The JI
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     
 

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Related articles in The JI
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Request Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Ettinger, R. A.
Right arrow Articles by Kwok, W. W.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Ettinger, R. A.
Right arrow Articles by Kwok, W. W.
The Journal of Immunology, 2006, 176: 1988-1998.
Copyright © 2006 by The American Association of Immunologists

Allelic Variation in Key Peptide-Binding Pockets Discriminates between Closely Related Diabetes-Protective and Diabetes-Susceptible HLA-DQB1*06 Alleles1

Ruth A. Ettinger2,*,{dagger}, George K. Papadopoulos{ddagger}, Antonis K. Moustakas§, Gerald T. Nepom*,{dagger} and William W. Kwok*,{dagger}

* Benaroya Research Institute at Virginia Mason, Seattle, WA 98101 and {dagger} Department of Immunology, University of Washington School of Medicine, Seattle, WA 98195; {ddagger} Laboratory of Biochemistry and Biophysics, Faculty of Agricultural Technology, Epirus Institute of Technology, Arta, Greece; and § Department of Biological Agriculture, Technological Educational Institute of Ionian Islands, Argostoli, Cephallonia, Greece


   Abstract
Top
 Abstract
Introduction
 Materials and Methods
 Results
 Discussion
 Disclosures
 References
 
HLA-DQA1*0102-DQB1*0602 is associated with protection against type 1 diabetes (T1D). A similar allele, HLA-DQA1*0102-DQB1*0604, contributes to T1D susceptibility in certain populations but differs only at seven amino acids from HLA-DQA1*0102-DQB1*0602. Five of these polymorphisms are found within the peptide-binding groove, suggesting that differences in peptide binding contribute to the mechanism of their association with T1D. In this study, we determine the peptide-binding motif for HLA-DQA1*0102-DQB1*0604 allelic protein (DQ0604) in comparison to the established HLA-DQA1*0102-DQB1*0602 (DQ0602) motif using binding assays with model peptides from T1D autoantigens and homology modeling using the coordinates of the DQ0602-hypocretin 1–13 crystal structure. The peptide binding preferences were deduced with a peptide from insulin that bound both with a 2- to 3-fold difference in avidity using the same amino acids in the peptide as anchors. Peptide binding differences directly influenced by the polymorphisms in or nearby pockets 1, 6, and 9 were observed. In pocket 1, DQ0604 was better able to accommodate aromatic residues due to the beta86 and beta87 polymorphisms. A negatively charged amino acid was preferred by DQ0604 in pocket 6 due to the positively charged beta30His. In pocket 9, DQ0604 preferred aromatic amino acids due to the beta9 and beta30 polymorphisms and had low tolerance of acidic residues. beta57Val in DQ0604 functions differently than beta57Ala, in that it pushes {alpha}76Arg outside of the pocket, preventing the formation of a salt bridge with an acidic amino acid in the peptide. This study furthers our understanding of the structure-function relationships of MHC class II polymorphisms.


   Introduction
Top
 Abstract
 Introduction
Materials and Methods
 Results
 Discussion
 Disclosures
 References
 
Human histocompatibility leukocyte Ag on chromosome 6p21 is the genetic factor with the strongest linkage to type 1 (insulin-dependent) diabetes (T1D)3 (1). Within the HLA gene complex, fine mapping with genetic markers indicates that the peak of association is as close as 85 kb centromeric to the MHC class II gene DQB1 and 160 kb centromeric to DRB1 (2). A primary association with the polymorphic MHC class II genes including DQB1, DQA1, and DRB1 has been suggested; however, the importance of each particular allele for T1D susceptibility or protection, may depend on the haplotype (3, 4, 5). The DRB1-DQA1-DQB1 haplotypes exhibit a hierarchy of genetic associations with T1D with relative risk ratios in the general range of <0.1 to >10, depending on the study (6). Although the distribution of MHC class II haplotypes and association with T1D varies between populations, in general, DR4-DQA1*0301-DQB1*0302 and DRB1*0301-DQA1*0501-DQB1*0201 are major susceptibility alleles, particularly when both haplotypes are carried, while DRB1*1501-DQA1*0102-DQB1*0602 is strongly associated with T1D protection (6, 7, 8, 9, 10).

The negative association with DQA1*0102-DQB1*0602 is age dependent, with an estimated risk of 0.03 at birth increasing to 1.1 at 35 years of age in the Swedish population (11). The protection associated with DQA1*0102-DQB1*0602 is dominant over the susceptibility of all other DQ haplotypes, including DQA1*0301-DQB1*0302 (7), particularly in the 0- to 14.99-year-old age group (11, 12). This protection in children is thought to occur even in the presence of islet cell Abs (13, 14, 15, 16). In contrast, DQA1*0102-DQB1*0604, which differs from DQA1*0102-DQB1*0602 at only seven amino acids, is not associated with T1D protection in most studies (6, 8, 10). In older individuals, the risk associated with DQB1*0604 increases (11). The presence of DQA1*0102-DQB1*0604, in combination with DQA1*0301-DQB1*0302, results in a synergistically increased effect on T1D susceptibility in Swedish and Finnish populations (12, 17).

A thorough understanding of the structure-function relationship attributed to the polymorphic differences in HLA-DQ is needed to deduce mechanisms of T1D susceptibility and protection linked to these proteins. In this study, we have focused on the polymorphisms that distinguish the DQA1*0102-DQB1*0604 allelic protein (DQ0604) from DQA1*0102-DQB1*0602 (DQ0602). DQB1*0602 and DQB1*0604 differ at seven amino acids; six are within the first external domain at beta9, beta30, beta57, beta70, beta86, and beta87, and one within the second external domain at beta130. The DQA1*0102 gene is in linkage disequilibrium with both DQB1*0602 and DQB1*0604; thus, the same {alpha}-chain complexes with either DQB1*0602 or DQB1*0604 polypeptide and the antigenic peptide, functioning as a ligand for the TCR.

Recently, a crystal structure for DQ0602 bound to a hypocretin peptide was determined (18). Hypocretin is an Ag that is absent in narcolepsy, a disease which is positively associated with DQB1*0602. The hypocretin 1–13 peptide side chains at amino acids 3, 6, 8, and 11 occupy the peptide-binding pockets 1, 4, 6, and 9, respectively. These amino acids correspond to a Leu in pocket 1, Thr in pocket 4, Val in pocket 6, and Ala in pocket 9, residues which are preferred by DQ0602 for peptide binding in these pockets (19). Two key structural features were hypothesized to be important for dominant protection against T1D. First, it was determined from the crystallography data that pocket 6 of the DQ0602-hypocretin 1–13 structure was ~42 Å3 bigger than the corresponding pocket in the DQ8 (DQ0302)-insulin B 9–23 structure (DQ8 is encoded by DQA1*0301-DQB1*0302) (20). Thus, it was suggested that the volume of pocket 6 in DQ0604 will fall in between the large volume of DQ0602 and the small volume of DQ0302. Furthermore, the large volume in pocket 6 of DQ0602 would allow more peptides to bind to DQ0602, implying that presentation of an expanded peptide repertoire is important for dominant protection. Second, crystallography of DQ0602 and DQ0302 showed that the noncovalent interactions between the beta57 polymorphism, invariant {alpha}76Arg, and the peptide residue in pocket 9 differed, as seen in all MHC class II-peptide complexes thus far. It was predicted that DQ0302 and DQ0604 would have similar peptide specificities due to the presence of a beta57non-Asp amino acid in both.

In this study, the peptide binding preferences for DQ0604 are determined using the insulin B 5–15 peptide that was used to determine the peptide-binding motif for DQ0602 (19). Structural interactions that control these preferences are proposed using homology molecular modeling based on the DQ0602 crystal structure. These experiments further the hypotheses set forth regarding the peptide-binding interactions that distinguish DQ0602 and DQ0604 and therefore lead to dominant protection and susceptibility in T1D. Our results demonstrate a role for the polymorphisms at beta9, beta30, beta57, beta86, and beta87 in controlling the peptide binding preferences in pockets 1, 6, and 9. These data further our understanding of DQ0602 and DQ0604, in conjunction with our previous studies identifying beta57 as a key determinant in {alpha}beta heterodimer stability and beta70 in TCR recognition (21, 22).


   Materials and Methods
Top
 Abstract
 Introduction
 Materials and Methods
Results
 Discussion
 Disclosures
 References
 
Cell lines

Homozygous EBV-transformed B lymphocyte cell lines (B-LCLs) from the Tenth International Histocompatibility Workshop include MGAR (DQA1*0102-DQB1*0602), AMAI (DQA1*0102-DQB1*0602), EMJ (DQA1*0102-DQB1*0604), WT47 (DQA1*0102-DQB1*0604), HOM-2 (DQA1*0101-DQB1*0501), KT3 (DQA1*0301-DQB1*0401), AMALA (DQA1*0501-DQB1*0301), JVM (DQA1*0501-DQB1*0301), EK (DQA1*0101-DQB1*0503), TEM (DQA1*0101-DQB1*0503), BSM (DQA1*0301-DQB1*0302), DEU (DQA1*0301-DQB1*0301), COX (DQA1*0501-DQB1*0201), OMW (DQA1*0103-DQB1*0603), and CB6B (DQA1*0103-DQB1*0603) (23). Other EBV-transformed B-LCLs used in this study include LG2 (DQA1*0101-DQB1*0501), HAS-15 (DQA1*0301-DQB1*0401), KAS011 (DQA1*0102-DQB1*0502), AZH (DQA1*0102-DQB1*0502), PRIESS (DQA1*0301-DQB1*0302), PF97387 (DQA1*0301-DQB1*0301), and MAT (DQA1*0501-DQB1*0201), and they were HLA typed by high-resolution oligonucleotide typing (Puget Sound Blood Center, Seattle, WA). BLS-1 was a gift from J. Lee (Memorial Sloan-Kettering Cancer Center, New York, NY) (24). BLS-1 is a HLA class II-null EBV-transformed B-LCL generated from the cells of a patient with bare lymphocyte syndrome (BLS). Cells were grown in IMDM with L-glutamine and 25 mM HEPES buffer (Invitrogen Life Technologies) supplemented with 10% FBS, 1 mM sodium pyruvate, 50 U/ml penicillin, and 50 µg/ml streptomycin.

Peptides

Peptides were synthesized at Benaroya Research Institute at Virginia Mason with an Applied Biosystems 432 Peptide Synthesizer (PerkinElmer) or synthesized by Genemed Synthesis. Peptides were biotinylated as described previously (25). The molecular mass of each peptide was determined by mass spectrometry (Protein and Carbohydrate Structure Facility, University of Michigan, Ann Arbor, MI and Genemed Synthesis). The sequence of all peptides is included in the tables and figures.

Peptide-binding assays

Measurement of binding of 10 µM biotinylated peptide to HLA-DQ on paraformaldehyde-fixed EBV-transformed B-LCLs was performed as described elsewhere (19). DQ0602 and DQ0604 were affinity purified from MGAR and EMJ EBV-transformed B-LCLs, respectively, as described for DQ0602 (19). Protein concentration was determined by Bradford microassay with BSA as the standard. Measurement of binding of 0.001–10 µM biotinylated peptide to purified DQ0602 and DQ0604 at pH 5.4 was essentially as described for DQ0602, except that, in the present report, BSA was added to the reaction mixture to determine nonspecific binding (19). In some experiments, the pH of the reaction mixture was varied from pH 4.0 to 8.0 using different 150 mM citrate-phosphate buffers for the pH range of 4.0–7.0, and 200 mM phosphate buffers for the pH range of 7.1–8.0, prepared as described previously (26). The binding of peptides to purified DQ0602 and DQ0604 was also evaluated in a competition assay with nonbiotinylated peptides competing for binding with 1.0 µM biotinylated insulin B 5–15 as described elsewhere (19). The nonbiotinylated peptides were tested at 0.1, 0.3, 1.0, 3.0, 10, 30, and 100 µM final concentrations in triplicate. The fluorescence values were plotted vs peptide concentration and the peptide concentration at which 50% inhibition in binding of the biotinylated peptide (IC50) occurred was extrapolated from the curve. Relative binding values were calculated by dividing the IC50 for insulin B 5–15 by the IC50 for the analog peptide. Binding of each peptide to DQ0602 and DQ0604 was compared within the same assay.

Molecular modeling

Models of DQ0602 and DQ0604 with the insulin B 5–15 peptide and its variants were prepared on a Silicon Graphics Indigo 2 work station using the program Insight II, version 2000 (Accelrys), essentially as previously described (22, 27). The crystal structure of DQ0602 complexed to the hypocretin 1–13 peptide was used as a base molecule for all of the simulation studies (18). The numbering scheme developed by Fremont et al. (28) for mouse H2-A MHC class II molecules was used here for HLA-DQ alleles. Essentially, it yields identical numbering in equivalent structural locations in all MHC class II alleles.


   Results
Top
 Abstract
 Introduction
 Materials and Methods
 Results
Discussion
 Disclosures
 References
 
Insulin B 5–15 peptide binds to DQ0604

The binding of insulin peptides, covering the primary structure of insulin, to several HLA-DQ alleles was previously described (19). The HLA-DQ alleles examined consisted of a panel representing at least one allele from each HLA-DQ serological specificity except DQ9. Insulin B 1–15 was identified as the insulin peptide that bound with highest avidity to DQ0602. In addition, insulin B 1–15 bound poorly to most HLA-DQ allelic proteins tested. The minimal binding epitope was insulin B 6–14, with insulin B 5–15 conferring maximal binding. The binding of insulin B 5–15 to DQ0602 led to the question of whether it would bind other HLA-DQ molecules within the DQ6 serological specificity. DQ0604 and the DQA1*0103-DQB1*0603 allelic protein expressed on EBV-transformed B-LCLs were examined in comparison with DQ0602 and eight other HLA-DQ alleles (Fig. 1). For each HLA-DQ genotype, two different EBV-transformed B-LCLs were included in the assay. Fig. 1 shows that DQ0604 bound insulin B 5–15 at a level 1.7- to 3.9-fold reduced compared with DQ0602, whereas the DQA1*0103-DQB1*0603 allelic protein bound insulin B 5–15 very poorly, at a level slightly above background. In agreement with our previous data (19), there is some binding to the DQA1*0101-DQB1*0501 allelic protein, which is 4.9- to 5.7-fold reduced compared with DQ0602. This binding was reduced to almost background for the DQ5 allelic proteins DQA1*0102-DQB1*0502 and DQA1*0101-DQB1*0503. Since DQB1*0502 complexes with the same {alpha}-chain as DQB1*0602 and DQB1*0604, the inability of the DQA1*0102-DQB1*0502 allelic protein to bind insulin B 5–15 is conferred by the polymorphisms in the beta-chain.


Figure 1
View larger version (27K):
[in this window]
[in a new window]
  		FIGURE 1. Binding of insulin B 5–15 to HLA-DQ alleles on EBV-transformed B-LCLs. Biotinylated insulin B 5–15 peptide (10 µM) was incubated with 1.5 x 106 paraformaldehyde-fixed B-LCLs in buffer at pH 5.4. Cells were washed to remove unbound peptide and lysed. HLA-DQ was extracted from the cell lysate by binding to SPVL3 coated on a microtiter plate. Biotinylated peptide bound to HLA-DQ was detected by fluorescence using a europium-labeled streptavidin system. Data are the means ± SD of triplicate determinations. The HLA-DQ genotypes of the homozygous B-LCLs are indicated on the y-axis and are presented by the cells listed in "Materials and Methods." The "no DQ" sample is the BLS-1 cell line.

 
The binding avidity of insulin B 5–15 to DQ0604 in comparison to DQ0602 was further characterized in binding assays with affinity-purified MHC proteins. Binding of biotinylated insulin B 5–15 to both DQ0602 and DQ0604 increased in a linear fashion from 0.001 to almost 1 µM (Fig. 2A). Within the linear range, binding of insulin B 5–15 to DQ0604 was 1.8- to 2.9-fold lower at each concentration (0.001, 0.01, 0.1, and 1 µM) compared with DQ0602. A similar difference in avidity of the insulin B 5–15 peptide for DQ0602 and DQ0604 was seen in a competition assay where different concentrations of nonbiotinylated insulin B 5–15 are competing for binding with 1.0 µM biotinylated insulin B 5–15 (Fig. 2B). The IC50 of insulin B 5–15 was always 2- to 3-fold lower for DQ0602 compared with DQ0604 and ranged from 1.5 to 3.4 µM for DQ0602 and 3.3 to 13 µM for DQ0604 (n = 10).


Figure 2
View larger version (22K):
[in this window]
[in a new window]
  		FIGURE 2. Comparison of binding of insulin B 5–15 to DQ0602 and DQ0604. A, Biotinylated insulin B 5–15 (0.001, 0.01, 0.1, 1.0, 10 µM) was incubated with 25 nM purified DQ0602, DQ0604, or BSA in pH 5.4 buffer for 48 h at 37°C. Results are plotted using a linear scale. B, Insulin B 5–15 (0.1, 0.3, 1.0, 3.0, 10, 30, 100 µM) was incubated with 25 nM purified DQ0602 or DQ0604 in the presence of biotinylated insulin B 5–15 (1.0 µM) in pH 5.4 buffer for 48 h at 37°C. Results are plotted using a log scale. C, Biotinylated insulin B 5–15 (1.0 µM) was incubated with 25 nM purified DQ0602 or DQ0604 in purified peptide binding buffer with varying pH (4.0, 4.5, 5.0, 5.5, 6.0, 6.5, 7.0, 7.5, 8.0) for 48 h at 37°C. Nonspecific binding was determined by incubation with BSA. A–C, After 48 h, HLA-DQ was captured from the reaction mixture on microtiter plates coated with SPVL3. Samples were washed to remove unbound peptide and HLA-DQ-bound biotinylated peptide was detected by fluorescence using a europium-labeled streptavidin system. Data are the means ± SD of triplicate determinations.

 
The effect of pH on the binding of insulin B 5–15 to DQ0602 and DQ0604 was examined. Interestingly, the pH optimum and shape of the profiles are different for DQ0602 and DQ0604 (Fig. 2C). Binding of insulin B 5–15 to DQ0602 was less affected by pH, with a 2.1-fold difference in pH over the range of 4.0–8.0 and a pH optimum of 5.5. In contrast, DQ0604 bound insulin B 5–15 poorly at pH 4.0, binding increased 19.4-fold in a linear fashion to pH 6.0, and then reached a plateau at an optimal level from pH 6.0 to pH 8.0. It should be noted that all other peptide binding experiments in this article were done at one acidic pH, that of pH 5.4, due to the acidic characteristics of the endosomal/lysosomal compartments where peptide-MHC class II binding generally occurs (29). However, other pH values may be physiologically relevant in the Ag processing and presentation pathway and as a result will regulate peptide binding levels.

Peptide-binding motif derivation for DQ0604 in comparison to DQ0602

The anchor residues for binding of insulin B 5–15 to DQ0602 and DQ0604 were defined by examining the effect of radical substitutions on binding. Arginine substitutions were used to map the primary anchors because of the effectiveness of using positively charged substitutions in identifying the amino acids in a peptide that are required for binding to HLA-DQ (19, 30, 31). Fig. 3 shows the binding of insulin B 5–15 Arg-substituted peptides to DQ0602 and DQ0604 expressed on EBV-transformed B-LCLs. The Arg-substituted peptides that showed dramatically reduced binding compared with the unsubstituted (wild-type) peptide were insulin B 5–15 6R, 8R, 9R, 11R, and 14R peptides. The pattern of binding of the insulin B 5–15 Arg-substituted peptides was very similar for DQ0602 and DQ0604, with one exception; insulin B 5–15 7R bound DQ0602 2.0-fold better while binding DQ0604 1.8-fold less well than wild type. The sequence of insulin B 5–15 is shown in Fig. 4 and the anchor positions at amino acids 6, 8, 9, 11, and 14 are designated as relative positions(p) 1 (p1), 3 (p3), 4 (p4), 6 (p6), and 9 (p9).


Figure 3
View larger version (15K):
[in this window]
[in a new window]
  		FIGURE 3. Effect of single Arg substitutions in insulin B 5–15 on binding to DQ0602 and DQ0604. Biotinylated insulin B 5–15 (wild-type (WT))- and insulin B 5–15 Arg (R)-substituted peptides (10 µM) were incubated with 1.5 x 106 paraformaldehyde-fixed B-LCLs in pH 5.4 buffer for 18 h at 37°C. Cells were washed to remove unbound peptide. HLA-DQ-bound biotinylated peptide was measured as described in figure legend 1. The HLA-DQ genotype of the homozygous B-LCLs is BLS-1 (none), MGAR (A1*0102-B1*0602), and EMJ (A1*0102-B1*0604). Data are the means ± SD of triplicate determinations.

 

Figure 4
View larger version (26K):
[in this window]
[in a new window]
  		FIGURE 4. DQ0602 and DQ0604 peptide-binding motifs. Motifs were determined within the context of the insulin B 5–15 peptide. Anchor positions are designated with arrows. Amino acids accepted for binding (relative binding capacity: >0.2) in each position are listed in order of preference from the best binder to the least well-accepted binder. In parentheses is the one letter code for the specific amino acid tested.

 
The amino acid requirements in each of the primary anchor positions for binding to DQ0602 and DQ0604 were investigated. Single amino acid substitutions in insulin B 5–15 were made to represent all of the general classes of amino acid side chains: glycine (Gly), aliphatic (Ala, Val, Leu, Ile), hydroxyl (Ser, Thr), amide (Asn, Gln), aromatic (Phe, Tyr, Trp), sulfur-containing (Cys, Met), cyclic imino acid (Pro), acidic (Asp, Glu), and basic (Lys, Arg, His). The effect of these substitutions in insulin B 5–15 at p1, p3, p4, p6, and p9 on relative binding to DQ0602 and DQ0604 is shown in Fig. 5. Fig. 4 summarizes the peptide-binding motif for DQ0602 and DQ0604. The amino acids are listed in order of preference from best tolerated to least well tolerated, with only those with a relative binding capacity >0.2 included.


Figure 5
View larger version (43K):
[in this window]
[in a new window]
  		FIGURE 5. Relative binding capacity of insulin B 5–15 analog peptides for DQ0602 and DQ0604. A, Amino acid 6 (p1)-substituted insulin B 5–15 peptides. B, Amino acid 8 (p3)-substituted insulin B 5–15 peptides. C, Amino acid 9 (p4)-substituted insulin B 5–15 peptides. D, Amino acid 11 (p6)-substituted insulin B 5–15 peptides. E, Amino acid 14 (p9)-substituted insulin B 5–15 peptides. Insulin B 5–15 analog peptides (0.1, 0.3, 1.0, 3.0, 10, 30, 100 µM) were incubated with 25 nM purified DQ0602 or DQ0604 and biotinylated insulin B 5–15 (1.0 µM) in pH 5.4 buffer for 48 h at 37°C. HLA-DQ-bound biotinylated peptide was measured as described in figure legend 2. Relative binding values were calculated by dividing the IC50 for insulin B 5–15 by the IC50 for the analog peptide. Inequality sign (<) indicates peptides that had an IC50 >100 µM and thus a relative binding capacity less than the value indicated on the graph. Two-fold or greater differences in relative binding capacity between DQ0602 and DQ0604 were designated with a star symbol (*) for peptides that prefer DQ0602 and an accent symbol ({wedge}) for peptides that prefer DQ0604.

 
The general motif derivation analysis indicated that DQ0602 is more tolerant of peptide substitutions than DQ0604. Fourteen insulin B 5–15-substituted peptides bind better to DQ0602 in pockets 1, 3, 4, and 6 (indicated by a star symbol in Fig. 5), while only the p1Phe and the p6Asp peptides were preferred by DQ0604 (indicated by an accent symbol). Pocket 9 has a more even distribution in number of amino acids preferred, with DQ0602 preferring Gly- and Ser-substituted insulin B 5–15 and DQ0604 preferring Leu- and aromatic (Tyr, Phe, Trp)-substituted insulin B 5–15 peptides.

Molecular modeling of insulin B 5–15 with DQ0602 and DQ0604

Models of DQ0602 and DQ0604 with insulin B 5–15 were created based on the crystal coordinates of DQ0602 (18). The structural simulation of the binding of insulin B 5–15 peptide in the groove of DQ0602 and DQ0604 revealed the means by which the different anchoring pocket residues are responsible for the different peptide residue preferences at these positions. The simulation of these complexes shows that this particular peptide fits well into the groove of either DQ allele (Fig. 6). There is no observed structural distortion, in comparison to the crystal structure of DQ0602, and the interactions of the antigenic peptide backbone with specific invariant residues (e.g., {alpha}62Asn, {alpha}68His, {alpha}69Asn, beta81His, beta82Asn) of both the DQ0602 and DQ0604 allele are present and unaltered, just as in all other MHC class II crystal structures recorded to date. There is a slight difference in the orientation of insulin B 5–15 peptide residues p1 to p5 between DQ0602 and DQ0604, and the surface of the proteins differ due to the beta70 polymorphism (Table I). The surface exposure of residue beta70Arg of DQ0604 is considerable (Fig. 6A), which most likely makes its contact with any cognate TCR unavoidable.


Figure 6
View larger version (71K):
[in this window]
[in a new window]
  		FIGURE 6. Molecular simulation of the binding of insulin B 5–15 peptide at pH 5.4 to DQ0604 (A) and DQ0602 (B), based on the crystal structure of DQ0602. TCR view of the {alpha}1beta1 domain in the same orientation and identical rendering pattern is shown for DQ0604 and DQ0602 in complex with the insulin B 5–15 peptide. The {alpha}1beta1 domain is shown in a transparent van der Waals surface representation, color coded according to the electrostatic potential (positive, blue; negative, red; neutral, gray). The peptide is shown in space filling mode (atomic color code: carbon, green; oxygen, red; nitrogen, blue; hydrogen, white), with the two histidines at p1 and p5 as positively charged residues (pH 5.4). The six amino acids that are different in this domain between DQ0602 and DQ0604, except beta30His, are shown in stick form with the same atomic color code as for the peptide, except that carbon atoms are in orange. The secondary structure of the domain is shown for orientation purposes, with {alpha} helix in red, beta-pleated sheet in turquoise, and random coil in gray. The depiction of the structures was performed with the program WebLabViewer of Accelrys.

 

View this table:
[in this window]
[in a new window]
  		Table I. The seven polymorphic amino acids that distinguish DQA1*0102-DQB1*0602 and DQA1*0102-DQB1*0604 and the corresponding amino acids at these positions in DQA1*0501-DQB1*0201 and DQA1*0301-DQB1*0302

 
The detailed examination of the simulated structures identifies polymorphisms that distinguish DQ0604 and DQ0602 in pockets 1, 6, and 9. In pocket 1 of DQ0604, the presence of beta86Gly renders this pocket more spacious and able to accommodate aromatic residues (Phe) even better than large aliphatic ones (Leu) (data not shown). In addition, the beta87 polymorphism, Tyr in DQ0604 vs Phe in DQ0602, allows for stronger aromatic-aromatic interactions because of the substantially larger permanent dipole moment of Tyr compared with Phe. Hence, DQ0604 was observed to better accommodate aromatic residues at p1 (Fig. 5A). By the same argument, smaller residues that are only slightly hydrophilic (Ala, Thr, Cys) are not well accommodated in p1 of DQ0604 compared with the same pocket in DQ0602, because water molecules are not energetically favored to occupy the space between the MHC class II residues lining the pocket and such a p1 residue of the antigenic peptide.

In pocket 6, while both alleles accept aliphatic residues, the DQ0604 allele also binds a p6Asp-substituted insulin B 5–15 peptide (Fig. 5D). This preference in all probability is due to the presence of beta30His at the "mouth" of pocket 6, which is positively charged at endosomal pH (Fig. 7). Furthermore, this negatively charged Asp at p6, would keep beta30His positively charged by induction, even at the slightly alkaline extracellular pH, once the DQ0604-peptide complex is on the cell membrane and exposed to the extracellular environment. The volume of this pocket in DQ0604 appears to be similar to that in DQ0602 (data not shown) and is sufficient in both for accommodating a bulky residue such as Leu, but not an inflexible aromatic hydrophobic residue such as Phe (Fig. 5D). The hydrophobicity of both pockets is a result of {alpha}65Val and {alpha}66Ala in addition to the methylene groups of the three Asn residues ({alpha}11, {alpha}62, and {alpha}69). The latter two form hydrogen bonds with the peptide backbone so their terminal amides do not determine the character of the pocket. The lower tolerance for substitutions at p6 by DQ0604 compared with DQ0602 is attributed to the polymorphisms at beta30 and beta9, which decrease the hydrophobic character of pocket 6 in DQ0604.


Figure 7
View larger version (45K):
[in this window]
[in a new window]
  		FIGURE 7. Details of the modeled structure of pocket 6 of DQ0604 with the p6Asp (D) substituted insulin B 5–15 peptide. The orientation is rotated by 20° with respect to the y-axis running through the middle of the photograph (right-hand side coming up, left-hand side going into the plane of the paper), in comparison to the photograph in Fig. 6, so that the interaction of the charged imidazole ring of beta30His with p6Asp can be shown in detail. Color code for surfaces and atoms of the antigenic peptide and the residues of the {alpha}1beta1 domain of DQ0604 is identical to that of Fig. 6.

 
Pocket 9 of DQ0604 accepts well aromatic and aliphatic residues, yet has low tolerance of acidic residues (Glu > Asp) (Fig. 5E). This ability is apparently due to the presence of three polymorphic residues at positions beta9, beta30, and beta57 (Table I), the combination of which shapes such an environment in pocket 9, that the binding of the said residues is promoted. Of the three aromatic amino acids, Tyr is best-suited as an anchor because its hydroxyl group interacts favorably with the hydroxyl groups of beta9Tyr and beta37Tyr and the imidazole group of beta30His via hydrogen bonding and aromatic-aromatic interactions (Fig. 8). The "walls" of the pocket are hydrophobic, hence the preference for hydrophobic residues of shorter length (Ala > Tyr > Leu > Phe > Trp). The low preference given to acidic residues in this pocket stems from the presence of beta57Val that pushes {alpha}76Arg outside the pocket, so that the formation of an energy-stabilizing salt bridge between this Arg residue and p9Asp is no longer tenable. By contrast, p9Glu, being one methylene group longer, can interact weakly with {alpha}76Arg, but the distance between the interacting group is longer because of the interfering beta57Val.


Figure 8
View larger version (47K):
[in this window]
[in a new window]
  		FIGURE 8. Details of the modeled structure of pocket 9 of DQ0604 with the p9Tyr (Y) substituted insulin B 5–15 peptide. Color code for surfaces and atoms of the antigenic peptide and the residues of the {alpha}1beta1 domain of DQ0604 is identical to that of Fig. 6.

 
Motif prediction and testing with glutamic acid decarboxylase 65 (GAD65), IA-2, and proinsulin peptides

Previously, in deducing the peptide-binding motif for DQ0602, the predictive power of the motif was examined by scanning GAD65, IA-2 (also called ICA512), and proinsulin for the DQ0602 peptide-binding motif (19). Autoantibodies against these three beta cell Ags are highly predictive of T1D, with 90% of newly diagnosed subjects having one of more of these autoantibodies (32). This analysis identified a total of 24 peptides, which were synthesized with an additional amino acid flanking each side of the 9-mer core. Of these peptides, 79% (19 of 24) bound to DQ0602 with an IC50 <100 µM (range, 0.7–90 µM). The same approach was applied to the DQ0604 motif. Twenty peptides were identified in human GAD65 (33), IA-2 (34), and proinsulin (35) that contain the peptide-binding motif deduced for DQ0604. Table II shows the IC50 of each of these peptides for binding to DQ0604 by competition with biotinylated insulin B 5–15. Of these peptides, 65% (13 of 20) bound to DQ0604 (IC50 < 100 µM). The IC50 value of the binders ranged from 1.9 to 86 µM, with insulin B 5–15 binding with an IC50 of 4.8 µM. When an entire Ag (HSV-2 VP16) was studied with a panel of 60 overlapping 20-mer, there was a 75% agreement between the presence and absence of the DQ0604 motif and peptide binding (data not shown). It is not known why certain peptides that contain the motif determined herein do not bind DQ0604 with an avidity comparable to insulin B 5–15. Factors that may play a role, but have not been investigated, are the presence of proline and glycine in the peptide and the effect of neighboring non-anchor amino acids on peptide binding.


View this table:
[in this window]
[in a new window]
  		Table II. Binding of human GAD65, IA-2, and proinsulin peptides selected with the DQ0604 peptide-binding motif to DQ0604 and DQ0602a

 
The binding specificity of the human GAD65, IA-2, and proinsulin peptides selected with the DQ0604 (Table II) and DQ0602 (Table III) motifs were tested by binding to DQ0602 for the DQ0604 motif peptides and DQ0604 for the DQ0602 motif peptides. Because of the overlap in motif, many of the peptides bound to both DQ0604 and DQ0602. This was particularly evident when the DQ0604 motif was used to select the peptides, since all of the peptides except for two had an IC50 for binding that was <2-fold different for DQ0602 and DQ0604. One of these two peptides, IA-2 586–596 actually had a lower IC50 for DQ0602. This was not surprising because this peptide was selected with the DQ0602 motif and demonstrated the highest binding avidity for DQ0602 of all of the tested peptides (Table III). The lone peptide that bound specifically to DQ0604 was GAD65 339–349. In contrast, the peptides selected with the DQ0602 motif identified six peptides (IA2 586–596, IA2 499–509, GAD65 334–344, GAD65 396–406, IA2 335–345, and GAD65 86–96) with more than 7-fold lower IC50 for DQ0602 compared with DQ0604.


View this table:
[in this window]
[in a new window]
  		Table III. Binding of human GAD65, IA-2, and proinsulin peptides selected with the DQ0602 peptide-binding motif to DQ0602 and DQ0604a

 
Testing of the pocket predictions with substituted GAD65 peptides

Two of the peptides in Tables II and III that demonstrated substantial binding specificity for either DQ0602 or DQ0604 were further analyzed. The objective was to test whether the deduced motifs could be used to turn a nonbinding peptide into a binder.

The first peptide examined was GAD65 339–349, which showed relatively strong binding to DQ0604 (IC50 = 4 µM with a relative binding capacity of 1.2 compared with insulin B 5–15) and no detectable binding to DQ0602 (IC50 > 100 µM). Direct binding of biotinylated GAD65 339–349 to DQ0602 and DQ0604 showed a binding differential similar to the competition assay (Fig. 9A). This peptide would be expected to bind DQ0604 well because of anchors that are well accepted at p1, p3, p4, p6, and p9 (relative binding capacity: 1.0, 1.0, 0.72, 0.28, 0.48, respectively) and not bind DQ0602 because of the Asp at p6 which is barely tolerated (relative binding capacity: 0.033). Also, the Leu at p9 is less well tolerated by DQ0602 (relative binding capacity: 0.15) relative to DQ0604 (relative binding capacity: 0.48). To test these predictions, amino acid substitutions were made in GAD65 339–349 at p6 and p9 and these peptides were tested for binding to DQ0602 (Fig. 9B). A dramatically increased binding to DQ0602, particularly at higher concentrations (>0.1 µM), was observed by changing p6 from an Asp to a Leu; changing p9 from a Leu to an Ala did not allow for binding, implying that the Asp at p6 may have a strong inhibitory effect on binding to DQ0602. Finally, changing both p6 and p9 to Leu and Ala resulted in increased binding compared with p6Leu alone at the low peptide concentrations of 0.001 µM (p6Leu, 1,939 ± 105; p6Leup9Ala, 5,366 ± 244) and 0.01 µM (p6Leu, 13,583 ± 527; p6Leup9Ala, 30,944 ± 980).


Figure 9
View larger version (30K):
[in this window]
[in a new window]
  		FIGURE 9. Testing of DQ0602 and DQ0604 pocket predictions with substituted GAD65 339–349 and 334–344 peptides. A, Binding of biotinylated GAD65 339–349 to DQ0602 and DQ0604. B, Binding of biotinylated GAD65 339–349 analog peptides to DQ0602. C, Binding of biotinylated GAD65 334–344 to DQ0602 and DQ0604. D, Binding of biotinylated GAD65 334–344 analog peptides to DQ0604. Biotinylated peptide (0.001, 0.01, 0.1, 1.0, 10 µM) was incubated with 25 nM purified DQ0602, DQ0604, or BSA in purified peptide binding buffer (pH 5.4) for 48 h at 37°C. HLA-DQ-bound biotinylated peptide was measured as described in figure legend 2. Nonspecific binding for each peptide was determined by incubation with BSA and was subtracted from the total binding with DQ0602 and DQ0604. Data are the means ± SD of triplicate determinations.

 
The second peptide examined was GAD65 334–344, which bound DQ0602 with an IC50 of 2.6 µM (relative binding capacity of 0.73 compared with insulin B 5–15) and DQ0604 with an IC50 > 100 µM. The substantially better binding of biotinylated GAD65 334–344 to DQ0602 compared with DQ0604 is shown in Fig. 9C. GAD65 334–344 contains a motif well-suited to binding DQ0602 with strong anchors in all five positions (relative binding capacity: 0.49, 1.0, 0.44, 0.52, and 1.0, respectively). This peptide is not expected to bind DQ0604 because of the Thr in p1 that is not tolerated (relative binding capacity: <0.1), and the Val in p6 that is less well accepted by DQ0604 (relative binding capacity: 0.21) than DQ0602 (relative binding capacity: 0.52). The p1 and p6 amino acids were both substituted with Leu, an amino acid that is better suited for DQ0604 pockets 1 and 6. Fig. 9D shows the binding of single-substituted peptides, p1Leu and p6Leu, and the double-substituted peptide to DQ0604. Each of the single-substituted peptides substantially increased binding to DQ0604. The double-substituted peptide resulted in increased binding at each concentration relative to the single-substituted peptides; however, the increase was not additive. These results substantiate the value of the motifs in predicting the peptide side chains preferred in pockets 1, 6, and 9 for DQ0602 and DQ0604.


   Discussion
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
Disclosures
 References
 
Using the techniques of peptide binding (direct and competitive) to allele-specific HLA-DQ-bearing B-LCLs and to purified HLA-DQ6 alleles, we have deciphered the peptide-binding motif of the DQ0604 allele (T1D susceptible/neutral). We in parallel compared this motif to that of the already established T1D-resistant allele DQ0602. The recently solved crystal structure of the latter molecule, in complex with the hypocretin 1–13 narcolepsy peptide (18), allowed us to very reliably model the binding of the insulin B 5–15 peptide to DQ0602 and DQ0604. This insulin peptide binds well (IC50 < 5 µM) to both DQ0602 and DQ0604.

The motif of the diabetes-susceptible/neutral DQ0604 allele differs in important respects from that of the closely related, yet diabetes-resistant, allele DQ0602: 1) a more pronounced preference by DQ0604 for aromatic residues than for DQ0602 in pocket 1. 2) An acidic residue is favored at p6 of DQ0604. 3) Aliphatic (Ala > Leu) and aromatic (Tyr > Phe > Trp) residues are favored at p9 for DQ0604 over small residues (Ala > Ser, Gly) for DQ0602. 4) Contrary to expectations, accommodation of acidic residues at p9 of DQ0604 (a beta57non-Asp allele) is very poor. This is in contrast to the well-characterized beta57non-Asp alleles, in which the positively charged {alpha}76Arg in the absence of beta57Asp is unopposed and thus prefers to interact with a negatively charged amino acid.

The differences in the peptide-binding motif of these two related HLA-DQ allelic protein can be directly attributed to five of the seven residue differences between DQ0602 and DQ0604, namely, beta9, beta30, beta57, beta86, and beta87, that are found in or nearby pockets 1, 6, and 9 (Table I). In pocket 1, the change in the residue preference of DQ0602 and DQ0604 is a consequence of the difference in residue at beta86 (Ala in DQ0602 vs Gly in DQ0604) and beta87 (Phe in DQ0602 vs Tyr in DQ0604). Thus, the absence of a methyl side chain at beta86 in pocket 1 of DQ0604 allows for the better accommodation of p1Phe. The Tyr at beta87, which is nearby pocket 1, has a significant permanent dipole moment associated with it, which allows it to interact well with p1Trp/Tyr/Phe (in decreasing order). It should be noted, however, that in HLA-DQ allelic protein, and likewise H2-A alleles, the p1 pocket does not show the exquisite dichotomy in preference to aromatic/aliphatic residues exhibited by HLA-DR allelic proteins that depend on the beta86Gly/Val dimorphism (20, 28, 36). beta86 is almost exclusively a dimorphism in DRB1, with beta86Asp in DRB1*1316 the exception, whereas in DQB1, beta86 is a trimorphism of Glu/Ala/Gly with 21, 15, and 9 alleles, respectively (37).

Pocket 6 of DQ0604 is controlled by beta30His, which is found only in four other HLA-DQ alleles besides DQB1*0604. beta30His is positively charged at endosomal pH and would remain so at extracellular pH in the presence of an acidic amino acid, thus in all probability forming a salt bridge with a p6Asp. The other DQB1 alleles that have His at beta30 are DQB1*0501, *0502, *0503, and *0504, thus they would also be expected to favor a negatively charged amino acid in p6. In agreement with this hypothesis, we have found that peptide binding to the allelic protein encoded by DQA1*0101-DQB1*0501 is enhanced by the presence of a Glu at p6 (R. Ettinger and W. Kwok, unpublished observation).

The pocket 9 preferences of DQ0604 are a result of the combination of beta9Tyr, beta30His, and beta57Val. The presence of beta57Val in DQ0604 in combination with {alpha}72Ile and {alpha}73Met results in an aliphatic barrier that a p9Asp residue from an antigenic peptide cannot easily overcome. The situation is slightly better with p9Glu due to the extra methylene group that confers greater ability of the terminal side chain carboxylate to reach the {alpha}76Arg guanidine side chain. This same aliphatic barrier along with residues beta37Tyr and beta61Trp promote van der Waals interactions between the side chains of these DQ0604 amino acids above and bulky aliphatic and aromatic p9 residues (Tyr > Leu > Phe > Trp). In particular, p9Tyr is further favored because in its presence the beta9Tyr and beta37Tyr residues form hydrogen bonds via their hydroxyl groups to the hydroxyl group p9Tyr. The presence of beta9Phe in DQ0602 precludes the latter possibility. In addition, there are aromatic-aromatic interactions among all of the above aromatic residues and beta30His, in pairs and extended partners. Thus far, the only other HLA-DQ molecules to accommodate hydrophobic residues in the p9 pocket, are the DQ2 allelic proteins, DQA1*0501-DQB1*0201 and DQA1*0201-DQB1*0202, which accept Tyr, Phe, and Trp with equal propensity (31, 38, 39). This is because of the combination of several unique, smaller, and flexible residues (beta30Ser, beta37Leu, beta57Ala) at close proximity to and in the p9 pocket. DQ0604 is the first beta57non-Asp HLA-DQ molecule that does not tolerate well acidic residues at p9 (30, 38, 40, 41, 42). In the three HLA-DQ molecules that have Ala at beta57 and have established peptide-binding motifs, all accept an acidic residue at p9. Our data suggest that beta57Val results in a poor propensity for p9 acidic residues and is supported by the observation that acidic residues are absent at p9 for DRA*0101-DRB1*0701 and DRA*0101-DRB1*1201 allelic proteins, two HLA-DR beta57Val-positive alleles for which a p9 has been identified and preferences examined (43, 44). The combination of beta9Tyr, beta30His, and beta57Val, which distinguish DQ0604 from DQ0602 in pocket 9, is found in only four other HLA-DQ alleles (DQB1*0501, *0608, *0617, and *0621) (37). It will be interesting to test whether the pocket 9 preferences in these alleles mimic DQ0604 or whether other polymorphisms will play a role producing different preferences.

In the first homology models of HLA-DQ alleles, it was hypothesized that pocket 4 is prominent and widely accommodating (45, 46). The large size has been verified since in the different DQ allele crystal structures (18, 20, 47). In the DQ0602 crystal structure, pocket 4 was the largest with a volume of ~100 Å3 (18). The experiments in this work show that both DQ0602 and DQ0604 can accommodate a number of substitutions well, including the bulky hydrophobic amino acid Phe; however, there are clear preferences in that Cys, Pro, Asp, and Arg are not tolerated by either allele. The difference in residue preferences at p4 (Gly and Asn preferred by DQ0602) between the two alleles is probably due to the presence of radically different residues in the vicinity of pocket 4 (beta71Arg vs Gly and beta30His vs Tyr).

Last, both alleles exhibit the peculiar selectivity at p3, first seen in the complex of DR3-CLIP and termed "p3 shelf" (48). In other words, even though this position is exposed to solvent, there is only a select group of residues that can occupy it in both DQ0602 and DQ0604 (Gly > Phe > Leu > Ala). By contrast, of the two serological DQ2 molecules (DQA1*0501/B1*0201 and DQA1*0201/B1*0202), only the latter exhibits the p3 shelf in not allowing Pro, His, Arg, and Lys at this position (38). No structural explanation has been proposed thus far for this phenomenon, which is worth taking into consideration in the design of allele-specific HLA-DQ ligands.

The pockets in the peptide-binding groove of DQ0604 are different from those of the two main alleles conferring susceptibility to T1D, DQA1*0301-B1*0302, and DQA1*0501-B1*0201. Just among the amino acids that are polymorphic between DQ0602 and DQ0604, there are obvious differences among DQ0604, DQ0302, and the DQA1*0501-DQB1*0201 allelic protein (DQ0201) at beta86 and beta87 in pocket 1, beta30 in pocket 6, and beta57 in pocket 9 (Table I). This is reflected in the observed preferences in the peptide-binding motifs for the three allelic proteins (30, 31, 38, 39, 40). Some of the more obvious differences regarding charged and bulky hydrophobic amino acids are that: 1) at p1, DQ0302 and DQ0201 can tolerate a negatively charged amino acid while DQ0604 does not. 2) At p4 and p7, DQ0201 favors negatively charged amino acids while DQ0302 and DQ0604 do not. 3) At p9, DQ0302 favors negatively charged residues and does not tolerate hydrophobic amino acids, while DQ0201 favors aromatic amino acids (Trp = Tyr = Phe) and tolerates negatively charged amino acids (Glu > Asp), and differently still, DQ0604 favors hydrophobic amino acids (Tyr > Phe > Trp) with low tolerance for negatively charged amino acids (Glu > Asp). Despite these differences, overlap in peptide binding specificity between DQ0302 and DQ0201 has been suggested (40, 49). DQ0302 and DQ0201 can bind many of the same peptide epitopes, although it has not been shown that they use the same anchor residues. A common motif was proposed for DQ0302 and DQ0201 based on pool sequencing with the following preferences: p1, bulky hydrophobic, polar residues; p4, aliphatic; p6/7, aliphatic, and p9, negatively charged. This common motif is acceptable for DQ0604 except at p9 where there is a low tolerance for negatively charged amino acids. Further study of additional MHC class II alleles would be helpful in evaluating whether indeed MHC class II T1D susceptibility alleles exhibit greater similarity in peptide binding preferences.

The synergy exhibited in certain populations for susceptibility to T1D by DQ0302 and DQ0604 most likely is not due to formation of DQA1*0301-DQB1*0604 and DQA1*0102-DQB1*0302 trans-heterodimers. These trans-heterodimers are not expected to form due to steric incompatibility resulting from structural differences at {alpha}45–53 and beta85–90, which were noted when the crystal structures of DQ0602 and DQ0302 were compared (18). Experimentally, it has been demonstrated that there are allele-specific restraints that prevent the cell surface expression of certain combinations of DQ{alpha} and DQbeta (50). DQA1*0101 does not pair with DQB1*0201 and DQB1*0302. Similarly, the beta-chain that pairs in cis with DQA1*0101, i.e., DQB1*0501, does not pair with DQA1*0301, DQA1*0401, and DQA1*0501.

The explanation that we can provide for the contrasting roles of the DQ0602 and DQ0604 allelic proteins in the pathogenesis of T1D are a consequence of the differences in antigenic peptide motifs, {alpha}beta heterodimer stability (21, 51), and TCR recognition (22). The dominant protection associated with DQA1*0102-DQbeta1*0602 may originate from the unusual {alpha}beta heterodimer stability of DQ0602 (51). This stability has been attributed to the salt bridge between {alpha}76Arg and beta57Asp plus an extra hydrogen bond that forms between beta57Asp and the main chain amide nitrogen at p9 (18, 21). In addition, the DQ0602 peptide-binding motif appears more accommodating than DQ0604 in pockets 1, 4, and 6 (Figs. 4 and 5). As a result, DQ0602 can bind more peptides, as suggested by the observation that all but one of the DQ0604 selected peptides bound DQ0602 with almost equal avidity (Table II). This is consistent with our previous study (19) in which DQ0602 was capable of binding five overlapping insulin peptides better or with equal avidity compared with seven other HLA-DQ allelic proteins. This presents the possibility that in vivo, DQ0602 can present a more diverse peptide repertoire leading to deletion of autoreactive T cells, immune tolerance, and the denial of these epitopes to T1D-susceptible HLA-DQ allelic proteins.

In the study presented here, there is one epitope from GAD65, IA-2, and proinsulin, that of GAD65 339–349, that binds with high avidity to DQ0604 and very low avidity to DQ0602. However, the overlapping epitope GAD65 334–344 is bound with high avidity to DQ0602 and very low avidity to DQ0604. It is thus possible that the antigenic peptide generated during processing contains both overlapping epitopes, in which case DQ0602 probably binds to the majority of the free peptide, denying the epitope to DQ0604. It is particularly noteworthy that HLA-DQ allelic proteins conferring susceptibility to T1D have also overlapping epitopes in the insulin B chain (core nonamer 13–21 for DQ0302 and 6–14 for DQ0604), while the protective allelic protein DQ0602 binds with higher avidity to the same insulin B 6–14 epitope (Refs.19 and 20 ; Fig. 3). These two cases support an earlier hypothesis that genetic protection is a result of the higher avidity of DQ0602 for the same peptide (52).

In conclusion, the T1D susceptibility associated with DQ0604 is likely to require an Ag-specific set of interactions to target the pancreatic beta cell. Ag-specific interactions are controlled by MHC peptide-binding groove polymorphisms and furthermore by polymorphisms on the surface of the groove such as the beta70Arg/Gly polymorphism, which dictates TCR recognition of the MHC class II-peptide complexes (22). Human GAD65, IA-2, and proinsulin have been found to colocalize to cells in thymus, spleen, and lymph nodes that express MHC class II (53), suggesting that these Ags will be presented by MHC class II proteins and contribute to the shaping of the T cell repertoire. In this study, we determined a peptide-binding motif for DQ0604 that differs from the DQ0602 motif. Although clearly there is significant overlap, the consequence in vivo is likely a different pancreatic beta cell peptide repertoire and subsequent T cell repertoire that is skewed toward recognition of the beta cell when DQ0604 is present. It remains therefore to delve deeper to determine the molecular link between MHC class II alleles associated with T1D susceptibility and destruction of the pancreatic beta cell by the immune system.


   Acknowledgments
 
We thank Patsy Byers for peptide synthesis, Demetrios Kyrkas and Dr. George Bodinas for laboratory assistance, and Dr. Åke Lernmark for critical reading of this manuscript.


   Disclosures
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 Disclosures
References
 
The authors have no financial conflict of interest.


   Footnotes
 
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1 This work was supported by National Institute of Health Grant AI-44443 (to W.W.K.), a Juvenile Diabetes Research Foundation International Grant (to W.W.K.), and a grant from the European Union (EPEAEK II scheme, Third European Union Regional Development Framework for Greece, Program "Archimedes") (to G.K.P.). Back

2 Address correspondence and reprint requests to Dr. Ruth A. Ettinger at her current address, University of Washington, R. H. Williams Laboratory, Box 357710, HSB K-165, 1959 NE Pacific Street, Seattle, WA 98195-7710. E-mail address: ettinger{at}u.washington.edu Back

3 Abbreviations used in this paper: T1D, type 1 diabetes; B-LCL, B lymphocyte cell line; BLS, bare lymphocyte syndrome; p, position; GAD, glutamic acid decarboxylase. Back

Received for publication October 18, 2005. Accepted for publication November 15, 2005.


   References
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 Disclosures
 References
 

  1. Onengut-Gumuscu, S., P. Concannon. 2002. Mapping genes for autoimmunity in humans: type 1 diabetes as a model. Immunol. Rev. 190: 182-194. [Medline]
  2. Herr, M., F. Dudbridge, P. Zavattari, F. Cucca, C. Guja, R. March, R. D. Campbell, A. H. Barnett, S. C. Bain, J. A. Todd, B. P. C. Koeleman. 2000. Evaluation of fine mapping strategies for a multifactorial disease locus: systematic linkage and association analysis of IDDM1 in the HLA region on chromosome 6p21. Hum. Mol. Genet. 9: 1291-1301. [Abstract/Free Full Text]
  3. Thorsby, E., K. S. Rønningen. 1993. Particular HLA-DQ molecules play a dominant role in determining susceptibility or resistance to type 1 (insulin-dependent) diabetes mellitus. Diabetologia 36: 371-377. [Medline]
  4. Kockum, I., R. Wassmuth, E. Holmberg, B. Michelsen, ÅA. Lernmark. 1993. HLA-DQ primarily confers protection and HLA-DR susceptibility in type 1 (insulin-dependent) diabetes studied in population-based affected families and controls. Am. J. Hum. Genet. 53: 150-167. [Medline]
  5. van der Auwera, B., C. van Waeyenberge, F. Schuit, H. Heimberg, C. Vandewalle, F. Gorus, J. Flament, The Belgian Diabetes Registry. 1995. DRB1*0403 protects against IDDM in Caucasians with the high-risk heterozygous DQA1*0301-DQB1*0302/DQA1*0501-DQB1*0201 genotype. Diabetes 44: 527-530. [Abstract]
  6. Rønningen, K. S., A. Spurkland, B. D. Tait, B. Drummond, C. Lopez-Larrea, F. S. Baranda, M. J. Menendez-Diaz, S. Caillat-Zucman, G. Beaurain, H.-J. Garchon, et al 1992. HLA class II associations in insulin-dependent diabetes mellitus among blacks, Caucasoids, and Japanese. K. Tsuji, and M. Aizawa, and T. Sasazuki, eds. HLA 1991 713-722. Oxford University Press,
  7. Baisch, J. M., T. Weeks, R. Giles, M. Hoover, P. Stastny, J. D. Capra. 1990. Analysis of HLA-DQ genotypes and susceptibility in insulin-dependent diabetes mellitus. N. Engl. J. Med. 322: 1836-1841. [Abstract]
  8. Rønningen, K. S., A. Spurrkland, T. Iwe, F. Vartdal, E. Thorsby. 1991. Distribution of HLA-DRB1, -DQA1 and -DQB1 alleles and DQA1-DQB1 genotypes among Norwegian patients with insulin-dependent diabetes mellitus. Tissue Antigens 37: 105-111. [Medline]
  9. Khalil, I., I. Deschamps, V. Lepage, R. al-Daccak, L. Degos, J. Hors. 1992. Dose effect of cis- and trans-encoded HLA-DQ {alpha}beta heterodimers in IDDM susceptibility. Diabetes 41: 378-384. [Abstract]
  10. Cucca, F., F. Muntoni, R. Lampis, F. Frau, L. Argiolas, M. Silvetti, E. Angius, A. Cao, S. De Virgilis, M. Congia. 1993. Combinations of specific DRB1, DQA1, DQB1 haplotypes are associated with insulin-dependent diabetes mellitus in Sardinia. Hum. Immunol. 37: 85-94. [Medline]
  11. Graham, J., I. Kockum, C. B. Sanjeevi, M. Landin-Olsson, L. Nyström, G. Sundkvist, H. Arnqvist, G. Blohme, F. Lithner, B. Littorin, et al 1999. Negative association between type 1 diabetes and HLA DQB1*0602-DQA1*0102 is attenuated with age at onset. Eur. J. Immunogenet. 26: 117-127. [Medline]
  12. Sanjeevi, C. B., M. Landin-Olsson, I. Kockum, G. Dahlquist, Å. Lernmark. 1995. Effects of the second HLA-DQ haplotype on the association with childhood insulin-dependent diabetes mellitus. Tissue Antigens 45: 148-152. [Medline]
  13. Huang, W., J-X. She, A. Muir, D. Laskowska, B. Zorovich, D. Schatz, N. K. Maclaren. 1994. High risk HLA-DR/DQ genotypes for IDD confer susceptibility to autoantibodies but DQB1*0602 does not prevent them. J. Autoimmun. 7: 889-897. [Medline]
  14. Pugliese, A., R. Gianani, R. Moromisato, Z. L. Awdeh, C. A. Alper, H. A. Erlich, R. A. Jackson, G. S. Eisenbarth. 1995. HLA-DQB1*0602 is associated with dominant protection from diabetes even among islet cell antibody-positive first-degree relatives of patients with IDDM. Diabetes 44: 608-613. [Abstract]
  15. Gianani, R., C. F. Verge, R. I. Moromisato-Gianani, L. Yu, Y. J. Zhang, A. Pugliese, G. S. Eisenbarth. 1996. Limited loss of tolerance to islet autoantigens in ICA+ first degree relatives of patients with type 1 diabetes expressing the HLA DQB1*0602 allele. J. Autoimmun. 9: 423-425. [Medline]
  16. Greenbaum, C. J., D. A. Schatz, D. Cuthbertson, A. Zeidler, G. S. Eisenbarth, J. P. Krischer, for the Diabetes Prevention Trial-Type 1 Study Group 2000. Islet cell antibody-positive relatives with human leukocyte antigen DQA1*0102, DQB1*0602: identification by the diabetes prevention trial-type 1. J. Clin. Endocrinol. Metab. 85: 1255-1260. [Abstract/Free Full Text]
  17. Hermann, R., C. S. Bartsocas, Gy. Solt,sz, A. Vazeou, P. Paschou, E. Bozas, A. Malamitsi-Puchner, O. Simell, M. Knip, J. Ilonen. 2004. Genetic screening for individuals at high risk for type 1 diabetes in the general population using HLA class II alleles as disease markers: a comparison between three European populations with variable rates of disease incidence. Diabetes Metab. Res. Rev. 20: 322-329. [Medline]
  18. Siebold, C., B. Hansen, J. R. Wyer, K. Harlos, R. E. Esnouf, A. Svejgaard, J. I. Bell, J. L. Strominger, E. Y. Jones, L. Fugger. 2004. Crystal structure of HLA-DQ0602 that protects against type 1 diabetes and confers strong susceptibility to narcolepsy. Proc. Natl. Acad. Sci. USA 101: 1999-2004. [Abstract/Free Full Text]
  19. Ettinger, R. A., W. W. Kwok. 1998. A peptide binding motif for HLA-DQA1*0102/DQB1*0602, the class II MHC molecule associated with dominant protection in insulin-dependent diabetes mellitus. J. Immunol. 160: 2365-2373. [Abstract/Free Full Text]
  20. Lee, K. H., K. W. Wucherpfennig, D. C. Wiley. 2001. Structure of a human insulin peptide—HLA-DQ8 complex and susceptibility to type 1 diabetes. Nat. Immunol. 2: 501-507. [Medline]
  21. Ettinger, R. A., A. W. Liu, G. T. Nepom, W. W. Kwok. 2000. beta57-Asp plays an essential role in the unique SDS stability of HLA-DQA1*0102/DQB1*0602 {alpha}beta protein dimer, the class II MHC allele associated with protection from insulin-dependent diabetes mellitus. J. Immunol. 165: 3232-3238. [Abstract/Free Full Text]
  22. Reichstetter, S., G. K. Papadopoulos, A. K. Moustakas, E. Swanson, A. W. Liu, S. Beheray, R. A. Ettinger, G. T. Nepom, W. W. Kwok. 2002. Mutational analysis of critical residues determining antigen presentation and activation of HLA-DQ0602 restricted T-cell clones. Hum. Immunol. 63: 185-193. [Medline]
  23. Kimura, A., R-P. Dong, H. Harada, T. Sasazuki. 1992. DNA typing of HLA class II genes in B-lymphoblastoid cell lines homozygous for HLA. Tissue Antigens 40: 5-12. [Medline]
  24. Hume, C. R., L. A. Shookster, N. Collins, R. O’Reilly, J. S. Lee. 1989. Bare lymphocyte syndrome: altered HLA class II expression in B cell lines derived from two patients. Hum. Immunol. 25: 1-11. [Medline]
  25. Kwok, W. W., G. T. Nepom, F. C. Raymond. 1995. HLA-DQ polymorphisms are highly selective for peptide binding interactions. J. Immunol. 155: 2468-2476. [Abstract]
  26. Gomori, G.. 1955. Preparation of buffers for use in enzyme studies. Methods Enzymol. 1: 138-146.
  27. Moustakas, A. K., J. Routsias, G. K. Papadopoulos. 2000. Modelling of the MHC II allele I-Ag7 of NOD mouse: pH-dependent changes in specificity at pockets 9 and 6 explain several of the unique properties of this molecule. Diabetologia 43: 609-624. [Medline]
  28. Fremont, D. H., D. Monnaie, C. A. Nelson, W. A. Hendrickson, E. R. Unanue. 1998. Crystal structure of I-Ak in complex with a dominant epitope of lysozyme. Immunity 8: 305-317. [Medline]
  29. Cresswell, P.. 1994. Assembly, transport, and function of MHC class II molecules. Annu. Rev. Immunol. 12: 259-293. [Medline]
  30. Kwok, W. W., M. E. Domeier, F. C. Raymond, P. Byers, G. T. Nepom. 1996. Allele-specific motifs characterize HLA-DQ interactions with a diabetes-associated peptide derived from glutamic acid decarboxylase. J. Immunol. 156: 2171-2177. [Abstract]
  31. Van de Wal, Y., Y. M. C. Kooy, J. W. Drijfhout, R. Amons, F. Koning. 1996. Peptide binding characteristics of the coeliac disease-associated DQ({alpha}1*0501, beta1*0201) molecule. Immunogenetics 44: 246-253. [Medline]
  32. Notkins, A. L., A. Lernmark. 2001. Autoimmune type 1 diabetes: resolved and unresolved issues. J. Clin. Invest. 108: 1247-1252. [Medline]
  33. Karlsen, A. E., W. A. Hagopian, C. E. Grubin, S. Dube, C. M. Disteche, D. A. Adler, H. Burmeier, S. Mathews, F. J. Grant, D. Foster, A. Lernmark. 1991. Cloning and primary structure of a human islet isoform of glutamic acid decarboxylase from chromosome 10. Proc. Natl. Acad. Sci. USA 88: 8337-8341. [Abstract/Free Full Text]
  34. Lan, M. S., J. Lu, Y. Goto, A. L. Notkins. 1994. Molecular cloning and identification of a receptor-type protein tyrosine phosphatase, IA-2, from human insulinoma. DNA Cell Biol. 13: 505-514. [Medline]
  35. Goeddel, D. V., A. Gray, A. Ullrich. 1980. Nucleotide sequence of human preproinsulin complementary DNA. Science 208: 57-59. [Abstract/Free Full Text]
  36. Stern, L. J., J. H. Brown, T. S. Jardetzky, J. C. Gorga, R. G. Urban, J. L. Strominger, D. C. Wiley. 1994. Crystal structure of the human class II MHC protein HLA-DR1 complexed with an influenza virus peptide. Nature 368: 215-221. [Medline]
  37. Robinson, J., M. J. Waller, P. Parham, N. de Groot, R. Bontrop, L. J. Kennedy, P. Stoehr, S. G. E. Marsh. 2003. IMGT/HLA and IMGT/MHC: sequence databases for the study of the major histocompatibility complex. Nucleic Acids Res. 31: 311-314. [Abstract/Free Full Text]
  38. Van de Wal, Y., Y. M. C. Kooy, J. W. Drijfhout, R. Amons, G. K. Papadopoulos, F. Koning. 1997. Unique peptide binding characteristics of the disease-associated DQ({alpha}1*0501, beta1*0201) vs the non-disease-associated DQ({alpha}1*0201,beta1*0202) molecule. Immunogenetics 46: 484-492. [Medline]
  39. Vartdal, F., B. H. Johansen, T. Friede, C. J. Thorpe, S. Stevanovic, J. E. Eriksen, K. Sletten, E. Thorsby, H.-G. Rammensee, L. M. Sollid. 1996. The peptide binding motif of the disease associated HLA-DQ ({alpha}1*0501/beta1*0201) molecule. Eur. J. Immunol. 26: 2764-2772. [Medline]
  40. Godkin, A., T. Friede, M. Davenport, S. Stevanovic, A. Willis, D. Jewell, A. Hill, H.-G. Rammensee. 1997. Use of eluted peptide sequence data to identify the binding characteristics of peptides to the insulin-dependent diabetes susceptibility allele HLA-DQ8 (DQ 3.2). Int. Immunol. 9: 905-911. [Abstract/Free Full Text]
  41. Kwok, W. W., M. E. Domeier, M. L. Johnson, G. T. Nepom, D. M. Koelle. 1996. HLA-DQB1 codon 57 is critical for peptide binding and recognition. J. Exp. Med. 183: 1253-1258. [Abstract/Free Full Text]
  42. Nepom, B. S., G. T. Nepom, M. Coleman, W. W. Kwok. 1996. Critical contribution of beta chain residue 57 in peptide binding ability of both HLA-DR and -DQ molecules. Proc. Natl. Acad. Sci. USA 93: 7202-7206. [Abstract/Free Full Text]
  43. Gelder, C., M. Davenport, M. Barnardo, T. Bourne, J. Lamb, B. Askonas, A. Hill, K. Welsh. 1998. Six unrelated HLA-DR-matched adults recognize identical CD4+ T cell epitopes from influenza A haemagglutinin that are not simply peptides with high HLA-DR binding affinities. Int. Immunol. 10: 211-222. [Abstract/Free Full Text]
  44. Falk, K., O. Rötzschke, S. Stevanovic, G. Jung, H.-G. Rammensee. 1994. Pool sequencing of natural HLA-DR, DQ, and DP ligands reveals detailed peptide motifs, constraints of processing, and general rules. Immunogenetics 39: 230-242. [Medline]
  45. Routsias, J., G. K. Papadopoulos. 1995. Polymorphic structural features of modelled HLA-DQ molecules segregate according to susceptibility or resistance to IDDM. Diabetologia 38: 1251-1261. [Medline]
  46. Paliakasis, K., J. Routsias, K. Petratos, C. Ouzounis, M. Kokkinidis, G. K. Papadopoulos. 1996. Novel structural features of the human histocompatibility molecules HLA-DQ as revealed by modeling based on the published structure of the related molecule HLA-DR. J. Struct. Biol. 117: 145-163. [Medline]
  47. Kim, C.-Y., H. Quarsten, E. Bergseng, C. Khosla, L. M. Sollid. 2004. Structural basis for HLA-DQ2-mediated presentation of gluten epitopes in celiac disease. Proc. Natl. Acad. Sci. USA 101: 4175-4179. [Abstract/Free Full Text]
  48. Ghosh, P., M. Amaya, E. Mellins, D. C. Wiley. 1995. The structure of an intermediate in class II MHC maturation: CLIP bound to HLA-DR3. Nature 378: 457-462. [Medline]
  49. Sidney, J., M-F. del Guercio, S. Southwood, A. Sette. 2002. The HLA molecules DQA1*0501/B1*0201 and DQA1*0301/B1*0302 share an extensive overlap in peptide binding specificity. J. Immunol. 169: 5098-5108. [Abstract/Free Full Text]
  50. Kwok, W. W., S. Kovats, P. Thurtle, G. T. Nepom. 1993. HLA-DQ allelic polymorphisms constrain patterns of class II heterodimer formation. J. Immunol. 150: 2263-2272. [Abstract]
  51. Ettinger, R. A., A. W. Liu, G. T. Nepom, W. W. Kwok. 1998. Exceptional stability of the HLA-DQA1*0102/DQB1*0602 {alpha}beta protein dimer, the class II MHC molecule associated with protection from insulin-dependent diabetes mellitus. J. Immunol. 161: 6439-6445. [Abstract/Free Full Text]
  52. Nepom, G. T.. 1990. A unified hypothesis for the complex genetics of HLA associations with IDDM. Diabetes 39: 1153-1157. [Abstract]
  53. Pugliese, A., D. Brown, D. Garza, D. Murchison, M. Zeller, M. Redondo, J. Diez, G. S. Eisenbarth, D. D. Patel, C. Ricordi. 2001. Self-antigen-presenting cells expressing diabetes-associated autoantigens exist in both thymus and peripheral lymphoid organs. J. Clin. Invest. 107: 555-564. [Medline]

Related articles in The JI:

IN THIS ISSUE

The JI 2006 177: 1373-1374. [Full Text]  



This article has been cited by other articles:


Home page
 J. Immunol.Home page
E. A. James, A. K. Moustakas, J. Bui, R. Nouv, G. K. Papadopoulos, and W. W. Kwok
The Binding of Antigenic Peptides to HLA-DR Is Influenced by Interactions between Pocket 6 and Pocket 9
J. Immunol., September 1, 2009; 183(5): 3249 - 3258.
[Abstract] [Full Text] [PDF]

Home page
 DiabetesHome page
X. Ge, J. D. Piganelli, H. M. Tse, S. Bertera, C. E. Mathews, M. Trucco, L. Wen, and W. A. Rudert
Modulatory Role of DR4- to DQ8-restricted CD4 T-Cell Responses and Type 1 Diabetes Susceptibility
Diabetes, December 1, 2006; 55(12): 3455 - 3462.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Related articles in The JI
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Request Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Ettinger, R. A.
Right arrow Articles by Kwok, W. W.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Ettinger, R. A.
Right arrow Articles by Kwok, W. W.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS