Research ArticleNephropathy

Low α-defensin gene copy number increases the risk for IgA nephropathy and renal dysfunction

See allHide authors and affiliations

Science Translational Medicine  29 Jun 2016:
Vol. 8, Issue 345, pp. 345ra88
DOI: 10.1126/scitranslmed.aaf2106

In defense of the kidney

Copy number variations play an important role in human disease development. In a new study, Ai et al. have investigated copy number variations of the α-defensin gene in Chinese patients with IgA nephropathy (IgAN) and healthy controls. They show that low copy number of the α-defensin gene increased the risk for IgAN development and renal degeneration. They replicated the risk association in a Caucasian cohort of IgAN patients. In addition, they demonstrated that α-defensin copy number variants showed negative correlations with serum IgA1 and galactose-deficient IgA1. By explaining 4.96% of disease risk and influencing renal dysfunction, α-defensin may be a potential therapeutic target in IgAN.

Abstract

Although a major source of genetic variation, copy number variations (CNVs) and their involvement in disease development have not been well studied. Immunoglobulin A nephropathy (IgAN) is the most common primary glomerulonephritis worldwide. We performed association analysis of the DEFA1A3 CNV locus in two independent IgAN cohorts of southern Chinese Han (total of 1189 cases and 1187 controls). We discovered three independent copy number associations within the locus: DEFA1A3 [P = 3.99 × 10−9; odds ratio (OR), 0.88], DEFA3 (P = 6.55 × 10−5; OR, 0.82), and a noncoding deletion variant (211bp) (P = 3.50 × 10−16; OR, 0.75) (OR per copy, fixed-effects meta-analysis). While showing strong association with an increased risk for IgAN (P = 9.56 × 10−20), low total copy numbers of the three variants also showed significant association with renal dysfunction in patients with IgAN (P = 0.03; hazards ratio, 3.69; after controlling for the effects of known prognostic factors) and also with increased serum IgA1 (P = 0.02) and galactose-deficient IgA1 (P = 0.03). For replication, we confirmed the associations of DEFA1A3 (P = 4.42 × 10−4; OR, 0.82) and DEFA3 copy numbers (P = 4.30 × 10−3; OR, 0.74) with IgAN in a Caucasian cohort (531 cases and 198 controls) and found the 211bp variant to be much rarer in Caucasians. We also observed an association of the 211bp copy number with membranous nephropathy (P = 1.11 × 10−7; OR, 0.74; in 493 Chinese cases and 500 matched controls), but not with diabetic kidney disease (in 806 Chinese cases and 786 matched controls). By explaining 4.96% of disease risk and influencing renal dysfunction in patients with IgAN, the DEFA1A3 CNV locus may be a potential therapeutic target for developing treatments for this disease.

INTRODUCTION

Immunoglobulin A nephropathy (IgAN) is the most common primary glomerulonephritis worldwide (1). The clinicopathologic patterns and the prognosis of IgAN show wide variability, ranging from asymptomatic to rapidly progressive forms (2). The pathogenesis of IgAN is not well understood, but it is generally considered to be a complex, multifactorial disorder, with both genetic and environmental factors contributing to its development (3). Thus far, there have been five genome-wide association studies (GWAS) for IgAN, and multiple susceptibility loci have been identified (48). However, these GWAS loci can only explain 5 to 7.6% of total disease risk (7), and none of them has an effect on disease prognosis.

Copy number variations (CNVs) have been shown to be a major source of genetic variation in the human genome (9, 10) and have significant phenotypic impact by altering gene dosage, disrupting coding sequences, or perturbing long-range gene regulation (11). Although CNVs have been suggested to play important roles in disease development (10, 12, 13), the associations of CNVs, particularly the role complex multiallelic CNVs may play in complex diseases, are poorly studied and understood, mainly because of experimental difficulties in measuring the copy numbers of such variants accurately (14). The α-defensin gene locus (DEFA1A3) is one of these complex multiallelic CNVs, that presents as tandem repeats of a 19-kb sequence unit, each of which contains one copy of DEFA1 or DEFA3 (differing by a single nonsynonymous coding variant within the third exon) as well as several internal biallelic polymorphisms (1517). The protein products of DEFA1A3, HNP1–3 (human neutrophil peptides 1 to 3), are the most abundant of neutrophil granule proteins, and there is evidence to suggest that neutrophils may play an important role in mediating glomerular injury in IgAN (1820). HNP1–3 have also been suggested to play a role in the regulation of the complement system and proinflammatory cytokine production, both of which have been shown to play important pathogenic roles in IgAN (2123). The DEFA1A3 CNV is located within the new IgAN susceptibility locus on 8p23.1 that was first discovered in the Chinese population and later validated by independent studies (4, 7, 8), which, together with its biological functions, suggests that the DEFA1A3 CNV locus may be involved in IgAN development.

To evaluate the role of DEFA1A3 CNV in IgAN, we performed a comprehensive association analysis of the DEFA1A3 CNV locus in two independent IgAN cohorts of southern Chinese Han, consisting of 1189 cases and 1187 healthy controls. We discovered three strong independent associations within the DEFA1A3 CNV locus and further demonstrated that the DEFA1A3 CNVs could also influence the prognosis of IgAN. For replication, we further confirmed the associations in Caucasian IgAN samples. We found an association with membranous nephropathy but not diabetic kidney disease in the Chinese population.

RESULTS

Distribution of DEFA1A3 CNVs in the Chinese population

Two independent cohorts of IgAN samples of southern Chinese Han were analyzed. The first cohort had 197 cases and 199 matched healthy controls in terms of age, gender, and geographic origin, and the second cohort included 992 cases and 988 healthy controls that were matched for geographic origin (table S1). We measured the copy numbers of DEFA1, DEFA3, and DEFA1A3 (DEFA1 + DEFA3) as well as the copy numbers of three biallelic polymorphisms [5–base pair (bp) indel (129bp versus 124bp), 7-bp duplication (282bp versus 275bp), and the 4-bp deletion (211bp versus 215bp)] using well-established paralog ratio test (PRT)–based methods (fig. S1) (16, 24). The PRT-based methods have been shown to have much greater accuracy for measuring the copy numbers of multiallelic CNVs than real-time polymerase chain reaction (PCR) analysis (25, 26) and have been successfully applied to studies of genes encoding chemokines (CCL3L1/CCL4L1) (27), immunoglobulin receptor (FCGR3) (28), and β-defensins (DEFB) (25, 29).

The copy number of DEFA1A3 ranged from 2 to 16 copies with a median copy number of 6 per genome in our Chinese samples (fig. S2 and table S2), which was consistent with the reported median copy numbers of 6 in the HapMap CHB (Chinese Han from Beijing) samples and 7 in CEU (CEPH Utah; United States) and ECACC (European Collection of Cell Cultures; UK) samples (15, 16), but was lower than the reported median copy number of 9 in the Japanese population (the JPT samples of the HapMap project) (22). In addition, the DEFA1A3 copy numbers of 24 HapMap CHB samples measured in this study showed a near-perfect correlation with the sequencing read depth of these samples from the 1000 Genome Project (R2 = 0.98, P < 0.0001) (fig. S3), confirming the accuracy of our copy number measurements. Correlation analysis revealed that the copy numbers of DEFA1, 124bp, 275bp, and 215bp correlated with each other (P < 0.0001) (table S3).

Association analysis of DEFA1A3 CNVs in a Chinese cohort of patients with IgAN

We tested the associations of the DEFA1A3 CNVs with susceptibility to IgAN using logistic regression analysis and by treating the two cohorts (see above) as independent samples through a fixed-effects meta-analysis. Although subpopulation structure existed between northern and southern Chinese Han populations, the samples from southern China were genetically homogeneous (30), and no control for population stratification was therefore included in our association analysis. We found highly significant associations for the copy numbers of DEFA1A3 [P = 3.99 × 10−9; odds ratio (OR), 0.88], DEFA1 (P = 6.71 × 10−5; OR, 0.92), DEFA3 (P = 6.55 × 10−8; OR, 0.82), 124bp (P = 2.89 × 10−8; OR, 0.87), 275bp (P = 2.51 × 10−7; OR, 0.89), and 211bp (P = 3.50 × 10−16; OR, 0.75) (OR per copy, adjusted for age and gender) without any evidence of genetic heterogeneity between the two independent cohorts (table S4). Stepwise conditional analysis demonstrated the independent associations of the 211bp, DEFA1A3, and DEFA3, and further multivariate analysis also confirmed the independence of these three associations: 211bp (P = 1.71 × 10−9; OR, 0.78), DEFA1A3 (P = 2.20 × 10−3; OR, 0.93), and DEFA3 (P = 3.33 × 10−3; OR, 0.85) (Table 1). All of the three associations remained statistically significant after Bonferroni correction (0.05/9 = 5.56 × 10−3). No evidence was found for interaction among the three independent CNVs (table S5).

Table 1. Logistic regression analysis of DEFA1A3 CNVs and single-nucleotide polymorphisms in the combined Chinese IgAN cohort (1189 cases and 1187 controls).

CI, confidence interval; NS, not significant (P ≥ 0.05).

View this table:

The patients carried significantly lower copy numbers of three CNVs than did the controls, indicating that low copy numbers of these variants had a risk effect for IgAN. On average, patients carried one copy of 211bp, six copies of DEFA1A3, and one copy of DEFA3, which were all significantly lower than the median copy numbers in the healthy controls, which were two copies of 211bp (P = 3.36 × 10−16), seven copies of DEFA1A3 (P = 4.74 × 10−8), and one copy of DEFA3 (P = 2.56 × 10−4) (Fig. 1 and table S6). We found that for all three CNVs, the risk was reduced progressively with the increase in copy number (Fig. 2). For example, individuals carrying zero copies of 211bp (about 26.8% of the population) had a 5.9-fold higher risk than those carrying five or more copies (about 1.9% of the population) (P = 2.17 × 10−6). The multivariant genetic score (GS) (the standardized weighted sum of the copy numbers) of the three variants (ranging from 1.35 to 19.67) (fig. S4) also showed strong association with IgAN susceptibility (P = 9.56 × 10−20). These three variants of the DEFA1A3 CNV locus could explain up to 4.96% of phenotypic variance cumulatively.

Fig. 1. Copy number distribution of the three variants of the DEFA1A3 CNV locus.

(A to C) The copy number distribution of the three variants of the DEFA1A3 CNV locus was examined in 1189 Chinese patients with IgAN and 1187 healthy controls. Wilcoxon rank sum tests were used to determine the significance of the distribution differences. The percentage of the Chinese cohort carrying a particular copy number for the three CNVs is shown on the y axis. Lower copies of DEFA1A3 CNVs were more frequent in IgAN patients than in controls: DEFA1A3 (A), DEFA3 (B), and 211bp (C).

Fig. 2. Distribution of OR values according to different copy numbers of DEFA1A3 CNVs.

(A to C) OR values were calculated from logistic regression analysis in the combined Chinese IgAN patient cohorts. The overall risk of IgAN decreased steadily with the increase in copy number for each variant: 211bp (A), DEFA1A3 (B), and DEFA3 (C).

Analysis of clinical phenotypes in a Chinese cohort of patients with IgAN

All of the 1189 IgAN patients analyzed in this study had clinical and pathological information available (summarized in table S7). In addition, of the 1189 IgAN patients, 382 patients had follow-up information with a median follow-up time of 4.5 years (interquartile range, 3.8 to 5.4 years). Of the 382 patients, 27 patients progressed to end-stage renal disease (ESRD) [estimated glomerular filtration rate (eGFR) <15 ml/min per 1.73 m2; or dialysis; or renal transplantation), and 9 patients were found to have a doubling of serum creatinine after diagnosis.

The analysis of renal outcomes (ESRD or doubling of serum creatinine after diagnosis) using the Cox regression model in the 382 IgAN patients revealed a significant association of the multivariant GS (ranging from 1.80 to 17.14) with survival in the IgAN cohort [P = 0.01; hazards ratio (HR), 0.79] after adjustment for known prognostic factors, including age, gender, proteinuria, hypertension, and serum creatinine at biopsy (Table 2) (31, 32). We further divided the 382 patients into five groups based on the quintile of their GS values and found that the patients in the first quintile of GS (average GS, 3.06) showed significantly worse survival than did those in the fifth quintile (average GS, 9.57) (P = 0.03; HR, 3.69; after adjustment for known prognostic factors) (Fig. 3 and Table 2). Although confirmation is needed from further studies, our results suggested that patients carrying low copy numbers of DEFA1A3 CNV had an increased risk of poor renal outcomes.

Table 2. Survival analysis of the GS of the DEFA1A3 CNVs in a Chinese cohort of IgAN patients (n = 382).
View this table:
Fig. 3. Survival analyses using Kaplan-Meier plots.

IgAN patients with follow-up data were divided into five groups based on the quintiles of GS: Q1 (first quintile), Q2 (second quintile), Q3 (third quintile), Q4 (fourth quintile), and Q5 (fifth quintile). A significant difference in patient survival was observed among the five groups of patients with different GS (log-rank, P = 0.0008).

The analysis of clinical phenotypes revealed the associations of the multivariant GS with hyperuricemia (P = 0.01) and thickening of the arterial wall (P = 0.008). However, the evidence could not survive the conservative Bonferroni correction for testing 15 clinical phenotypes (table S8), and the findings need to be confirmed by further studies.

DEFA1A3 CNVs may be a driver of the GWAS locus on 8p23.1

We genotyped rs2738048 and rs12716641 [the previously reported susceptible single-nucleotide polymorphisms (SNPs) within DEFA1A3 CNV region by GWAS of the Chinese population] (4, 8) in the same 1189 cases and 1187 controls and performed association analysis using the same method described previously. We confirmed the associations of rs2738048 (P = 9.58 × 10−4; OR, 0.80) and rs12716641 (P = 6.99 × 10−6; OR, 0.72) in our samples (Table 1), but the associations were weaker than those of 211bp, DEFA1A3, and DEFA3. Furthermore, controlling the association effect of either 211bp or DEFA1A3 fully abolished the association at rs2738048, whereas all the associations of the DEFA1A3 CNVs remained highly significant after controlling the association effect at rs2738048 (table S9). Consistently, the genotypes of rs2738048 showed strong correlation with the copy numbers of 211bp and DEFA1A3 (fig. S5 and table S3). However, the association effect of rs12716641 remained significant even after controlling the association effect for all three risk variants of DEFA1A3 (P = 2.54 × 10−4; OR, 0.76) (table S9). These results demonstrated that the previously reported association at rs2738048 was due to the association effect of the DEFA1A3 CNV locus, whereas the association effect of rs12716641 was independent of the DEFA1A3 CNV.

Replication study in Caucasian patients with IgAN and Chinese non-IgAN patients with kidney disease

For replication, we investigated the associations of 211bp, DEFA1A3, and DEFA3 copy numbers in a Caucasian IgAN cohort of 531 cases and 198 controls using the same genotyping method. Both DEFA1A3 (P = 4.42 × 10−4; OR, 0.82) and DEFA3 (P = 4.30 × 10−3; OR, 0.74) showed significant associations in the Caucasian cohort with consistent genetic effects as in the Chinese cohort (Table 3). However, 211bp was much rarer in the Caucasian population, and consequently, the association of 211bp did not reach statistical significance (P = 7.56 × 10−2), although its genetic effect was consistent between Caucasian (OR, 0.79) and Chinese samples (OR, 0.75) (Table 3 and fig. S6).

Table 3. Logistic regression analysis of three DEFA1A3 CNVs in Chinese and Caucasian IgAN cohorts.
View this table:

We further investigated the associations of 211bp, DEFA1A3, and DEFA3 in two Chinese cohorts with non-IgAN renal disease, that is, diabetic kidney disease (806 cases and 786 age- and gender-matched controls) and membranous nephropathy (493 cases and 500 age- and gender-matched controls) using the same method. We detected significant association of the 211bp variant with membranous nephropathy (P = 1.11 × 10−7; OR, 0.74 per copy) but no association with diabetic kidney disease (Table 4).

Table 4. Logistic regression analysis of DEFA1A3 CNVs in Chinese patients with diabetic kidney disease (DKD) or membranous nephropathy (MN).
View this table:

Association analysis of DEFA1A3 CNVs with HNP1–3 and IgA1

To investigate the effect of DEFA1A3 CNVs on HNP1–3 expression, we first measured the copy numbers of DEFA1A3 CNVs (211bp, DEFA1A3, and DEFA3) as well as the amounts of HNP1–3 in serum and urine from 96 IgAN patients and 62 healthy controls. There were no significant differences in copy number distributions of DEFA1A3 CNVs between patients and controls (Wilcoxon test, P > 0.05). In both serum and urine, HNP1–3 were significantly higher in patients than in healthy controls (P < 0.001) (Fig. 4), which was consistent with the fact that HNP1–3 proteins were largely produced by activated neutrophils during an inflammatory response (33). However, we could not find any correlation between DEFA1A3 CNVs and the amounts of either serum or urine HNP1–3 proteins (table S10). In addition, we also analyzed HNP1–3 in the neutrophils isolated from 83 IgAN patients and 79 healthy controls after stimulating neutrophils with LPS. Whereas there were no differences in DEFA1A3 copy number distributions between the patients and controls (Wilcoxon test, P > 0.05), the amount of extracellular HNP1–3 after stimulation with LPS (100 ng/ml for 6 hours) was significantly lower in neutrophils isolated from patients than controls (P = 0.006) (Fig. 4). However, we did not detect any association between the DEFA1A3 CNVs and the amount of extracellular HNP1–3 (table S10).

Fig. 4. HNP1–3 protein concentrations in serum, urine, and neutrophils.

(A to C) Concentration of HNP1–3 proteins in serum (A), urine (B), and produced by neutrophils (C) after stimulation with lipopolysaccharide (LPS). Wilcoxon rank sum tests were used to determine the significance of the differences. Serum and urine HNP1–3 concentrations were both significantly higher in IgAN patients compared to healthy controls (P < 0.001), whereas HNP1–3 produced by neutrophils after stimulation with LPS in vitro were significantly lower in IgAN patients compared to healthy controls (P < 0.01).

Given that the formation of galactose-deficient IgA1-containing immune complexes was the key pathogenic factor contributing to the development of IgAN (34), we also investigated the effects of DEFA1A3 CNVs on serum IgA1 levels and the proportion of galactose-deficient IgA1 in 96 IgAN patients and 62 controls. As expected, the serum IgA1 levels and proportion of galactose-deficient IgA1 were both significantly higher in IgAN patients than in healthy controls (P < 0.01) (fig. S7). Although there was no significant evidence for individual DEFA1A3 CNVs, the multivariant GS of these variants did show significant negative association with both serum IgA1 levels (P = 0.02) and the proportion of galactose-deficient IgA1 (P = 0.03) (table S11). These results were suggestive and need to be confirmed by further studies.

DISCUSSION

Our current study has discovered three independent risk factors for IgAN within the DEFA1A3 CNV locus, the copy numbers of DEFA1A3, DEFA3, and 211bp, by analyzing two independent Chinese IgAN cohorts and further confirming the association in a Caucasian IgAN cohort. Although each extra copy of these CNVs was only associated with moderate reduction of IgAN risk (by 12 to 25%), the overall impact of the whole DEFA1A3 CNV locus was strong, because each person carried multiple copies of these variants. By explaining 4.96% of risk variance, the DEFA1A3 CNV locus is the strongest genetic susceptibility factor that has been discovered for IgAN so far. Our study also revealed that the low copy number of DEFA1A3 CNV was associated with poor renal function in IgAN patients as well as high serum IgA1 levels and an increase in the proportion of galactose-deficient IgA1, a known pathogenic risk factor for IgAN. Consistently, our study has also demonstrated that HNP1–3 expression was elevated during an inflammatory response as seen in the serum and urine samples of patients; neutrophils from healthy individuals produced more HNP1–3 than did those from IgAN patients when stimulated by LPS. Together, our genetic and expression analysis suggested that HNP1–3 may provide a protective effect against development of IgAN and the progression of renal dysfunction.

Our study has also revealed a strong association of the DEFA1A3 CNV (211bp) with membranous nephropathy, but not diabetic kidney disease, although our diabetic kidney disease cohort had a bigger sample size and thus more power than did the membranous nephropathy cohort to detect the association. The genetic effect (OR) of 211bp was consistent between membranous nephropathy (OR, 0.74) and IgAN (OR, 0.75) without evidence of heterogeneity (I2 = 0; Q test = 0.84), suggesting the shared protective effect of 211bp variant against both membranous nephropathy and IgAN. Both IgAN and membranous nephropathy are common causes of primary glomerulonephritis, and both are mediated by immune complexes. In contrast, diabetic kidney disease is a chronic microvascular complication of diabetes due to metabolic dysregulation (35). Although the precise immunopathogenesis of IgAN and membranous nephropathy remains unclear, HLA-DQA1 has been shown to be associated with both IgAN and membranous nephropathy (6, 36). Together, these results suggest that there are shared pathogenic pathways between IgAN and membranous nephropathy and that the DEFA1A3 CNV locus may have a broad impact on the development of primary glomerulonephritis.

It is interesting to see the independent association effect of DEFA3 copy number beyond that of DEFA1A3 copy number. DEFA3 is a more recent human-specific variant that arose from DEFA1 through a nonsynonymous coding mutation in the third exon (15). Previous studies have suggested that HNP3 was less active than HNP1, although HNP3 was expressed at about twice the level of HNP1 (33, 37, 38). It is intriguing that the 211bp copy number also showed an independent association effect. The 211bp is a 4-bp deletion variant (TATC) within the second intron of DEFA1A3. Its molecular function is unclear, but the analysis by Human Splicing Finder (39) suggests that this 4-bp deletion may create a new splice acceptor site (CAG), raising the possibility that this 4-bp deletion may create new splice isoforms, which may in turn influence the expression and activities of the HNP1–3 proteins.

Our large-scale protein analysis, however, failed to reveal a direct correlation between the DEFA1A3 copy numbers and HNP1–3 levels in serum and urine or the production of HNP1–3 by neutrophils upon LPS stimulation. Previous studies of DEFA1A3 expression and its correlation with its copy number were limited. The mRNA analysis in leukocytes and lymphoblastoid cell lines (15, 22) as well as plasma protein analysis in patients with severe sepsis did not show correlation (40), but a small study (eight subjects) showed the correlation of HNP1–3 amounts inside neutrophils with the copy numbers of DEFA1A3 (17). The failure to show a correlation between the DEFA1A3 copy numbers and expression of HNP1–3 in the previous and current studies is probably due to the fact that HNP1–3 proteins are mainly produced by the bone marrow precursors of neutrophils, promyelocytes, and early myelocytes (33). As demonstrated in other CNV loci (41, 42), a full understanding of the haplotype structure (allelic structure) of the whole DEFA1A3 CNV locus may also be necessary for demonstrating the correlation.

DEFA1A3 CNVs could potentially influence IgAN development and renal disease progression in several ways. First, because DEFA1A3 is primarily expressed in promyelocytes within the bone marrow, it may have physiological effects on the bone marrow microenvironment (43, 44). Second, recent studies of IgAN pathogenesis have highlighted the important role of impaired immune regulation along the mucosa–bone marrow axis and have suggested that galactose-deficient IgA1 seems to be overproduced in the bone marrow (4547). DEFA1A3 CNVs may influence the impaired mucosa–bone marrow axis through its impact on the overproduction of galactose-deficient IgA1. Third, HNP1–3 plays a very important role in innate immunity, and neutrophils are the first defenders against infection. DEFA1A3 CNVs may influence the dysregulated mucosal immune response in IgAN by enhancing inflammatory signals (48).

Our study has demonstrated that the DEFA1A3 CNVs may be the primary driver of the previously reported association within the GWAS locus on 8p23.1 (4). The DEFA1A3 CNVs showed stronger associations than the previously reported SNP rs2738048; controlling the association effect of DEFA1A3 CNVs fully abolished the association at rs2738048, but not vice versa. So, the association at rs2738048 was secondary to the association effect of DEFA1A3 CNVs. In addition, our study has also demonstrated that the newly reported susceptibility signal rs12716641 (8) was independent of DEFA1A3 CNVs. A previous study reported a positive correlation between DEFA1A3 copy number and rs2738048 but did not evaluate the association of DEFA1A3 with IgAN (49). Recently, a newly published IgAN GWAS revealed another independent SNP rs10086568 within the GWAS locus on 8p23.1 in a Caucasian population (7). It would be interesting to investigate whether the association of this SNP is also due to the effect of DEFA1A3 CNVs in a future study.

Two major limitations of this study should be noted. First, our current findings are largely of genetic associations. The potential mechanisms underlying the involvement of the DEFA1A3 in the development of IgAN are interesting and biologically plausible but do need to be elucidated by future functional investigations. Second, the current study did not investigate the haplotype structure of DEFA1A3 CNVs because current genotyping and sequencing technologies could not determine the haplotype structure of multiallelic CNVs with long repeats in unrelated samples. Rapid development of long-read next-generation sequencing analysis may offer opportunities to characterize the haplotype structure of DEFA1A3 CNVs and understand in detail how different DEFA1A3 CNVs work together to influence the level and functionality of HNP proteins and susceptibility to developing IgAN.

In summary, our study discovered the association of the DEFA1A3 CNV locus with IgAN risk and renal disease progression and is one of the few cases where the association of a complex multiallelic CNV has been demonstrated robustly. Our study has demonstrated that this multiallelic CNV is the strongest genetic susceptibility locus for IgAN identified so far and may have a broad impact on the development of primary glomerular diseases. By demonstrating its protective effect against IgAN development and renal disease progression, our study has suggested the possibility of HNP1–3 as a candidate therapeutic target and prognostic marker.

MATERIALS AND METHODS

Study design

The study was designed to investigate the effect of DEFA1A3 CNV locus on IgAN development by analyzing the association between the copy numbers of DEFA1A3 CNVs with IgAN susceptibility and renal disease progression. First, two independent cohorts of southern Chinese Han, consisting of 1189 IgAN cases and 1187 controls, were used for the association analysis. Second, for replication, the three risk variants of the DEFA1A3 CNV locus discovered in Chinese subjects with IgAN were further analyzed in additional samples from Caucasian IgAN patients and controls (531 cases and 198 age- and gender-matched controls), Chinese subjects with diabetic kidney disease and controls (806 cases and 786 age- and gender-matched controls), and Chinese subjects with membranous nephropathy and controls (493 cases and 500 age- and gender-matched controls). Clinicopathologic information and follow-up data were collected from Chinese IgAN patients to further assess the clinical impact of DEFA1A3 CNVs on renal disease progression. We also investigated the relationship between the DEFA1A3 CNV locus and the previously published GWAS locus on 8q23.1 by genotyping the reported SNPs (rs2738048, rs12716641) in the same Chinese cohorts. Last, we investigated the effects of DEFA1A3 CNVs on the expression of HNP1–3 and the proportion of galactose-deficient IgA1. There was no randomization or blinding.

Study subjects

All IgAN cases were histopathologically diagnosed as primary IgAN by renal biopsy according to the following criteria: (i) immunofluorescence showing at least 2+ (scale, 0 to 3+) mesangial deposition of IgA, with IgA comprising the dominant immunoglobulin deposited in the glomeruli, and (ii) excluding individuals with cirrhosis, Henoch-Schönlein purpura nephritis, hepatitis B–associated glomerulonephritis, HIV infection, and systemic lupus erythematosus. All cases of diabetic kidney disease were diagnosed as persistent albuminuria (>30 mg/g) and/or low eGFR (<60 m/min per 1.73 m2) in patients with diabetes (50). Diabetes was defined as fasting plasma glucose (≥7.0 mM) and/or 2-hour postprandial plasma glucose (≥11.1 mM), by the use of hypoglycemic agents despite fasting plasma glucose, or any self-reported history of diabetes. All membranous nephropathy cases were adult biopsy–proven idiopathic membranous nephropathy. All the healthy controls were clinically verified with normal urinalysis (without red blood cells and protein in urine) and normal serum creatinine levels. Gender, age, geographical origin, and ethnicity information were collected from both cases and controls through questionnaires. All subjects completed a written informed consent form, and the study protocol was approved by the Institutional Review Board at The First Affiliated Hospital of Sun Yat-sen University.

Copy number measurements

Using the established PRT-based methods (16, 51), we measured the copy numbers of DEFA1A3 (DEFA1 + DEFA3) (by MLT1A0 and DEFA4 PRT) and the copy numbers of DEFA1 and DEFA3 (by DefHae3 ratio), as well as the allelic copy numbers of three internal biallelic polymorphisms: 5-bp indel (129bp versus 124bp, measured by Indel5 ratio), 7-bp duplication (282bp versus 275bp, measured by7bpdup ratio), and 4-bp deletion (211bp versus 215bp, measured by Del4 ratio) (fig. S1). These internal variants were chosen as part of PRT methodology, which can help to determine the copy number more accurately (16). Previous studies have suggested that internal variants may influence the expression and functionality of HNP proteins (15) and showed good correlation with major haplotype classes of DEFA1A3 locus at least in Europeans (51). Briefly, we combined the information from multiplex PRTs (MLT1A0 and DEFA4 PRTs) and four allelic ratio measurements (DefHae3/Indel5/7bpdup/Del4) into a maximum-likelihood framework to assign each sample to an integer copy number for all the DEFA1A3 CNVs with high confidence, as previously described (16).

For the newly characterized 4-bp deletion (TATC) variant within the second intron of DEFA1A3, the Del4 assay was developed in this study. The PCR for the Del4 assay was performed by using 1 μM of the primers TGCTCTCATTTTTGCATTCC and NED-TTTCTCCAAAGACTTGATTCCAA and 27 cycles at 95°C for 30 s, 57°C for 30 s, and 68°C for 60 s, followed by a 70°C hold for 40 min to generate amplicon products of 211bp/215bp from 10-ng genomic DNA input. The products were mixed with 10 μl of HiDi formamide with ROX 500 marker (Applied Biosystems) for fragment analysis using ABI 3730xl DNA Analyzer. GeneMapper 4.1 (Applied Biosystems) was used to extract the peak areas of the separated PCR products. The ratio between the 211bp and 215bp products (Del4 ratio) was estimated, and the copy numbers of the 211bp and 215bp alleles were calculated by using this Del4 ratio and the total copy number of DEFA1A3.

For the PRT analysis of the total number of DEFA1A3, seven CEPH/UTAH samples were included as reference samples of known copy number: NA11931 (seven copies), NA07347 (eight copies), NA11930 (five copies), NA11993 (seven copies), NA06993 (eight copies), NA12248 (six copies), and NA12249 (eight copies). The copy numbers of these reference samples had been determined unequivocally in previous studies using both segregation and read depth analyses (16, 51). The total DEFA1A3 copy numbers were determined by calibration of MLT1A0/DEFA4 PRTs against these reference samples, and the copy numbers of the other variants (DEFA1 versus DEFA3, 129bp versus 124bp, 282bp versus 275bp, and 211bp versus 215bp) were determined by using the total DEFA1A3 copy numbers and the ratio between the copy numbers of two variants. Copy number measurements were carried out by the Nottingham group (Chinese IgAN cohort I), the Guangzhou group (Chinese IgAN cohort II, Chinese DKD and MN cohort), and the Leicester group (UK IgAN cohort) using the same methods (table S1).

Genotyping analysis of rs2738048 and rs12716641

Genotyping was performed using the TaqMan assay from ABI and established methods (see the Supplementary Materials).

Measurement of HNP1–3 proteins in serum and urine

The serum and urine levels of HNP1–3 were measured with the HNP1–3 ELISA (enzyme-linked immunosorbent assay) Test Kit (Hycult Biotech) according to the manufacturer’s instructions. Serum was diluted by 1:50 in appropriate buffer, whereas urine was analyzed without dilution. Both serum and urine samples were transferred in duplicate samples to microtiter plates for incubation procedures. Absorbance was measured at 450 nm using the SpectraMax Plus 384 Microplate Reader (Molecular Devices). The concentration of HNP1–3 was calculated according to the standard curve.

Isolation, culture, and stimulation of human neutrophils

Whole blood (5 ml) was collected using EDTA-coated Vacutainer and processed immediately for neutrophil purification. Neutrophils were isolated by single-step centrifugation of whole blood onto Polymorphprep (Axis-Shield) according to the manufacturer’s recommendation. Neutrophils were resuspended in RPMI 1640 medium supplemented with 10% fetal bovine serum at a final concentration of 1 × 106/ml. The purity of the neutrophils was on average >95%. Trypan blue exclusion showed greater than 96% viability. Neutrophils were maintained at 37°C in the presence of 5% CO2 and then stimulated with LPS from Escherichia coli (100 ng/ml) (Sigma, catalog no. L4391) for 6 hours. Neutrophils from similar cultures not exposed to stimulants were used as negative controls. After incubation, cells were centrifuged, and the culture supernatants were stored at −80°C until use for the determination of HNP1–3. The concentration of HNP1–3 in culture supernatants was measured with the HNP1–3 ELISA Kit (Hycult Biotech) as previously described. The ratio for HNP1–3 level in stimulated supernatants normalized with that of negative controls was used in cell culture experiments.

Serum IgA1 and galactose-deficient IgA1

Serum IgA1 and galactose-deficient IgA1 [using Helix pomatia agglutinin (HPA), a lectin that binds terminal GalNAc in galactose-deficient IgA1] were measured by ELISA. The serum samples were diluted by 1:40,000 for serum IgA1 and 1:500 for galactose-deficient IgA1. For the measurement of serum IgA1, a standard consisting of native IgA1 (Calbiochem) was used at concentrations from 600 ng/ml to 4.68 ng/ml. Rabbit anti-human IgA F(ab′)2 and horseradish peroxidase–labeled F(ab′)2 fragment of goat IgG anti-human IgA1 were purchased from Dako and Southern Biotech, respectively. Neuraminidase and biotin-labeled HPA were purchased from Roche and Sigma, respectively. The absorbance was measured at 490 nm (A490) with the SpectraMax Plus 384 Microplate Reader (Molecular Devices). The serum concentration of IgA1 was calculated according to the standard curve. The relative galactose-deficient IgA1 level (proportion of serum IgA1) was calculated as the A490 value of HPA over the A490 level of IgA1.

Statistical analysis

Analysis of DEFA1A3 CNVs and SNPs. Wilcoxon rank sum tests and χ2 tests were used to determine the distribution differences in DEFA1A3 CNVs and SNPs between cases and controls. Spearman correlation analyses were used to test the correlations between DEFA1A3 CNVs and SNPs. Association analyses were performed using PROC LOGISTIC in SAS 9.2 software (SAS Institute), and the fixed-effects meta-analysis was performed in PLINK (52). The degree of heterogeneity was estimated using the heterogeneity index (I2) and Cochran’s Q test. Conditional association analyses were performed using a stepwise logistic regression model. A multivariate association analysis was conducted to confirm the independence of associations. Interaction analysis was performed by adding the product term of two CNVs into the logistic regression analysis. A composite GS of the three risk-associated variants (DEFA1A3, DEFA3, and 211bp) was built by calculating the standardized weighted sum of copy numbers, where the copy number of each DEFA1A3 CNV was weighted by its predicted effect size (natural log of OR from multivariant logistic regression analysis), and the sum of the weighted copy numbers of the three variants was then divided by the average effect size to rescale the score (standardization). The percentage of the total variance explained was estimated by calculating Nagelkerke’s pseudo-R2 using the fmsb package, where the diploid copies of independent DEFA1A3 CNVs and the affection status were entered into the glm function in R (version 2.15.1). OR values were measured as OR per copy of DEFA1A3 CNV. P < 0.05 after Bonferroni correction for multiple testing was considered as statistically significant.

Analysis of clinical phenotypes and DEFA1A3 CNVs. A linear regression model was applied for association analysis of quantitative phenotypes, whereas a logistic regression model was used for analyzing categorical variables, and an ordinal logistic regression model was used for ordinal variables. Survival analysis was performed using both univariate (for the GS only) and multivariate (GS and known prognostic factors) Cox regression analysis and Kaplan-Meier method. HRs were reported to evaluate the prognostic effects.

Association analysis of HNP1–3 and IgA1 expression with DEFA1A3 CNVs. Linear regression analyses were applied to detect associations between DEFA1A3 CNVs and expression level of HNP1–3 and IgA1 in samples adjusted by disease status. We assigned normal transformed values of HNP1–3 or IgA1 as independent variables, whereas three DEFA1A3 CNVs (DEFA1A3, DEFA3, and 211bp, or the combined GS) were treated as dependent variables. Standardized β and P values were reported to evaluate the associations. All the statistical analyses of clinical phenotypes and HNP1–3 and IgA1 expression were performed using SPSS 16.0 software (SPSS) and SAS 9.2 software (SAS Institute).

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/8/345/345ra88/DC1

Methods

Fig. S1. Schematic map of the DEFA1A3 CNV.

Fig. S2. Distribution of DEFA1A3 CNVs in the combined Chinese IgAN cohort.

Fig. S3. Correlation of the DEFA1A3 copy numbers measured by two methods.

Fig. S4. Distribution of GSs in the combined Chinese IgAN cohorts.

Fig. S5. Distribution of the average copy numbers of DEFA1A3 CNVs according to the genotypes of rs2738048 in the combined Chinese IgAN cohorts.

Fig. S6. Distribution of three risk variants of DEFA1A3 CNVs in the Caucasian IgAN cohort (531 cases and 198 controls) and Chinese IgAN cohorts (1189 cases and 1187 controls).

Fig. S7. Serum IgA1 and proportion of galactose-deficient IgA1.

Table S1. Summary of the study cohorts for the DEFA1A3 CNVs association analysis.

Table S2. Copy number distribution of DEFA1A3 CNVs in the combined IgAN cohorts (1189 cases and 1187 controls).

Table S3. Spearman correlation analysis between DEFA1A3 CNVs and SNPs in the combined Chinese IgAN cohorts (1189 cases and 1187 controls).

Table S4. Logistic regression analysis of DEFA1A3 CNVs and SNPs in the two independent Chinese IgAN cohorts.

Table S5. Interaction analysis of three DEFA1A3 CNVs in the combined Chinese IgAN cohorts (1189 cases and 1187 controls).

Table S6. Comparison of the copy numbers of DEFA1A3 CNVs between cases and controls in the combined Chinese IgAN cohorts.

Table S7. The clinical and pathological features at the time of diagnosis for 1189 Chinese IgAN patients.

Table S8. Association analysis between the GSs of DEFA1A3 CNVs and clinicopathologic features in 1189 Chinese IgAN patients.

Table S9. Logistic regression analysis of rs2738048, rs12716641, and DEFA1A3 CNVs in the combined Chinese IgAN cohort (1189 cases and 1187 controls).

Table S10. Multivariate linear regression analysis between serum HNP1–3, urine HNP1–3, neutrophil extracellular HNP1–3 (LPS-stimulated), and DEFA1A3 CNVs.

Table S11. Linear regression analysis of serum IgA1 and proportion of galactose-deficient IgA1 with the copy numbers of DEFA1A3 CNVs in 158 subjects.

REFERENCES AND NOTES

  1. Acknowledgments: We thank the staff in The First Affiliated Hospital of Sun Yat-sen University for the help with sample collection, DNA extraction, and sample storage. We thank J. Feehally from University of Leicester for providing the Caucasian samples. Funding: This study was supported by a grant from Guangdong Department of Science and Technology Translational Medicine Center (2011A080300002), the Specialized Research Fund for the Doctoral Program of Higher Education of China (20130171130008), the Science and Technology Planning Project of Guangdong Province, China (2013B051000019), Guangzhou Committee of Science and Technology, China (2012J5100031), Young Scientists Fund of National Natural Science Foundation of China (81200489), Young Scholars Fund for the Doctoral Program of Higher Education of China (20120171120087), a grant from the Biotechnology and Biological Sciences Research Council (BB/1006370/1; to J.A.L.A.), and the Agency for Science and Technology and Research (A*STAR) of Singapore. O.M. is supported by the Division of Higher Education, Kingdom of Saudi Arabia (S4674). The funding was also provided for the UK work from David Mayer Research Fund. Author contributions: J.B., D.G., J.F., X.T., X.D., J.C., J.W., and T.L. performed clinical characterization and recruitment of subjects and contributed samples; P.Y., W.L., and Z.Z. performed DNA extraction; Z.A., O.M., D.G., W.L., J.Y., Qin Z., and S.F. performed the experiments; Z.A., J.-N.F., and Qian Z. analyzed and interpreted the data; H.M., M.L., S.F., and R.X. contributed to analytical support and discussion; X.Y., J.L., and J.A.L.A. provided the platform, carried out the study protocol, and organized the study; Z.A., M.L., J.L., and X.Y. prepared the manuscript; and J.L. and X.Y. conceived and supervised the project. Competing interests: X.Y. has a pending patent entitled “Therapeutic target of IgA nephropathy and its detection method,” licensed to The First Affiliated Hospital, Sun Yat-sen University (patent #201510966895.9). X.Y. consults for Baxter, Fresenius, 3 Shine, MSD, Kabi, Novartis, Aventis, Kirin, and Roche. The other authors declare that they have no competing interests.
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article