Abstract
Objective. Identification of an association between IRF5 rs2004640 and systemic sclerosis (SSc) has highlighted a key role for type 1 interferon (IFN). Additional functional IRF5 variants have been identified as autoimmune susceptibility factors. Our aim was to investigate whether IRF5 haplotypes confer susceptibility to SSc, and to perform genotype haplotype-phenotype correlation analyses.
Methods. We genotyped IRF5 rs377385, rs2004640, and rs10954213 in 1623 individuals of French European Caucasian origin. SSc patient subphenotypes were analyzed according to cutaneous subsets and for SSc-related pulmonary fibrosis.
Results. Case-control studies of single markers revealed an association between IRF5 rs3757385, rs2004640, and rs10954213 variants and SSc. We identified an IRF5 risk haplotype “R” (padj = 0.024, OR 1.23, 95% CI 1.07–1.40) and a mirrored protective haplotype “P” (padj = 8.8 × 10−3, OR 0.78, 95% CI 0.68–0.90) for SSc susceptibility. Genotype-phenotype correlation analyses failed to detect any association with a single marker. By contrast, phenotype-haplotype correlation analysis was able to detect intra-cohort association and to discriminate SSc patients with from those without the following clinical traits: “R” and/or “P” haplotypes identified diffuse cutaneous SSc (p = 0.0081) and fibrosing alveolitis (p = 0.018).
Conclusion. IRF5 haplotypes are more informative than single markers, suggesting that they could be helpful for risk stratification of SSc patients. Our study provides further evidence of a key role of IRF5 in SSc severity.
Systemic sclerosis (SSc) is an orphan complex multiorgan disease affecting the microvascular network, the immune system, and connective tissue. Progressive organ failure makes SSc a severe chronic disease leading to major disability or death1. To date no drug has proven its efficacy to counteract the generalized multiorgan fibrotic process2. Recent data have highlighted the poor efficacy of current available drugs, in particular in the 2 major causes of death, interstitial lung disease3 and pulmonary arterial hypertension4.
The type 1 interferon (IFN) pathway has been postulated to play a key role in autoimmune diseases5. Indeed, an increased expression of type 1 IFN genes by peripheral blood mononuclear cells, referred to as an IFN signature, has been detected in multiple autoimmune diseases such as systemic lupus erythematosus (SLE), rheumatoid arthritis (RA), and Sjögren’s syndrome (SS) and in a subgroup of patients with SSc6–9. The interferon regulatory factors (IRF) are major regulators of genes activated by the type 1 IFN. Recently, the interferon regulatory factor 5 gene (IRF5) has been identified as an autoimmune disease susceptibility gene in the background of SS, SLE, RA, and SSc10–13. Our original finding of an association between the IRF5 and SSc provided additional support for the pivotal role of IRF5 and the type 1 IFN pathway in SSc14. We found that the functional IRF5 rs2004640 single-nucleotide polymorphism (SNP), which creates a donor splice site in intron 1 of the IRF5 gene resulting in transcription of the alternative exon 1B, displayed association with SSc in 2 independent patient cohorts from French Caucasian populations14. Further, our data provided new clues for a role of IRF5 and IFN pathways in fibrosing alveolitis. Recently, different additional functional IRF5 variants have been convincingly identified as autoimmune genetic factors. Among these IRF5 variants, the rs10954213 SNP alters the polyadenylation site of IRF5, the rare A allele leading to a shorter and more stable mRNA15. A 5-bp biallelic insertion deletion polymorphism (CGGGG indel) located in the first intron of the IRF5 gene, resulting in increased binding of the transcription factor SP1 to the risk allele of the 5-bp indel, was also reported to confer susceptibility to autoimmunity16–18. Interestingly, the functional CGGGG indel polymorphism could be unambiguously inferred from haplotype reconstructions due to the high linkage disequilibrium with the 3 following SNP: rs377385, rs2004640, and rs1095421317. Therefore, the aim of our study was to investigate whether IRF5 haplotypes determined by those 3 SNP may be also involved in the genetic background of SSc and to account for the respective subphenotypes.
MATERIALS AND METHODS
Study population and design
We performed a large case-control association study including 827 SSc patients and 989 controls from the French SSc network as described14,19. All SSc patients were classified according to the cutaneous subtype classification of LeRoy, et al20, and phenotypically assessed as recommended21. Controls consisted of healthy unrelated individuals ethnically matched to the SSc cases. All subjects were of white French Caucasian ethnicity, defined by all 4 grandparents being white French Caucasian. The ethics committee of Cochin Hospital, Paris, approved the study and all subjects gave their written informed consent.
As reported, SSc cases with co-occurrence of a known IRF5-associated autoimmune disease (SLE and/or SS and/or RA and/or multiple sclerosis) were excluded prior to the analyses14. As reported, SSc patients with any ground-glass opacity on thoracic high-resolution computed tomography scan were considered to have interstitial lung involvement22,23. Pulmonary vascular and heart involvements were assessed by cardiac echocardiography and right-heart catheterization in cases suspected of having pulmonary arterial hypertension24.
Autoantibody analysis
SSc patients were tested for antinuclear antibodies (ANA) using indirect immunofluorescence (IIF) and HEp-2 cells as antigen substrate (Antibodies Inc., Davis, CA, USA). Specific SSc antibodies were systematically assessed; anticentromere antibodies (ACA) were determined by their distinctive IIF pattern on HEp-2 cells. Anti-topoisomerase I antibodies were determined by counter-immunoelectrophoresis.
Genotyping
The 3 SNP were genotyped using a competitive allele-specific polymerase chain reaction (PCR) system (Kaspar genotyping, Kbioscience, Hoddeston, UK) and the Taqman SNP genotyping assay-allelic discrimination method (Applied Biosystems, Foster City, CA, USA) as described14,19. IRF5 rs2004640 was previously genotyped for the same individuals in our cohort25. The average genotype completeness for the 3 SNP tested was 89% for the SSc and control samples. The accuracy was > 99%, by duplicate genotyping of 10% of all samples. The CGGGG indel polymorphic marker was amplified by fluorescent PCR and genotyped on an ABI-Prism 3100 device (Applied Biosystems, Courtaboeuf, France) and analyzed with Genotyper software in 100 (i.e., 200 chromosomes) randomly selected individuals homozygous for IRF5 risk haplotype and in 40 (80 chromosomes) individuals homozygous for the protective haplotype.
Statistical analysis
Single-marker and haplotype analyses were performed in the SSc patients and controls having full genotype information for the 3 IRF5 SNP (743 SSc patients, 880 controls). Statistical analyses were performed using the R software (version 2.6.0). The level of significance for all tests corresponds to a type I error rate α = 5%. The Bonferroni correction was applied for all the tests for a “hypothesis-generating step” comparing SSc cases and controls (10 phenotypic subsets; p values adjusted). All odds ratios are provided with 95% confidence interval. Conformity to Hardy-Weinberg equilibrium was tested using a classical 1 degree of freedom chi-square test in controls. Individual association analyses of IRF5 SNP with SSc were performed by comparing cases and controls with Fisher’s exact test on genotypes. Corresponding odds ratios were assessed using standard logistic regression analysis with the most frequent genotype taken as the reference. The same procedure was applied in subgroups stratified according to SSc phenotypes. An analysis of haplotype diversity was realized for the 3 SNP with the expectation-maximization (EM) algorithm as implemented in the haplo.stats R library. Specific p values and odds ratios were obtained by comparing each haplotype with the more common haplotype in the population, applying Fisher’s exact test.
RESULTS
IRF5 single-marker analysis
Genotype frequencies of the 3 IRF5 SNP were in Hardy-Weinberg equilibrium in the control population.
Among the 829 patients and 989 controls investigated, complete genotyping data for the 3 IRF5 variants were available for 743 and 880 individuals, respectively. Allelic and genotypic frequencies of the IRF5 rs3757385, rs2004640, and rs10954213 SNP were in good agreement with frequencies reported among the European Caucasian population17. The case-control association study involving 743 SSc cases and 880 controls revealed an association between IRF5 rs3757385, rs2004640, and rs10954213 variants and SSc (Table 1). The magnitude of the association was most important for IRF5 homozygous genotypes for the risk allele of rs3757385 and rs2004640 SNP, suggesting a dosage effect (p = 0.009, OR 1.51, 95% CI 1.10–2.08 and p = 3.5 × 10−5, OR 1.81, 95% CI 1.36–2.40, respectively).
We tested whether the associated SNP constituted an independent effect. Pair-wise correlation coefficient (r2) analysis found rs3757385 in relatively high linkage disequilibrium with rs2004640 (r2 = 0.57) and rs10954213 (r2 = 0.56). Conversely, rs2004640 was not in strong linkage disequilibrium with rs10954213 (r2 = 0.34). Multiple logistic regression analysis of the 3 SNP revealed no independent effect: the rs2004640 was the only variant that remained significantly associated conditionally to the 2 others (p = 0.0086).
IRF5 haplotype analysis
Among the 8 IRF5 haplotypes defined by the 3 IRF5 rs3757385, rs2004640, and rs10954213 SNP, 4 common haplotypes (frequency > 5% in both controls and cases) were predicted from our sample. Of the 4 haplotypes, 2 were at a frequency of 25% or more in the controls. As previously observed in a European Caucasian population17, these 2 common haplotypes C-T-A and A-G-G were present at a frequency of 44% and 33%, respectively, in our control chromosomes. Taking into account the IRF5 risk alleles rs3757385*C, rs2004640*T, and rs10954213*A, we tested for association of the IRF5 C-T-A risk haplotype (“R” haplotype). The IRF5 “R” haplotype was associated with SSc as its frequency was significantly increased in SSc patients (49%) compared to controls (44%) (padj = 0.024, OR 1.23, 95% CI 1.07–1.40). We also observed that the C-T-A “mirror” haplotype, i.e., the A-G-G haplotype, acts with a protective (“P” haplotype) effect, its frequency being decreased in the SSc population (27.8%) compared to controls (32.9%) (padj = 8.8 × 10−3, OR 0.78, 95% CI 0.68–0.90; Table 2).
Phenotype-genotype correlation in SSc
We previously reported results identifying IRF5 rs2004640 as a new SSc susceptibility gene, with a strong effect in the genetic background of diffuse cutaneous SSc (dcSSc) and SSc-related fibrosing alveolitis14. Following those results we investigated the effect of the IRF5 “R” and “P” haplotypes in subsets of SSc [dcSSc, limited cutaneous SSc (lcSSc), and patients with and without fibrosing alveolitis].
In this study, we observed a strong association of the IRF5 rs2004640 TT genotype with both dcSSc (padj = 2.6 × 10−3, OR 2.01, 95% CI 1.39–3.05) and SSc-related pulmonary fibrosis (padj = 3.7 × 10−4, OR 2.19, 95% CI 1.50–3.20). The TT risk genotype was also strongly associated with the subset of severe organ involvement, compared to controls (padj = 2.2 × 10−4, OR 2.16, 95% CI 1.50–3.09). The IRF5 rs3757385 variant was also found to be associated with the different SSc subsets; conversely, no association was detected with rs10954213 (Table 1).
However, regarding the phenotype-genotype correlation, intra-cohort analyses failed to detect any significant differences comparing the allele and genotype frequencies of the 3 IRF5 SNP and the 3 subsets (i.e., cutaneous phenotype, pulmonary fibrosis, and severe organ involvement; data not shown).
Phenotype-haplotype correlation in SSc
Regarding the IRF5 C-T-A “R” haplotype, we found an association restricted to dcSSc (padj = 0.037, OR 1.32, 95% CI 1.10–1.59). The IRF5 A-G-G “P” haplotype was found to act with a protective effect restricted to the diffuse cutaneous subtype and the subset of SSc patients with fibrosing alveolitis (Table 2). When phenotype-haplotype correlation was investigated, intra-cohort analyses revealed a significantly increased frequency of the “R” haplotype in dcSSc compared to lcSSc (p = 0.0081). Conversely, the “P” haplotype was found to be underrepresented in the subset of SSc patients having fibrosing alveolitis compared to those without (p = 0.018); the frequency of the “R” haplotype did not differ between the 2 SSc subsets (Table 2). As fibrosing alveolitis is reported to be linked to dcSSc, we next compared the frequencies of the IRF5 “P” haplotype between patients with and those without fibrosing alveolitis in the lcSSc subset: we observed a decreased frequency of the “P” haplotype in the patient subset without fibrosing alveolitis; however, this did not achieve statistical significance (data available upon request from the author).
As previously stated, the susceptibility CGGGG indel allele was inferred from the 2 IRF5 “R” and “P” haplotypes. Among a group of 100 SSc patients and controls who were homozygous for the “R” haplotype, 100% were carrying the CGGGG insertion allele; conversely, among 40 individuals homozygous for the IRF5 “P” haplotype, none had the CGGGG indel susceptibility allele, corroborating that the “R” haplotype harbors the CGGGG indel risk allele.
DISCUSSION
We previously reported an association between the functional IRF5 rs2004640 polymorphism and SSc14. Identification of IRF5 as a genetic susceptibility factor shared by various autoimmune diseases is an important step for understanding type I IFN-driven autoimmunity. Further, this has generated new clues about interstitial lung disease in the context of SSc, which has become a leading cause of death, without established therapeutic options. However, to date, many IRF5 functional variants have been identified as risk factors for different autoimmune diseases and remain to be investigated in SSc. Our study is the first to provide evidence for association between 2 IRF5 haplotypes and SSc. We identify an IRF5 risk haplotype (“R”) defined by the alleles rs3757385*C, rs2004640*T, and rs10954213*A (C-T-A) that confers susceptibility to SSc. The IRF5 “R” haplotype was found to confer susceptibility to defined subsets of SSc including dcSSc and SSc-related pulmonary fibrosis. These findings illustrate that IRF5 may contribute to the typical heterogeneity of SSc. This observation is in good agreement with our original study14, as the main subphenotype associations were the diffuse cutaneous and fibrosing alveolitis subsets. In this sample, intra-cohort single-marker analyses failed to provide evidence for using one of the 3 IRF5 variants as a marker for a particular SSc subset. This discrepancy with our previous findings could be related to the cross-sectional design of the study and a smaller sample size (743 SSc cases, 880 controls) leading to a weaker power to detect such intra-cohort associations. Of interest, phenotype-haplotype correlation analysis did detect intra-cohort association signals. Indeed, IRF5 “R” and/or “P” haplotypes were able to discriminate dcSSc and fibrosing alveolitis. Those results emphasize that IRF5 haplotype analyses are more informative than single-marker analyses of rs3757385, rs2004640, and rs10954213, suggesting the haplotype biomarker could be helpful for risk stratification of patients with Ssc.
It is noteworthy that the same risk haplotype has been reported to be associated with high serum levels of type 1 IFN in patients with SLE26. It could be hypothesized that the IRF5 “R” haplotype harbors one or multiple causal polymorphisms. Interestingly, a 5-bp biallelic insertion-deletion polymorphism (CGGGG indel) located in the first intron of the IRF5 gene was recently reported to be associated with multiple autoimmune diseases16–18. The CGGGG risk insertion, which results in increased binding of the SP1 transcription factor, has recently been reported to regulate the expression of IRF5 mRNA27,28. The CGGGG risk insertion suggested to be the causal variant of IRF5 in SLE and in primary SS could be unambiguously inferred from haplotype reconstructions with the 3 IRF5 SNP tested in our study17. Hence, the IRF5 “R” haplotype is the exclusive haplotype carrying the CGGGG insertion risk allele, suggesting a possible contribution of this functional variant in the genetic background of SSc17.
As previously reported, the susceptibility CGGGG indel allele was unambigously inferred from the 2 IRF5 “R” and “P” haplotypes in a French Caucasian population. This haplotype was also reported to confer susceptibility to multiple sclerosis in the European Caucasian population17. However, resequencing of the IRF5 gene in patients homozygous for the “R” haplotype is required for a conclusion that the CGGGG insertion deletion is the definite IRF5 causal variant. Further, it is notable that multiple IRF5 SNP including rs2070197 and rs12539741, located in the 3’UTR, were reported to confer susceptibility to SLE and may be involved in SSc susceptibility15. Of interest, those SNP are in perfect linkage disequilibrium with several additional Hapmap SNP distributed over a 10-kb region located 3’ of the IRF5 gene that contains the transportin 3 (TNPO3) gene. The reported pair-wise linkage disequilibrium between these SNP and the CGGGG indel is very low, suggesting that the SNP could tag an independent SLE signal association; this remains to be investigated in SSc16.
Genetic studies of SSc susceptibility have made great progress, and considering the current efforts, more genetic risk factors will be identified. In contrast, the field of the genetics of SSc heterogeneity is relatively unexplored. Following our results, IRF5 should be investigated in genetics of SSc phenotypic heterogeneity. Identification of the IRF5 gene as a susceptibility factor for multiple autoimmune disorders emphasizes that similar immunogenetic mechanisms underlie autoimmune diseases. It is also important for understanding type I IFN-driven autoimmunity — IFN regulatory factors (IRF), such as IRF5, coordinate the expression of type I IFN, which are assumed to participate in the pathogenesis of connective tissue disorders including SLE, SS, and SSc6,7,9. IRF5 also coordinates the expression of interleukin 629, which was found to be elevated in plasma serum of patients with dcSSc and SSc-related pulmonary fibrosis30,31. Increased understanding of the role of genetics in disease outcomes in SSc may promote the development of new therapies.
Our study provides evidence for IRF5 as a new SSc gene playing a key role in disease severity, and supports the pivotal role of type 1 IFN in the pathophysiology of SSc. Further research is needed on the pathogenesis of this severe autoimmune disease, for which there is no cure.
Acknowledgments
The authors thank Dr. Joëlle Benessiano, Centre de Ressource Biologique, Hôpital Bichat, Paris, France; Établissement Français du Sang (EFS); and Prof. Bernard Granchamp and Dr. Nadem Soufir for their assistance in setting up the control samples.
Footnotes
-
Supported by Association des Sclérodermiques de France, Institut national de la santé et de la recherche médicale (INSERM), Agence Nationale pour la Recherche (ANR; no. R07094KS), and Groupe Français de Recherche sur la Sclérodermie.
- Accepted for publication December 29, 2009.