High-density Association Mapping and Interaction Analysis of PLA2R1 and HLA Regions with Idiopathic Membranous Nephropathy in Japanese

Although recent studies showed anti-PLA2R antibody plays a crucial role in idiopathic membranous nephropathy (IMN), detailed HLA mapping and interaction between the HLA genes and PLA2R1 have not been investigated in IMN. We genotyped across the PLA2R1 gene and the HLA region, using 183 IMN patients and 811 healthy controls. Five SNPs around the PLA2R1 gene were significantly associated with IMN. In addition to the two SNPs previously reported to be strongly associated with IMN, rs3749119 and rs35771982 (OR 3.02 and 2.93, P = 3.24E-14 and 4.64E-14, respectively), two novel intronic SNPs (rs2715928 and rs16844715) were also identified as IMN-associated SNPs (OR = 2.30 and 2.51, P = 3.15E-10 and 5.66E-13, respectively). In the HLA gene analysis, DRB1*1501 and DQB1*0602 were strongly associated with IMN (P = 1.14E-11 and 1.25E-11, respectively). The interaction was strongest between HLA-DRB1*15:01 - HLA-DQB1*06:02 and the intronic SNP rs2715928 (OR = 17.53, P = 4.26E-26). Furthermore, positive interaction was also observed between HLA-DRB1*15:01 - HLA-DQB1*06:02 and the missense SNP rs35771982 (OR = 15.91, P = 2.76E-29), which is in strong linkage disequilibrium with 5′UTR SNP rs3749119, and intronic SNP rs16844715 (OR = 15.91, P = 2.30E-26) for IMN. Neither HLA-DRB1*15:01 nor HLA-DQB1*06:02 was associated with steroid responsiveness, overall survival and renal survival during the observation period of mean 11 years though limited number of analysis.

In Japan, IMN was reported to account for 77.9% of total MN patients, while MN was present in 36.8% of primary nephrotic syndrome patients 8 . However, not many genetic studies of IMN in Japanese population have been conducted. Although previous studies have reported the effect of single locus and the genetic interaction of SNPs in PLA2R1 and HLA-DQA1, no study has performed the high-density association mapping of both the PLA2R1 gene and HLA genes to date for identifying the primary polymorphisms and it is still not known the interaction effect of PLA2R1 risk variants and fully detailed analysis of HLA-gene alleles.

Results
PLA2R1 association with IMN. Association analysis of PLA2R1 SNPs in the first set of IMN patients and healthy controls. Fine mapping of PLA2R1 SNPs was performed in the first stage of the study including 53 IMN patients and 419 healthy controls. The characteristics and clinical information of 53 IMN cases were described in Table 1. Of the 15 SNPs genotyped, two SNPs failed in genotyping and were excluded from the study. Significant deviations from Hardy-Weinberg equilibrium (P < 0.05) were not observed for any of the 13 SNPs. Of the13 SNPs included in the association analysis, we found 9 SNPs significantly associated with IMN (P < 0.05) (bold in Supplementary Table 1). When we corrected for the multiple testing, 7 SNPs survived to be significant ( Table 2).
Replication in the second data set and combined analysis. Seven significant SNPs in the first set were attempted to replicate in a total of 130 IMN patients and 392 healthy controls. Five SNPs were successfully replicated in the second set.
All the SNPs that were consistently significant in all data sets were selected in haplotype analysis. Analysis of linkage disequilibrium (LD) pattern showed high LD (r 2 = 0.8) between rs35771982, missense SNP located in exon 5, and rs3749119 located in 5′ UTR region of PLA2R1. Risk haplotype composed of risk alleles (G for rs35771982, A for rs2715928 and C for rs16844715) exhibited strong association with IMN (P = 7.30E-13, OR = 2.29), while protective haplotype including protective alleles (C for rs35771982, G for rs2715928 and T for rs16844715) was found to be protective to IMN (P = 1.84E-14, OR = 0.34). (see Supplementary Table S2).
Association of HLA genes with IMN. Association analysis of HLA genes in the first set of IMN patients and healthy controls. In the discovery stage by full detail analysis, HLA-A, -B, -C, -DRB1, -DQB1 and -DPB1 genotypes were determined in a total of 53 IMN patients and 419 healthy controls. Regarding HLA class I genes, HLA-A*33:03 showed marginal association (P = 0.03, OR = 0.40) with IMN (see Supplementary Table S3). Tendency of negative association was observed in HLA-B*07:02 with IMN while positive association was found with HLA-B*35:01 (see Supplementary Table S4). HLA-C*0704 exhibited a high odds ratio for IMN (P = 5.79E-03, OR = 5.89; see Supplementary Table S5). However, no alleles mentioned above remained significant after correction for multiple comparisons.
Regarding HLA class II genes, HLA-DRB1*15:01 was the most strongly associated allele (P = 7.72E-5, OR = 2.85), and it remained to be significant when P-value was corrected for the number of alleles tested (see Supplementary Table S6). Negative association was also observed in HLA-DRB1*04:05. On the other hand, four HLA-DQB1 alleles showed significant associations with IMN. Among them, HLA-DQB1*06:02 survived to be significant after correction for multiple comparisons (P = 5.12E-4, OR = 2.60) (see Supplementary Table S7). No association was observed between HLA-DPB1 alleles and IMN (see Supplementary Table S8).
Replication in the second sample set and the combined set. The replication study of two HLA class II genes, HLA-DRB1 and HLA-DQB1, was performed in an independent set of 130 IMN patients and 392 healthy controls. A significant positive association was observed for HLA-DRB1*15:01 and HLA-DQB1*06:02 with IMN in the replication stage with P = 1.71E-9, OR = 3.36 and P = 5.14E-10, OR = 3.56, respectively (Tables 3 and 4). Except for HLA-DRB1*13:02 and HLA-DQB1*06:04 which showed marginal association with IMN, none of the significant HLA-DRB1 and HLA-DQB1 alleles in first set was found to have P-value < 0.05.
Genetic interaction analysis between PLA2R1 and HLA risk alleles. We analyzed the total of 177 IMN cases and 792 healthy controls to investigate the interactions between HLA-DRB1*15:01-HLA-DQB1*06:02 and PLA2R1 risk alleles.
HLA and clinical outcome. During the observation period [median 11 years, interquartile range (IQR) 8.5-13 years], 50% increase of serum creatinine was found in eight patients. Among them, four patients developed to end stage renal disease. Ten patients died of heart failure, infection, and traffic accident.
Supplementary Tables S9 and S10 shows overall survival and renal survival according to risk HLA alleles, HLA DRB1*15:01 and HLA DQB1*06:02. Neither risk alleles were associated with clinical outcome. Combination of risk alleles of HLA genes and PLA2R1 were not associated with clinical outcome.

SNP Position
Risk allele

Discussion
The genetic association of PLA2R1 and HLA-DQA1 risk alleles with IMN in the Caucasian populations was reported by a GWAS that included three independent GWAS in three populations 4 . Recent studies following this paper selected several SNPs as representative from PLA2R1 and the HLA regions 7,[9][10][11] . The diversity of HLA genes is well-known in human genetics and tightly bound to disease appearance and severity. Admittedly, it has been reported that IMN in Japanese has a better course and outcome 12 . The purpose of this study was to clarify: (i) the primary risk SNPs of PLA2R1 in Japanese population, (ii) HLA typing in full detail, (iii) the interaction between the risk alleles of PLA2R1 and HLA genes, and (iv) clinical outcome according to risk alleles. We found 7 SNPs within PLA2R1 gene confirmed to be significantly associated with IMN, including a non-synonymous SNP (H [His] ⇒ D [Asp]) and a 5′ UTR SNPs reported in previous studies and additional 5 SNPs. In Japanese population, rs35771982 of PLA2R1 is reported to be most strongly associated with IMN. In agreement with Coenen et al. 9 and Liu et al. 6 we found that G allele of non-synonymous SNP rs35771982 (G/C) showed significantly increased risk of developing IMN. While Coenen et al. 9 also reported that C allele of 5′ UTR rs3749119 raised the risk of IMN, our data showed rs35771982 is in strong LD with 5′ UTR SNP rs3749119. Although Kim et al. 5 reported that C allele of rs35771982 elevated the risk of IMN, our genotyping data is robust considering that we obtained the significant association of rs1511223 located in 3′ UTR with IMN which similar to a report by Saeed et al. 11 .
Interestingly, our study identified the new significant associations of two intronic SNPs, rs2715928 and rs16844715. Both are coincidentally located in the first intron of PLA2R1. Our findings suggest that risk haplotypes from the combination of common variants and SNP-SNP interaction within PLA2R1 region may not explain the low occurrence of IMN in general population. In other words, HLA risk alleles possibly explain IMN development together with PLA2R1 risk alleles.
With respect to HLA association with IMN, recent studies have reported the strong association of HLA-DQA1 SNP rs2187668 with IMN in Caucasian and Chinese populations 4,7 . HLA is highly polymorphic region spanning approximately 7.6 megabase pairs (Mb) of genomic sequence on the short arm of chromosome 6 13 . It has been reported that rs2187668 is a tag SNP for HLA-DRB1*03:01 in northern European populations, and the haplotype including DRB1*03:01 was associated with MN [14][15][16] .
Our study identified the new significant association of HLA-DRB1*15:01 and DQB1*06:02 with IMN in Japanese population. These alleles are well known to form a common haplotype, HLA-DRB1*15:01-DQB1*06:02, in Japanese population 17    HLA-DRB1 could be associated with anti-PLA2R. Anti-PLA2R antibodies detected in 26 of 37 patients with IMN were reported to be IgG4 antibodies 2 . IgG4 co-localized with PLA2R in the immune complex deposition on the glomerular basement membrane in patients with IMN, but not in those with secondary MN. Significantly higher frequency of HLA-DRB1*15 was reported in primary sclerosing cholangitis patients with increased levels of IgG4 24 . HLA association with IgG4-related IMN has not been identified yet, and how or whether IgG4 may relate to IMN is still uncertain.
With respect to HLA class I, no association of HLA class I alleles with IMN has been reported to date. In our analysis of the discovery set, we observed a weak association of HLA-B*07:02 with IMN before correction appearing in 1 of 53 IMN patients (1.9%) and 53 of 419 healthy individuals (12.6%). Additionally, our results showed a weak association of HLA-C*01:02 with IMN before correction, although the association ceased to be significant after multiple correction. We also showed a potential risk of HLA-C*07:04 with Japanese IMN patients. HLA-C*07:04 was reported to increase relative risk for carbamazepine-induced cutaneous adverse reactions in Japanese 25 . These HLA class I alleles are expected to be assessed in larger number of samples.
Our results showed more than additive effects on the risk of IMN among the individuals with risk alleles of   The associations among renal survival, overall survival, steroid responsiveness, and HLA genes were analyzed in this study. This is because the HLA region may explain that both mild severity of IMN in Japanese and the impact of PLA2R on clinical course of IMN. However, even the strongest association of HLA-DRB1*15:01 and DQB1*06:02 with IMN was not related to renal survival, overall survival and steroid responsiveness. The fact provided that HLA-DRB1*15:01 and DQB1*06:02 were only associated with IMN development and that they did not influence the better prognosis particularly observed in Japanese IMN patients. This means that HLA-DRB1*15:01 and -DQB1*06:02 are possibly the risk alleles in the other populations as well.
There are some limitations in the present study. Firstly, titers of PLA2R1 antibody was not measured because the DNA samples were obtained more than ten years ago and most of the patients were either deceased or unable for follow-up. However, median of observation period was as long as 11 years, which is sufficient to evaluate overall and renal survival though the number is small. In addition, although anti-PLA2R antibodies are found in Japanese IMN patients, the prevalence is lower than that of other countries 26 . Our report partially explains why PLA2R1 serum level is not so frequently increased in Japanese IMN. The second limitation is the relatively small sample number, because IMN is not a common disease. Japan Renal Biopsy Registry in 2009 and 2010 announced that annual occurrence of total MN is approximately 200 27 . The sample number of the present study is comparable to that of nationwide, and our results are statistically robust.
In summary, our study has identified PLA2R1 risk variants, the HLA risk alleles and haplotypes for association with IMN in Japanese population. Our study is the first report to perform high-density association mapping including the HLA region and reveal the increased risk of interaction effect between PLA2R1 risk variants and HLA risk haplotype in IMN. Individuals with PLA2R1 and HLA-DRB1*15:01-DQB1*06:02 risk alleles have higher risk of developing IMN while they were not associated with overall and renal survival. We also found novel two intronic SNPs (rs2715928 and rs16844715) and confirmed previous findings of PLA2R1 association with IMN.

Methods
Human subjects and sample collection. This study included a total of 994 subjects comprised of healthy controls and cases with (biopsy-proven) IMN cases was established by kidney biopsy together with other routine clinical procedures and the patients with malignancy, chronic infectious disease including hepatitis B and C viruses, and drug induced secondary MN were excluded from the study. The second set of study that included 130 IMN cases from BioBank Japan and 392 healthy controls was recruited for replication purposes from The Tokyo University Hospital and the Pharma SNP Consortium (Tokyo, Japan). Both the discovery and replication studies included the Japanese individuals.