H.Hinchcliffe, C.E.Chalfant, R.E.Brown, <strong>and</strong> D.J.Patel.<br />
2013. Non-vesicular trafficking by a ceramide-1-phosphate transfer protein regulates<br />
eicosanoids. Nature.<br />
Simonsen, A. <strong>and</strong> H.Stenmark. 2001. PX domains: attracted by phosphoinositides. Nat. Cell<br />
Biol. 3, E179-E182.<br />
Skehel, P.A., K.C.Martin, E.R.K<strong>and</strong>el, <strong>and</strong> D.Bartsch. 1995. A VAMP-binding protein from<br />
Aplysia required for neurotransmitter release. Science 269, 1580-1583.<br />
Stenmark, H., R.Aasl<strong>and</strong>, B.H.Toh, <strong>and</strong> A.D'Arrigo. 1996. Endosomal localization of the<br />
autoantigen EEA1 is mediated by a zinc-binding FYVE finger. J. Biol. Chem. 271,<br />
24048-24054.<br />
Tessier, M. <strong>and</strong> J.R.Woodgett. 2006. Role of the Phox homology domain <strong>and</strong> phosphorylation<br />
in activation of serum <strong>and</strong> glucocorticoid-regulated kinase-3. J. Biol. Chem. 281, 23978-<br />
23989.<br />
Tidhar, R. <strong>and</strong> A.H.Futerman. 2013. The complexity of sphingolipid biosynthesis in the<br />
endoplasmic reticulum. Biochim. Biophys. Acta 1833, 2511-2518.<br />
Tomishige, N., K.Kumagai, J.Kusuda, M.Nishijima, <strong>and</strong> K.Hanada. 2009. Casein kinase<br />
I{gamma}2 down-regulates trafficking of ceramide in the synthesis of sphingomyelin.<br />
Mol. Biol. Cell 20, 348-357.<br />
Tuuf, J. <strong>and</strong> P.Mattjus. 2007. Human <strong>glycolipid</strong> transfer protein-Intracellular localization <strong>and</strong><br />
effects on the sphingolipid synthesis. Biochim. Biophys. Acta 1771, 1353-1363.<br />
Tuuf, J., L.Wistbacka, <strong>and</strong> P.Mattjus. 2009. The <strong>glycolipid</strong> transfer protein interacts with the<br />
vesicle-associated membrane protein-associated protein VAP-A. Biochem. Biophys. Res.<br />
Commun. 388, 395-399.<br />
Vieira, O.V., P.Verkade, A.Manninen, <strong>and</strong> K.Simons. 2005. FAPP2 is involved in the<br />
transport of apical cargo in polarized MDCK cells. J. Cell Biol. 170, 521-526.<br />
Vihervaara, T., M.Jansen, R.L.Uronen, Y.Ohsaki, E.Ikonen, <strong>and</strong> V.M.Olkkonen. 2011.<br />
Cytoplasmic oxysterol-binding proteins: sterol sensors or transporters? Chem. Phys.<br />
Lipids 164, 443-450.<br />
Wagner, S.A., P.Beli, B.T.Weinert, M.L.Nielsen, J.Cox, M.Mann, <strong>and</strong> C.Choudhary. 2011. A<br />
proteome-wide, quantitative survey of in vivo ubiquitylation sites reveals widespread<br />
regulatory roles. Mol. Cell Proteomics. 10, M111.<br />
Accepted Manuscript<br />
25<br />
Page 26 of 34
Warnock, D.E., M.S.Lutz, W.A.Blackburn, W.W.Young, Jr., <strong>and</strong> J.U.Baenziger. 1994.<br />
Transport of newly synthesized glucosylceramide to the plasma membrane by a non-<br />
Golgi pathway. Proc. Natl. Acad. Sci. U. S. A 91, 2708-2712.<br />
Weir, M.L., H.Xie, A.Klip, <strong>and</strong> W.S.Trimble. 2001. VAP-A binds promiscuously to both v-<br />
<strong>and</strong> tSNAREs. Biochem. Biophys. Res. Commun. 286, 616-621.<br />
West, G., M.Nylund, J.P.Slotte, <strong>and</strong> P.Mattjus. 2006. Membrane interaction <strong>and</strong> activity of<br />
the <strong>glycolipid</strong> transfer protein. Biochim. Biophys. Acta 1748, 1732-1742.<br />
West, G., Y.Nymalm, T.T.Airenne, H.Kidron, P.Mattjus, <strong>and</strong> T.T.Salminen. 2004.<br />
Crystallization <strong>and</strong> X-ray analysis of bovine <strong>glycolipid</strong> transfer protein. Acta Crystallogr.<br />
D. Biol. Crystallogr. 60, 703-705.<br />
West, G., L.Viitanen, C.Alm, P.Mattjus, T.A.Salminen, <strong>and</strong> J.Edqvist. 2008. Identification of<br />
a glycosphingolipid transfer protein GLTP1 in Arabidopsis thaliana. FEBS J. 275, 3421-<br />
3437.<br />
Wong, M., R.E.Brown, Y.Barenholz, <strong>and</strong> T.E.Thompson. 1984. Glycolipid transfer protein<br />
from bovine brain. Biochemistry 23, 6498-6505.<br />
Wright, C.S., Q.Zhao, <strong>and</strong> F.Rastinejad. 2003. Structural analysis of lipid complexes of GM2-<br />
activator protein. J. Mol. Biol. 331, 951-964.<br />
Wyles, J.P., C.R.McMaster, <strong>and</strong> N.D.Ridgway. 2002. Vesicle-associated membrane proteinassociated<br />
protein-A (VAP-A) interacts with the oxysterol-binding protein to modify<br />
export from the endoplasmic reticulum. J. Biol. Chem. 277, 29908-29918.<br />
Yamada, K., A.Abe, <strong>and</strong> T.Sasaki. 1986. Glycolipid transfer protein from pig brain transfers<br />
<strong>glycolipid</strong>s with beta-linked sugars but not with alpha-linked sugars at the sugar-lipid<br />
linkage. Biochim. Biophys. Acta 879, 345-349.<br />
Yamada, K., A.Abe, <strong>and</strong> T.Sasaki. 1985. Specificity of the <strong>glycolipid</strong> transfer protein from<br />
pig brain. J. Biol. Chem. 260, 4615-4621.<br />
Yamada, K. <strong>and</strong> T.Sasaki. 1982a. A rat brain cytosol protein which accelerates the<br />
translocation of galactosylceramide, lactosylceramide <strong>and</strong> glucosylceramide between<br />
membranes. Biochim. Biophys. Acta 687, 195-203.<br />
Yamada, K. <strong>and</strong> T.Sasaki. 1982b. Rat liver <strong>glycolipid</strong> transfer protein. A protein which<br />
facilitates the translocation of mono- <strong>and</strong> dihexosylceramides from donor to acceptor<br />
liposomes. J. Biochem. (Tokyo) 92, 457-464.<br />
Zhai, X., M.L.Malakhova, H.M.Pike, L.M.Benson, H.R.Bergen, III, I.P.Sugar, L.Malinina,<br />
D.J.Patel, <strong>and</strong> R.E.Brown. 2009. Glycolipid acquisition by human <strong>glycolipid</strong> transfer<br />
protein dramatically alters intrinsic tryptophan fluorescence: insights into <strong>glycolipid</strong><br />
binding affinity. J. Biol. Chem. 284, 13620-13628.<br />
Zhai, X., W.E.Momsen, D.A.Malakhov, I.A.Boldyrev, M.M.Momsen, J.G.Molotkovsky,<br />
H.L.Brockman, <strong>and</strong> R.E.Brown. 2013. GLTP-fold interaction with planar<br />
phosphatidylcholine surfaces is synergistically stimulated by phosphatidic acid <strong>and</strong><br />
phosphatidylethanolamine. J. Lipid Res. 54, 1103-1113.<br />
Zou, X., T.Chung, X.Lin, M.L.Malakhova, H.M.Pike, <strong>and</strong> R.E.Brown. 2008. Human<br />
<strong>glycolipid</strong> transfer protein (GLTP) genes: organization, transcriptional status <strong>and</strong><br />
evolution. BMC. Genomics 9, 72.<br />
Zou, X., Y.Gao, V.R.Ruvolo, T.L.Gardner, P.P.Ruvolo, <strong>and</strong> R.E.Brown. 2011. Human<br />
<strong>glycolipid</strong> transfer protein gene (GLTP) expression is regulated by Sp1 <strong>and</strong> Sp3:<br />
involvement of the bioactive sphingolipid ceramide. J. Biol. Chem. 286, 1301-1311.<br />
Accepted Manuscript<br />
26<br />
Page 27 of 34
Figure legends<br />
Fig. 1. A) Schematic presentation of GLTP, FAPP1 <strong>and</strong> FAPP2 sequences, showing the<br />
important PH <strong>and</strong> <strong>glycolipid</strong> binding domains. The binding sites of PI4P within the PH<br />
domain <strong>and</strong> the FFAT-like motifs within GLTP are highlighted. B) The consensus sequence<br />
of the FFAT motif is shown in bold (Loewen et al., 2003), together with the FFAT-like<br />
sequences in GLTP <strong>and</strong> FAPP2 in italic letters (Mikitova <strong>and</strong> Levine, 2012), <strong>and</strong> a second<br />
FFAT-like motif in the GLTP sequence written in normal letters (Tuuf et al., 2009).<br />
Figure 2. Structural features of GLTP. (A) A typical all alpha-helical GLTP fold with the 8<br />
helices colored as follows; helix 1 (dark blue), helix 2 (light blue), helix 3 (dark green), helix<br />
4 (light green), helix 5 (yellow), helix 6 (dark yellow), helix 7 (orange) <strong>and</strong> helix 8 (red). The<br />
location of the FFAT-like motif is in purple on helix 1. (B) The GSL sugar-binding pocket of<br />
GLTP (1BV7) with gangliosides GM 3 bound. Note that only the glucose residue of the<br />
ganglioside GM 3 head group is shown. The GSL molecule has the sphingosine chain directed<br />
outwards <strong>and</strong> a long amide linked acyl chain inside the tunnel. Amino acid residues Asp-48,<br />
Asn-52 (above) <strong>and</strong> His-140 (below) are shown in different shads of blue, <strong>and</strong> Trp-96 in red,<br />
are all responsible for stabilizing the <strong>glycolipid</strong>s inside the structure. Additional hydrophobic<br />
interactions between amino acids aligning the interior of the tunnel <strong>and</strong> with the acyl chain(s)<br />
of the lipid further locks the lipid in position. The yellow lines represent hydrogen bonds<br />
between the three amino acids <strong>and</strong> the GSL.<br />
Figure 3. (A) Structural presentation of sphingosine-in <strong>and</strong> sphingosine-out forms of the<br />
Accepted Manuscript<br />
glycosphingolipid binding of GLTP. The far end opening of the hydrophobic cavity can be<br />
seen in the left picture (black arrow). GalCer with N-linked linoleic acid (18:2 Δ 9,12 ) <strong>and</strong> with<br />
the sphingosine chain in a sphingosine-out conformation (blue) next to GalCer with N-linked<br />
oleic acid (18:1 Δ 9 ) in a sphingosine-in conformation (yellow). (B) Hydrophobicity scale<br />
coloring of GLTP amino acids, red most hydrophobic <strong>and</strong> white hydrophilic. A putative<br />
membrane interaction domain with high hydrophobicity is indicated with a black arrow (left<br />
picture. The right picture is an 180 degree rotation, revealing a red area originating from the<br />
amino acids lining the far end of the hydrophobic tunnel (white arrow).<br />
27<br />
Page 28 of 34
Figure 4. Schematic illustration of the hypothetical GLTP (in cyan) positioning at a<br />
phosphtidylcholine membrane surface with a glucosylceramide bound to GLTP (only one<br />
membrane leaflet is shown), the right picture is an 180 degree rotation. The yellow colored<br />
amino acids indicate the location of the FFAT-like domain in GLTP. The phosphatidylcholine<br />
molecules are rendered in the van der Waals volume. Lipid coloring: black, carbon; red,<br />
oxygen; orange, phosphorus <strong>and</strong> white, hydrogen. The approximate length of a membrane<br />
leaflet is indicated.<br />
Accepted Manuscript<br />
28<br />
Page 29 of 34
Accepted Manuscrip<br />
Figure 1<br />
Page 30 of 34
Figure 2<br />
Accepted Manuscript<br />
Page 31 of 34
Accepted Manuscrip<br />
Figure 3<br />
Page 32 of 34
Accepted Manuscrip<br />
Figure 4<br />
Page 33 of 34
Accepted Manuscrip<br />
Graphical Abstract<br />
Page 34 of 34