MHC Class II Risk Alleles and Amino Acid Residues in Idiopathic Membranous Nephropathy

CLINICAL RESEARCH www.jasn.org

† ‡ Zhao Cui,* Li-jun Xie,* Fang-jin Chen, Zhi-yong Pei,§ Li-jie Zhang,§ Zhen Qu,* Jing Huang,* Qiu-hua Gu,* Yi-miao Zhang,* Xin Wang,* Fang Wang,* Li-qiang Meng,* | † Gang Liu,* Xu-jie Zhou,* Li Zhu,* Ji-cheng Lv,* Fan Liu, Hong Zhang,* Yun-hua Liao, ‡ †† ‡‡ Lu-hua Lai, Pierre Ronco,¶** and Ming-hui Zhao*

*Renal Division, Department of Medicine, Peking University First Hospital, Institute of Nephrology, Peking University, Key Laboratory of Renal Disease, Ministry of Health of China, Beijing, China; †Renal Division, Department of Medicine, First Afﬁliated Hospital of Guangxi Medical University, Nanning, China; ‡State Key Laboratory for Structural Chemistry of Unstable and Stable Species, Beijing National Laboratory for Molecular Sciences, College of Chemistry and Molecular Engineering and Center for Theoretical Biology, Peking University, Beijing, China; §Beijing Computing Center, Beijing, China; |Key Laboratory of Genomic and Precision Medicine, Beijing Institute of Genomic, Chinese Academy of Science, Beijing, China; ¶Sorbonne Universités, Université Pierre et Marie Curie Univ Paris 06, Paris, France; **Institut National de la Santé et la Recherche Médicale, Unité Mixte de Recherche 1155, Paris, France; ††Assistance Publique-Hôpitaux de Paris, Service de Néphrologie et Dialyses, Hôpital Tenon, Paris, France; and ‡‡Peking-Tsinghua Center for Life Sciences, Beijing, China

ABSTRACT Epitopes of phospholipase A2 receptor (PLA2R), the target antigen in idiopathic membranous nephropathy (iMN), must be presented by the HLA–encoded MHC class II molecules to stimulate autoantibody production. A genome–wide association study identified risk alleles at HLA and PLA2R loci, with the top variant rs2187668 within HLA-DQA1 showing a risk effect greater than that of the top variant rs4664308 within PLA2R1. How the HLA risk alleles affect epitope presentation by MHC class II molecules in iMN is unknown. Here, we genotyped 261 patients with iMN and 599 healthy controls at the HLA-DRB1, HLA- DQA1, HLA-DQB1, and HLA-DPB1 loci with four-digit resolution and extracted the encoded amino acid sequences from the IMGT/HLA database. We predicted T cell epitopes of PLA2R and constructed MHC- DR molecule-PLA2R peptide-T cell receptor structures using Modeler. We identified DRB1*1501 (odds ratio, 4.65; 95% confidence interval [95% CI], 3.39 to 6.41; P,0.001) and DRB1*0301 (odds ratio, 3.96; 95% CI, 2.61 to 6.05; P,0.001) as independent risk alleles for iMN and associated with circulating anti–PLA2R antibodies. Strong gene-gene interaction was noted between rs4664308(AA) and HLA-DRB1*1501/ DRB1*0301. Amino acid positions 13 (P,0.001) and 71 (P,0.001) in the MHC-DRb1 chain independently associated with iMN. Structural models showed that arginine13 and alanine71, encoded by DRB1*1501, and lysine71, encoded by DRB1*0301, facilitate interactions with T cell epitopes of PLA2R. In conclusion, we identified two risk alleles of HLA class II genes and three amino acid residues on positions 13 and 71 of the MHC-DRb1 chain that may confer susceptibility to iMN by presenting T cell epitopes on PLA2R.

J Am Soc Nephrol 28: 1651–1664, 2017. doi: https://doi.org/10.1681/ASN.2016020114

Idiopathic membranous nephropathy (iMN) is the Received February 1, 2016. Accepted September 21, 2016. most common cause of nephrotic syndrome in Z.C., L.-j.X., and F.-j.C. contributed equally to this work. adults, deﬁned histopathologically by the presence Published online ahead of print. Publication date available at of immune complexes on the extracapillary side of www.jasn.org. glomerular basement membrane. It is now recog- Correspondence: Dr. Ming-hui Zhao, Renal Division and In- nized as an organ–speciﬁc autoimmune disease, in stitute of Nephrology, Peking University First Hospital, Beijing, which circulating antibodies to a conformation- China 100034. Email: [email protected] dependent epitope on the target antigen, the Copyright © 2017 by the American Society of Nephrology

J Am Soc Nephrol 28: 1651–1664, 2017 ISSN : 1046-6673/2805-1651 1651 CLINICAL RESEARCH www.jasn.org

M-type receptor for phospholipase A2 (PLA2R), are detectable fibronectin type II domain-CTLD1 region. Fresquet et al.11 in about 70% of patients from various ethnic groups.1 further identified two peptides from the ricin domain, Autoantibody generation is suggested to rely on the stimu- with a longer sequence covering both peptides (31-mers) pro- lation of antigen–specific CD4+ T cells on antigen presentation ducing 85% inhibition of autoantibody binding to PLA2R. by MHC class II molecules. These molecules are encoded by Seitz-Polski et al.12 recently identified reactive epitopes in HLA genes. The association between iMN and HLA class II CysR, CTLD1, and CTLD7 domains. CysR appears as the pri- genes has been revealed by a genome–wide association study mary dominant epitope, with evidence for epitope spreading (GWAS) in European white ancestry. This study identified toward CTLD1 and CTLD7. However, which epitopes and rs2187668 within HLA-DQA1 to be the lead variant associated with iMN, with higher risk effect compared with top variant rs4664308 of PLA2R1.2 The associations with these two risk alleles were validated in our recent study of the Chinese population.3 The more important observation was that these risk alleles were closely associated with circulating anti–PLA2R antibodies and glomerular PLA2R expression. Our data support an interaction between HLA- DQA1 and PLA2R1 contributing to the development of iMN, likely by affecting the presence of autoantibodies. It is postu- lated that sequence variants within HLA-DQA1 alleles that are unique to iMN lead to the presentation of such peptides to immunocompetent cells, resulting in autoantibody formation.2 HLA locus is considered the most complex one in human genome. In a joint GWAS study performed by Stanescu et al.2 in three different populations of white European patients, 282 single- nucleotide polymorphisms (SNPs) within HLA locus (chromosome 6) presented significant associations with iMN. SNP rs2187668 within HLA-DQA1 was the most significant one.2 Because each amino acid residue on MHC is typically assigned to multiple HLA classic alleles, the disease risk effect at certain amino acid positions would go undetected if only HLA classic alleles and single SNPs were assessed.4–7 Thus, additional genetic and pathogenic investigations on HLA protein– coding alleles are urgently warranted. PLA2R1 is a 180-kD membrane receptor witha largeextracellular regioncomprising ten distinct domains, namely a cysteine- rich domain (CysR), a fibronectin type II domain, and eight distinct C–type lectin – 8 domains (CTLDs; CTLD1 CTLD8). Figure 1. Associations of HLA class II alleles with iMN. (A) Association results for HLA Several epitopes on PLA2R1 targeted by class II alleles analysis in Chinese patients with iMN. (B) After adjusting for DRB1*1501, anti-PLA2R1 antibodies were revealed DRB1*0301 still showed significantly positive association with iMN. (C) After adjusting recently.9 Kao et al.10 first located the for both DRB1*1501 and DRB1*0301, no HLA allele showed significant association immunodominant epitope in the CysR- with iMN.

1652 Journal of the American Society of Nephrology J Am Soc Nephrol 28: 1651–1664, 2017 www.jasn.org CLINICAL RESEARCH how they are presented by MHC class II molecules are not with DRB1*0301 (P,0.001) (Figure 1B). After controlling for elucidated. DRB1*0301, similarly, DRB1*1501 showed significant posi- In this study, we thoroughly investigated the HLA class II tive association with the disease (P,0.001). Thus, gene profile at four-digit resolution from 261 patients with DRB1*1501 and DRB1*0301 were two independent risk iMN and 599 ethnically matched controls and explored the HLA alleles for iMN. However, when we controlled for iMN-HLA associations with alleles, haplotypes, amino acid DRB1*1501 and DRB1*0301, no HLA allele showed signifi- positions, and residues. With this line of evidence, we further cant association with the disease (Figure 1C). propose models for the presentation of PLA2R epitopes by In our patients, 73.6% (192 of 261) carried DRB1*1501 or susceptible HLA alleles, which strongly supports a causative DRB1*0301, whereas in healthy controls, only 33.1% (198 of link between the presence of HLA risk alleles and the devel- 599) carried DRB1*1501 or DRB1*0301 (chi squared =120.4; opment of PLA2R antibodies and iMN. P,0.001). Thirteen percent (34 of 261) of the patients carried both DRB1*1501 and DRB1*0301, whereas only 0.7% (four of 599) of the healthy individuals did so (chi squared =65.7; RESULTS P,0.001). The top SNP rs2187668 in HLA-DQA1, which is known Population Characteristics as the tag SNP of DRB1*0301, was in moderate linkage The patient group comprised 261 patients with biopsy-proven disequilibrium (LD) with DRB1*0301 (r2=0.46, D9=0.73) in iMN. Patients with secondary causes, such as infections, like our population; 79.2% of the patients having rs2187668(A) hepatitis B and hepatitis C virus, syphilis, lupus, malignancy, carried DRB1*0301. Although unconditional association and heavy metal poisoning, were excluded. Circulating anti– analysis showed a significantly positive association between PLA2R antibodies were assessed in all patients at the time of rs2187668(A) and iMN (odds ratio [OR], 1.94; 95% confi- renal biopsy, with a positive rate of 66.3%. Clinical and path- dence interval, 1.09 to 3.45; P=0.03), stepwise conditional ologic characteristics of patients with iMN are shown in analysis revealed that rs2187668(A) lost its association with Supplemental Table 1. The control group comprised 599 iMN (P=0.60) after adjusting for DRB1*0301. healthy, voluntarily recruited blood donors. All participants were self–reported Chinese Han living in the north of China. LD among HLA Alleles BecausetheHLAregioncontainsthedensestLDinthehuman Association of HLA Alleles with iMN genome, LD was analyzed among the four susceptible alleles Using sequencing-based typing, we have successfully genotyped that we had identified (Table 2). DRB1*0301 was in low LD 142 classic HLA alleles at four-digit resolution, including 47 with DQB1*0602 (D9=0.56, r2=0.01) and DQA1*0102 DRB1, 19 DQA1, 21 DQB1, and 55 DPB1 alleles. P value of (D9=0.89, r2=0.02). In contrast, the other three susceptible ,0.001 was adopted as the threshold for statistical significance alleles (DRB1*1501, DQB1*0602, and DQA1*0102) were in in classic HLA allele association analysis. Association results are overt LD. shown in Supplemental Table 2. Unconditional association analysis revealed a significantly HLA DR-DQ Haplotypes Associated with iMN positive association between iMN and DRB1*1501, The association of HLA DR-DQ haplotypes with iMN was DQB1*0602, DRB1*0301, and DQA1*0102 and a significantly analyzed using the expectation-maximization algorithm via negative association between iMN and DRB1*0901, the R Project. Three hundred thirty distinct HLA DR-DQ DQB1*0303, DQA1*0301, and DQA1*0302 (Figure 1A, Table haplotypes were found in patients and healthy controls, and 1). Stepwise conditional analysis revealed that, after adjusting therefore, we set P,0.05/330 (,0.001) as the threshold for for DRB1*1501, the most significant association was observed statistical significance.

Table 1. Association results for HLA class II alleles analysis in Chinese patients with iMN (four digits; P,0.001) Genotype Frequency in Genotype Frequency Allele Frequency in Allele Frequency in Allele OR P Value Patients (n=261), % in Controls (n=599), % Patients (2n=23261), % Controls (2n=23599), % Susceptible alleles DRB1*1501 62.40 26.21 37.55 14.69 4.65 ,0.001 DQB1*0602 54.26 29.55 32.38 15.44 2.82 ,0.001 DRB1*0301 24.42 7.53 12.07 3.84 3.96 ,0.001 DQA1*0102 62.20 43.57 40.04 26.21 2.11 ,0.001 Protective alleles DRB1*0901 5.81 28.21 3.07 14.94 0.16 ,0.001 DQB1*0303 8.14 27.38 4.22 14.44 0.24 ,0.001 DQA1*0301 4.65 20.87 2.49 11.02 0.19 ,0.001 DQA1*0302 6.20 19.53 3.10 10.35 0.28 ,0.001

J Am Soc Nephrol 28: 1651–1664, 2017 MHC II in Idiopathic Membranous Nephropathy 1653 CLINICAL RESEARCH www.jasn.org

Table 2. LD analysis among the susceptible HLA class II P,0.001). GLU had no significant association with the disease alleles in patients with iMN (Figure 4, Table 4). Allele-1 Allele-2 D9 r2 After adjusting for position 13, position 71 still showed fi P, DRB1*1501 DQB1*0602 0.69 0.45 signi cant effects for iMN ( 0.001) (Figure 2B). Conversely, DRB1*1501 DQA1*0102 0.77 0.38 after adjusting for position 71, position 13 still had an effect DQB1*0602 DRB1*0301 0.56 0.01 (P,0.001) (Figure 2C). Thus, positions 13 and 71 on DRb1 DQB1*0602 DQA1*0102 0.73 0.32 chain were independently at risk for iMN. After adjusting for DRB1*0301 DQA1*0102 0.89 0.02 both positions 13 and 71, there was no additional association signal for the amino acid positions on MHC-DRb1, -DQa1, DRB1*1501-DQA1*0102-DQB1*0602 haplotype showed -DQb1 and -DPb1 chains (Figure 2D). the strongest risk association with iMN (score =8.80; ALA71 and ARG13 are encoded by DRB1*1501 allele, and P,0.001). DRB1*0301-DQA1*0501-DQB1*0201 haplotype LYS71 is encoded by DRB1*0301 allele (Table 4); 81.67% of was another risk haplotype for iMN (score =6.85; P,0.001) the patients carried at least one of these three susceptible (Table 3). amino acid residues compared with 42.07% of the healthy DRB1*0901-DQA1*0302-DQB1*0303 haplotype also individuals (chi squared =114.44; P,0.001). showed association with iMN (score =25.37; P,0.001) but The two protective amino acid residues ARG71 and PHE13 with opposite effect (Table 3). are encoded by DRB1*0901 (Table 4); 10.34% of the patients carried this combination compared with 35.23% of the Association of MHC Amino Acid Residues with iMN healthy individuals (chi squared =56.22; P,0.001). The amino acid residues of MHC-DRb1, -DQa1, -DQb1, and -DPb1 chains were analyzed for potential associations with Genetic Interaction Analyses between PLA2R and HLA iMN. We successfully imputed 307 polymorphic amino acid Risk Alleles positions among DRb1, DQa1, DQb1, and DPb1onthebasis We selected rs4664308 located in the first intron of PLA2R1, of the corresponding HLA class II alleles at four-digit resolu- which was reported to be strongly associated with iMN in the tion. Thus, P value of ,0.001 was adopted as the threshold for white GWAS study2 and our Beijing Asian population study.3 statistical significance. Gene-gene interaction analyses were performed between Unconditional analysis revealed that the most significant rs4664308 and HLA-DRB1*1501/DRB1*0301 (Table 5). The association was mapped to the amino acid position 13 on combination of risk genotypes of rs4664308 and HLA- MHC-DRb1chainencodedbyexon2ofHLA-DRB1 DRB1*1501 [rs4664308(AA) + DRB1*1501(+)] conferred a (P,0.001) (Figure 2A). The amino acid residue on position 32.40-fold higher risk (P,0.001) for the development of 13 participates in the formation of a b-sheet in the fourth iMN compared with the protective genotype at both loci. peptide–binding pocket of MHC-DRb1 chain (Figure 3). The combination of risk genotypes of rs4664308 and Among the six possible amino acid residues, arginine HLA-DRB1*0301 [rs4664308(AA) + DRB1*0301(+)] (ARG), tyrosine (TYR), histidine, serine, glycine, and phenyl- conferred a 29.95-fold higher risk (P,0.001) for the develop- alanine (PHE), at this position, ARG13 showed a risk effect for ment of iMN compared with the protective genotype at both iMN (OR, 2.89; P,0.001), whereas PHE13 showed a protec- loci. tive effect (OR, 0.24; P,0.001). The other four residues had Both rs3749117 and rs35771982, encoding the nonsynon- no significant effect (Figure 4, Table 4). ymous amino acids M292V and H300D of PLA2R, respec- A similar level of significance (P,0.001) was observed for tively, also showed interaction with DRB1*1501/DRB1*0301 the amino acid residue on position 71 in the MHC-DRb1 (Supplemental Tables 3 and 4). chain (Figure 2A). This amino acid residue forms part of an a-helix in the fourth and seventh peptide–binding pockets of Association of HLA Alleles and Circulating Anti–PLA2R MHC-DRb1 chain (Figure 3). Four possible amino acid res- Antibodies idues, alanine (ALA), lysine (LYS), glutamic acid (GLU), and In patients with circulating anti–PLA2R antibodies, ARG, were identified at this position. LYS71 (OR, 2.76; DRB1*1501 (OR, 6.06; P,0.001) and DRB1*0301 (OR, P,0.001) and ALA71 (OR, 3.09; P,0.001) showed risk effects 4.92; P,0.001) showed high risk effects for the disease. In for iMN. ARG71 showed a protective effect (OR, 0.31; patients without anti-PLA2R antibodies, DRB1*1501 (OR,

Table 3. Extended haplotype analysis among HLA-DRB1, -DQA1, -DQB1, and -DPB1 in Chinese patients with iMN (P,0.001) Haplotype Frequency Haplotype Frequency Extended Haplotype Haplotype Score P Value in Patients, % in Controls, % DRB1*1501-DQA1*0102-DQB1*0602 26.20 9.00 8.80 ,0.001 DRB1*0301-DQA1*0501-DQB1*0201 10.73 2.92 6.85 ,0.001 DRB1*0901-DQA1*0302-DQB1*0303 1.91 9.18 25.37 ,0.001

1654 Journal of the American Society of Nephrology J Am Soc Nephrol 28: 1651–1664, 2017 www.jasn.org CLINICAL RESEARCH

3.14; P,0.001) still showed positive association with iMN, albeit to a much lower level of signiﬁcance, but DRB1*0301 (OR, 2.54; P=0.35) lost its risk effect (Table 6). In the patients carrying DRB1*1501 or DRB1*0301, the prevalence of anti- PLA2R antibody was higher (72.10%) than in those without these susceptible alleles (50.00%; chi squared =10.95; P=0.001). ARG13, ALA71, and LYS71, three susceptible amino acid residues on MHC- DRb1, also showed positive association with the presence of circulating anti– PLA2R antibodies (chi squared =9.75; P=0.002). In patients carrying ARG13, ALA71, or LYS71, 71.30% of the patients had positive circulating anti–PLA2R antibodies. In contrast, only 48.21% of the patients without these amino acid residues developed circulating anti–PLA2R antibodies.

Modeling of PLA2R Epitopes Presented by Susceptible MHC Class II Molecules We used SYFPEITHI online database to predict T cell epitopes on the extracellular domain (amino acid residues 21–1397) of PLA2R presented by DRB1*1501.13 Pep- tide 285 in CTLD1 domain and peptide 1130 in CTLD7 domain were the best ﬁtting peptides, with equal scores of 34 (Table 7). MHC class II molecule DR2b is encoded by DRB1*1501 and DRA*0101. The structural models of DR2b-peptide 285 (SKTVEVWMGLNQLDE)-TCR were constructed using the Modeler program.14–17 The optimum structure is showninFigure5A.Inthismodel, the amino acid residues of peptide 285, valine288,tryptophan291,glycine293,leu- cine294,andglutamine296, are presented in the peptide-binding pockets 1, 4, 6, 7, and 9, respectively, of DR2b (Table 7). ALA71 in DR2b b-chain makes pocket 4largerthanthatinotherMHCDR

acid position 71, position 13 still showed sig- Figure 2. Associations of MHC class II molecules with iMN. (A) Unconditional analysis nificant association with the disease. (D) After results showed that amino acid positions 13 and 71 on HLA-DRb1weremostsignif- adjusting for both positions 13 and 71, no icantly associated with iMN risk. (B) After adjusting for amino acid position 13, position aminoacidpositionshowedsignificant asso- 71 still showed significant association with the disease. (C) After adjusting for amino ciation with iMN.

J Am Soc Nephrol 28: 1651–1664, 2017 MHC II in Idiopathic Membranous Nephropathy 1655 CLINICAL RESEARCH www.jasn.org

815 (PWLFYQDAEYLFHTF)-TCR structural models were constructed and screened, with the optimum structure shown in Figure 5E. Phenylalanine818,asparticacid821,glutamic acid823, tyrosine824, and phenylalanine826 are located in pockets 1, 4, 6, 7, and 9, respectively, of DR3 molecule (Table 7). The amidogen nitrogen atom of LYS71 and the guanidino nitrogen atom of ARG74 of DR3 b-chain interact with the carboxyl oxygen atom of aspartic acid821 on peptide 815 at distances of 2.7 and 2.9 Å, respectively (Figure 5F). The optimum structure of DR3-peptide 1194 Figure 3. Three–dimensional ribbon models for the risk HLA-DR (SFTFWKDEESSLLGD)-TCR is shown in Figure 5G. Phe- molecules. Amino acid position 71 of HLA-DRb chain is located nylalanine1197,asparticacid1200,glutamicacid1202,ser- on peptide-binding pockets 4 and 7, and amino acid position 13 ine1203,andleucine1205 are presented in pockets 1, 4, 6, 7, is located on peptide-binding pocket 4. (A) HLA-DR2b: the and 9, respectively, of DR3 molecule (Table 7). The amido- structure is on the basis of Protein Data Bank entry 1BX2. The a- gen nitrogen atom of LYS71 and the guanidino nitrogen b and -chains are encoded by HLA-DRA*0101 and DRB1*1501, atom of ARG74 of DR3 b-chain interact with the carboxyl respectively, with a direct view of the peptide-binding groove. oxygen atom of aspartic acid at distances of 2.8 and 2.6 Å, ALA71 (blue) and ARG13 (orange) were identified through asso- 1200 respectively (Figure 5H). ciation analysis. (B) HLA-DR3: the structure is on the basis of Protein Data Bank entry 1A6A. The a-andb-chains are encoded – by HLA-DRA*0101 and DRB1*0301, respectively, with a direct Modeling of a Major Antibody Binding Peptide (31- view of the peptide-binding groove. LYS71 (yellow) and serine13 mers) of PLA2R Presented by Susceptible MHC Class II (green) are highlighted. Molecules Peptide WQDKGIFVIQSESLKKCIQAGKSVLTLENCK, corresponding to the amino acid sequences 35–65, was reported as a molecules, so that the indole ring of tryptophan291 can stretch major antibody–binding peptide of PLA2R in the N–terminal into this pocket. Distance between the guanidino nitrogen CysR domain.11 We used the SYFPEITHI database to atom of ARG13 and the carbonyl oxygen atom of trypto- predict the epitopes presented by DR2b and DR3 molecules phan291 backbone in pocket 4 is 3.0 Å, which facilitates the on this immunodominant peptide. We identified two potential formation of hydrogen bond. Furthermore, tryptophan291 T cell epitopes with equal scores of 24 for DRB1*1501, and PHE26 and TYR78 in pocket 4 form strong hydrophobic GIFVIQSESLKKCIQ and KKCIQAGKSVLTLEN, and two pos- interactions with each other (Figure 5B). sible T cell epitopes with equal scores of 18 for DRB1*0301, The structural models of DR2b-peptide 1130 DKGIFVIQSESLKKC and SESLKKCIQAGKSVL, which over- (NANMTWYAAIKTCLM)-TCR were also constructed. The lapped with those predicted for DRB1*1501. Results are shown optimum structure is shown in Figure 5C. The amino acid in Figure 6 and Table 7. residues, methionine1133, tyrosine1136, ala- nine1138, isoleucine1139, and threonine1141, are presented by pockets 1, 4, 6, 7, and 9, respectively, of DR2b molecule (Table 7). ALA71 makes pocket 4 larger and accommodates the phenyl ring of tyrosine1136 stretching into this pocket. Distance between the oxygen atom of tyrosine1136 and the guanidino nitrogen atom of ARG13 is 3.0 Å, which facilitates hydrogen bond formation. tyro- sine1136 and PHE26 and TYR78 of pocket 4 produce strong hydrophobic interactions with each other (Figure 5D). We also predicted two peptides defining potential PLA2R epitopes presented by DRB1*0301, with equal scores of 33. Pep- tide 815 is located between CTLD4 and CTLD5 domains, and peptide 1194 is in Figure 4. Effects of individual amino acid residue at positions 13 and 71 on HLA-DRb1. CTLD7 domain (Table 7). The amino acid residue frequencies of positions 13 and 71 on HLA-DRb1 in the pa- MHC molecule DR3 is encoded by tients and controls are plotted, and the univariate ORs are indicated. GLY, glycine; DRB1*0301 and DRA*0101. DR3-peptide HIS, histidine; SER, serine.

1656 Journal of the American Society of Nephrology J Am Soc Nephrol 28: 1651–1664, 2017 www.jasn.org CLINICAL RESEARCH

Table 4. Effects of individual amino acid residues at positions 13 and 71 on MHC-DRb1 chain (P,0.001) Position and Patient Control Observed Classical HLA-DRB1 Alleles OR P Value Amino Acid Frequency, % Frequency, % DRb1: 13 TYR 5.71 8.84 0.62 .0.99 SER 32.28 23.91 1.52 0.19 ARG 43.70 20.23 1501, 1502, 1504, 1602 2.89 ,0.001 HIS 5.12 11.12 0.43 0.08 GLY 7.68 16.56 0.42 0.002 PHE 5.51 19.31 0101, 0102, 0901, 1001 0.24 ,0.001 DRb1: 71 ARG 41.00 71.57 0101, 0102, 0403, 0404, 0405, 0406, 0407, 0410, 0.31 ,0.001 0701, 0801, 0802, 0803, 0804, 0809, 0901, 1001, 1101, 1104, 1106, 1201, 1202, 1312, 1401, 1403, 1405, 1454, 1602 LYS 12.60 5.35 0301, 0401 2.76 ,0.001 GLU 4.33 5.03 0.86 .0.99 ALA 41.73 18.06 1501, 1502 3.09 ,0.001 SER, serine; HIS, histidine; GLY, glycine.

Table 5. Analysis of gene-gene interaction: ORs for iMN according to SNP rs4664308 (PLA2R1) and HLA-DRB1*1501/ DRB1*0301 combinations HLA-DRB1*1501 HLA-DRB1*0301 SNP rs4664308 (PLA2R1) (2) (+) (2)(+) GG (low risk) No. of patients/controls 2/30 2/10 3/37 1/3 OR (95% CI) 1 — 1 — P value Reference 0.30 Reference 0.33 GA (moderate risk) No. of patients/controls 21/133 14/38 23/153 12/18 OR (95% CI) — 5.53 (1.17 to 26.22) — 8.22 (2.06 to 32.84) P value 0.38 0.02 0.43 0.002 AA (high risk) No. of patients/controls 51/137 108/50 125/173 34/14 OR (95% CI) 5.58 (1.79 to 24.21) 32.40 (7.45 to 140.94) 8.91 (2.69 to 29.55) 29.95 (7.91 to 113.37) P value 0.01 ,0.001 ,0.001 ,0.001 95% CI, 95% conﬁdence interval; —, P.0.05.

PLA2R Sequence Variants in MHC Presentation findings provide important clues to the triggering of the Process immune response and the presentation of recently defined Using the SYFPEITHI database, we also assessed the sequence var- epitopes.10–12 iantsofPLA2R1,includingD200Y,M292V,H300D,R387H,R387C, We found that DRB1*1501 allele and its haplotype R404H, Y499C, V680L, L687P, and G1106S.18 The predicted T cell DRB1*1501-DQA1*0102-DQB1*0602 showed the strongest epitopes of PLA2R presented by DR2b or DR3 did not change after association with iMN. Two risk amino acid residues, ALA71 the one by one amino acid substitution. Amino acid substitution of and ARG13 of HLA-DRb1, were also positively associated M292V in peptide 285 [SKTVEVW(M/V)GLNQLDE] did not in- with the presence of circulating anti–PLA2R antibodies. fluence the presentation modeling by DR2b. The structure of These findings are consistent with the results of old studies DR2b-peptide 285 with the V substitution [SKTVEVW(V) in Japanese patients,6,7 whereas in whites, DRB1*1501 was not GLNQLDE]-TCR is shown in Supplemental Figure 1. found to be associated with iMN.4,5 We speculate that ethnic disparity may contribute to this difference. DRB1*1501 is a common risk allele for many autoimmune diseases, like DISCUSSION Goodpasture disease, multiple sclerosis, pemphigus vulgaris, and SLE.19–21 DRB1*1501 still showed positive association In this study, we identified several independent risk allelotypes with patients without anti-PLA2R antibodies, which suggests and their bearing haplotypes as well as critical amino acid that DRB1*1501 may also present non-PLA2R epitopes in- residues associated with iMN in the Chinese population. These volved in the pathogenesis of iMN.

J Am Soc Nephrol 28: 1651–1664, 2017 MHC II in Idiopathic Membranous Nephropathy 1657 CLINICAL RESEARCH www.jasn.org

Table 6. Association results for HLA class II alleles analysis with circulating anti–PLA2R antibodies Overall Analysis Anti-PLA2R (+) Versus Healthy Controls Anti-PLA2R (2) Versus Healthy Controls HLA Allele OR P Value OR P Value OR P Value DRB1*1501 4.65 ,0.001 6.06 ,0.001 3.14 ,0.001 DQB1*0602 2.82 ,0.001 3.43 ,0.001 1.94 0.12 DRB1*0301 3.96 ,0.001 4.92 ,0.001 2.54 0.35 DQA1*0102 2.11 ,0.001 2.39 ,0.001 1.66 0.51

Table 7. Peptides on PLA2R presented by the susceptible MHC class II molecules Amino Acid Risk MHC-DR Molecules PLA2R Domain Score Pocket 1 Pocket 4 Pocket 6 Pocket 7 Pocket 9 HLA-DR2b Peptide 285 (SKTVEVWMGLNQLDE) CTLD1 34 V W G L Q Peptide 1130 (NANMTWYAAIKTCLM) CTLD7 34 M Y A I T Peptide (31-mers; GIFVIQSESLKKCIQ) Ricin B–type lectin 24 V S S L K Peptide (31-mers; KKCIQAGKSVLTLEN) Ricin B–type lectin 24 I G S V T HLA-DR3 Peptide 815 (PWLFYQDAEYLFHTF) Between CTLD4 and CTLD5 33 F D E Y F Peptide 1194 (SFTFWKDEESSLLGD) CTLD7 33 F D E S L Peptide (31-mers; DKGIFVIQSESLKKC) Ricin B–type lectin 18 I ISEL Peptide (31-mers; SESLKKCIQAGKSVL) Ricin B–type lectin 18 L C Q A K V, valine; W, tryptophane; G, glycine; L, leucine; Q, glutamine; M, methionine; Y, tyrosine; A, alanine; I, isoleucine; T, threonine; S, serine; K, lysine; F, phenylalanine; D, aspartic acid; E, glutamic acid; C, cysteine.

DRB1*0301 allele and its bearing haplotype DRB1*0301- T-helper cells, in turn, stimulates antibody–producing B cells DQA1*0501-DQB1*0102 were also at high risk for to release antibodies to those specific antigens. Strong gene- iMN. LYS71 of b-chain of DR3 molecule encoded by gene interactions were identified between rs4664308(A) DRB1*0301 was identified as a risk amino acid residue for (top intronic SNP within PLA2R1) and HLA-DRB1*1501/ this disease. In the patients with anti-PLA2R antibodies, DRB1*0301. It is remarkable that rs3749117 and rs35771982, DRB1*0301 remained strongly associated with the disease, which code for nonsynonymous substitution in the CTLD1 and LYS71 was also confirmed as a risk amino acid residue, domain of PLA2R where lie both B cell and T cell (peptide whereas in those without anti-PLA2R antibodies, 285) epitopes, also showed interactions with HLA- DRB1*0301 association lost significance, which may suggest DRB1*1501/DRB1*0301. Close associations were also shown that this allelotype is mostly associated with PLA2R-related between circulating anti–PLA2R antibodies and DRB1*1501/ iMN. The identification of DRB1*0301 as a major risk allele DRB1*0301. As a whole, these data suggest a major role for is consistent with the previous GWAS study, which identified HLA-DRB1*1501/DRB1*0301 in the pathogenesis of iMN the top SNP rs2187668 (A/G) within HLA-DQA1,2 because through T cell epitope presentation. rs2187668(A) is the tag SNP of DRB1*0301.20,22 After ad- T cell epitopes on PLA2R have not been identified yet. On justing for DRB1*0301, rs2187668(A) lost its association the basis of risk allele sequences, we predicted two possible with iMN, which indicates that DRB1*0301 may account PLA2R epitopes presented by the susceptible allele DRB1*1501, for the majority or even all of the association signal of peptide 285 (SKTVEVWMGLNQLDE) in CTLD1 domain rs2187668(A). Our results are also in agreement with those and peptide 1130 (NANMTWYAAIKTCLM) in CTLD7 do- of old serologic studies, which showed that the HLA-B8/DR3 main, and two epitopes presented by the susceptible allele serotype (corresponding to the haplotype HLA-B*0801- DRB1*0301, peptide 815 (PWLFYQDAEYLFHTF) in the DRB1*0301-DQA1*0501-DQB1*0102) was associated with region between CTLD4 and CTLD5 and peptide 1194 theoccurrenceofiMNinwhites.4,5,23 HLA-B8/DR3 (SFTFWKDEESSLLGD) in CTLD7 domain as well. These haplotype was also associated with high production of findings fit well with recent data showing B cell–reactive epi- TNF-a in vivo and in vitro,24 which suggests that the in- topes (recognized by circulating antibodies) in CTLD1 and creased spontaneous release of TNF-a might participate in CTLD7,10,11,12 CTLD7 being an immunodominant domain the onset of iMN. associated with a severe clinical phenotype in patients with The major function of HLA class IIgene complex is specu- iMN.12 Sequence variants of PLA2R1 were shown to have little lated to encode protein products within MHC class II mole- influence on T cell epitope prediction. In particular, amino cules and participate in the immune system by presenting acid substitution M292V in peptide 285 did not influence the antigens mostly to T lymphocytes. The multiplication of presentation modeling by DR2b. These results strengthen the

1658 Journal of the American Society of Nephrology J Am Soc Nephrol 28: 1651–1664, 2017 www.jasn.org CLINICAL RESEARCH

view that rare variants in the coding sequence of PLA2R1 are unlikely to explain the pathogenesis of iMN.18 The molecular analysis at the amino acid level presented in this study should give new impetus to the search for mimicking sequences among antigens of microbial or other environmental origin. Molecular mimicry between these external antigens andPLA2Rcouldleadtoproductionof anti-PLA2R antibodies in patients genetically predisposed to autoimmunity and might well account for the increasing inci- dence of iMN worldwide. Alternatively, immunization could be due to the unmask- ing of hidden cryptic epitopes undergoing conformational changes on exposure to toxic environmental factors.

hydrogen bond. Furthermore, tryptophan291 and PHE26 and TYR78 in pocket 4 form strong hydrophobic interactions with each other. (C) Overall structure of the DR2b-peptide 1130(NANMTWYAAIKTCLM)-TCR molecule.

Methionine1133,tyrosine1136, alanine1138,iso- leucine1139, and threonine1141 are presented by pockets 1, 4, 6, 7, and 9, respectively. (D) Tyro-

sine1136 is in pocket 4. ALA71 makes pocket 4 larger and accommodates the phenyl ring of

TYR1136 stretching into this pocket. Distance between the oxygen atom of TYR1136 and the guanidino nitrogen atom of ARG13 is 3.0 Å, which facilitates hydrogen bond formation.

TYR1136 and PHE26 and TYR78 of pocket 4 produce strong hydrophobic interactions with each other. (E) Overall structure of the DR3-peptide 815(PWLFYQDAEYLFHTF)-TCR

molecule. Phenylalanine818, aspartic acid821, glutamic acid823, tyrosine824, and phenylala- nine826 are located in pockets 1, 4, 6, 7, and 9, respectively. (F) Aspartic acid821 is in pocket 4. The amidogen nitrogen atom of LYS71 and the guanidino nitrogen atom of ARG74 of Figure 5. Structural models of MHC class II-PLA2R peptide-TCR molecule. Gray DR3 b-chain interact with the carboxyl oxy- cartoon is peptide, cyan cartoon is MHC DR molecule, and magenta cartoon is TCR gen atom of aspartic acid821 on peptide 815 at molecule. Magenta sticks are amino acid residues of pocket 1, yellow sticks are amino distances of 2.7 and 2.9 Å, respectively. acid residues of pocket 4, blue sticks are amino acid residues of pocket 6, gray sticks are (G) Overall structure of the DR3-peptide 1194 amino acid residues of pocket 7, and white sticks are amino acid residues of pockets 9. (SFTFWKDEESSLLGD)-TCR molecule. Phenylal-

The distance between amino acid residues is shown by red dotted lines, and the unit is anine1197,asparticacid1200,glutamicacid1202, Angstroms. (A) Overall structure of the DR2b-peptide 285 (SKTVEVWMGLNQLDE)- serine1203, and leucine1205 are presented in TCR molecule. Valine288, tryptophan291, glycine293, leucine294, and glutamine296 are pockets 1, 4, 6, 7, and 9, respectively. (H) Aspartic presented in pockets 1, 4, 6, 7, and 9, respectively. (B) Tryptophan291 is in pocket 4. acid1202 is in pocket 4. The amidogen nitrogen ALA71 in DR2b b-chain makes pocket 4 larger than in other MHC DR molecules, so atom of LYS71 and the guanidino nitrogen atom that the indole ring of tryptophan291 can stretch into this pocket. Distance between of ARG74 of DR3 b-chain interact with the the guanidino nitrogen atom of ARG13 and the carbonyl oxygen atom of trypto- carboxyl oxygen atom of aspartic acid1200 at phan291 backbone in pocket 4 is 3.0 Å, which facilitates the formation of distances of 2.8 and 2.6 Å, respectively.

J Am Soc Nephrol 28: 1651–1664, 2017 MHC II in Idiopathic Membranous Nephropathy 1659 CLINICAL RESEARCH www.jasn.org

A distinguishing feature of HLA-DR2b encoded by DRB1*1501 is a large, primarily hydrophobic pocket 4 of peptide-binding sites, which accommodates aromatic amino acids.25,26 This large space is mostly created by ALA71 of DR2b b-chain.27 The benzene ring of critical amino acid residues, tryptophan291 in PLA2R peptide 285 and tyrosine1136 in PLA2R peptide 1130, is just in the large pocket 4 of DR2b molecules. The guanidino nitrogen atom of another risk amino acid residue, ARG13, facilitates the formation of hydrogen bond with the oxygen atom of trypto- phan291 and tyrosine1136. In the two PLA2R peptides 815 and 1194 presented by the susceptible allele DRB1*0301, the critical amino acid residues, aspartic acid821 and aspartic acid1200, interact with LYS71 and ARG74 on pocket 4 of DR3 b-chain. This interaction occurs through the carboxyl oxygen atom of aspartic acid on the one hand and the amidogen nitrogen atom of LYS71 and the guanidino nitrogen atom of ARG74 encoded by DRB1*0301 on the other hand. Although the CysR domain did not show up as potential T cell epitope in this studyon the basis of sequence analysis of the most signiﬁcantly associated allelotypes, we modeled interaction of risk alleles with a longer peptide (31-mers) in the CysR

DR2b-peptide(KKCIQAGKSVLTLEN)-TCR.

Isoleucine52,glycine55, serine57, valine58,and threonine60 are presented in pockets 1, 4, 6, 7, and 9, respectively. (D) Glycine55 is in pocket 4. The distance between the oxygen atom of

glycine55 main chain and the guanidino nitrogen atom of ARG13 on DR2b b-chain was 2.9 Å, which facilitates hydrogen bond for- Figure 6. Structural models of MHC class II-peptide (WQDKGIFVIQSESLKKCI- mation. (E) Overall structure of DR3-peptide QAGKSVLTLENCK)-TCR molecule. Gray cartoon is peptide, cyan cartoon is MHC DR (DKGIFVIQSESLKKC)-TCR. Isoleucine40, iso- molecule, and magenta cartoon is TCR molecule. Magenta sticks are amino acid leucine43,serine45,glutamine46, and leucine48 residues of pocket 1, yellow sticks are amino acid residues of pocket 4, blue sticks are arepresentedinpockets1,4,6,7,and9, amino acid residues of pocket 6, gray sticks are amino acid residues of pocket 7, and respectively. (F) Isoleucine43 is in pocket 4. white sticks are amino acid residues of pockets 9. The distance between amino acid Isoleucine43 interacts with PHE26 and TYR78 residues is shown by red dotted lines, and the unit is angstroms. (A) Overall structure of through hydrophobic interaction. (G) Overall the DR2b-peptide(GIFVIQSESLKKCIQ)-TCR molecule. Valine42,serine45,serine47, structure of DR3-peptide(SESLKKCIQAGKSVL)- leucine48, and lysine50 are presented in pockets 1, 4, 6, 7, and 9, respectively. (B) TCR. Leucine48,cystine51,glutamine53, alanine54, Serine45 is in pocket 4. Distance between the oxygen atom of the serine side chain and lysine56 are presented in pockets 1, 4, 6, 7, and the guanidino nitrogen atom of ARG13 on DR2b b-chain is 2.8 Å, which facilitates and 9, respectively. (H) Cystine51 is in pocket 4. the formation of hydrogen bond. Serine45 and PHE26 and TYR78 of pocket 4 also Cystine51 interacts with PHE26 and TYR78 produce strong hydrophobic interactions with each other. (C) Overall structure of through hydrophobic interaction.

1660 Journal of the American Society of Nephrology J Am Soc Nephrol 28: 1651–1664, 2017 www.jasn.org CLINICAL RESEARCH domain, because this peptide was recently characterized as epitopes on PLA2R. This study provides a first step producing 85% inhibition of autoantibody binding to toward a PLA2R-HLA-DR fitting model, which will need PLA2R.11 Several interactions were identified, notably in further investigations, because this approach may reveal pocket 4. In particular, the amino acid residues, isoleucine43 key events in the pathogenesis. and cystine51, in the 31-mer peptide could interact with two TYRs encoded by DRB1*0301 (at positions 26 and 78 of DR3 b-chain) through hydrophobic interaction, albeit with a rela- CONCISE METHODS tively low presenting score. DRB1*1501 and DRB1*0301 are two independent risk Study Participants HLA alleles for iMN. The structure of HLA-DR2b and DR3 A total of 860 participants, comprising 261 patients with biopsy- and the risk amino acid residues at positions 13 and 71 of proven iMN and 599 racially matched healthy controls, were enrolled DRb1 are entirely different. For PLA2R antigen presentation, in this study. All of the participants were of Chinese Han ethnicity. DRB1*1501 and DRB1*0301 may present different epitopes Birth places of all patientswere checked by theirnational identification under different antigen presentation modes. Consequently, number: 96.6% (252 of 261) of the patients were from north China the mechanisms for DRB1*1501 and DRB1*0301 contribut- (Beijing, 55.9%; Hebei, 16.5%; Shanxi, 9.1%; inner Mongolia, 4.6%; ing to the onset of iMN might be different, which will need Liaoning, 2.7%; Shandong, 2.3%; Henan, 2.3%; Heilongjiang, 1.5%; further investigations. Tianjin, 0.8%;and Jilin,0.8%),1.1% (threeof 261)of the patientswere This study has several genetic weaknesses. Because of the from south China (Guizhou, 0.8%; and Guangxi, 0.4%), and 2.3% (six relatively small sample size of patients and controls, this study of 261) of the patients were from central China (Anhui, 1.9%; and was not powered to detect association signals for rare alleles. Hubei, 0.4%). All of the controls came from the Beijing area. The Although .95% of the patients were from north China, there diagnosis of iMN was established by exclusion of secondary causes, is genetic heterogeneity among these populations,28 which such as hepatitis B and hepatitis C virus infections, SLE, malignancy, could not be addressed by genetic ancestry assessment in drugs, and heavy metal poisoning. this study. One must be aware that subtle stratification even This study was performed in compliance with the Declaration of with closely related Han Chinese may introduce serious con- Helsinki and approved through the ethics committee of Peking Uni- founding in case-control analyses of this region. The lack of versity First Hospital. Written informed consent was obtained from all independent replication is another limitation to the study. participants. However, several studies corroborate the association with DRB1*1501 in the Japanese population6,7 and DRB1*0301 Detection of Circulating Anti–PLA2R Antibodies in the white European population.23 We focused on a single Serum samples from all patients were collected on the day of renal top SNP (rs2187668) in HLA-DQA1, because of the 282 as- biopsy. Circulating anti–PLA2R autoantibodies were detected sociated SNPs in the HLA region identified in the white Eu- by a direct immunofluorescence assay with the use of the human ropean population,2 only a few among the top SNPs were embryonic kidney cell line that was transiently transfected conserved in the Chinese population.3 Additional studies with full-length cDNA encoding a PLA2R1 isoform (FA1254– should include a fine mapping of the HLA region in the 1005–50; EUROIMMUN AG, Lubeck, Germany). The detection Chinese to provide a more comprehensive picture of the ge- was performed on an immunofluorescence assay Mosaic slide netic associations. Because search for PLA2R antigen could following the standard instructions.3,30 Antibody positivity not be performed in biopsies, it is likely that the true prev- was defined as a green staining on the cells at the serum dilutions alence of PLA2R-related iMN was underestimated.29 Lastly, $1/10. structural models were on the basis of the identification of T cell epitopes predicted from the SYFPEITHI online database Four–Digit Resolution Genotyping of HLA Alleles and constructed using the Modeler program. The predicted Peripheral blood samples (10 ml) were collected from patients and epitopes and critical amino acid residues need further inves- controls with anticoagulant EDTA. Genomic DNA was isolated from tigations on the basis of ELISPOT analysis and amino acid peripheral whole blood using the Puregene Blood Core Kit C (catalog mutation experiments. However, these findings provide use- no. 158389; QIAGEN Science). ful information for subsequent epitope mapping studies and HLA class II genes were typed on an Applied Biosystems Incor- mimicry analyses. porated Company 3130XL platform using SeCoreTM Sequencing Kits In conclusion, our study identified two independent risk (Invitrogen, Brown Deer, WI). Sequencing analysis software was used alleles, HLA-DRB1*1501 and -DRB1*0301, and their bear- to process the raw data and generate the sequence files. Allele typing ing haplotypes for iMN onset. The association between reports were issued using uType SBT HLA software. The typing of risk HLA alleles and the presence of circulating anti– DRB1 alleles was performed through the bidirectional sequencing of PLA2R antibodies was also confirmed. Risk amino acid res- exon 2, the typing of DQB1 was performed through the bidirectional idues on positions 13 and 71 at peptide-binding pocket 4 of sequencing of exons 2 and 3, and the typing of DPB1 was performed HLA-DRb1 may confer susceptibility to iMN by critically through the bidirectional sequencing of exons 2–4. For DQA1 typing, contributing to the presentation of the possible T cell the alleles were typed using electrophoresis.

J Am Soc Nephrol 28: 1651–1664, 2017 MHC II in Idiopathic Membranous Nephropathy 1661 CLINICAL RESEARCH www.jasn.org

SNP Selection and Genotyping Modeler Bothrs4664308locatedinthe intron1ofPLA2Randrs2187668within The amino acid sequence of PLA2R was extracted from Uniprot HLA-DQA1, which were reported to be strongly associated with iMN (http://www.uniprot.org/; identification Q13018). We used the in white and Asian population studies,2,3 were selected in our study. SYFPEITHI online database (http://www.syfpeithi.de/) to predict Two SNPs correlated with rs4664308, which encoded nonsynony- the T cell epitopes of PLA2R. All predicted MHC class II ligands mous amino acid (rs3749117, M292V; and rs35771982, H300D) in are 15-mers, consisting of three N–terminal flanking residues, the the extracellular CTLD1 of PLA2R, were also selected. We did not nonamer core sequence located within the binding groove, and include rs11244 in HLA-DOB and rs2301271 in HLA-DQB2, which three C–terminal flanking residues. Thus, anchor residue P1 ap- were identified in the white population, because they were not asso- pears in position 4 of the peptides predicted with SYFPEITHI.13 ciated with iMN in our Han population.3 The protein crystal structure of 1YMM (http://www.rcsb.org/ All 199 patients with iMN and 398 healthy controls with sufficient pdb/home/home.do) was used as a template to construct DR2b- DNA samples were genotyped by PCR Sanger sequencing technology. peptide-TCR structural models using the Modeler program (version The sequence was detected by an Applied Biosystems Incorporated 9V8; http://salilab.org/modeller/).14–17,34 The crystal structures Company 3730XL DNA Analyzer. of 1A6A and 2WBJ were used as templates to construct DR3-peptide- TCR structural models using the Modeler program.35,36 Modeler refine Determination of Amino Acid Sequence of DRb1, was set very slow, optimization repeat was six, and ending model was set DQa1, DQb1, and DPb1 at 800. On the basis of four–digit resolution HLA alleles, we extracted amino We moved the peptides step by step and sought the optimum acid sequences of the DRb1, DQa1, DQb1, and DPb1 chains from structure. The conformations of 800 compounds were exported the IMGT/HLA public database: http://www.ebi.ac.uk/ipd/imgt/hla/. from every move and manually evaluated according to the following criteria. We measured the distance between amino acid residues of Association Analyses PLA2R peptides and corresponding peptide binding pockets at the fi # # All genotyping datasets were subjected to the same rigorous quality rst step (DR2b: distance pocket 1 4 Å, distance pocket 4 4Å, # # assessments both before and after patients and control subjects were distance pocket 7 4 Å, DR3: distance pocket 1 4 Å, distance pocket # # # compared. Individuals for whom consistent genotypes could not be 4 4 Å, distance pocket 6 4 Å, and distance pocket 9 4Å).We obtained for each of the loci were removed from this study. Hardy– chose the position with the shortest distance as the optimum posi- Weinberg equilibrium was assessed at each locus in controls only. tion. Then, we measured the distance between the guanidino nitrogen HLA alleles were excluded from the analysis when observed to be atom of ARG on position 13 encoded by DRB1*1501 and the oxygen out of Hardy–Weinberg equilibrium (P,0.001) or have a minor allele atom of the main chain of the residues located in pocket 4 and the frequency below 0.001. Power and sample size calculation (PS version distance between the side chain of LYS on position 71 encoded by 3.1.2) was performed to calculate the effect sizes (Supplemental DRB1*0301 and the oxygen atom of the main chain of the amino acid Table 5).28 residues located in pocket 4. We set the distance below 3.5 Å and Unconditional analysis and stepwise conditional analysis for chose the conformations with shortest distance as the optimum HLA class II alleles and amino acid positions of HLA-DRb1, structure. DQa1, DQb1, and DPb1 were performed through logistic regression using PLINK software version 1.9 (http://zzz.bwh.harvard. edu/plink/).31 Haplotypes of HLA class II alleles were constructed ACKNOWLEDGMENTS using the expectation-maximization algorithm with the R project 3.1.1 (Haplo.Stats package; http://www.r-project.org).32 Differences were considered significant when the P value was ,0.05 for distinct The authors appreciate the technical support provided by Ping Hou. alleles, distinct haplotype, or distinct amino acid positions. Part of the analysis was performed on the Computing Platform of the The markers were obtained and encoded to determine the as- Center for Life Science. fi sociation analysis. The biallelic markers with two HLA alleles or This work was nancially supported by Natural Science Founda- two–residue amino acid positions were encoded as alleles 1 and tion of China grant 81321064 (to the Innovation Research Group), 2. Multiallelic (three or more alleles) markers were encoded as the Outstanding Young Scholar grant 81622009, and other programs presence and absence of each allele of multiresidue amino acid grants 81330020 and 81370801 and National Key Technology R&D positions.33 A log–likelihood ratio test, estimating the log-likelihood Program grant 2013BAI09B14. P.R. is a recipient of European Research difference reflecting the effects at a single amino acid position, was Council ERC-2012-ADG_20120314 grant 322947 and 7th Framework used to evaluate the iMN association at an amino acid position. A Programme of the European Community contract 2012-305608 logistic regression analysis was used to evaluate the association with (European Consortium for High-Throughput Research in Rare iMN risk. Kidney Diseases). Statistical analysis was performed using the chi-squared test for categorical variables. Gene interaction analyses were conducted by Fisher exact test. Statistical analysis was performed with the DISCLOSURES SPSS 13.0 software package (SPSS Inc., Chicago, IL). None.

1662 Journal of the American Society of Nephrology J Am Soc Nephrol 28: 1651–1664, 2017 www.jasn.org CLINICAL RESEARCH

REFERENCES 18. Coenen MJ, Hofstra JM, Debiec H, Stanescu HC, Medlar AJ, Stengel B, Boland-Augé A, Groothuismink JM, Bockenhauer D, Powis SH, Mathieson PW, Brenchley PE, Kleta R, Wetzels JF, Ronco P: Phos- 1. Beck LH Jr., Bonegio RG, Lambeau G, Beck DM, Powell DW, Cummins pholipase A2 receptor (PLA2R1) sequence variants in idiopathic TD, Klein JB, Salant DJ: M-type phospholipase A2 receptor as target membranous nephropathy. JAmSocNephrol24: 677–683, 2013 antigen in idiopathic membranous nephropathy. N Engl J Med 361: 19. Masterman T, Ligers A, Olsson T, Andersson M, Olerup O, Hillert J: 11–21, 2009 HLA-DR15 is associated with lower age at onset in multiple sclerosis. 2. Stanescu HC, Arcos-Burgos M, Medlar A, Bockenhauer D, Kottgen Ann Neurol 48: 211–219, 2000 A, Dragomirescu L, Voinescu C, Patel N, Pearce K, Hubank M, 20. Fernando MM, Stevens CR, Sabeti PC, Walsh EC, McWhinnie AJ, Shah Stephens HA, Laundy V, Padmanabhan S, Zawadzka A, Hofstra JM, A, Green T, Rioux JD, Vyse TJ: Identification of two independent risk Coenen MJ, den Heijer M, Kiemeney LA, Bacq-Daian D, Stengel B, factors for lupus within the MHC in United Kingdom families. PLoS Powis SH, Brenchley P, Feehally J, Rees AJ, Debiec H, Wetzels JF, Genet 3: e192, 2007 Ronco P, Mathieson PW, Kleta R: Risk HLA-DQA1 and PLA(2)R1 21. Yang R, Cui Z, Zhao J, Zhao MH: The role of HLA-DRB1 alleles on alleles in idiopathic membranous nephropathy. NEnglJMed364: susceptibility of Chinese patients with anti-GBM disease. Clin Immunol 616–626, 2011 133: 245–250, 2009 3. Lv J, Hou W, Zhou X, Liu G, Zhou F, Zhao N, Hou P, Zhao M, Zhang H: 22. Miretti MM, Walsh EC, Ke X, Delgado M, GriffithsM,HuntS,MorrisonJ, Interaction between PLA2R1 and HLA-DQA1 variants associates with Whittaker P, Lander ES, Cardon LR, Bentley DR, Rioux JD, Beck S, anti-PLA2R antibodies and membranous nephropathy. JAmSoc Deloukas P: A high-resolution linkage-disequilibrium map of the Nephrol 24: 1323–1329, 2013 human major histocompatibility complex and first generation of tag 4. Papiha SS, Pareek SK, Rodger RS, Morley AR, Wilkinson R, Roberts single-nucleotide polymorphisms. Am J Hum Genet 76: 634–646, DF, Kerr DN: HLA-A, B, DR and Bf allotypes in patients with idio- 2005 pathic membranous nephropathy (IMN). Kidney Int 31: 130–134, 23. Sekula P, Li Y, Stanescu HC, Wuttke M, Ekici AB, Bockenhauer D, Walz 1987 G, Powis SH, Kielstein JT, Brenchley P, Eckardt KU, Kronenberg F, 5. Klouda PT, Manos J, Acheson EJ, Dyer PA, Goldby FS, Harris R, Kleta R, Köttgen A; GCKD Investigators: Genetic risk variants for Lawler W, Mallick NP, Williams G: Strong association between membranous nephropathy: extension of and association with other idiopathic membranous nephropathy and HLA-DRW3. Lancet 2: chronic kidney disease aetiologies [published online ahead of print 770–771, 1979 February 4, 2016]. Nephrol Dial Transplant 6. Ogahara S, Naito S, Abe K, Michinaga I, Arakawa K: Analysis of HLA 24. Lio D, Candore G, Colombo A, Colonna Romano G, Gervasi F, Marino class II genes in Japanese patients with idiopathic membranous glo- V, Scola L, Caruso C: A genetically determined high setting of TNF- merulonephritis. Kidney Int 41: 175–182, 1992 alpha influences immunologic parameters of HLA-B8,DR3 positive 7. Hiki Y, Kobayashi Y, Itoh I, Kashiwagi N: Strong association of HLA-DR2 subjects: Implications for autoimmunity. Hum Immunol 62: 705–713, and MT1 with idiopathic membranous nephropathy in Japan. Kidney 2001 – Int 25: 953 957, 1984 25. Smith KJ, Pyrdol J, Gauthier L, Wiley DC, Wucherpfennig KW: Crystal 8. Ancian P, Lambeau G, Mattéi MG, Lazdunski M: The human 180-kDa structure of HLA-DR2 (DRA*0101, DRB1*1501) complexed with a fi receptor for secretory phospholipases A2. Molecular cloning, identi - peptide from human myelin basic protein. JExpMed188: 1511–1520, cation of a secreted soluble form, expression, and chromosomal lo- 1998 – calization. JBiolChem270: 8963 8970, 1995 26.ScharfSJ,FriedmannA,BrautbarC,SzaferF,SteinmanL,HornG, 9. Beck LH Jr.: The dominant humoral epitope in phospholipase A2 Gyllensten U, Erlich HA: HLA class II allelic variation and suscepti- p receptor-1: Presentation matters when serving up a slice of . JAmSoc bility to pemphigus vulgaris. Proc Natl Acad Sci USA 85: 3504 –3508, – Nephrol 26: 237 239, 2015 1988 fi 10. Kao L, Lam V, Waldman M, Glassock RJ, Zhu Q: Identi cation of the 27. Zerva L, Cizman B, Mehra NK, Alahari SK, Murali R, Zmijewski CM, immunodominant epitope region in phospholipase A2 receptor- Kamoun M, Monos DS: Arginine at positions 13 or 70-71 in pocket 4 of mediating autoantibody binding in idiopathic membranous nephropa- HLA-DRB1 alleles is associated with susceptibility to tuberculoid lep- – thy. J Am Soc Nephrol 26: 291 301, 2015 rosy. JExpMed183: 829–836, 1996 ’ 11. Fresquet M, Jowitt TA, Gummadova J, Collins R, O Cualain R, 28. Gharavi AG, Kiryluk K, Choi M, Li Y, Hou P, Xie J, Sanna-Cherchi S, fi McKenzie EA, Lennon R, Brenchley PE: Identi cation of a major epitope MenCJ,JulianBA,WyattRJ,NovakJ,HeJC,WangH,LvJ,ZhuL, recognized by PLA2R autoantibodies in primary membranous ne- Wang W, Wang Z, Yasuno K, Gunel M, Mane S, Umlauf S, Tikhonova – phropathy. J Am Soc Nephrol 26: 302 313, 2015 I, Beerman I, Savoldi S, Magistroni R, Ghiggeri GM, Bodria M, 12. Seitz-Polski B, Dolla G, Payre C, Girard CA, Polidori J, Zorzi K, Birgy- Lugani F, Ravani P, Ponticelli C, Allegri L, Boscutti G, Frasca G, Barelli E, Jullien P, Courivaud C, Krummel T, Benzaken S, Bernard AmoreA,PeruzziL,CoppoR,IzziC,ViolaBF,PratiE,SalvadoriM, G, Burtey S, Mariat C, Esnault VL, Lambeau G: Epitope spreading of Mignani R, Gesualdo L, Bertinetto F, Mesiano P, Amoroso A, Scolari autoantibody response to PLA2R associates with poor prognosis in F,ChenN,ZhangH,LiftonRP:Genome-wideassociationstudy membranous nephropathy. J Am Soc Nephrol 27: 1517–1533, identifies susceptibility loci for IgA nephropathy. Nat Genet 43: 2015 321–327, 2011 13. Rammensee H, Bachmann J, Emmerich NP, Bachor OA, StevanovicS: 29. Ronco P, Debiec H: Pathophysiological advances in membranous ne- SYFPEITHI: Database for MHC ligands and peptide motifs. Immuno- phropathy: Time for a shift in patient’scare.Lancet 385: 1983–1992, genetics 50: 213–219, 1999 2015 14. Martí-Renom MA, Stuart AC, Fiser A, Sánchez R, Melo F, Sali A: Com- 30. Hoxha E, Harendza S, Zahner G, Panzer U, Steinmetz O, Fechner K, parative protein structure modeling of genes and genomes. Annu Rev Helmchen U, Stahl RA: An immunofluorescence test for phospholipase- Biophys Biomol Struct 29: 291–325, 2000 A₂-receptor antibodies and its clinical usefulness in patients with 15. Webb B, Sali A: Comparative protein structure modeling using membranous glomerulonephritis. Nephrol Dial Transplant 26: 2526– MODELLER. Curr Protoc Bioinformatics 47: 1–32, 2014 2532, 2011 16. Sali A, Blundell TL: Comparative protein modelling by satisfaction of 31. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D, spatial restraints. J Mol Biol 234: 779–815, 1993 Maller J, Sklar P, de Bakker PI, Daly MJ, Sham PC: PLINK: A tool set for 17. Fiser A, Do RK, Sali A: Modeling of loops in protein structures. Protein whole-genome association and population-based linkage analyses. Sci 9: 1753–1773, 2000 Am J Hum Genet 81: 559–575, 2007

J Am Soc Nephrol 28: 1651–1664, 2017 MHC II in Idiopathic Membranous Nephropathy 1663 CLINICAL RESEARCH www.jasn.org

32. Long JC, Williams RC, Urbanek M: An E-M algorithm and testing strategy low-afﬁnity crossreactivity to common microbial peptides. Immunity for multiple-locus haplotypes. Am J Hum Genet 56: 799–810, 1995 30: 348–357, 2009 33. Kim K, Bang SY, Lee HS, Okada Y, Han B, Saw WY, Teo YY, Bae SC: The 36. Ghosh P, Amaya M, Mellins E, Wiley DC: The structure of an in- HLA-DRb1 amino acid positions 11-13-26 explain the majority of termediate in class II MHC maturation: CLIP bound to HLA-DR3. Nature SLE-MHC associations. Nat Commun 5: 5902, 2014 378: 457–462, 1995 34. Hahn M, Nicholson MJ, Pyrdol J, Wucherpfennig KW: Unconventional topology of self peptide-major histocompatibility complex binding by a human autoimmune T cell receptor. Nat Immunol 6: 490–496, 2005 See related editorial, “Genetic Complexities of the HLA Region and Idiopathic 35. Harkiolaki M, Holmes SL, Svendsen P, Gregersen JW, Jensen LT, Membranous Nephropathy,” on pages 1331–1334. McMahon R, Friese MA, van Boxel G, Etzensperger R, Tzartos JS, Kranc K, Sainsbury S, Harlos K, Mellins ED, Palace J, Esiri MM, van der Merwe This article contains supplemental material online at http://jasn.asnjournals. PA, Jones EY, Fugger L: T cell-mediated autoimmune disease due to org/lookup/suppl/doi:10.1681/ASN.2016020114/-/DCSupplemental.

1664 Journal of the American Society of Nephrology J Am Soc Nephrol 28: 1651–1664, 2017 Supplementary Table 1. Clinical and pathological characteristics of iMN patients with and without anti-PLA2R antibodies.

Anti-PLA2R Anti-PLA2R Clinical parameter Overall P positive (n=173) negative (n=88)

Gender (male/female) 147/114 101/72 46/42 0.35

Age (yr) 48.74±15.61 50.91±15.88 44.48±14.28 0.002

Serum albumin (g/L) 25.89±5.87 24.89±5.35 27.86±6.37 <0.0001

Proteinuria (g/24h) 5.93±4.83 6.79±5.34 4.26±3.04 <0.0001

Serum creatinine (μmol/L) 70.84±25.11 73.04±26.81 66.51±20.86 0.03 eGFR 117.20±38.60 113.25±38.18 125.17±38.38 0.02

Triglyceride (mmol/L) 2.79±2.12 2.68±1.87 2.73±2.55 0.85

Total cholesterol (mmol/L) 7.25±2.34 7.32±2.34 7.10±2.34 0.49

Pathology stage, n (%)

I-MN 134(51.3%) 81(46.8%) 53(60.2%)

II-MN 112(42.9%) 83(48.0%) 29(33.0%) 0.07

III-MN 15(5.7%) 9(5.2%) 6(6.8%) eGFR=175*POWER(Scr/88.1,-1.234)*POWER(Age,-0.179) Supplementary Table 2. Association results for HLA class II allele analysis in

Chinese patients with iMN Allele frequency Allele frequency OR P in cases in controls DRB1*1501 0.3755 0.1469 4.65 1.67×10-21

DRB1*0901 0.03065 0.1494 0.16 1.29×10-13

DQB1*0602 0.3238 0.1544 2.82 2.84×10-10

DQB1*0303 0.04215 0.1444 0.24 7.54×10-10

DQA1*0301 0.0249 0.1102 0.19 2.43×10-9

DRB1*0301 0.1207 0.0384 3.96 3.95×10-9

DQA1*0302 0.03065 0.1035 0.28 3.53×10-6

DQA1*0102 0.4004 0.2621 2.11 1.55×10-5

DPB1*0402 0.05192 0.1064 0.42 0.00502

DQB1*0501 0.03065 0.06928 0.39 0.01351

DQA1*0201 0.04789 0.08598 0.49 0.03962

DQB1*0605 0.009579 0 26.01 0.05294

DRB1*0803 0.01533 0.04591 0.35 0.10199

DQA1*0505 0.113 0.06678 1.78 0.10843

DRB1*1201 0.01724 0.04508 0.35 0.15093

DQB1*0401 0.01533 0.0384 0.4 0.29328

DQA1*0401 0.003831 0.01503 0.18 0.37346

DRB1*1454 0 0.01169 0.08 0.37996

DRB1*0405 0.01724 0.04257 0.41 0.38934

DQB1*0302 0.02299 0.04591 0.47 0.49727

DQA1*0501 0.1207 0.08932 1.49 0.58263

DQB1*0609 0 0.009182 0.1 0.88542

DPB1*3301 0 0.03685 0.02 1

DRB1*0407 0 0.001669 0.46 1

DRB1*0801 0 0.001669 0.46 1 DRB1*1106 0 0.001669 0.46 1

DQB1*0203 0.001916 0.0008347 2.32 1

DQB1*0305 0.001916 0.0008347 2.32 1

DRB1*0907 0.001916 0.0008347 2.32 1

DRB1*1103 0.001916 0.0008347 2.32 1

DPB1*10001 0 0.001675 0.46 1

DPB1*1101 0 0.001675 0.46 1

DPB1*2002 0 0.001675 0.46 1

DPB1*20401 0 0.001675 0.46 1

DPB1*2401 0 0.001675 0.46 1

DPB1*5701 0 0.001675 0.46 1

DPB1*6801 0 0.001675 0.46 1

DPB1*4801 0.001923 0.0008375 2.32 1

DPB1*3801 0.003846 0 11.65 1

DRB1*0410 0.001916 0.001669 1.39 1

DPB1*3601 0.001923 0.001675 1.39 1

DPB1*0302 0 0.002513 0.33 1

DPB1*0601 0 0.002513 0.33 1

DPB1*3501 0 0.002513 0.33 1

DPB1*4101 0 0.002513 0.33 1

DPB1*4501 0 0.002513 0.33 1

DQA1*0506 0 0.003339 0.26 1

DPB1*2001 0 0.00335 0.26 1

DPB1*2101 0.003846 0.001675 2.32 1

DPB1*10501 0.001923 0.002513 1.39 1

DPB1*1901 0.005769 0.0008375 5.45 1

DRB1*1407 0.003831 0.002504 1.66 1

DRB1*1503 0.003831 0.002504 1.66 1 DRB1*1504 0.007663 0.0008347 7.06 1

DPB1*1601 0 0.004188 0.21 1

DPB1*2301 0 0.004188 0.21 1

DPB1*2402 0 0.004188 0.21 1

DPB1*10401 0.005769 0.001675 3.26 1

DPB1*1001 0.003846 0.002513 1.66 1

DRB1*0404 0 0.005008 0.18 1

DPB1*2601 0.001923 0.004188 0.63 1

DRB1*0402 0.005747 0.003339 1.29 1

DRB1*0102 0.003831 0.004174 1.05 1

DPB1*1801 0 0.005863 0.15 1

DRB1*1403 0.003831 0.005008 0.89 1

DRB1*1404 0.005747 0.005843 1.08 1

DPB1*0101 0.003846 0.0067 0.68 1

DRB1*0802 0.005747 0.006678 0.68 1

DQB1*0402 0.009579 0.005843 1.4 1

DRB1*1312 0.005747 0.008347 0.85 1

DQA1*0503 0.01149 0.007513 1.6 1

DQA1*0508 0.005747 0.01085 0.6 1

DQA1*0105 0.007663 0.01002 0.83 1

DRB1*1104 0.01724 0.008347 2.14 1

DRB1*0403 0.003831 0.01503 0.31 1

DQB1*0604 0.009579 0.01336 0.77 1

DPB1*13501 0.01538 0.01173 1.47 1

DRB1*0401 0.007663 0.01586 0.53 1

DPB1*1401 0.01731 0.01256 1.37 1

DRB1*1001 0.009579 0.01669 0.61 1

DQB1*0603 0.02299 0.01252 1.75 1 DRB1*1301 0.02299 0.01419 1.55 1

DRB1*0101 0.01149 0.02254 0.56 1

DPB1*0901 0.03269 0.01424 2.26 1

DRB1*0406 0.01149 0.02504 0.41 1

DRB1*1602 0.01916 0.0217 0.95 1

DRB1*1401 0.0249 0.0192 1.35 1

DRB1*1302 0.01533 0.03172 0.49 1

DRB1*1502 0.02874 0.03005 1.01 1

DPB1*1701 0.01923 0.03518 0.58 1

DPB1*0202 0.02885 0.0335 0.82 1

DRB1*1405 0.03257 0.03255 1.03 1

DQA1*0101 0.02874 0.03589 0.84 1

DPB1*1301 0.03462 0.03434 1.06 1

DQA1*0303 0.0249 0.04257 0.59 1

DQB1*0202 0.03448 0.04341 0.82 1

DPB1*0301 0.03846 0.04606 0.89 1

DQA1*0601 0.03448 0.04841 0.68 1

DQA1*0104 0.06513 0.04257 1.6 1

DRB1*1202 0.03257 0.05676 0.59 1

DQB1*0601 0.06322 0.05008 1.23 1

DQB1*0502 0.06322 0.05342 1.21 1

DQB1*0503 0.05939 0.06427 0.87 1

DRB1*1101 0.07854 0.06344 1.31 1

DQA1*0103 0.07854 0.06427 1.11 1

DRB1*0701 0.05364 0.08765 0.61 1

DPB1*0401 0.1596 0.134 1.36 1

DQB1*0301 0.1648 0.1836 0.77 1 Supplementary Table 3. Analysis of gene-gene interactions: odds ratios for iMN according to SNP rs3749117 (PLA2R1) and

HLA-DRB1*1501/DRB1*0301 combinations.

HLA-DRB1*1501 HLA-DRB1*0301 SNP rs3749117 (PLA2R1) (-) (+) (-) (+)

No. of cases/controls 3/22 0/8 2/28 1/2 CC Odds ratio (95% CI) 1 - 1 - (Low risk) P-value (ref) 0.56 (ref) 0.26

No. of cases/controls 23/138 16/39 26/155 13/22 TC Odds ratio (95% CI) - - - 8.27 (1.69-40.57) (Moderate risk) P-value 1.0 0.16 0.38 0.007

No. of cases/controls 47/139 106/51 122/179 31/11 TT Odds ratio (95% CI) - 15.24 (4.36-53.29) 9.54 (2.23-40.80) 39.46 (8.04-193.66) (High risk) P-value 0.21 1.58×10-7 1.07×10-4 7.72×10-9 Supplementary Table 4. Analysis of gene-gene interactions: odds ratios for iMN according to SNP rs35771982 (PLA2R1) and

HLA-DRB1*1501/DRB1*0301 combinations.

HLA-DRB1*1501 HLA-DRB1*0301 rs35771982(PLA2R1) (-) (+) (-) (+)

No. of cases/controls 3/22 0/8 2/28 1/1 CC Odds ratio (95% CI) - - - - (Low risk) P-value (ref) 0.56 (ref) 0.18

No. of cases/controls 23/138 16/39 26/156 13/21 GC Odds ratio (95% CI) - - - 8.67 (1.76-42.61) (Moderate risk) P-value 1.0 0.16 0.38 0.003

No. of cases/controls 47/140 106/51 122/179 31/12 GG Odds ratio (95% CI) - 15.24 (4.36-53.29) 9.54 (2.23-40.80) 36.17 (7.44-175.90) (High risk) P-value 0.21 1.58×10-7 1.07×10-4 1.23×10-8 Supplementary Table 5. Study power as a function of disease allele frequency and odd ratios.

Allele OR for a risk allele

frequency 1.2 1.4 1.6 1.8 2.0 3.0 4.0 5.0

0.1% 0.06 0.07 0.08 0.09 0.12 0.21 0.22 0.40

0.5% 0.06 0.08 0.11 0.14 0.17 0.35 0.53 0.67

1% 0.07 010 0.14 0.19 0.39 0.54 0.77 0.99

3% 0.08 0.16 0.26 0.38 0.51 0.90 0.99 1.00

5% 0.10 0.21 0.37 0.54 0.92 1.00 1.00 1.00

10% 0.13 0.33 0.57 0.77 0.99 1.00 1.00 1.00

15% 0.16 0.42 0.70 0.88 1.00 1.00 1.00 1.00 Supplementary Figure 1. The structure of DR2b-peptide 285

(SKTVEVW(M/V)GLNQLDE)-TCR with amino acid substitution M292V. Gray cartoon is peptide, cyan cartoon is MHC DR2b molecule, and magenta cartoon is TCR molecule. Blue stick is methionine, and aubergine stick is valine. Amino acid substitution of M292V in peptide 285 [SKTVEVW(M/V)GLNQLDE] does not influence the presentation modeling by DR2b. S, serine; K, lysine; T, threonine; V, valine; E, glutamic acid; W, tryptophan; M, methionine; G, glycine; L, leucine; N, asparagine; Q, glutarnine; D, aspartic acid.