Introduction

IgA nephropathy (IgAN; MIM 161950) is one of the most common glomerulonephritis throughout the world, which is characterized by predominant IgA deposition in the glomerular mesangial area.1, 2 The pathogenesis of IgAN is not clear, but it is commonly thought to be connected with both genetic and environmental factors.3, 4 Genetic risk factors are considered to be involved in IgAN, according to the ethnic and geographic distributions, familial clustering, as well as the inter-individual variation in disease course and prognosis.5, 6, 7

In the past two decades, a number of single-locus genetic association studies have been performed for IgAN. The candidate genes involved in adaptive and innate immunity, glycosylation of IgA1 and renin–angiotensin system were investigated,8 including human leukocyte antigen,9, 10 megsin,11, 12 uteroglobin,13selectins14 and several cytokines and inflammatory factors.15, 16, 17 However, most of these studies suffer the limitations of small sample size and insufficient methodologies, and the results have not always been reproducible in independent samples.18 This lack of reproducibility may be due to genetic heterogeneity among different ethnic groups, or may reflect inadequate statistical power of IgAN cohorts.

In addition, individual single-nucleotide polymorphism (SNP) may not exhibit large independent increments in disease risk. The emerging concept suggested that multiple genes may interact with each other in complex ways to induce the occurrence and phenotypic variability of common complex diseases. Gene–gene interactions may be of particular importance in complex diseases.19 Comparing to a single candidate gene approach, the analysis of multiple candidate gene variants and their interactions might present a more powerful approach to study complex diseases.20As for IgAN, till now, there was only one study using the method of gene–gene interaction (FAMHAP and DBMA), which found the interaction between variants of two glycosyltransferase genes (C1GALT1 and ST6GALNAC2) in IgAN.21

MDR (multifactor-dimensionality reduction), as a novel computational algorithm for testing gene–gene interactions,22 has been used in many common complex multifactorial diseases, such as breast cancer, lupus nephritis and rheumatoid arthritis,22, 23, 24 which is thought to be the reliable method to study the gene–gene interaction.

In our study, to understand the genetic effect on IgAN in Chinese Han population, we surveyed the SNPs of 24 candidate genes that involved in the pathways implicated in the development and progress of IgAN, in a large IgAN cohort comprising of 527 cases and 543 age- and gender-matched controls, investigated the interactive effect of these genes on IgAN by using the MDR method. Our results indicated that the variants in C1GALT1 and IL5RA had a potential genetic interactive effect in terms of predisposition of IgAN.

Materials and methods

Study population

Patients were recruited from the Department of Nephrology, The First Affiliated Hospital of Sun Yat-sen University in Guangzhou, China. A total of 527 patients with primary IgAN were enrolled in this study. IgAN was diagnosed by renal biopsy following the criteria of the World Health Organization. Patients with evidence of systemic diseases such as diabetes, chronic liver disease and systemic lupus erythematosus were excluded. Pregnant women were also excluded. The 543 healthy gender- and age-matched unrelated subjects with no history of renal disease or hypertension were recruited as controls on voluntary basis from Guangdong province. The study was approved by the Ethics Committee of The First Affiliated Hospital of Sun Yat-sen University (Guangzhou, China) and all participants gave an written informed consent to participate in our study.

Selection of candidate genes and SNPs

Totally 73 SNPs in 24 candidate genes with potential involvement in the development and progression of IgA nephropathy were selected. In all, 48 tagSNPs of them were chosen to perform the multiplex genotyping with SNP stream Ultra-High Throughput machine (Beckman Coulter, Fulleron, CA, USA), according to the primer compatibility of multiplex PCR.

DNA extraction and genotyping

Venous blood samples were obtained from all subjects. The genomic DNA was extracted using QIAamp DNA Blood Midi Kit (QIAGEN, Hilden, Germany) according to the manufacturer’s protocol. SNP stream Ultra-High Throughput machine (Beckman Coulter) was used to genotype 48 SNPs simultaneously. Sequences surrounding the SNPs were obtained from GenBank (http://www.ncbi.nlm.nih.gov/Genbank/index.html) and submitted to Autoprimer.com (Beckman Coulter). For each SNP, three primers were designed, two for PCR amplification and internal primer with a 5′ DNA sequence tag. Pairs of primers were used to initiate PCR amplification. The internal primers were used to initiate a sequencing reaction that added one labeled base for the alternative nucleotides of each SNP to have distinct labels. The labeled products were separated on an SNP-IT plate consisting of 384 mini-arrays with 48-spots each (Beckman Coulter).

Association testing

The Hardy–Weinberg equilibrium was tested by χ2 statistic. Only SNPs that passed the HWE test in controls were further analyzed. The allele and genotype frequencies in patients and controls were also compared by χ2-test. Logistic regression analyses were performed to detect single-locus associations with dominant, recessive and additive genetic models. Corrected P-values (Pc) were calculated using the Bonferroni inequality method. All of these analyses were conducted with R software, version 2.10.1 (http://www.R-project.org/).

Interaction analysis

The evaluation of gene–gene interactions was performed using the method of MDR, as outlined by Moore,25 which has been implemented in the open-source MDR software package, version 0.6.1 (Hanover, NH, USA), available at http://www.epistasis.org.

Results

Allele and genotype frequencies in IgAN patients and controls

The demographic information of recruited cohort of IgAN patients and healthy controls were shown in Table 1. There were no significant differences between the cases and healthy controls with regard to mean age or gender distribution. Totally, 31 SNPs were successfully genotyped, and detailed information of the SNPs was shown in Table 2. All of them were in Hardy–Weinberg equilibrium in each of the groups. Table 3 summarized the allele and genotype frequencies in IgAN patients and controls, respectively. There were three SNPs that exhibited suggestive association with IgAN by the genotypic test, including tumor necrosis factor (TNF)-α (rs1800629, P=0.014), ST6GALNAC2 (rs9890158, P=0.031) and C1GALT1 (rs1008898, P=0.027). However, none of the associations was found after using the Bonferroni’s inequality method.

Table 1 Characteristics of the study population
Table 2 Detailed information of the 31 successfully genotyped SNPs
Table 3 Allele and genotype frequencies in IgAN patients and controls

Association of SNPs with inheritance models in IgAN

We also tested for association according to different inheritance models using logistic regression (Table 4). We found that the G allele at rs9890158 (ST6GALNAC2) was significantly associated with IgAN under an additive model (P=0.008) and the T allele at rs1008898 (C1GALT1) was significantly associated under a recessive model (P=0.008). However, none of the associations could withstand a multiple comparison adjustment using the Bonferroni’s inequality method.

Table 4 Analysis of IgAN risk according to the mode of inheritance of the alleles

Gene–gene interaction study (MDR) of IgAN

We explored whether SNP–SNP interactions among our candidate loci contribute to the susceptibility of IgAN in this cohort. Table 5 summarized the results of the MDR analysis. The best model in each dimension was shown along with its TA, CVC consistency and significance P-value, as determined by permutation testing. The two-way combination of C1GALT1 (rs1008898 G/T) and IL5RA (rs340833 A/G) was significantly associated with IgAN (P<0.001). In this model, the combination of C1GALT1-GT, IL5RA-GG and coexistence of C1GALT1-TT, IL5RA-GG or AG genotypes were the only high-risk combinations (Figure 1). A higher-level interaction models (third- and fourth-level interaction terms) did not reveal other significant patterns of genetic interactions.

Table 5 Results of multifactor-dimensionality reduction analysis of interactions between SNPs–SNPs contributing to susceptibility IgAN
Figure 1
figure 1

IL5RA-A/G and C1GALT1-G/T combinations associated with IgAN.

Each multi-locus genotype combination is considered high risk when the ratio of cases to controls exceeds a threshold T, equal to the ratio of cases to controls in each population; otherwise, the cell is classified as low risk. High-risk combinations are depicted as darkly shaded cells, low-risk combinations as lightly shaded cells. For each cell, the left bar indicates the cases, the right bar the controls. A full color version of this figure is available at the Journal of Human Genetics journal online.

Discussion

In this study, we selected SNPs of 24 candidate genes, which were related to the development and progression of IgAN, to perform an association study in southern Chinese Han population. Most of these selected genes have been studied previously in different populations (Caucasian and Asian) in IgAN, and many of these genes were found having some association with the susceptibility and progression of IgAN.11, 13, 16, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38 However, in our study, we could not replicate any of the previously described association signals in the 24 high-priority candidate genes, although the relatively larger sample size (>1000 individuals) were used in our study comparing to the previous studies, which tended to produce more reliable results owing to increasing the power to detect authentic associations and decreasing the rates of type I errors.39, 40 In addition, through extensive literature review, the associations of some candidate genes have not always been reproducible in previous studies. For example, polymorphisms in the selectin gene cluster were found to be associated with IgAN in Japanese patients, but not in Caucasians from Canada, France and Finland.26 This lack of reproducibility may be due to genetic heterogeneity among different ethnic people, or may reflect inadequate statistical power of IgAN cohorts.

The analyses of alternative inheritance models in our study detected a suggestive association of rs9890158 in ST6GALNAC2 under an additive model (risk allele: G, P=0.008) and rs1008898 in C1GALT1 under a recessive model (risk allele: T, P=0.008). Recently published association study in Han Chinese population (including 670 IgAN patients and 494 healthy controls) identified the association of ST6GALNAC2 and C1GALT1 genes with IgAN.37, 41 Our results also showed marginal association of these two genes. However, considering a large number of SNPs and models tested, our findings were not robust to Bonferroni’s correction for multiple testing. The reason for the failure of replication here may be due to the potential genetic heterogeneity between different ethnic population as well as different geographic populations of China. Further studies will be needed to investigate this possibility.

As we know, IgAN is the multifactorial complex disease where genetic and environmental factors contributed equally to this disease. It is thought that there were multiple genes involved in the genetics of IgAN, where some of the candidate genes may have the strong effects as ‘susceptibility loci’ for development and progression, and some genes may have modest effects as ‘modifier genes’ for endophenotypic expression.3 Thus, gene–gene interaction perhaps has significant effect on the genetic background of IgAN. Till now, there was only one study about the gene–gene interaction of IgAN,21 which also have the limitation of only focusing on the IgA1 glycosylation in IgAN. So it is important for us to perform the study of the gene–gene interaction in IgAN by checking relatively more candidate genes.

To explore this possibility we utilized the MDR approach,22 a novel computational algorithm for testing gene–gene interactions. MDR as a method for reducing the dimensionality of multi-locus information has reasonable power to identify interactions among two or more loci in relatively small samples and can improve the identification of polymorphism combinations associated with disease risk.20, 42 This method has been used in some common complex diseases and has discovered that the genes jointly contribute to the susceptibility of these diseases.22, 23, 24

In our study, when pairwise interactions were analyzed, a higher risk for susceptibility of IgAN was observed for the simultaneous presence of C1GALT1-GT and IL5RA-GG genotypes, as well as for C1GALT1-TT and IL5RA-GG or AG genotypes. Three- and four-way interaction models did not lead to more specific genetic patterns of interaction. As outlined by Moore and Williams,43 it is difficult to make inferences about the biologic significance of a statistical model of interactions; nonetheless, these two genes (C1GALT1 and IL5RA) have more biologic plausibility with IgAN, which indicated that they were potentially important for the susceptibility of IgAN.

IL5RA encodes the α chain of the interleukin 5 receptor, and signaling through this receptor may be related to modulating Th1/Th2 polarity and immune regulation.44, 45 Also, a study found that IL-5RA expressing B-1a B cells were involved in innate mucosal IgA antibody responses.46 Although there were only one study that showed the association with IL-5RA and IgAN, the polymorphisms of this gene were found having association with some auto-immune diseases, such as asthma47, 48 and atopic dermatitis,49 suggesting that this gene may have some influence on the pathogenesis of immune response for IgAN.

C1GALT1 (core 1 synthase, glycoprotein-N-acetylgalactosamine 3-β-galactosyltransferase) were the coding genes for the key glycosyltransferase (β1,3-galactosyltransferase) in IgA1 O-glycosylation process, which mediate the addition of galactose onto GalNAc residues of IgA1.50 Accumulating evidence suggests that aberrant glycosylation of the hinge region of IgA1 has the influence on the pathogenesis of IgAN,51, 52 including glomerular deposition53, 54 and injury to renal intrinsic cells.55, 56 Till now, there were several studies that found the polymorphism of C1GALT1 related to IgAN, in both the Caucasian and Asia population.21, 57, 58

In our study, although the selected candidate genes have no association with IgAN individually, but after using the MDR method, we can find that the interaction of these two genes (IL5RA and C1GALT1) contribute to the susceptibility of IgAN, which from another aspect demonstrated that IgAN is a complex disease, where the genetic factor contributed to this disease. The genes involved in the disease may have some minor effect, but the interaction of these genes may have more influence on the genetics of IgAN. Also, these two genes indicated that immune response and glycosylation of IgA1 may have some effect on IgAN.

The limitation of our study is that we only selected limited SNPs of some candidate genes from the pathogenesis of IgAN, which may lose some information of other potential susceptibility genes. As of now several GWAS (genome-wide association study) for IgAN have emerged59, 60, 61, 62 and we can comprehensively and accurately get more information about the susceptibility genes for IgAN. More candidate genes will be selected to study the gene–gene interaction through the comprehensive aspect from the whole genome level, where we can provide more solid information of the genetic background of IgAN. This will be the future work for us to do.

In conclusion, using MDR method, we explored the genetic interactive effect of variants in IL5RA and C1GALT1 on IgAN in a large Southern Chinese Han population. Our results suggested that variants in IL5RA and C1GALT1 had interaction on the genetic susceptibility of IgAN, which also demonstrated that IgAN may be associated with specific genetic polymorphisms involved in immune regulation and IgA1 O-glycosylation.