Article Text
PDF
Abstract
Objective Recent genome-wide association studies suggested the PRDM1-ATG5 gene region as a systemic lupus erythematosus (SLE)-associated locus both in Caucasian and Chinese populations; however, the candidate gene was still obscure and the possible functional significance needed to be determined.
Methods In this study, by a multistage integrative strategy, the authors first performed a case–control association study involving 1745 individuals in the Chinese population by genotyping nine single nucleotide polymorphisms within this region, and a meta-analysis was conducted. Correlation between associated genotypes and expression levels of messenger RNA in B-cell lines from 210 unrelated HapMap data was examined, and was validated in vitro. To determine the biological significance, a genetic association study was also checked in a pathway-based manner and the significant associations were validated in a second 844 Chinese cohort.
Results A peak of association was found in the intergenic region (p=0.036–3.26×10−4). Meta-analysis consolidated the association between rs548234 and SLE (OR 1.254, p=1.28×10−16). Significant positive correlations with ATG5 expression were identified, suggesting ATG5 as a candidate gene in the region. Epstein–Barr virus B-cell-based downstream gene expression analysis supported a functional effect of rs548234 and rs6937876, and in-vitro experiments confirmed the regulatory effect of rs6937876 in B-cell populations. Finally, an autophagy pathway-based genetic association study identified ATG7 (p=1.12×10−4) and IRGM (p=0.015) as novel candidate genes, and gene–gene interactions were observed between ATG5, ATG7 and IRGM.
Conclusion These data may demonstrate that autophagy is involved in the pathogenesis of SLE and imply a common biological pathway in autoimmunity.
Statistics from Altmetric.com
With the progress of genome-wide association studies (GWAS), the number of convincingly established genetic associations with systemic lupus erythematosus (SLE) has increased sharply over the past few years. These studies have improved our understanding of the genetic basis of SLE, especially as they have identified and robustly replicated several novel loci, which implicated newer layers of immunological disturbances or pathways involved in disease pathogenesis.1,–,3 However, GWAS always identified some significantly associated region rather than straightforward candidate genes, such as TRAF1-C5, C8orf13-BLK and PRDM1-ATG5.3,–,9 In addition, once a single nucleotide polymorphism (SNP) association is observed and confirmed, much work remains to be done to establish which genetic variants in the region are actually responsible for the association.10
As mentioned above, the PRDM1-ATG5 gene region was reported to be associated with SLE both in Caucasian and Chinese populations;5,–,7 however, the candidate gene was still obscure. ATG5 was suggested to be the candidate gene,5 for the most significantly associated SNP rs573775 was located in ATG5. Later GWAS from a Chinese population denied the association between polymorphisms in ATG5 and SLE, but replicated the association between the intergenic region of PRDM1-ATG5 (rs548234 and rs6568431) and SLE.7 At the same time, from Caucasian replication data, both PRDM1 (rs6568431) and ATG5 (rs2245214) were suggested as candidate genes, because rs6568431 was more close to PRDM1 and rs6568431 has an r2 of less than 0.1 with rs2245214.6
Considering gene function, both PRDM1 (PR domain containing 1, with ZNF domain, also known as Blimp-1) and ATG5 (APG5 autophagy 5-like) play important roles in immunity. Blimp-1 enables secretion and effector functions but inhibits proliferation. Blimp-1-deficient T cells exhibit hyperproliferation in vivo, which results in autoimmunity. Genetic ablation of Blimp-1 in B cells prevents mature B cells from differentiating into either short-lived or long-lived plasma cells, which results in dramatically reduced antibody titres.11 12 Autophagy-related gene (ATG) 5 is a gene product required for the formation of autophagosomes.13 Autophagy is a phylogenetically ancient mechanism by which the cell can degrade and dispose of intracellular constituents or intracellular infectious agents in a regulated manner. Recent studies indicate that autophagy participates in trafficking events that activate innate and adaptive immunity.14,–,16 What is more, recent genome-wide scans have uncovered strong genetic associations between genes involved in autophagy, IRGM and ATG16L1, and susceptibility to Crohn's disease (CD).17,–,21
In this study, we aim to clarify which is the candidate gene within the PRDM1-ATG5 locus and try to find the possibly responsible functional correlations in a Chinese population.
Materials and methods
Patients and controls
A total of 1745 subjects, who were of Han ethnicity living in the north of China, was enrolled in the initial case–control study. The mean age of 807 SLE patients was 33.7±12.3 years and 89.1% were women. All SLE patients met American College of Rheumatology (ACR) classification according to the revised SLE criteria of the ACR.22 23 Nine hundred and thirty-eight healthy controls were geographically and ethnically matched (35.1±10.5 years, 59.0% women). Positive findings were replicated in a second northern Chinese Han cohort, including 385 SLE patients (31.5±11.7 years, 86.5% women) and 459 controls (33.7±10.6 years, 72.1% women).
The study was approved by the medical ethics committee of Peking University. All patients gave informed consent.
SNP selection
A total of nine SNP in the PRDM1-ATG5 region were selected for SLE association study. To select the SNP, initially, for replication, we selected four SNP that were reported to be associated with SLE, including rs548234 and rs6568431 in the intergenic region (candidate gene PRDM1 or ATG5), rs2245214 and rs573775 in ATG5.5,–,8 To validate further the association and determine the candidate gene in this region, five common SNP with minor allele frequency greater than 5% in Asian population from different linkage disequilibrium (LD) blocks were selected. Among the five newly selected SNP, rs811925 and rs573869 were located in the PRDM1 gene, rs6937876 in the intergenic region was in the same block with rs6568431, rs4945747 in ATG5 was in the same block with rs2245214, and rs2757133 in ATG5 was in the same block with rs573775. The nine tag SNP thus represented five LD blocks in the region covering both PRDM1 and ATG5 (figure 1).
Linkage disequilibrium (LD) plot of selected polymorphisms within the PRDM1-ATG5 gene region in a control population in the present study. The colours of the diamonds represent the r2 values. Blocks connecting pairs of single nucleotide polymorphisms (SNP) are shaded according to the strength of the LD between the SNP, from 0.0 (white) to 1.0 (black), as measured by the squared correlation statistic (r2 within the diagonal boxes). Rs548234 was not in LD with any of the other SNP (r2=0).
To validate further the genetic association of ATG5 and to analyse whether an implied autophagy pathway was involved in SLE, three SNP (IRGM rs10065172, rs13361189 and ATG16L1 rs2241880) with identified association with CD were examined in SLE patients.17,–,20 More recent data suggested that ATG7 plays an essential role independent of ATG5 in autophagy,24 25 thus a tag SNP (rs11706903) of ATG7, which had pair-wise LD values with all other SNP within ATG7 of more than 0.7, was also included to reveal a more close relationship.
Genotyping
TaqMan allele discrimination assays (Applied Biosystems, Foster City, California, USA) were used to determine the genotypes according to the manufacturer's instructions. The variants were detected using an ABI Prism 7500 Sequence Detection System (Applied Biosystems).
Association of genotype with expression in HapMap population
A correlation of messenger RNA expression with genetic variants was aimed to generate and test association hypotheses. Normalised mRNA data from Epstein–Barr virus (EBV)-transformed lymphoblastoid cell lines derived from 210 healthy unrelated HapMap individuals were obtained from the database of the Gene Expression Variation (GENEVAR) project at the Wellcome Trust Sanger Institute (http://www.sanger.ac.uk/humgen/genevar/).4 26
In-vitro blood cell isolation and mRNA quantification
T-cell and B-cell collection and mRNA quantification were performed as previously described.27 In brief, peripheral blood mononuclear cells were isolated using Ficoll, and then highly purified T and B cells were isolated using the EasySep CD3 and CD19 positive selection cocktail (Stem Cell Technologies, Vancouver, British Columbia, Canada). Isolation of RNA from T and B-cell populations in individuals was performed using the AllPrep DNA/RNA/Protein Mini Kit (Qiagen, Valencia, CA, USA) as described in protocols and procedures. Complementary DNA were synthesised from total RNA via a cDNA synthesis kit (Invitrogen, Carlsbad, California, USA). Quantitative real-time (RT)–PCR on cDNA samples by the SYBR green RT–PCR method (Applied Biosystems; primers see in supplementary material, available online only). All amplification was done in quadruplicate. Quantification of gene expression was made relative to an endogenous reference gene by calculating the differences in Ct (ΔCt) and relative values determined by 2(−ΔΔCt).
Statistical analysis
The genotype frequencies of SNP were tested for Hardy–Weinberg equilibrium separately in cases and controls. Associations between disease and SNP were analysed by χ2 tests or by logistic regression analysis. LD was tested using Haploview (version 4.2, http://www.broadinstitute.org/scientific-community/science/programs/medical-and-population-genetics/haploview/downloads). Meta-analysis was carried out using the Mantel–Haenszel approach. Statistical power was estimated with the power and sample size calculations software (http://biostat.mc.vanderbilt.edu/PowerSampleSize). Gene interaction analyses were conducted by logistic regression and χ2 test (see in supplementary material, available online only). Statistical analyses were performed with SPSS12.0 software or GraphPad PRISM 5.00. A two-tailed p value of less than 0.05 was considered statistically significant. For a replication, the unadjusted p values were reported.
Results
Association of polymorphisms in PRDM1-ATG5 region with SLE in Chinese Han population
A total of nine SNP in the PRDM1-ATG5 region were selected and genotyped in 1745 Chinese Han population, including 807 SLE patients and 938 healthy controls. Deviation from Hardy–Weinberg equilibrium was not observed for any of the SNP in the patients or controls. Figure 1 shows a LD plot constructed from the Chinese control data.
At the allele level, results of the case–control association analyses are shown in table 1. Three of the nine SNP exhibited evidence of an association with SLE (p=0.05), and rs6937876 showed a marginal significance. rs548234 showed the highest association with SLE susceptibility (OR 1.311, 95% CI 1.131 to 1.520, p=3.26×10−4). rs548234 was not in LD with any of the other SNP (r2=0), which may suggest an independent effect (figure 1). In the multivariate analysis (see supplementary material, available online only), rs548234 and rs6937876 remained significant when controlled for the nine SNP, gender and age by logistic regression analysis. Both dominant (OR 1.563, 95% CI 1.263 to 1.933, p=3.89×10−5) and additive (OR 1.552, 95% CI 1.249 to 1.927, p=7.20×10−5) models showed significant associations in rs548234. The additive model provided showed significant associations in rs6937876 (OR 1.701, 95% CI 1.114 to 2.599, p=0.014).
Meta-analysis of SNP within PRDM1-ATG5 in SLE studies
Allelic association of nine SNP in the PRDM1-ATG5 region with SLE
When possible, we performed a meta-analysis with previous reports. In a Chinese population (in comparison, the results of earlier studies were show in table 1), the rs548234 allele C conferred a 1.254-fold risk of SLE (p=1.28×10−16),5 the rs6568431 (in LD with rs6937876) allele G conferred a 1.202-fold risk of SLE (p=9.08×10−5),5 but rs573775 (located in ATG5) was not associated with SLE (p=0.194).5 Whereas in Caucasians, the rs6568431 allele G conferred a 1.20-fold risk of SLE (p=7.10×10−10),6 8 the rs2245214 (located in ATG5) allele C conferred a 1.15-fold risk of SLE (p=1.20×10−5)6 8 and the rs573775 allele T conferred a 1.19-fold risk of SLE (p=1.36×10−7).5 8 The different associations regarding polymorphisms in ATG5 seemed not to be derived from insufficient power, for the current Mantel–Haenszel analysis had a 0.974 power to detect a 1.2-fold risk for rs573775 in the Chinese population. We then calculated the population-attributable risk proportion (PARP) to assess the contribution of rs548234 to the development of SLE, and the PARP for rs548234 was 0.063 in the Chinese population.
Association of polymorphisms in intergenic region of PRDM1-ATG5 with PRDM1 and ATG5 mRNA expression
Regarding the possibility that the intergenic region may be a regulatory factor for gene expression4 28; we thus examined possible correlations between gene expression (PRDM1 and ATG5) and the genotype of intergenic SNP. As both PRDM1 and ATG5 were reported to be expressed in B cells, we analysed a dataset of gene expression in EBV-transformed lymphoblastoid B-cell lines from unrelated HapMap individuals. The correlation between SNP genotype and gene expression was assessed using the pooled dataset of all three HapMap ethnic groups (JPT+CHB, CEU and YRI, n=210). Twelve SNP in the intergenic region that is common (minor allele frequency (MAF)>0.05) to all three HapMap ethnic groups were thus selected. These 12 SNP were in two LD blocks identified from the genotype data of CHB and JPT HapMap (block 1: 7 kb, rs548234, rs693612, rs9480642; block 2: 19 kb, rs12661970, rs6937876, rs1008945, rs12213031, rs6568431, rs6901662, rs742109, rs10400411, rs1885449). No association between different genotypes of 12 SNP and PRDM1 gene expression was observed (p>0.05, data not shown). However, significant associations were observed between ATG5 expression and genotypes of rs548234 (p=2.71×10−3), rs693612 (p=0.013), rs9480642 (p=4.09×10−3), rs6937876 (p=5.38×10−4). rs548234 and rs6937876 showed the most significance within the respective blocks.
Association of possible functional SNP rs548234 and rs6937876 with downstream ATG5 activation
Based on the above genetic association data and the possible effect of rs548234 and rs6937876 on ATG5 expression, we further examined associations between genotypes and the downstream gene expression of ATG5. It was previously reported that ATG5 cross-talked with many other autophagy molecules,14 16 necessitated interferon (IFN) alpha and interleukin 12 production,29 engaged in apoptosis regulation,30,–,32 and its signalling may depend on the nuclear factor κ B (NF-κB) pathway.33,–,36 The expression of 97 genes involved in these downstream-effect pathways was thus checked. Among the 97 genes examined, 31 genes correlated with ATG5 expression in genome-wide significance (p<1.06×10−6, see in supplementary material, available online only). Autophagy and the NF-κB signalling pathway showed stronger association within the associated pathways, and 24 genes were associated with rs6937876 and 20 genes were associated with rs548234 genotypes after multiple comparisons adjustment (table 2).
Correlations of rs548234 and rs6937876 with expression levels in transformed B cells from 210 unrelated HapMap samples
In-vitro confirmation of associations between rs548234 and rs6937876 with PRDM1, ATG5, ATG3 and IRF5 expression
To confirm the reliability of the electronic expression model in vitro, T and B cells from 20 SLE as well as 20 control individuals were isolated, and PRDM1, ATG5, ATG3 (which also showed the most significant associations in the autophagy pathway) and IRF5 (a highly SLE associated gene) mRNA were detected by quantitative RT-PCR. In the B-cell population, it was observed that ATG5 gene expression correlated with ATG3 and IRF5 (p<0.05) and was higher in SLE patients than in controls (figure 2). Samples with homozygous CC for rs548234 and GG for rs6937876 were few; genotypes of SNP were assessed under dominant models (CC+CT vs TT, and GG+GA vs AA) for association with gene expression. As in figure 2, in both normal healthy controls and SLE patients, genotypes of rs6937876 were associated with the expression of ATG5, ATG3 and IRF5 in the B-cell population, but such association was not observed in the T-cell population. Similar to the expression pattern from EBV-transformed lymphoblastoid B-cell lines, no associations between genotypes and PRDM1 gene expression were observed, but for rs548234, no significant associations between genotypes and gene expressions were observed (p>0.05, data not shown).
In-vitro confirmation of associations between rs6937876 with ATG5, ATG3 and IRF5 expression. PRDM1, ATG5, ATG3 and IRF5 expression in normal controls (n=20) and systemic lupus erythematosus (SLE) patients (n=20) from (A) B-cell population and (B) T-cell population; in B-cell population (combined n=40); (C) ATG3 and (D) IRF5 expression were correlated with ATG5 expression; genotypes of rs6937876 were correlated with ATG5, ATG3 and IRF5 expression both in (E) normal controls and (F) SLE patients. Each symbol represents an individual, and horizontal lines indicate mean values.
Association of SNP of genes involved in autophagy pathway with SLE
Considering the shared molecular mechanisms in immunological pathways between different autoimmune diseases,1 2 37 38 we further tested gene polymorphisms in the autophagy pathway with SLE. Initial case–control study combined replication validated the associations between rs548234 (p=2.81×10−5), rs6937876 (p=1.10×10−3) and SLE. In addition, it suggested that ATG7 (rs11706903, p=1.12×10−4) and IRGM (rs10065172, p=0.015 and rs13361189, p=0.030) are novel SLE loci (table 3).
Genetic analysis of autophagy pathway in SLE
Genetic interaction analysis between autophagy genes
To detect the possible genetic interaction effect among the associated autophagy genes, multiplicative interactions were tested in the compiled dataset of five SLE-associated SNP in the autophagy pathway. For pairwise tests, we observed multiplicative interactions, especially between rs548234 (ATG5) and rs6937876 (ATG5) (p=1.57×10−5), rs6937876 (ATG5) and rs13361189 (IRGM) (p=4.19×10−4), rs11706903 (ATG7) and rs13361189 (IRGM) (p=3.10×10−3), rs10065172 (IRGM) and rs13361189 (IRGM) (p=3.06×10−9).
Additive interactions were next analysed by direct counting and the χ2 test. Under dominant models, the overall significance for the difference in risk genotype counts between patients and controls in all groups was high. The risk genotype combination contributed the most to the overall interaction. The relative excess risk due to interaction for rs548234 and rs6937876 was 0.607, for rs6937876 and rs13361189 it was 0.016 and for rs11706903 and rs13361189 it was 0.442. What is more, interactions between ATG5 (rs6937876) and ATG7 (rs11706903) were also revealed (relative excess risk due to interaction 0.104) (see in supplementary material, available online only).
Discussion
In the present study, we demonstrated that polymorphisms in the intergenic region of PRDM1-ATG5 were associated with SLE susceptibility, presumably through upregulating gene expression. It was reported that complement-inactivated sera from patients with SLE can significantly activate autophagy,39 and elevated ATG5 expression was observed in autoimmune demyelination and multiple sclerosis (MS), both in a mouse model of autoimmune demyelination as well as blood and brain tissues from MS cases.27 These data were consistent with the current hypothesis. Together with former GWAS and replication data, the correspondence between the two independent associations (gene expression and disease susceptibility) strongly indicated the presence of disease-causing regulatory variants in this region.
The pathogenic mechanisms of SLE are incompletely understood, but are postulated to involve multiple diverse aspects in the dysregulated immune response system. Autophagy may participate in the pathogenesis of SLE at multiple levels.1 14 15 40 Autophagy may be involved in the activation of innate immunity by delivering viral nucleic acids to endosomal compartments containing Toll-like receptor 7, which signals the induction of type 1 IFN production.29 33 41 We demonstrated a correlation between genotypes and mRNA expressions involved in autophagy and the type 1 IFN pathway, which was consistent with previous reports; plasmacytoid dendritic cells from ATG5 knockout mice produce less IFNα in response to viral infections.29 It may underscore a role for viral and bacterial triggers in SLE.42 Autophagy may be involved in adaptive immunity by delivering endogenously synthesised microbial antigens and self-antigens to late endosomes, where they are loaded onto MHC class II molecules for presentation to CD4 T cells.43 It was reported that ATG5 independently influence life and death decisions of the cell by both ‘autophagic’ cell death and apoptotic death pathways.31 44 The present data also provide some clues in that genetic variants of ATG5 are associated with both changes in autophagy and the apoptosis pathway. The lack of efficient apoptotic cell clearance may overcome tolerance to self-antigens and lead to SLE.43 In signalling transduction, the NF-κB signalling pathway is involved in regulating many aspects of cellular activity, in stress, injury and especially in pathways of the immune response. It was reported that it connects autophagy to major stress pathways.36 In the current study, we found the second prominent altered pathway was the NF-κB signalling pathway. Such data together with recent reports that a number of genetic associations of the IFNα pathway and NF-κB signalling pathways indicated that autophagy may be involved in such a link.1 5 6 45,–,47
We checked the correlation between genotypes and mRNA expression of many related genes, if not all; however, it indicated that rs6937876 and rs548234 had a functional significance. The low PARP for rs548234 may underlie a gene–environment interaction. Immortalised lymphoblasts that are of restricted lineage could more readily be studied without the environmental influences and transcriptome diversity found in a mixed lymphocyte population in vivo. Of note, in vitro, we confirmed the correlation with regard to rs6937876. Although we included SLE patients in active status and with few medications, we cannot totally rule out possible confounding factors, thus the associations seemed to be less significant than those in normal controls. However, the same patterns observed both in normal controls and SLE patients could consolidate such correlations. Such correlation was just observed in B cells, which may indicate a cell line-specific effect, the same can be seen from the effect of CD associated variations, which were restricted to Paneth cells.21 48,–,50 However, rs548234, which showed the most significant disease association and second significant expression correlation in EBV B cells, was not correlated with expression in vitro in the current study. A former study indicated that autophagy had characteristics of lineage specificities.21 48,–,50 Therefore, a lack of correlation for rs548234 may be derived from an inhomogeneous cell population or limited samples in the current investigation. Further exploration of a more specific cell lineage, such as CD4 or CD8 T cells, may be needed.43
The intricate interplay and effect of autophagy mirrors complexities in deciphering the roles of autophagy in human diseases. We thus further checked genetic variants of other genes involved in autophagy. It definitely supports that the complexities of the autophagy pathway may engage the complicated pathogenesis of SLE. A recent large-scale replication study on rheumatoid arthritis (RA) also associated disease susceptibility with the PRDM1-ATG5 intergenic region,45 and this may further suggest a common mechanism.
As discussed, variations of ATG5 might affect multiple aspects of immunity. Therefore, manipulation of macroautophagy could have therapeutic merit for patients affected by SLE. Rapamycin, a drug used to treat SLE, and demonstrated to be effective was found to interfere with the autophagy pathway.51 52 Further precise pharmacological mechanisms and more selective methods need to be developed.
In the current study, we suggested that ATG5 was the candidate in the PRDM1-ATG5 locus; the possible functional variants rs548234 and rs6937876 act in multiple facets to effect on the pathogenesis of SLE. A further pathway-based study supported evidence of a new layer of mechanisms—autophagy in the pathogenesis of the disease. The overall data derived from integrative strategies by multiple methodologies (including case–control study, meta-analysis, computer-based expression correlation, in-vitro confirmation, downstream effect validation, pathway-based genetic association, gene–gene interaction investigation) supported our conclusion. The current data reinforced the notion of common biological events in autoimmunity (between CD and SLE, and possibly RA and MS). The study provided interesting clues for elucidating the mechanism of autophagy in the pathogenesis of SLE and other autoimmune diseases, and emphasised a new layer of biological pathway that can be targeted for therapy in the future. Further more widespread replication and precise functional investigation will be needed.
Acknowledgments
The authors are grateful for the participation of all of the patients and control subjects.
References
Supplementary materials
Web Only Data
Files in this Data Supplement:
Footnotes
-
Funding This work was supported by grants from the National Natural Science Foundation of China (no 30801022, No 30825021) and the Foundation of Ministry of Health of China (no 200802052).
-
Competing interests None.
-
Patient consent Obtained.
-
Ethics approval This study was conducted with the approval of the medical ethics committee of Peking University.
-
Provenance and peer review Not commissioned; externally peer reviewed.