Summary
Goodpasture's disease is closely associated with HLA, particularly DRB1*1501. Other susceptible or protective HLA alleles are not clearly elucidated. The presentation models of epitopes by susceptible HLA alleles are also unclear. We genotyped 140 Chinese patients and 599 controls for four‐digit HLA II genes, and extracted the encoding sequences from the IMGT/HLA database. T‐cell epitopes of α3(IV)NC1 were predicted and the structures of DR molecule‐peptide—T‐cell receptor were constructed. We confirmed DRB1*1501 (OR = 4·6, P = 5·7 × 10−28) to be a risk allele for Goodpasture's disease. Arginine at position 13 (ARG13) (OR = 4·0, P = 1·0 × 10−17) and proline at position 11 (PRO11) (OR = 4·0, P = 2·0 × 10−17) on DR β1, encoded by DRB1*1501, were associated with disease susceptibility. α 134–148 (HGWISLWKGFSFIMF) was predicted as a T‐cell epitope presented by DRB1*1501. Isoleucine137, tryptophan140, glycine142, phenylalanine143 and phenylalanine145, were presented in peptide‐binding pockets 1, 4, 6, 7 and 9 of DR2b, respectively. ARG13 in pocket 4 interacts with tryptophan140 and forms a hydrogen bond. In conclusion, we propose a mechanism for DRB1*1501 susceptibility for Goodpasture's disease through encoding ARG13 and PRO11 on MHC‐DR β1 chain and presenting T‐cell epitope, α 134–148, with five critical residues.
Keywords: anti‐ glomerular basement membrane disease, epitope, Goodpasture's disease, HLA, MHC
Introduction
Goodpasture's disease, also called anti‐glomerular basement membrane (GBM) disease, is a rare but severe autoimmune disease characterized by the presence of anti‐GBM autoantibodies and the rapid progression of glomerulonephritis, frequently accompanied by pulmonary haemorrhage.1, 2 The major target antigen is non‐collagenous domain 1 (NC1) of the α3 chain of collagen IV, which intertwines, forming an α345 triple helical network in the basement membrane of glomerular and alveolar capillaries. The α3NC1 domain harbours two well‐defined B‐cell epitopes: EA and EB. 3, 4 The linear peptide containing EB, designated as P14 (α3127–148) (TDIPPCPHGWISLWKGFSFIMF), was demonstrated as a T‐cell epitope5 and was recognized by antibodies from patients with anti‐GBM disease.6
This disease is an ideal model for studies on mechanisms of autoimmune disease aetiology, in which genetic predispositions act as ‘first hits’ that combine with ‘second hits’ such as infection and chemical substances.7 Studies on four well‐controlled Caucasian groups reported that most (approximately 80%) patients with anti‐GBM disease carry the human leucocyte antigen (HLA)‐DRB1*1501 allele.8 This finding was confirmed in Asian populations.9, 10, 11 The critical T‐cell epitope presented by this molecule is α3136–146, and α3136–146‐specific CD4+ T‐cell clones generated from DRB1*1501 transgenic mice could transfer the disease to naive DRB1*1501 transgenic mice.12 However, other HLA‐DRB1 alleles also have hierarchical influence on disease susceptibility.8, 13 Furthermore, HLA‐DRB1 alleles present different linkage disequilibria with other HLA class II alleles, i.e. the DQA1 and DQB1 alleles.14 These linkages might interfere with the results interpretation for HLA‐DRB1 and conceal significant findings for other alleles.
HLA II molecules possess distinctive peptide‐binding characteristics that selectively bind and present antigen‐derived peptides to CD4+ T cells and influence autoimmune responses.15, 16 Structural comparisons of HLA II molecules associated with anti‐GBM disease are helpful in interpreting allele analysis findings by distinguishing the involvement of these alleles from linkage disequilibrium.8 Defining the pathogenic interactions of HLA molecules with autoantigenic peptides is also important for developing more specific immunotherapies,17 such as synthetic peptides that block these interactions, induce tolerance, or act as T‐cell receptor antagonists. However, few studies have analysed the susceptible amino acid residues on HLA molecules on Goodpasture's disease.
This study enrolled the largest cohort of 140 patients with Goodpasture's disease and 599 geographically matched controls. We typed the HLA‐DRB1, ‐DQA1, ‐DQB1 and ‐DPB1 allele genotypes at four‐digit resolution and investigated the disease associations with HLA class II alleles, haplotypes and risk amino acid residues on the HLA‐DRβ1, ‐DQα1, ‐DQβ1 and ‐DPβ1 chains of HLA molecules. We made a model of the P14 peptide bound to the DR2b molecule to infer elements critical for binding, which may provide information for the future design of immunotherapies.
Materials and methods
Participants
The patient group comprised 140 Chinese patients with anti‐GBM disease who presented with rapidly progressive glomerulonephritis with or without pulmonary haemorrhage. Diagnosis was established in all cases based on the detection of serum anti‐GBM autoantibodies through ELISA using bovine α3(IV)NC1 (Euroimmun, Lübeck, Germany) and recombinant human α3(IV)NC1 as solid‐phase ligands18 and detection of linear immunoglobulin G deposits along the GBM under direct immunofluorescence in all renal biopsies. The control group comprised 599 healthy Chinese individuals voluntarily recruited as blood donors and geographically matched with the patients. This research was performed in compliance with the Declaration of Helsinki and approved by the Peking University First Hospital ethics committee. Written informed consent was obtained from all participants. All of the authors vouch for the accuracy and completeness of the data.
HLA allele genotyping
Peripheral blood samples (10 ml) were collected from patients and controls using anticoagulant EDTA. Genomic DNA was obtained from peripheral blood leucocytes using the Puregene Blood Core Kit C (Cat. No. 158389, Qiagen Science, Hilden, Germany).
HLA‐DRB1, ‐DQB1 and ‐DPB1 sequencing was performed on an Applied Biosystems 3130xl platform using SeCore® Sequencing Kits (Invitrogen, Carllsbad, CA). HLA allele typing reports were issued using utype SBT HLA software. HLA‐DRB1, ‐DQB1 and ‐DPB1 typing was performed using bidirectional sequencing of exon 2, exons 2 and 3, and exons 2, 3 and 4, respectively; HLA‐DQA1 alleles were typed using electrophoresis.
Extraction of amino acid sequences of HLA‐DRβ1, ‐DQα1, ‐DQβ1, and ‐DPβ1 chains
Based on four‐digit resolution HLA alleles, the amino acid sequences of the HLA‐DRβ1, ‐DQα1, ‐DQβ1 and ‐DPβ1 chains were extracted from the IMGT/HLA public database (http://www.ebi.ac.uk/ipd/imgt/hla/).
Association analysis
All genotyping data sets were subjected to the same rigorous quality assessments both before and after cases and controls were compared. Individuals for whom consistent genotypes could not be obtained for each locus were removed from the study. Missing genotypes were reconstructed; Hardy–Weinberg equilibrium was assessed at each locus. HLA alleles were excluded from the analysis if they were out of Hardy–Weinberg equilibrium (P < 0·001) or had minor allele frequency < 0·001. Unconditional and stepwise conditional analyses for HLA class II alleles and the amino acid positions of HLA‐DRB1, ‐DQA1, ‐DQB1 and ‐DPB1 were performed through logistic regression using PLINK version 1.9 (http://www.cog-genomics.org/plink/1.9/).19 HLA class II allele haplotypes were constructed using the EM algorithm with R Project 3.1.1 (haplo.stats package) (http://www.r-project.org).20 Differences were considered significant when P < 0·05 per distinct allele or per distinct amino acid position.
Markers were obtained and encoded to determine association analysis. Bi‐allelic markers with two HLA alleles or two‐residue amino acid positions were encoded as alleles 1 and 2. Multi‐allelic (three or more alleles) markers were encoded as the presence and absence of each allele of multi‐residue amino acid positions. A log likelihood ratio test, estimating the log likelihood differences reflecting the effects at a single amino acid position, was used to evaluate the anti‐GBM disease association at an amino acid position. Associations with anti‐GBM disease risk were evaluated using logistic regression analysis.21
Modeller
The amino acid sequence of type IV collagen α3 chain was extracted from Protein (http://www.ncbi.nlm.nih.gov/protein). We used syfpeithi online database (http://www.syfpeithi.de/) to predict the T‐cell epitopes of α3(IV)NC1. All predicted MHC class II ligands are 15‐mers, consisting of three N‐terminal flanking residues, the nonamer core sequence located within the binding groove, and three C‐terminal flanking residues. Hence, anchor residue P1 appears in position 4 of the peptides predicted with syfpeithi.22 The protein crystal structure of 1YMM (http://www.rcsb.org/pdb/home/home.do) was used as a template to construct DR2b‐peptide‐TCR structural models using the modeller program (version 9V8, http://salilab.org/modeller/).23, 24, 25, 26, 27
We moved the peptides (PCPHGWISLWKGFSFIMF) ‘step by step’ and sought the optimum structure. modeller refine was set very slowly and optimization was repeated six times. Ending model was set as 800. The conformations of 800 compounds were exported from every ‘move’ and manually evaluated according to the following criteria. We measured the distance between amino acid residues of α3(IV)NC1 peptides and corresponding peptide binding pockets at the first step (DR2b: distance‐pocket 1 ≤ 4 Å, distance‐pocket 4 ≤ 4 Å, distance‐pocket 7 ≤ 4 Å). We chose the position with the shortest distance as the optimum position. Then we measured distance between the guanidino nitrogen atom of arginine on position 13 encoded by DRB1*1501 and the carbonyl oxygen atom of the main chain of the amino acid residue located in pocket 4. We set the distance below 3·5 Å and chose the conformations with shortest distance as the optimum structure.
Results
Population characters
The patient group comprised 140 Chinese Han patients with biopsy‐proven anti‐GBM disease. The control group comprised 599 healthy, voluntarily recruited Chinese Han blood donors who had been racially matched with the patients.
Association of HLA alleles with Goodpasture's disease
A total of 141 distinct HLA alleles were identified in the patients and controls: 46 DRB1, 19 DQA1, 22 DQB1 and 54 DPB1. The threshold for statistical significance was set as P < 0·05/141 (3·6 × 10−4) in classic HLA allele association analysis.
Unconditional analysis revealed a positive association between the disease and DRB1*1501 (OR = 4·6, P = 5·7 × 10−28), DQB1*0602 (OR = 3·3, P = 2·0 × 10−17) and DQA1*0502 (OR = 30·7, P = 7·0 × 10−7) alleles. There were negative associations between the disease and DQB1*0303 (OR = 0·3, P = 8·6 × 10−6) and DRB1*0901 (OR = 0·4, P = 1·6 × 10−4) alleles (Table 1, Fig. 1a).
Table 1.
Association results for HLA allele analysis in Chinese patients with anti‐glomerular basement membrane disease (P < 3·6 × 10−4)
| Allele | Allele frequency in cases (n = 140 × 2) (%) | Allele frequency in controls (n = 599 × 2) (%) | Genotype frequency in cases (n = 140) (%) | Genotype frequency in controls (n = 599) (%) | OR | P‐value | Sensitivity (%) | Specificity (%) |
|---|---|---|---|---|---|---|---|---|
| Susceptible allele | ||||||||
| DRB1*1501 | 43·8 | 14·7 | 78·6 | 26·2 | 4·6 | 5·7 × 10−28 | 78·6 | 73·8 |
| DQB1*0602 | 38·4 | 15·4 | 63·6 | 29·5 | 3·3 | 2·0 × 10−17 | 63·6 | 70·5 |
| DQA1*0502 | 2·6 | 0·1 | 5·0 | 0·2 | 30·7 | 7·0 × 10−7 | 5·0 | 99·8 |
| Protective allele | ||||||||
| DQB1*0303 | 4·7 | 14·4 | 9·3 | 27·4 | 0·3 | 8·6 × 10−6 | 27·4 | 92·7 |
| DRB1*0901 | 6·5 | 14·9 | 12·9 | 28·1 | 0·4 | 1·6 × 10−4 | 28·1 | 87·1 |
Figure 1.
Associations of HLA class II alleles with anti‐glomerular basement membrane (GBM) disease. (a) Unconditional analysis results showing significant associations between DRB1*1501, DQB1*0602, DQA1*0502, DRB1*0901, and DQB1*0303 and anti‐GBM disease. (b) No HLA allele was associated with anti‐GBM disease after conditioning on DRB1*1501. (c) DRB1*1501 was associated with anti‐GBM disease risk after conditioning on DQB1*0602. [Colour figure can be viewed at wileyonlinelibrary.com]
After adjusting for DRB1*1501 under stepwise conditional analysis, no other HLA allele showed an association with the disease, reflecting the strong influence of DRB1*1501 on anti‐GBM disease (Fig. 1b). When adjusting for DQB1*0602, the close association between DRB1*1501 and anti‐GBM disease (P = 1·7 × 10−9) remained (Fig. 1c).
Linkage disequilibrium among HLA alleles
Linkage disequilibrium was analysed among the three susceptibility alleles (DRB1*1501, DQB1*0602, DQA1*0502) and two protective alleles (DRB1*0901, DQB1*0303) described above.
Risk allele, DRB1*1501 was in overt linkage disequilibrium with DQB1*0602 (D’ = 0·56, r 2 = 0·31). The protective allele DRB1*0901 was in overt linkage disequilibrium with DQB1*0303 (D’ = 0·56, r 2 = 0·31). However, the susceptibility allele DQA1*0502 was not in linkage disequilibrium with the other two susceptibility alleles or the protective alleles (Table 2).
Table 2.
Linkage analysis among the significant HLA alleles with anti‐glomerular basement membrane disease
| Allele‐1 | Allele‐2 | D′ | r 2 |
|---|---|---|---|
| DRB1*1501 | DQB1*0602 | 0·56 | 0·31 |
| DRB1*1501 | DQB1*0303 | 0·82 | 0·02 |
| DRB1*1501 | DQA1*0502 | 0·24 | 0·001 |
| DQB1*0602 | DRB1*0901 | 0·69 | 0·02 |
| DQB1*0602 | DQA1*0502 | 0·46 | 0·01 |
| DQB1*0303 | DRB1*0901 | 0·64 | 0·39 |
| DQB1*0303 | DQA1*0502 | 1·00 | 0·001 |
| DQA1*0502 | DRB1*0901 | 0·15 | 0 |
HLA haplotypes associated with Goodpasture's disease
The association of HLA haplotypes with Goodpasture's disease was analysed using the EM algorithm via R Project.3.1.1.20 Distinct HLA DR‐DQ haplotypes were found in our cases and healthy controls, and we set P = 1·0 × 10−4 as the threshold for statistical significance.
The analysis revealed positive associations between HLA haplotypes and anti‐GBM disease. When two loci were analysed together, DRB1*1501‐DQB1*0602 haplotype was positively associated with the disease (score = 8·8, P = 9·7 × 10−19). When three loci were analysed in combination, DRB1*1501‐DQA1*0102‐DQB1*0602 was positively associated with the disease (score = 4·6, P = 1·0 × 10−6). When four loci were analysed, DRB1*1501‐DQA1*0102‐DQB1*0602‐DPB1*0201 was positively associated with the disease (score = 3·5, P = 4·2 × 10−4) (Table 3).
Table 3.
Extended haplotype analysis among HLA‐DRB1, ‐DQA1, ‐DQB1 and ‐DPB1 in Chinese patients with anti‐glomerular basement membrane disease (P < 1·0 × 10−4)
| Extended haplotype | Haplotype frequency in cases (n = 140 × 2) (%) | Haplotype frequency in controls (n = 599 × 2) (%) | Haplotype score | P‐value |
|---|---|---|---|---|
| Susceptible haplotype | ||||
| DRB1*1501‐DQB1*0602 | 29·0 | 9·5 | 8·8 | 9·7 × 10−19 |
| DRB1*1501‐DRB1*0601 | 2·7 | 0·8 | 4·1 | 3·8 × 10−5 |
| DRB1*1501‐DQA1*0102‐DQB1*0602 | 19·0 | 9·0 | 4·6 | 1·0 × 10−6 |
| DRB1*1501‐DQA1*0102‐DQB1*0602‐DPB1*0201 | 8·4 | 3·4 | 3·5 | 4·2 × 10−4 |
| Protective haplotype | ||||
| DRB1*0901‐DQB1*0303 | 2·2 | 10·0 | −4·3 | 1·6 × 10−5 |
| DRB1*0901‐DQA1*0302‐DQB1*0303 | 1·1 | 9·2 | −4·5 | 7·3 × 10−6 |
| DRB1*0901‐DQA1*0302‐DQB1*0303‐DPB1*0501 | 0·4 | 5·6 | −3·5 | 4·2 × 10−4 |
HLA haplotype analysis also revealed negative associations between haplotypes and anti‐GBM disease. DRB1*0901‐DQB1*0303 (haplotype score = −4·3, P = 1·6 × 10−5), DRB1*0901‐DQA1*0302‐DQB1*0303 (score = −4·5, P = 7·3 × 10−6), and DRB1*0901‐DQA1*0302‐DQB1*0303‐DPB1*0501 (score = −3·5, P = 4·2 × 10−4) all showed negative associations with the disease (Table 3).
Association of HLA amino acid residues with Goodpasture's disease
The amino acid residues of the HLA‐DRβ1, ‐DQα1, ‐DQβ1 and ‐DPβ1 chains were analysed for associations with anti‐GBM disease. We successfully imputed 152 polymorphic amino acid positions among DRβ1, DQα1, DQβ1 and DPβ1 based on the corresponding HLA alleles at four‐digit resolution. Hence we set P < 0·05/152 (3·3 × 10−4) as the threshold for statistical significance.
Unconditional analysis revealed that the most significant association was mapped to the amino acid position 13 (P = 1·3 × 10−24) on HLA‐DRβ1 chain (Table 4, Fig. 2a), which locates in the fourth peptide‐binding pocket. Arginine (ARG) at position 13 (ARG13) showed a risk effect for the disease (OR = 4·0, P = 1·0 × 10−17). Serine (SER) 13 and phenylalanine (PHE) 13 were protective against the disease (SER13: OR = 0·4, P = 4·9 × 10−6; PHE13: OR = 0·4, P = 1·6 × 10−4) (Table 4, Fig. 3).
Table 4.
Effects of individual amino acids at positions 13 and 11 on HLA‐DRβ1 (P < 3·3 × 10−4)
| Position | Amino acid | Case frequency (%) | Control frequency (%) | Observed classical HLA‐DRB1alleles | OR | P‐value |
|---|---|---|---|---|---|---|
| DRβ1: 13 | TYR | 5·4 | 13·7 | 0·5 | 0·03 | |
| SER | 9·3 | 21·2 | 0301, 1101, 1104, 1106, 1301, 1302, 1312, 1401, 1403, 1405, 1454 | 0·4 | 4·9 × 10 −6 | |
| ARG | 48·8 | 5·9 | 1501, 1502, 1504, 1602 | 4·0 | 1·0 × 10 −17 | |
| HIS | 16·5 | 20·0 | 1·8 | 0·003 | ||
| GLY | 10·3 | 15·0 | 0·5 | 0·006 | ||
| PHE | 9·7 | 23·9 | 0101, 0102, 0901, 1001 | 0·4 | 1·6 × 10 −4 | |
| DRβ1: 11 | ASP | 6·4 | 18·0 | 0901 | 0·4 | 3·0 × 10 −4 |
| GLY | 5·4 | 13·7 | 0·5 | 0·03 | ||
| LEU | 2·2 | 5·1 | 0·8 | 0·67 | ||
| PRO | 48·8 | 5·9 | 1501, 1502, 1504, 1602 | 4·0 | 1·0 × 10 −17 | |
| SER | 19·6 | 36·2 | 0301, 0801, 0802, 0803, 0804, 0809, 1101, 1104, 1106, 1201, 1202, 1301, 1302, 1312, 1401, 1403, 1404, 1405, 1454 | 0·4 | 2·5 × 10 −9 | |
| VAL | 17·6 | 20·8 | 1·6 | 0·02 |
PHE, phenylalanine; GLY, glycine; HIS, histidine; ARG, arginine; SER, serine; TYR, tyrosine; ASP, aspartic acid; LEU, leucine; PRO, proline; VAL, valine. P < 3·3*10−4 are shown in bold.
Figure 2.
Association between amino acid positions on MHC II molecules and anti‐glomerular basement membrane (GBM) disease. (a) Unconditional analysis results showing that amino acid positions 13 and 11 on HLA‐DR β1 were most closely associated with anti‐GBM disease risk. (b) Controlling for amino acid position 13 could completely eliminate the association of position 11 with the disease. (c) Controlling for amino acid position 11 could completely eliminate the association of position 13 with the disease. (d) No amino acid position was associated with the disease after conditioning on both positions 13 and 11 on HLA‐DR β1. [Colour figure can be viewed at wileyonlinelibrary.com]
Figure 3.
Effects of individual amino acids at positions 11 and 13 on MHC‐DR β1. The amino acid frequencies of amino acid positions 11 and 13 on HLA‐DR β1 in the cases and controls are plotted, and the univariate odds ratios (OR) are listed. PHE, phenylalanine; GLY, glycine; HIS, histidine; ARG, arginine; SER, serine; TYR, tyrosine; ASP, aspartic acid; LEU, leucine; PRO, proline; VAL, valine. [Colour figure can be viewed at wileyonlinelibrary.com]
A similar level of significance was observed at position 11 (P = 2·8 × 10−24) on DRβ1 (Table 4, Fig. 2a), which locates in the sixth peptide‐binding pocket. Proline (PRO) 11 increased anti‐GBM disease risk (OR = 4·0, P = 2·0 × 10−17), while SER11 and aspartic acid (ASP) 11 were negatively associated with the disease (SER11: OR = 0·4, P = 2·5 × 10−9; ASP11: OR = 0·4, P = 3·0 × 10−4) (Table 4, Fig. 3).
DRB1*1501 encodes both PRO11 and ARG13. DRB1*0901 encodes both ASP11 and PHE13. There was extremely close correlation (LOD = 98) between positions 11 and 13 on DRβ1. Conditional analysis adjusting for position 11 or 13 completely eliminated the association of the other position with disease (Fig. 2b,c). No amino acid position was associated with disease risk after conditioning on both positions 13 and 11 (Fig. 2d).
Modelling of P14 presented by susceptible DR2b molecule
We used the syfpeithi online database22 to predict T‐cell epitopes of α3(IV)NC1 presented by DRB1*1501. Peptide (HGWISLWKGFSFIMF), α3134–148, was found with the highest score of 30. It is inside the T‐cell epitope α3129–148, which is revealed previously from anti‐GBM patients.5
We constructed the structure of DR2b‐peptide (PCPHGWISLWKGFSFIMF) using the modeller program.23, 24, 25, 26 The optimum structure is shown in Fig. 4(a). The amino acid residues, isoleucine137, tryptophan140, glycine142, phenylalanine143 and phenylalanine145, were presented in peptide‐binding pockets 1, 4, 6, 7, 9 of DR2b separately (Table 5, Fig. 4a). The distance between guanidino nitrogen atom of ARG13, encoded by DRB1*1501, and the carbonyl oxygen atom of tryptophan140 backbone in pocket 4 is 3·3 Å, which facilitates the formation of hydrogen bond. The 71st amino acid residue of DR2b β‐chain, ALA71, is a small residue and has no interaction with peptide. However, alanine makes pocket 4 larger than other MHC DR molecules, hence the indole ring of tryptophan140 could stretch into this pocket. Furthermore, tryptophan140 on the peptide and two residues in pocket 4 of DRB1*1501, PHE26 and TYR78, form strong hydrophobic interactions with each other (Table 5, Fig. 4b).
Figure 4.
Structure models of DR2b‐peptide (PCPHGWISLWKGFSFIMF)‐TCR. Grey cartoon is peptide, cyan cartoon is MHC DR molecule, and magenta cartoon is TCR molecule. Magenta sticks are amino acids of pocket 1, yellow sticks are amino acids of pocket 4, blue sticks are amino acids of pocket 6, grey sticks are amino acids of pocket 7, and white sticks are amino acids of pockets 9. The distance between amino acids is shown by red dotted line, and the unit is Å. (a). Overall structure of DR2b‐peptide(PCPHGWISLWKGFSFIMF)‐TCR; (b). tryptophan140 was in pocket 4. [Colour figure can be viewed at wileyonlinelibrary.com]
Table 5.
Amino acids of peptide (PCPHGWISLWKGFSFIMF) in corresponding pockets of DR2b molecule
| Position | Residue encoded by HLA‐DRB1*1501 | Residue of peptide (PCPHGWISLWKGFSFIMF) in pocket | |
|---|---|---|---|
| Pocket 1 | 85 | V | I (α 137) |
| 89 | F | ||
| 86 | V | ||
| Pocket 4 | 13 | R | W (α 140) |
| 71 | A | ||
| 78 | Y | ||
| 70 | Q | ||
| 74 | A | ||
| 26 | F | ||
| Pocket 6 | 9 | W | G (α 142) |
| 11 | P | ||
| 30 | Y | ||
| Pocket 7 | 28 | D | F (α 143) |
| 61 | W | ||
| 71 | A | ||
| 47 | F | ||
| 67 | I | ||
| Pocket 9 | 9 | W | F (α 145) |
| 60 | Y | ||
| 57 | D | ||
| 37 | S | ||
| 38 | V |
V, valine; F, phenylalanine; R, arginine; A, alanine; Y, tyrosine; Q, glutarmine; W, tryptophan, P, proline, D, aspartic acid, S, serine, I, ieucine, G, glycine.
Discussion
In this study, we confirmed DRB1*1501 allele and its bearing haplotype (DRB1*1501‐DQA1*0102‐DQB1*0602‐DPB1*0201) were associated with susceptibility to Goodpasture's disease in a large cohort of Chinese patients. Furthermore, we conducted an initial case–control study on HLA analysis at amino acid level. The results demonstrated the effect of coding variants from HLA class II genes and revealed that ARG13 on peptide‐binding pocket 4 on DRβ1, encoded by DRB1*1501, interacts with tryptophan140 on α3134–148 and facilitates the T‐cell epitope presentation of Goodpasture's disease. These findings may clarify the mechanism of DRB1*1501 susceptibility and provide useful information for the future design of immunotherapies through amino acid mutations.
The peptide α3129–148 (IPPCPHGWISLWKGFSFIMF) was recognized by T cells from patients with Goodpasture's disease.5 This peptide was predicted again in the current study as the most likely T‐cell epitope presented by DRB1*1501. High affinity of P14 containing core sequence binding to HLA‐DR15 had also been demonstrated.28, 29 ARG13 in pocket 4 encoded by DRB1*1501 interacts with tryptophan140 on P14 and forms hydrogen bond. ALA71 creates larger space of pocket 4 and facilitates the indole ring of tryptophan140 stretching into the pocket. Furthermore, PHE26 and TYR78 also form strong hydrophobic interactions with tryptophan140 on P14. These risk residues encoded by DRB1*1501 and the molecule structure of MHC‐DR2b help us to understand the mechanism of antigen presentation process for P14 on Goodpasture's disease.
Critical amino acid residues on α3127–148 presented by DRB1*1501 was deduced by molecule modelling. We found isoleucine (I)137, tryptophan (W)140, glycine (G)142, phenylalanine (F)143, phenylalanine (F)145, were presented in peptide‐binding pocket 1, 4, 6, 7, 9 of DR2b separately. Ooi et al.12 characterized a nephritogenic T‐cell epitope on α3(IV)NC1 using HLA‐DRB1*1501 transgenic mice and found that by substitution of V137, W140, G142 or F143 resulted in the loss of T‐cell epitope activity. Goodpasture antigen is a highly conservative protein during evolution. V137 corresponds to I137 in human α3(IV)NC1. Hence four of the critical residues had been proven for T‐cell epitope through animal experiments. The last three residues, glycine142, phenylalanine143 and phenylalanine145, were exactly the critical motif for B‐cell recognition of P14, which was revealed in our previous study from patients with Goodpasture's disease.6 All these studies defined the critical residues on region α3127–148 and obtained highly consistent results. We further deduced the presentation pattern for these critical residues by DRB1*1501. Functional assays are required to confirm the importance of the proposed critical amino acids on T‐cell activation. These findings may be important for further investigations on the disease aetiology and the potential immunotherapies using these residues.
We also revealed a high‐risk association between DQA1*0502 and Goodpasture's disease and few linkage with DRB1*1501. Although DQA1*0502 did not show significant association after conditioning on DRB1*1501, we could not exclude DQA1*0502 as a susceptibility allele for the following reasons. First, DQA1*0502 is a rare allele in healthy individuals. We used the Allele Frequency Net Database (http://www.allelefrequencies.net/) and evaluated 10 independent European studies with a total sample size of 1307 Caucasians, and found that DQA1*0502 allele frequency was 0·038%. No finding on DQA1*0502 has been reported among Chinese populations. Therefore, the 5% allele frequency obtained from patients with anti‐GBM disease in this study is a striking high frequency. Second, DQA1*0502 represents a single C‐to‐G transversion in codon 59 (exon 2), and the results showing an amino acid residue change from proline to arginine, previously demonstrated to be conserved in human DQα1 polypeptides,30 would alter the HLA‐DQ molecule structure and function and influence antigen‐presenting process. Hence, we proposed that DQA1*0502 might be a more specific risk allele for anti‐GBM disease.
Our results revealed that DQB1*0602 is a risk allele for Goodpasture's disease; however, the significant association of this allele may reflect strong linkage with DRB1*1501. Recent studies on DRB1*1501 and DQB1*0602 transgenic mice with ‘humanized’ multiple sclerosis offered a rationale for the pathogenic involvement of DQB1*0602 in susceptibility to multiple sclerosis.31, 32 However, the susceptible role for DQB1*0602 to Goodpasture's disease is insufficient because of the lack of evidence.
In our study, DRB1*0901 was negatively associated with Goodpasture's disease. The haplotypes bearing DRB1*0901 were also protective against the disease. PHE13 and ASP11 are both encoded by DRB1*0901 and are negatively associated with the disease. Since our analysis accounts for a large predominance of DRB1*1501 alleles in patients, its effect on the analysis of other HLA alleles could not be excluded. We used the total number of non‐HLA DRB1*1501 alleles as denominator and still found a negative association of DRB1*0901 (OR = 0·41, P = 0·02) with the disease. These provide strong evidence supporting DRB1*0901 as a protective allele. However, DRB1*07 has a protective effect in Caucasian patients.8 We suggest that it probably reflects a racial difference. In two studies involving 105 Caucasian patients with Goodpasture's disease,33, 34 the DRB1*0701 allele frequency was 2·4% (5/105 × 2), comparable with 5·4% in the cases of the present study (χ 2 = 2·72, P = 0·099). However, among healthy individuals, DRB1*0701 allele frequency was higher in Caucasian population than Chinese population (13·3 versus 10·3%, χ 2 = 142·02, P = 9·63 × 10−33), which we determined using the Allele Frequency Net Database and involved 1 362 965 healthy Caucasian and 8784 healthy Chinese individuals. DRB1*0901 is a rare allele reported in Caucasian populations with a frequency of 0·9%; in contrast, it is commonly carried in Chinese individuals with 13·0% frequency (χ 2 = 14722·08, P = 0). DRB1*0901 allele frequency was 0·5% in 105 Caucasian patients with anti‐GBM disease and 6·4% in the cases of the present study (χ 2 = 11·41, P = 7·32 × 10−4). This racial difference may contribute to the finding that DRB1*0901 is a protective allele against the disease, which needs further validations from Chinese populations.
In summary, DRB1*1501 and its encoding ARG13 and PRO11 in HLA‐DRβ1 chain are associated with the susceptibility to Goodpasture's disease. The presenting epitope is α3134–148 with critical residues of isoleucine137, tryptophan140, glycine142, phenylalanine143 and phenylalanine145.
Disclosures
All of the authors declare no competing interests.
Acknowledgements
The authors greatly appreciate the technical support provided by Ping Hou. Part of the analysis was performed on the Computing Platform of the Center for Life Science. This work is supported by grants from the Natural Science Foundation of China to the Innovation Research Group (81621092), the Outstanding Young Scholar (81622009), and other programmes (81330020, 81370801, 81400703).
References
- 1. Goodpasture EW. Landmark Publication from The American Journal of the Medical Sciences. Am J Med Sci 2009; 338:148–51. [DOI] [PubMed] [Google Scholar]
- 2. Wilson CB, Dixon FJ. Anti‐glomerular basement membrane antibody‐induced glomerulonephritis. Kidney Int 1973; 3:74–89. [DOI] [PubMed] [Google Scholar]
- 3. Pedchenko V, Bondar O, Fogo AB, Vanacore R, Voziyan P, Kitching AR et al Molecular architecture of the Goodpasture autoantigen in anti‐GBM nephritis. N Engl J Med 2010; 363:343–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Cui Z, Zhao M, Jia X, Wang M, Hu S, Wang S et al Antibodies to α5 chain of collagen IV are pathogenic in Goodpasture's disease. J Autoimmun 2016; 70:1–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Cairns LS, Phelps RG, Bowie L, Hall AM, Saweirs WW, Rees AJ et al The fine specificity and cytokine profile of T‐helper cells responsive to the alpha3 chain of type IV collagen in Goodpasture's disease. J Am Soc Nephrol 2003; 14:2801–12. [DOI] [PubMed] [Google Scholar]
- 6. Jia XY, Cui Z, Yang R, Hu SY, Zhao MH. Antibodies against linear epitopes on the Goodpasture autoantigen and kidney injury. Clin J Am Soc Nephrol 2012; 7:926–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Couser WG, Johnson RJ. The etiology of glomerulonephritis: roles of infection and autoimmunity. Kidney Int 2014; 86:905–14. [DOI] [PubMed] [Google Scholar]
- 8. Phelps RG, Rees AJ. The HLA complex in Goodpasture's disease: a model for analyzing susceptibility to autoimmunity. Kidney Int 1999; 56:1638–53. [DOI] [PubMed] [Google Scholar]
- 9. Kitagawa W, Imai H, Komatsuda A, Maki N, Wakui H, Hiki Y et al The HLA‐DRB1*1501 allele is prevalent among Japanese patients with anti‐glomerular basement membrane antibody‐mediated disease. Nephrol Dial Transplant 2008; 23:3126–9. [DOI] [PubMed] [Google Scholar]
- 10. Yang R, Cui Z, Zhao J, Zhao MH. The role of HLA‐DRB1 alleles on susceptibility of Chinese patients with anti‐GBM disease. Clin Immunol 2009; 133:245–50. [DOI] [PubMed] [Google Scholar]
- 11. Luo H, Chen M, Cui Z, Yang R, Xu PC, Zhou XJ et al The association of HLA‐DQB1, ‐DQA1 and ‐DPB1 alleles with anti‐ glomerular basement membrane (GBM) disease in Chinese patients. BMC Nephrol 2011; 12:21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Ooi JD, Chang J, O'Sullivan KM, Pedchenko V, Hudson BG, Vandenbark AA et al The HLA‐DRB1*15:01‐restricted Goodpasture's T cell epitope induces GN. J Am Soc Nephrol 2013; 24:419–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Ooi JD, Holdsworth SR, Kitching AR. Advances in the pathogenesis of Goodpasture's disease: from epitopes to autoantibodies to effector T cells. J Autoimmun 2008; 31:295–300. [DOI] [PubMed] [Google Scholar]
- 14. Begovich AB, McClure GR, Suraj VC, Helmuth RC, Fildes N, Bugawan TL et al Polymorphism, recombination, and linkage disequilibrium within the HLA class II region. J Immunol 1992; 148:249–58. [PubMed] [Google Scholar]
- 15. Madden DR. The three‐dimensional structure of peptide‐MHC complexes. Annu Rev Immunol 1995; 13:587–622. [DOI] [PubMed] [Google Scholar]
- 16. Stern LJ, Wiley DC. Antigenic peptide binding by class I and class II histocompatibility proteins. Structure 1994; 2:245–51. [DOI] [PubMed] [Google Scholar]
- 17. Gaur A, Fathman CG. Immunotherapeutic strategies directed at the trimolecular complex. Adv Immunol 1994; 56:219–65. [DOI] [PubMed] [Google Scholar]
- 18. Cui Z, Zhao MH. Avidity of anti‐glomerular basement membrane autoantibodies was associated with disease severity. Clin Immunol 2005; 116:77–82. [DOI] [PubMed] [Google Scholar]
- 19. Purcell S, Neale B, Todd‐Brown K, Thomas L, Ferreira MA, Bender D et al PLINK: a tool set for whole‐genome association and population‐based linkage analyses. Am J Hum Genet 2007; 81:559–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Long JC, Williams RC, Urbanek M. An E‐M algorithm and testing strategy for multiple‐locus haplotypes. Am J Hum Genet 1995; 56:799–810. [PMC free article] [PubMed] [Google Scholar]
- 21. Kim K, Bang SY, Lee HS, Okada Y, Han B, Saw WY et al The HLA‐DRbeta1 amino acid positions 11‐13‐26 explain the majority of SLE‐MHC associations. Nat Commun 2014; 5:5902. [DOI] [PubMed] [Google Scholar]
- 22. Rammensee H, Bachmann J, Emmerich NP, Bachor OA, Stevanovic S. SYFPEITHI: database for MHC ligands and peptide motifs. Immunogenetics 1999; 50:213–9. [DOI] [PubMed] [Google Scholar]
- 23. Marti‐Renom MA, Stuart AC, Fiser A, Sanchez R, Melo F, Sali A. Comparative protein structure modeling of genes and genomes. Annu Rev Biophys Biomol Struct 2000; 29:291–325. [DOI] [PubMed] [Google Scholar]
- 24. Webb B, Sali A. Comparative protein structure modeling using MODELLER. Curr Protoc Bioinformatics 2014; 47:5–6. [DOI] [PubMed] [Google Scholar]
- 25. Fiser A, Do RK, Sali A. Modeling of loops in protein structures. Protein Sci 2000; 9:1753–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Sali A, Blundell TL. Comparative protein modelling by satisfaction of spatial restraints. J Mol Biol 1993; 234:779–815. [DOI] [PubMed] [Google Scholar]
- 27. 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 2005; 6:490–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Phelps RG, Jones VL, Coughlan M, Turner AN, Rees AJ. Presentation of the Goodpasture autoantigen to CD4 T cells is influenced more by processing constraints than by HLA class II peptide binding preferences. J Biol Chem 1998; 273:11440–7. [DOI] [PubMed] [Google Scholar]
- 29. Phelps RG, Jones V, Turner AN, Rees AJ. Properties of HLA class II molecules divergently associated with Goodpasture's disease. Int Immunol 2000; 12:1135–43. [DOI] [PubMed] [Google Scholar]
- 30. Zimmerman PA, Phadke PM, Lee A, Elson LH, Aruajo E, Guderian R et al Migration of a novel DQA1* allele (DQA1*0502) from African origin to North and South America. Hum Immunol 1995; 42:233–40. [DOI] [PubMed] [Google Scholar]
- 31. Ramagopalan SV, Maugeri NJ, Handunnetthi L, Lincoln MR, Orton SM, Dyment DA et al Expression of the multiple sclerosis‐associated MHC class II Allele HLA‐DRB1*1501 is regulated by vitamin D. PLoS Genet 2009; 5:e1000369. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Kaushansky N, Altmann DM, David CS, Lassmann H, Ben‐Nun A. DQB1*0602 rather than DRB1*1501 confers susceptibility to multiple sclerosis‐like disease induced by proteolipid protein (PLP). J Neuroinflammation 2012; 9:29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Fisher M, Pusey CD, Vaughan RW, Rees AJ. Susceptibility to anti‐glomerular basement membrane disease is strongly associated with HLA‐DRB1 genes. Kidney Int 1997; 51:222–9. [DOI] [PubMed] [Google Scholar]
- 34. Dunckley H, Chapman JR, Burke J, Charlesworth J, Hayes J, Haywood E et al HLA‐DR and ‐DQ genotyping in anti‐GBM disease. Dis Markers 1991; 9:249–56. [PubMed] [Google Scholar]