Abstract

We analyzed a comprehensive set of single-nucleotide polymorphisms (SNPs) and length polymorphisms in the interferon regulatory factor 5 (IRF5) gene for their association with the autoimmune disease systemic lupus erythematosus (SLE) in 485 Swedish patients and 563 controls. We found 16 SNPs and two length polymorphisms that display association with SLE (P < 0.0005, OR > 1.4). Using a Bayesian model selection and averaging approach we identified parsimonious models with exactly two variants of IRF5 that are independently associated with SLE. The variants of IRF5 with the highest posterior probabilities (1.00 and 0.71, respectively) of being causal in SLE are a SNP (rs10488631) located 3′ of IRF5, and a novel CGGGG insertion-deletion (indel) polymorphism located 64 bp upstream of the first untranslated exon (exon 1A) of IRF5. The CGGGG indel explains the association signal from multiple SNPs in the IRF5 gene, including rs2004640, rs10954213 and rs729302 previously considered to be causal variants in SLE. The CGGGG indel contains three or four repeats of the sequence CGGGG with the longer allele containing an additional SP1 binding site as the risk allele for SLE. Using electrophoretic mobility shift assays we show increased binding of protein to the risk allele of the CGGGG indel and using a minigene reporter assay we show increased expression of IRF5 mRNA from a promoter containing this allele. Increased expression of IRF5 protein was observed in peripheral blood mononuclear cells from SLE patients carrying the risk allele of the CGGGG indel. We have found that the same IRF5 allele also confers risk for inflammatory bowel diseases and multiple sclerosis, suggesting a general role for IRF5 in autoimmune diseases.

INTRODUCTION

Systemic lupus erythematosus (SLE) (OMIM # 152700) is regarded as the prototype autoimmune disease and is characterized by inflammation in multiple organs and the production of autoantibodies against nucleic acid and its associated proteins. The etiopathogenesis of the disease is incompletely understood, but an increased expression of type I interferons (IFN) and type I IFN regulated genes is found in patients with SLE (1). The ongoing type I IFN production in SLE is most probably induced by nucleic acid containing immune complexes that are internalized by plasmacytoid dendritic cells (PDC) via the Fc-gamma receptor IIa, followed by engagement of endosomal Toll-like receptors TLR7 or TLR9 (2,3). The type I IFN production triggered via the TLR signaling pathway is dependent on the adaptor molecule MyD88, the transcription, factor interferon regulatory factor 5 (IRF5) and several other molecules (4,5). Besides its function in the type I IFN response to viral infections, IRF5 regulates the expression of a number of other genes coding for molecules involved in cell-cycle regulation, cell adhesion, apoptosis and immune response, all of potential importance in autoimmunity (4,6).

Our initial finding of an association between SLE and three single-nucleotide polymorphisms, SNPs rs729302, rs2004640 and rs752637 in the IRF5 gene in patients from three Nordic countries (7) indicated that the type I IFN and TLR signaling pathways are involved in the pathogenesis of SLE. Our original association finding has been convincingly replicated by analysis of patients with SLE from several European populations, and in patients with SLE from North and South American and Asian (Korean) populations (8–12).

To elucidate the function of IRF5 in the pathogenesis of SLE has proven to be a challenge. The SNP rs2004640 that gave the strongest signal of association with SLE in our original study is located in a splice junction of an alternative exon 1B of IRF5, and the major allele (T) of this SNP creates a splice donor site for exon 1B, which results in a low level of transcription of a unique IRF5 mRNA (8,13). In addition to the SNP rs2280714 downstream of IRF5 (14), several additional SNPs in the 3′ end of the IRF5 gene have been found to be associated with the expression levels of IRF5 in lymphoblastoid cell lines (15). Of these multiple SNPs, a functional role has been suggested for SNP rs10954213 located in the 3′-UTR of IRF5, because the major allele (A) of SNP rs10954213 is correlated with a truncated IRF5 transcript with elevated expression levels in peripheral blood mononuclear cells (PBMCs) (15,16). Recently, three closely linked SNPs, rs2070197 located in the 3′-UTR of IRF5, and rs10488631 and rs12539741 located downstream of IRF5 that give very strong signals of association with SLE were identified (15). A functional role for these SNPs cannot be suggested based on their sequence context. The same study also identified a 30 bp insertion/deletion polymorphism (indel) in exon 6 of IRF5 that deletes 10 amino acids in a domain of the IRF5 protein enriched in proline (P), glutamate (E), serine (S) and threonine (T). Although this 30 bp indel is an interesting candidate for a functional variant of IRF5 in SLE, it does not show a direct association with SLE. This PEST domain of IRF5 is a random coil, linking two structural domains of the IRF5 protein, and thus deletion of 10 amino acids from this region might not affect the function of IRF5 (http://www.predictprotein.org). A model with these three variants that describes risk and protective haplotypes of IRF5 in SLE was proposed based on conditional logistic regression analysis of SNPs in IRF5 that show association with SLE and by considering that SNPs rs2004640 and rs10954213, as well as that of the 30 bp indel could be functional (15). However, the genotype data underlying this model was incomplete, and indicates that there may be additional variants of IRF5 which contribute to the association between IRF5 and SLE. The purpose of the present study was therefore to identify a comprehensive set of genetic variants of IRF5 and to use association tests with SLE to predict by statistical means how many functional variants are likely to exist in the gene and which variants are the best candidates for being functional.

RESULTS

To obtain complete information of all variants of the IRF5 gene we re-sequenced the whole IRF5 gene, including the introns, as described previously (15). We identified 48 polymorphic sites of which 30 SNPs and four indels that were observed in more than one individual. No missense SNPs in the protein coding region of the IRF5 gene were discovered. We designed a panel of polymorphisms in the IRF5 gene to be tested for their association with SLE, which included 24 of these SNPs and three length polymorphisms. Five of the 30 SNPs had to be omitted from the panel for genotyping because design of Golden Gate assays was not feasible for them, and one SNP and a 2 bp length polymorphism were omitted because they were in perfect LD with another SNP included in the panel. In addition, the panel included the SNP rs729302 9 kb upstream of IRF5 identified in our original study (7) and 19 SNPs selected from dbSNP that were located within 5 kb regions upstream and downstream of IRF5. This set of 47 polymorphisms was genotyped in 485 Swedish patients with SLE and 563 controls using the Golden Gate assay from Illumina (17). Data for 33 polymorphisms with allele frequencies >1% that passed genotyping quality control were subjected to further analysis. The Materials and Methods section and Supplementary Materials, Table S1 provides additional information on the SNP selection and the genotyping results.

To estimate the significance of the association with SLE for each individual IRF5 polymorphism, we used Fisher’s exact test to compare the allele counts in the patients and controls. In this analysis, 18 of genotyped polymorphisms yielded a strong signal of association with SLE with a P-value <0.0005 and an odds ratio (OR) >1.4 (Table 1). These association signals remain significant after adjustment for multiple testing, even using the conservative Bonferroni correction. We observed a particularly strong association signal for a CGGGG indel polymorphism, which is located 64 bp upstream of the first untranslated exon (exon 1A) of IRF5, with P = 4.6 × 10−9 and OR = 1.69 (95% CI 1.42–2.02). The CGGGG indel is diallelic and part of a polymorphic repetitive DNA-stretch that consists of either three or four CGGGG units. In addition, we identified five SNPs, rs3778754, rs11761199, rs7808907, rs7800687 and a novel SNP (IRF5-15-1) located in intron 2 of IRF5, as well as a G indel located in a homopolymer stretch of four or five G nucleotides in intron 8 of IRF5, which all displayed strong association signals with SLE (P = 10−4–10−7) (Fig. 1). None of these seven polymorphisms have previously been analyzed for their association with SLE. As expected, our analysis also identified 11 SNPs that have previously been found to be associated with SLE by us and others (Table 1) (7,8,12,15,16).

Figure 1.

Schematic view of the IRF5 gene with all successfully genotyped SNPs and indels. The untranslated exons of IRF5 are marked with white boxes and the translated exons are marked with dark boxes. The bars in the graph correspond to the negative log 10 of the P-value for each marker in a Fisher’s exact test of association with SLE. The lower part of the Figure shows the pair-wise LD-values (r2) between the polymorphisms.

Figure 1.

Schematic view of the IRF5 gene with all successfully genotyped SNPs and indels. The untranslated exons of IRF5 are marked with white boxes and the translated exons are marked with dark boxes. The bars in the graph correspond to the negative log 10 of the P-value for each marker in a Fisher’s exact test of association with SLE. The lower part of the Figure shows the pair-wise LD-values (r2) between the polymorphisms.

Table 1.

Association of IRF5 polymorphisms with systemic lupus erythematosus

Markera Positionb Major: minor allelesc Risk allele Allele countsg Minor allele frequencies    
    Cases Controls Cases Controls P-valuesd ORe 95% CI 
rs729302 128162911 A:C 719:259 712:366 0.27 0.34 2.70E-04 1.42 (1.17–1.72) 
rs11768806 128167079 C:T 846:130 900:178 0.13 0.17 0.041 1.29 (1.00–1.64) 
rs12537192 128167644 T:C 954:24 1033:23 0.02 0.02 0.77 1.13 (2.01–0.63) 
rs4728142 128167918 A:G 541:435 473:605 0.55 0.44 1.80E-07 1.59 (1.89–1.33) 
rs1874330 128169175 T:C 833:141 880:196 0.14 0.18 0.023 1.32 (1.04–1.67) 
rs3778754 128169503 G:C 569:407 502:572 0.42 0.53 1.80E-07 1.59 (1.33–1.89) 
rs3757386 128171248 C:T 888:90 948:126 0.09 0.12 0.062 1.31 (0.98–1.74) 
rs6968563 128171682 T:C 915:45 998:54 0.05 0.05 0.68 1.10 (0.73–1.65) 
CGGGG indelf 128171882 4 / 3 CGGGG 546:412 458:586 0.43 0.56 4.60E-09 1.69 (1.42–2.02) 
rs6953165 128172161 C:G 910:42 1009:53 0.04 0.05 0.60 1.14 (0.75–1.72) 
rs2004640 128172252 T:G 613:363 557:519 0.37 0.48 5.70E-07 1.57 (1.31–1.87) 
rs3807307 128173153 C:T 562:412 495:577 0.42 0.54 2.20E-07 1.59 (1.33–1.89) 
rs752637 128173371 C:T 693:285 676:398 0.29 0.37 1.80E-04 1.43 (1.18–1.71) 
rs3807306 128174631 T:G 608:370 550:528 0.38 0.49 3.90E-07 1.58 (1.32–1.87) 
rs11767834 128175227 C:T 946:28 1041:33 0.03 0.03 0.80 1.07 (0.64–1.78) 
rs11761199 128175786 G:A 569:407 515:555 0.42 0.52 4.90E-06 1.51 (1.26–1.79) 
IRF5-10-1 128175834 G:A 880:98 932:144 0.10 0.13 0.024 1.37 (1.04–1.80) 
rs6975315 128176436 G:A 955:21 1052:26 0.02 0.02 0.77 1.13 (0.62–2.01) 
rs7808907 128178035 C:T 603:373 557:521 0.38 0.48 4.40E-06 1.51 (1.26–1.80) 
rs1874328 128179055 A:G 381:589 423:655 0.39 0.39 1.00 (0.83–1.19) 
IRF5-15-1 128179567 C:T 757:221 887:141 0.23 0.14 2.20E-07 1.84 (2.32–1.46) 
IRF5-18-1 128180688 G:A 952:14 1056:16 0.01 0.01 1.03 (0.50–2.12) 
30 bp indelf 128181356 Ins/del Ins 459:495 517:521 0.48 0.50 0.45 1.07 (0.90–1.28) 
G indelf 128182387 4 / 5 G 5 G 735:211 665:107 0.22 0.14 8.70E-06 1.78 (2.29–1.37) 
IRF5-22-3 128182426 C:A 936:42 1026:52 0.04 0.05 0.60 1.13 (0.74–1.71) 
rs10954213 128183378 A:G 692:286 674:404 0.29 0.37 8.60E-05 1.45 (1.20–1.74) 
rs11770589 128183439 G:A 471:507 539:539 0.48 0.50 0.40 1.08 (0.90–1.28) 
rs10954214 128183584 T:C 734:244 725:351 0.25 0.33 1.50E-04 1.45 (1.19–1.76) 
rs13242262 128185315 T:A 689:285 671:401 0.29 0.37 1.20E-04 1.44 (1.19–1.73) 
rs7800687 128187670 G:C 732:244 727:351 0.25 0.33 1.80E-04 1.44 (1.19–1.75) 
rs10488630 128187899 A:G 369:607 401:677 0.38 0.37 0.82 1.02 (1.22–0.85) 
rs10488631 128188134 T:C 754:220 941:133 0.23 0.12 9.40E-10 2.07 (2.62–1.63) 
rs2280714 128188676 T:C 734:244 728:350 0.25 0.32 2.10E-04 1.44 (1.18–1.74) 
Markera Positionb Major: minor allelesc Risk allele Allele countsg Minor allele frequencies    
    Cases Controls Cases Controls P-valuesd ORe 95% CI 
rs729302 128162911 A:C 719:259 712:366 0.27 0.34 2.70E-04 1.42 (1.17–1.72) 
rs11768806 128167079 C:T 846:130 900:178 0.13 0.17 0.041 1.29 (1.00–1.64) 
rs12537192 128167644 T:C 954:24 1033:23 0.02 0.02 0.77 1.13 (2.01–0.63) 
rs4728142 128167918 A:G 541:435 473:605 0.55 0.44 1.80E-07 1.59 (1.89–1.33) 
rs1874330 128169175 T:C 833:141 880:196 0.14 0.18 0.023 1.32 (1.04–1.67) 
rs3778754 128169503 G:C 569:407 502:572 0.42 0.53 1.80E-07 1.59 (1.33–1.89) 
rs3757386 128171248 C:T 888:90 948:126 0.09 0.12 0.062 1.31 (0.98–1.74) 
rs6968563 128171682 T:C 915:45 998:54 0.05 0.05 0.68 1.10 (0.73–1.65) 
CGGGG indelf 128171882 4 / 3 CGGGG 546:412 458:586 0.43 0.56 4.60E-09 1.69 (1.42–2.02) 
rs6953165 128172161 C:G 910:42 1009:53 0.04 0.05 0.60 1.14 (0.75–1.72) 
rs2004640 128172252 T:G 613:363 557:519 0.37 0.48 5.70E-07 1.57 (1.31–1.87) 
rs3807307 128173153 C:T 562:412 495:577 0.42 0.54 2.20E-07 1.59 (1.33–1.89) 
rs752637 128173371 C:T 693:285 676:398 0.29 0.37 1.80E-04 1.43 (1.18–1.71) 
rs3807306 128174631 T:G 608:370 550:528 0.38 0.49 3.90E-07 1.58 (1.32–1.87) 
rs11767834 128175227 C:T 946:28 1041:33 0.03 0.03 0.80 1.07 (0.64–1.78) 
rs11761199 128175786 G:A 569:407 515:555 0.42 0.52 4.90E-06 1.51 (1.26–1.79) 
IRF5-10-1 128175834 G:A 880:98 932:144 0.10 0.13 0.024 1.37 (1.04–1.80) 
rs6975315 128176436 G:A 955:21 1052:26 0.02 0.02 0.77 1.13 (0.62–2.01) 
rs7808907 128178035 C:T 603:373 557:521 0.38 0.48 4.40E-06 1.51 (1.26–1.80) 
rs1874328 128179055 A:G 381:589 423:655 0.39 0.39 1.00 (0.83–1.19) 
IRF5-15-1 128179567 C:T 757:221 887:141 0.23 0.14 2.20E-07 1.84 (2.32–1.46) 
IRF5-18-1 128180688 G:A 952:14 1056:16 0.01 0.01 1.03 (0.50–2.12) 
30 bp indelf 128181356 Ins/del Ins 459:495 517:521 0.48 0.50 0.45 1.07 (0.90–1.28) 
G indelf 128182387 4 / 5 G 5 G 735:211 665:107 0.22 0.14 8.70E-06 1.78 (2.29–1.37) 
IRF5-22-3 128182426 C:A 936:42 1026:52 0.04 0.05 0.60 1.13 (0.74–1.71) 
rs10954213 128183378 A:G 692:286 674:404 0.29 0.37 8.60E-05 1.45 (1.20–1.74) 
rs11770589 128183439 G:A 471:507 539:539 0.48 0.50 0.40 1.08 (0.90–1.28) 
rs10954214 128183584 T:C 734:244 725:351 0.25 0.33 1.50E-04 1.45 (1.19–1.76) 
rs13242262 128185315 T:A 689:285 671:401 0.29 0.37 1.20E-04 1.44 (1.19–1.73) 
rs7800687 128187670 G:C 732:244 727:351 0.25 0.33 1.80E-04 1.44 (1.19–1.75) 
rs10488630 128187899 A:G 369:607 401:677 0.38 0.37 0.82 1.02 (1.22–0.85) 
rs10488631 128188134 T:C 754:220 941:133 0.23 0.12 9.40E-10 2.07 (2.62–1.63) 
rs2280714 128188676 T:C 734:244 728:350 0.25 0.32 2.10E-04 1.44 (1.18–1.74) 

aMarkers with dbSNP build 127 rs number or in-house id.

bPosition according to genome build 36.

cMajor allele in cases.

dP-values are based on comparison of allele counts between cases and controls by Fisher’s exact test.

eOR calculated for the risk allele.

fThe alleles for the indels are three or four CGGGG units, 30 bp insertion/deletion of ‘GCCGCCTACTCTGCAGCCGCCCACTCTGCA’ and four or five nucleotides G, respectively.

gA subset of the samples from Lund (133/144) and Uppsala (77/136) were included in our original study on SLE (7), and all the samples from Stockholm and 120 samples from Uppsala were included in our more recent study (15).

Given that the genotypes at several polymorphisms in IRF5 are significantly associated with SLE, and the fact that the genotypes at different polymorphisms are not independent of one another, but are correlated to varying degrees (Fig. 1) we performed conditional logistic regression analysis to assess whether a single polymorphism can account for all observed association signals, or whether more than one polymorphism is required. Similar P-values as the ones obtained by Fischer’s exact text for the individual markers were obtained using an additive logistic regression approach (Table 2). We conditioned on the two polymorphisms that gave the most significant evidence of association individually (CGGGG indel and SNP rs10488631), conditioning on either of them one alone or both jointly. Conditioning on the CGGGG indel, we found that the association signals from each of the seven SNPs located in the 5′ promoter and first intron of IRF5 as well as the signals from SNPs rs10954213 and rs10954214 located in the 3′-UTR of IRF5 and the signal from the SNP rs2280714 located 5 kb downstream of IRF5 are abolished by this analysis (Table 2). Thus the CGGGG indel polymorphism accounts for the association signal observed from two of the SNPs that were previously suggested to be functional in SLE by affecting the expression of the IRF5 gene. These are the SNP rs2004640, which drives transcription of an alternate IRF5 transcript (8) and the SNP rs10954213 that alters a polyadenylation site of IRF5 (15,16). The signal from the SNP rs2280714 that has earlier been shown to be associated with the expression levels of IRF5 in lymphoblastoid cell lines (14) was also abolished. After logistic regression analysis conditional on the SNP rs10488631, the association signals from the SNPs IRF5-15-1 in intron 2, and the G indel polymorphism in intron 8 of IRF5, which both are in high LD with the SNP rs10488631 with r2-values >0.98 were abolished, while association signals from the seven SNPs in the 5′-region of IRF5 were retained. Conditional analysis using the CGGGG indel and the SNP rs10488631 together abolished the association signals from all markers in the IRF5 gene (Table 2). Logistic regression analysis conditional on either of the two strongly associated polymorphisms suggests a model according to which two groups of SNPs represented by the CGGGG indel and the SNP rs10488631, respectively, are independently associated with SLE. A recent study proposes a similar model according to which the SNP rs729302 located 9 kb upstream of the IRF5 gene appears to tag a potential protective haplotype and the SNP rs10488631 tags a risk haplotype (12). However, in our analysis the association signal from SNP rs729302 was abolished by the CGGGG indel.

Table 2.

Single marker logistic regression analysis and conditional logistic regression analysis

Marker Single marker P-valuesa P-values for conditional logistic regressionb 
  CGGGG indel rs10488631 CGGGG indel and rs10488631 
rs729302 3.4E-04 0.22 0.04 0.63 
rs11768806 0.04 0.94 0.35 0.71 
rs12537192 0.68 0.99 0.98 0.83 
rs4728142 3.2E-07 0.31 6.2E-04 0.63 
rs1874330 0.02 0.97 0.26 0.71 
rs3778754 3.2E-07 0.28 2.9E-03 0.89 
rs3757386 0.07 0.93 0.35 0.88 
rs6968563 0.63 0.51 0.99 0.48 
CGGGG indel 1.7E-08 NA 5.8E-05 NA 
rs6953165 0.52 0.66 0.91 0.60 
rs2004640 7.0E-07 0.23 3.0E-03 0.83 
rs3807307 4.1E-07 0.33 3.4E-03 0.79 
rs752637 1.6E-04 0.47 0.03 0.94 
rs3807306 7.6E-07 0.24 3.1E-03 0.84 
rs11767834 0.79 0.28 0.76 0.68 
rs11761199 7.6E-06 0.58 0.02 0.51 
IRF5-10-1 0.02 0.53 0.16 0.80 
rs6975315 0.69 0.96 0.99 0.93 
rs7808907 5.4E-06 0.25 0.01 0.96 
rs1874328 0.98 0.01 0.01 0.94 
IRF5-15-1 7.1E-07 8.9E-04 NA NA 
IRF5-18-1 0.93 0.61 0.86 0.57 
30 bp indel 0.44 7.9E-03 0.05 0.89 
G indel 2.0E-05 5.7E-03 0.61 0.92 
IRF5-22-3 0.55 0.65 0.94 0.58 
rs10954213 1.2E-04 0.41 0.03 0.89 
rs11770589 0.39 0.01 0.03 0.89 
rs10954214 1.9E-04 0.30 0.03 0.69 
rs13242262 1.4E-04 0.45 0.04 0.94 
rs7800687 2.3E-04 0.32 0.03 0.71 
rs10488630 0.81 0.02 8.9E-03 0.81 
rs10488631 6.4E-09 1.9E-05 NA NA 
rs2280714 2.5E-04 0.35 0.04 0.76 
Marker Single marker P-valuesa P-values for conditional logistic regressionb 
  CGGGG indel rs10488631 CGGGG indel and rs10488631 
rs729302 3.4E-04 0.22 0.04 0.63 
rs11768806 0.04 0.94 0.35 0.71 
rs12537192 0.68 0.99 0.98 0.83 
rs4728142 3.2E-07 0.31 6.2E-04 0.63 
rs1874330 0.02 0.97 0.26 0.71 
rs3778754 3.2E-07 0.28 2.9E-03 0.89 
rs3757386 0.07 0.93 0.35 0.88 
rs6968563 0.63 0.51 0.99 0.48 
CGGGG indel 1.7E-08 NA 5.8E-05 NA 
rs6953165 0.52 0.66 0.91 0.60 
rs2004640 7.0E-07 0.23 3.0E-03 0.83 
rs3807307 4.1E-07 0.33 3.4E-03 0.79 
rs752637 1.6E-04 0.47 0.03 0.94 
rs3807306 7.6E-07 0.24 3.1E-03 0.84 
rs11767834 0.79 0.28 0.76 0.68 
rs11761199 7.6E-06 0.58 0.02 0.51 
IRF5-10-1 0.02 0.53 0.16 0.80 
rs6975315 0.69 0.96 0.99 0.93 
rs7808907 5.4E-06 0.25 0.01 0.96 
rs1874328 0.98 0.01 0.01 0.94 
IRF5-15-1 7.1E-07 8.9E-04 NA NA 
IRF5-18-1 0.93 0.61 0.86 0.57 
30 bp indel 0.44 7.9E-03 0.05 0.89 
G indel 2.0E-05 5.7E-03 0.61 0.92 
IRF5-22-3 0.55 0.65 0.94 0.58 
rs10954213 1.2E-04 0.41 0.03 0.89 
rs11770589 0.39 0.01 0.03 0.89 
rs10954214 1.9E-04 0.30 0.03 0.69 
rs13242262 1.4E-04 0.45 0.04 0.94 
rs7800687 2.3E-04 0.32 0.03 0.71 
rs10488630 0.81 0.02 8.9E-03 0.81 
rs10488631 6.4E-09 1.9E-05 NA NA 
rs2280714 2.5E-04 0.35 0.04 0.76 

aP-value for association with SLE based on logistic regression.

bP-value for the association conditional on the indicated marker. NA indicates that the marker can not be distinguished from the conditional model.

A limitation of our application of conditional logistic regression analysis is that only a subset of all possible models were examined and only a single model (or a few models) is obtained that gives the best fit to the analyzed data. The selection of the SNPs to be used as conditioning variables, either based on their strongest association signal, as we did here, or on potential function as was done in our previous study (15), is somewhat arbitrary and may affect the result of the analysis. To overcome these limitations, we used a Bayesian model selection and averaging approach (18) to statistically select those IRF5 polymorphisms that are jointly most significantly associated with SLE and to estimate their effects. The advantage of the Bayesian model selection procedure is that it does not require a priori selection of the number and identity of variants that are used as the conditioning variables. Instead, all possible models (i.e. subsets of genetic variants) can be analyzed and compared with each other using the Bayesian information criterion (BIC), which is an appropriate approximation to the Bayes factor on regular models (19,20) such as these (18). The BIC can be used to assess whether the dimension of a model (i.e. the number of SNPs) is justified for the amount of phenotypic variation that their genotypes jointly explain. The BIC allows computation of the approximate posterior probabilities of competing models, using in the computation either all possible models or only parsimonious ones, such as those within Occam’s window (21). The posterior probabilities of competing models can then be used as weights in a parameter estimation procedure that explicitly allows for model uncertainty (20,22). The posterior probabilities of the effect of individual SNPs can be obtained by summing the posterior probabilities of all models that contain a given SNP.

Similar P-values for the individual SNPs were obtained using this additive measured genotype association analysis as with Fisher’s exact test and conditional logistic regression (Tables 1–3). According to determination of the posterior probabilities for all variants, based on all competing models within Occam’s window (21), two variants, the CGGGG indel and the SNP rs10488631 show very high posterior probabilities of association with SLE (0.71 and 1.00, respectively) (Table 3). These two polymorphisms are thus the most probable functional variants of IRF5, although we cannot exclude the SNP IRF5-15-1 or the G indel, due to their extremely high genotype correlation with the SNP rs10488631. Given that the SNP rs10488631 is located downstream of IRF5 and that the SNP IRF5-15-1 is located in intron 2, the latter may well be the better candidate for follow-up functional studies. Only models containing exactly two variants were identified as being parsimonious, suggesting that among the genotyped variants there are exactly two sites that are independently functional in SLE. The most probable model contains the SNP rs10488631 (=IRF5-15-1 or G indel) together with the previously unknown CGGGG indel (posterior probability = 0.71), followed by the SNP rs10488631 in combination with the SNPs rs3807306 (0.09), rs4728142 (0.07), rs3778754 (0.05), rs10488630 (0.04) or rs2004640 (0.04) in order of probability.

Table 3.

Bayesian model selection analysis

Marker Single marker P-values Posterior probability of association 
  All markersan = 923 Uncorrelated markersbn = 943 
rs729302 2.9E-04 
rs11768806 0.037 – 
rs12537192 0.68 
rs4728142 2.1E-07 0.074 – 
rs1874330 0.020 
rs3778754 2.0E-07 0.050 – 
rs3757386 0.069 
rs6968563 0.63 – 
CGGGG indel 1.0E-08 0.71 0.90 
rs6953165 0.53 
rs2004640 4.4E-07 0.036 0.051 
rs3807307 2.6E-07 – 
rs752637 1.4E-04 
rs3807306 4.9E-07 0.089 – 
rs11767834 0.79 
rs11761199 5.7E-06 – 
IRF5-10-1 0.020 
rs6975315 0.69 
rs7808907 3.9E-06 – 
rs1874328 1.00 – 
IRF5-15-1 4.2E-07 – – 
IRF5-18-1 0.93 
30 bp indel 0.45 
G indel 1.2E-05 – – 
IRF5-22-3 0.55 – 
rs10954213 1.0E-04 
rs11770589 0.40 
rs10954214 1.6E-04 – 
rs13242262 1.2E-04 – 
rs7800687 2.0E-04 – 
rs10488630 0.79 0.040 0.049 
rs10488631 2.4E-09 
rs2280714 2.1E-04 – 
Marker Single marker P-values Posterior probability of association 
  All markersan = 923 Uncorrelated markersbn = 943 
rs729302 2.9E-04 
rs11768806 0.037 – 
rs12537192 0.68 
rs4728142 2.1E-07 0.074 – 
rs1874330 0.020 
rs3778754 2.0E-07 0.050 – 
rs3757386 0.069 
rs6968563 0.63 – 
CGGGG indel 1.0E-08 0.71 0.90 
rs6953165 0.53 
rs2004640 4.4E-07 0.036 0.051 
rs3807307 2.6E-07 – 
rs752637 1.4E-04 
rs3807306 4.9E-07 0.089 – 
rs11767834 0.79 
rs11761199 5.7E-06 – 
IRF5-10-1 0.020 
rs6975315 0.69 
rs7808907 3.9E-06 – 
rs1874328 1.00 – 
IRF5-15-1 4.2E-07 – – 
IRF5-18-1 0.93 
30 bp indel 0.45 
G indel 1.2E-05 – – 
IRF5-22-3 0.55 – 
rs10954213 1.0E-04 
rs11770589 0.40 
rs10954214 1.6E-04 – 
rs13242262 1.2E-04 – 
rs7800687 2.0E-04 – 
rs10488630 0.79 0.040 0.049 
rs10488631 2.4E-09 
rs2280714 2.1E-04 – 

aAll markers except the SNP 15-1, which was excluded based on its high genotype correlation (P > 0.98) with the SNP rs10488631) and the G indel because of its lower genotyping success rate (<85%).

bUncorrelated markers only, the markers (–) were excluded based on P > 0.9.

The previous Bayesian model selection analysis included some SNPs with relatively high genotype correlations among each other, which potentially could cause problems due to multicollinearity in joint analyses. To confirm the previous finding, we repeated the analysis on a subset of 17 polymorphisms that do not have any high pair-wise genotype correlation among each other (correlation coefficient ≥0.9). All but the variant giving the most significant evidence of association in marginal testing were excluded from each set of variants with high intra-set genotype correlations (Table 3). Joint analysis on this set of 17 polymorphisms confirmed the findings from the analysis of all 31 variants, with the best single model again being the one including the CGGGG indel and the SNP rs10488631 (or the SNP IRF5-15-1), with posterior probabilities of 0.90 and 1.00, respectively. Thus the results from the Bayesian model selection procedure corroborate those obtained by conditional logistic regression analysis described above.

The two alleles of the CGGGG indel polymorphisms contain either three or four repeats of the sequence CGGGG. The insertion of one CGGGG repeat is the risk allele (4x CGGGG) for SLE and is predicted with an extremely high probability score (94.5%) (23) to have three so called GC-boxes ‘GGGCGGG’ that are binding sites for the transcription factor SP1 and related factors (24), while the shorter (3x CGGGG) allele is predicted to have two binding sites for SP1. Using electrophoretic mobility shift assays (EMSA) we observed a higher level of binding of protein to the 4x CGGGG allele of IRF5 than to the 3x CGGGG allele (Fig. 2A), which supports the sequence-based prediction of SP1 binding to the CGGGG alleles of IRF5. We also performed two experiments to compare the expression levels of IRF5 between individuals carrying the 4x CGGGG and 3x CGGGG alleles. Using minigene reporter assays with promoters cloned from individuals homozygous for either the 4x or the 3x allele of the CGGGG indel, we observed two- to ten-fold higher expression levels of IRF5 transcript containing exon 1a from promoters with the 4x CGGGG allele than from the corresponding promoters with the 3x CGGGG allele of IRF5 (Fig. 2B). Using immunoblot analysis we show that three SLE patients, who are homozygous for the 4x CGGGG allele express a higher level of IRF5 protein in PBMCs than a patient homozygous for the 3x CGGGG allele and a patient heterozygous for the CGGGG indel (Fig. 2c).

Figure 2.

(A)Electrophoretic mobility shift assay for the two alleles of the CGGGG indel showing stronger protein binding to the 4x allele compared with the 3x allele. Lanes 1–3 contain: (1) Labeled probe; (2) Labeled probe and nuclear extract; (3) Labeled probe, nuclear extract and 100-fold excess of an unlabeled competitor probe for the 4x allele. Lanes 4–6 contain: (4) Labeled probe; (5) Labeled probe and nuclear extract; (6) Labeled probe, nuclear extract and 100-fold excess of an unlabeled competitor probe for the 3x allele. (B) Relative levels of IRF5 mRNA expressed from minigene constructs of promoters cloned from four individuals homozygous for the 4x CGGGG and two individuals for the 3x CGGGG allele of the CGGGG indel polymorphisms. The relative expression levels of IRF5 transcript from exon 1a are shown after normalization with respect to transfected pcDNA3 vector or beta actin levels from three independent experiments. (C) Immunoblot analysis of IRF5 expressed in peripheral blood mononuclear cells (PBMCs). Lanes 1–3 contain PBMC samples from homozygotes for the 4x CGGGG indel; Lane 4 contains PBMCs from a heterozygote for the CGGGG indel; Lane 5 contains PBMCs from a homozygote for the 3x CGGGG indel.

Figure 2.

(A)Electrophoretic mobility shift assay for the two alleles of the CGGGG indel showing stronger protein binding to the 4x allele compared with the 3x allele. Lanes 1–3 contain: (1) Labeled probe; (2) Labeled probe and nuclear extract; (3) Labeled probe, nuclear extract and 100-fold excess of an unlabeled competitor probe for the 4x allele. Lanes 4–6 contain: (4) Labeled probe; (5) Labeled probe and nuclear extract; (6) Labeled probe, nuclear extract and 100-fold excess of an unlabeled competitor probe for the 3x allele. (B) Relative levels of IRF5 mRNA expressed from minigene constructs of promoters cloned from four individuals homozygous for the 4x CGGGG and two individuals for the 3x CGGGG allele of the CGGGG indel polymorphisms. The relative expression levels of IRF5 transcript from exon 1a are shown after normalization with respect to transfected pcDNA3 vector or beta actin levels from three independent experiments. (C) Immunoblot analysis of IRF5 expressed in peripheral blood mononuclear cells (PBMCs). Lanes 1–3 contain PBMC samples from homozygotes for the 4x CGGGG indel; Lane 4 contains PBMCs from a heterozygote for the CGGGG indel; Lane 5 contains PBMCs from a homozygote for the 3x CGGGG indel.

DISCUSSION

Our statistical analysis of a comprehensive set of polymorphisms in the IRF5 gene defines four IRF5 polymorphism, to be subjected to functional analyses to elucidate the role of IRF5 in SLE. The CGGGG indel located in the promoter region of IRF5, only 64 bp upstream of exon 1a of IRF5, is an unequivocal candidate for a causal variant of IRF5, according to the Bayesian model selection method in combination with the preliminary experimental data presented here. The sequence-based prediction of an additional SP1 binding site in the 4x allele of the CGGGG indel polymorphism that confers risk for SLE was verified experimentally by increased protein binding to this allele. The IFN stimulated response elements (ISREs) are known to interact with SP1 binding sites to increase the expression of the IFN inducible protein kinase regulated by RNA (PKR) gene (25,26). This could also be the case for IRF5, which contains an ISRE motif and is known to be induced by IFNs (27). A role for the CGGGG indel as regulator of IRF5 expression is supported by our finding of increased transcription of IRF5 mRNA from exon 1a using minigene constructs with promoters containing the 4x allele of the CGGGG indel compared with promoters containing the 3x allele of the CGGGG indel and by increased levels of IRF5 protein observed in three individuals homozygous for the risk allele of the CGGGG indel compared with a homozygote for the protective allele and a heterozygote. Previously, two SNPs in IRF5 have been suggested to be functional in SLE by affecting the expression of IRF5. The major allele (T) of the SNP rs2004640, which is located at the splice junction of the alternative exon 1b of IRF5 has been shown to drive transcription of a unique IRF5 mRNA in PBMCs (13,15). However, this transcript is expressed at a very low level relative to the transcript from exon 1a (13,15), and the expression of IRF5 mRNA containing exon 1a has been reported to be independent of the genotype of the SNP rs2004640 (13). These findings speak against a major functional role for the SNP rs2004640 as regulator of gene expression in SLE, and support a functional role for the CGGGG indel based on its position close to exon 1a of IRF5. However, the complex expression pattern of IRF5 with multiple splice variants (27) warrants further investigation in SLE. The association signals with SLE observed for the SNP rs2004640 in many studies (8–12) could be due to its relatively high LD with the CGGGG indel. The major allele (A) of the SNP rs10954213 located in the 3′-UTR of IRF5 has been shown to be correlated with a truncated IRF5 transcript with elevated expression levels in PBMCs (15,16). The minigene promoter construct that we used here to test the expression of IRF5 mRNA does not contain the 3′-UTR of IRF5, and hence the elevated expression level of IRF5 mRNA that we observe using a promoter with 4x CGGGG units is independent of the genotype of the SNP rs10954213. Moreover, the Bayesian model averaging approach that we used in our study to analyze the association data, considers the risk for SLE of all possible allele combinations, and does not support a major disease-causing role for either of these SNPs alone or in combination with the risk allele of the CGGGG indel.

The second functional variant located in the IRF5 gene could be the SNP 15-1 in intron 2, the G indel in intron 8 or the SNP rs2070197 in the 3′-UTR of IRF5. These SNPs in IRF5 are in almost perfect LD with the SNP rs10488631 located ∼5 kb downstream of IRF5 and according to the data from the HapMap project with 11 additional SNPS within a 100 kb region downstream of IRF5 which contains the transportin 3 (TNPO3) gene. Using statistical methods we cannot distinguish between these highly correlated polymorphisms that yield strong signals of association with SLE. Based on bioinformatics, no obvious functional role for any of these SNPs in 3′ end of the IRF5 gene and the TNPO3 gene region can be predicted.

We attempted to identify distinct functions in SLE for the polymorphisms represented by the SNP rs10488631 and the CGGGG indel by mapping their risk alleles against the 11 ACR criteria used to classify patients with SLE, but no significant signals of association with any of the ACR criteria fulfilled by the patients were observed. Interestingly, however, in two other autoimmune diseases, inflammatory bowel diseases (IBDs) (28) and multiple sclerosis (MS) (G. Kristjansdottir, J.K. Sandling, A. Bonetti, I.M. Roos, L. Milani, C. Wang, S. Gustafsdottir, S. Sigurdsson, A. Lundmark, P.J. Tienari, K. Koivisto, I. Elovaara, T. Pirttilä, M. Reunanen, L. Peltonen, J. Saarela, J. Hillert, T. Olsson, U. Landegren, A. Alcina, O. Fernández, L. Leyva, M. Guerrero, M. Lucas, G. Izquierdo, F. Matesanz, and A-C. Syvänen, manuscript in preparation) we have observed association signals from the CGGGG indel, but no association for the SNP rs10488631 or its proxies. In an earlier study on rheumatoid arthritis (RA) we observed association of the SNP rs3807306 (29), which is correlated with the CGGGG indel (r2 = 0.61). Two recent genome-wide association studies failed to detected an association with RA for the SNP rs10488631 or its proxies (30, 31), while one of the studies confirmed the association of SNP rs3807306 with RA (30). Our results therefore suggest a general function for IRF5 as regulator of the autoimmune response, with the 4x allele of the CGGGG indel as a putative causal IRF5 allele in at least SLE, IBD and MS, and possibly in RA. On the other hand, one or more of the three SLE-associated IRF5 polymorphisms represented by the SNP rs10488631 or SNPs downstream of IRF5 may contribute to the disease manifestations that are specific for SLE. The function of the risk allele of the CGGGG indel in autoimmune diseases could be to up-regulate the expression of one or several of the IRF5 isoforms, which could facilitate expression of genes encoding type I IFNs and other proinflammatory and immunostimulatory cytokines. Increased IRF5 levels can furthermore enhance apoptosis (32,33), resulting in increased release of nucleic acids that activate the TLRs 7–9 (2,3) and of autoantigens, thus increasing the risk for the development of autoimmunity.

MATERIALS AND METHODS

Study subjects

Our study included 485 Swedish patients with SLE and 563 controls. The patients originate from the rheumatology clinics at the Lund, Uppsala and Karolinska (Stockholm) University Hospitals in Sweden. Each of the patients fulfilled at least four of the classification criteria for SLE as defined by the American College of Rheumatology (ACR) (34). The samples from Uppsala consisted of 136 patients with SLE and 155 controls. The samples from Stockholm consisted of 205 patients (35) and 213 controls, and the samples from Lund consisted of 144 patients and 195 controls (36). The controls were from the same regions as the patients and were matched for age and sex. The samples from Lund (120 out of 144 samples) and Uppsala (77 out of 136 samples) were included in our original study on SLE (7), and the samples from Stockholm (205/205) and Uppsala (120/136) in were included in our more recent study (16). All study subjects provided informed consent to participation in the study, and the study was approved by the regional ethical boards.

Genotyping

The panel of polymorphisms in the IRF5 gene that was genotyped included 24 SNPs and three length polymorphisms identified by sequencing the exons, introns and 1 kb of the region upstream of the first exon of the IRF5 gene in 40 Swedish patients with SLE and eight controls (15). Four of the 30 originally identified SNPs were located within a 60 bp distance of another SNP and were omitted from genotyping because design of Golden Gate assays was not feasible for them. Assay design for the SNP rs1874327 failed for other reasons. Two SNPs, rs2070197 and rs12539741, in the 3′-region of IRF5 were omitted because they are known to be in close to perfect LD (r2 > 0.95) with the SNP rs10488631 included in the panel (15). In addition, the panel included the SNP rs729302 located 9 kb upstream of IRF5 identified in our original study (7) and 19 SNPs selected from dbSNP that were located within 5 kb regions upstream and downstream of IRF5 (Supplementary Material, Table S1). The SNPs were genotyped in 250 ng of DNA extracted from blood samples of the study subjects using the Illumina Golden Gate assay according to protocols provided by Illumina (17) (Illumina, San Diego, USA). Out of the 44 genotyped SNPs, 40 (91%) were successfully genotyped in 1060 out of 1096 samples (97% sample success rate). Four SNPs were failed due to poor genotype clustering. Of the successfully genotyped SNPs, 10 had minor allele frequencies below 1%. Thirty-two samples were genotyped in duplicates with the Golden Gate assay with 99.98% consistency between genotype calls. Eight of the IRF5 SNPs had previously been genotyped in 60% of the samples (15) using a different genotyping method (SNPstream, Beckman Coulter). For these 5321 genotypes determined in duplicate with two different genotyping methods, the reproducibility was 99.5%.

The 5 bp indel (CGGGG indel) and the 30 bp indel in exon 6 of IRF5 were genotyped by PCR amplification followed by size separation by electrophoresis using 4% agarose gels, or using an ABI 3770 capillary sequencer (Applied Biosystems, Foster City, USA). A single base pair indel (G indel) and a two base pair indel (dbSNP ID rs3834330) were genotyped by sequencing of PCR fragments (Macrogen, South Korea). The indel rs3834330 (TA/–) was sequenced in 96 patients and 96 controls to estimate the LD of the indel with the genotyped SNPs. The indel rs3834330 was in perfect LD (r2 = 1) with SNP rs3807134 and the genotyped SNP rs3757386, and was therefore not genotyped in the whole sample set. The G indel was sequenced in all samples with 84% success. All genotyped SNPs and indels conformed to Hardy–Weinberg equilibrium in the control samples (P > 0.01). Supplementary Material, Table S1 includes additional genotype information for the SNPs and indels, and Supplementary Material, Table S2 includes the primers used for PCR amplification and sequencing of indels. All oligonucleotides used were obtained from Integrated DNA Technologies (IDT Inc, Coralville, IA, USA).

Statistical analysis

Genotype data quality was verified for each marker by testing for Hardy–Weinberg equilibrium in the controls samples using Fisheŕs exact test at P < 0.01. The allele counts in the cases and controls were compared using Fisher’s exact test to estimate the significance of the association with SLE for each individual marker. Estimation of linkage disequilibrium, calculation of ORs and conditional logistic regression analysis were performed using the software PLINK (37). A Bayesian model selection approach was used to statistically identify those genetic variants (from among all genotyped ones) that are jointly most strongly associated with the SLE disease status phenotype (18). Model selection was based on an underlying additive measured genotype association model in which the dosage of the rarer allele (0, 1 or 2 copies) was used as a linear predictor of phenotypic status, and this fixed effects model was incorporated into a standard variance components-based genetic analysis approach, as implemented in the SOLAR software package (18,38). Thirty-one of the IRF5 polymorphisms for which complete genotype data was available in 923 individuals were included in the analysis. In the joint analysis the marker G indel was omitted (success rate 84%) because only individuals with complete genotype information can be included in the analysis. The SNP IRF5-15-1 was omitted from the analysis because of its high correlation with the SNP rs10488631.

Electrophoretic mobility shift assays

Pairs of complementary 5'-biotinylated and unmodified 37 bp oligonucleotides were designed for the two alleles of the CGGGG indel polymorphism (Supplementary Material, Table S2). The complementary oligonucleotides were annealed in 10 mM Tris–HCl, pH 7.5, 50 mm NaCl and 1 mm EDTA to form double stranded probes for the EMSA reaction. Twenty fmoles of the biotinylated double stranded probes were incubated with 2 µl of nuclear extract prepared from blood cells, using the NE-PER Nuclear and Cytoplasmic Extraction Reagents Kit (Pierce Biotechnology, Rockford, IL, USA), in a freshly made binding buffer containing 12 mm HEPES, pH 7.4, 5 mm MgCl2, 60 mm KCl, 1% glycerol, 0.05% NP−40, 50 µg/µl BSA, 1 mm DTT, 0.5 mm EDTA with 50 ng/µl of poly(dI-dC)·poly(dI-dC) (GE Healthcare, Piscataway, NJ, USA) and Halt™ Protease Inhibitor Cocktail (Pierce Biotechnology) in a final volume of 20 µl for 20 min. The reaction mixtures were analyzed by 6% polyacryl-amide gel electrohoresis (PAGE) and transferred to Hybond NT membranes, according to the manufacturer’s instruction (LightShift Chemiluminescent EMSA kit, Pierce). The Chemiluminescent Nucleic Acid Detection Module (Pierce Biotechnology) was used to detect the biotinylated oligonucleotides on the membranes using a ChemiDoc XRS system (Bio-Rad Laboratories, Hercules, CA, USA).

Minigene reporter assays

DNA fragments for IRF5 5′-UTR minigene constructs were generated by PCR from genomic DNA from six SLE patients using a primer located 1.4 kb upstream of exon 1a and a primer in exon 2 (Supplementary Material, Table S2). DNA fragments were amplified using the Expand Long Template PCR kit (Roche Applied Sciences, Indianapolis, IN, USA) following the manufacturer’s instructions. The thermocycling conditions were10 min at 94°C, 10 cycles of 20 s at 94°C, 30 s at 56°C, 8 min at 68°C, 3 single cycles of 25 s at 94°C, 30 s at 56°C and 8 min at 68°C, increasing each subsequent extension step by 5 s, then 15 cycles of 25 s at 94°C, 30 s at 56°C and 8 min 20 s at 68°C. PCR products were initially cloned into the pCR-Blunt II TOPO vector (Invitrogen, Carlsbad, CA, USA). After verification by sequencing, the fragments were subcloned into a pcDNA3 vector using NotI-EcoRI digest. The pcDNA3 constructs were transfected into 293T cells lacking endogenous IRF5 expression using the Lipofectamine 2000 reagent (Invitrogen). Twenty-four hours after transfection, cDNA corresponding to exon 1a-associated transcripts were synthesized from 1 µg of DNAse-treated total RNA using AMV reverse transcriptase (Invitrogen). Fifty nanogram of cDNA was subjected to 35 PCR cycles using Taq DNA polymerase (Invitrogen). The primers used in real-time PCR were derived either from sequences of the pcDNA3 vector or specific sequences recognizing exon 1a-associated transcripts amplified through exon 2 (27). Real-time PCR assays were carried out with the ABI Prism 7300 Sequence Detector (Applied Biosystems) using the Power SYBR® Green PCR Master Mix at 2x concentration and 150 µM of forward and reverse primers. The specificity of each reaction was examined by dissociation curve analysis generated in the range of 60–95°C using the geometric mean of pcDNA3 vector or beta-actin for standardization.

Immunoblot analysis

Cell lysates from purified PBMC pellets were prepared by resuspending in equal volumes of SDS sample buffer (50 µl ß-ME were added to 950 µl Laemmli sample buffer). Samples were denatured at 100°C for 5 min and stored at −20°C. Proteins were then separated by 10% SDS–PAGE, transferred onto nitrocellulose membrane and probed with mouse anti-human IRF5 antibodies at a 1:1000 dilution (2E3-1A11, Novus Biologicals, Littlelton, CO, USA) and rabbit anti-alpha-tubulin antibodies at a 1:1000 dilution (Cell Signaling Technology, Danvers, MA, USA). HRP-conjugated anti-mouse (1:1000) and anti-rabbit (1:2000) secondary antibodies were used. Immunoreactive protein complexes were visualized with enhanced chemiluminescence using ECL reagents (GE Healthcare). The membranes were stripped and reprobed using Restore Western Blot stripping buffer (Pierce Biotechnology).

SUPPLEMENTARY MATERIAL

Supplementary Material is available at HMG Online.

FUNDING

The study was supported by a Target Identification in Lupus (TIL) grant from the Alliance for Lupus Research, USA, by grants from the Swedish Research Council for Medicine, Agnes and Mac Rudberg’s Foundation and from the Knut and Alice Wallenberg Foundation, the Swedish Rheumatism Association, Centre of Gender related Medicine at Karolinska Institutet, the Swedish Heart-Lung Foundation, the Åke Wiberg foundation, the King Gustaf V 80-year Foundation and Ulla and Roland Gustafsson Foundation. Development of SOLAR is supported by a grant from the US National Institutes of Mental Health (MH059490). The study was is part supported by a grant to B.J.B from the US National Institute of Arthritis and Musculoskeletal and Skin Diseases (R03AR054070). Funding to pay Open Access publication charges for this article was provided by the Swedish Research Council for Medicine.

ACKNOWLEDGEMENTS

We thank Anders Lundmark for excellent technical assistance with genotyping. Genotyping was performed using equipment available at the SNP Technology Platform in Uppsala (www.genotyping.se). We thank Dr Timothy W. Behrens and Dr Robert R Graham for helpful discussion.

Conflict of Interest statement. None declared.

REFERENCES

1
Baechler
E.C.
Batliwalla
F.M.
Karypis
G.
Gaffney
P.M.
Ortmann
W.A.
Espe
K.J.
Shark
K.B.
Grande
W.J.
Hughes
K.M.
Kapur
V.
, et al.  . 
Interferon-inducible gene expression signature in peripheral blood cells of patients with severe lupus
Proc. Natl Acad. Sci. USA
 , 
2003
, vol. 
100
 (pg. 
2610
-
2615
)
2
Rönnblom
L.
Eloranta
M.L.
Alm
G.V.
The type I interferon system in systemic lupus erythematosus
Arthritis Rheum.
 , 
2006
, vol. 
54
 (pg. 
408
-
420
)
3
Yasuda
K.
Richez
C.
Maciaszek
J.W.
Agrawal
N.
Akira
S.
Marshak-Rothstein
A.
Rifkin
I.R.
Murine dendritic cell type I IFN production induced by human IgG-RNA immune complexes is IFN regulatory factor (IRF)5 and IRF7 dependent and is required for IL-6 production
J. Immunol.
 , 
2007
, vol. 
178
 (pg. 
6876
-
6885
)
4
Takaoka
A.
Yanai
H.
Kondo
S.
Duncan
G.
Negishi
H.
Mizutani
T.
Kano
S.
Honda
K.
Ohba
Y.
Mak
T.W.
, et al.  . 
Integral role of IRF-5 in the gene induction programme activated by Toll-like receptors
Nature
 , 
2005
, vol. 
434
 (pg. 
243
-
249
)
5
Honda
K.
Taniguchi
T.
IRFs: master regulators of signalling by Toll-like receptors and cytosolic pattern-recognition receptors
Nat. Rev. Immunol.
 , 
2006
, vol. 
6
 (pg. 
644
-
658
)
6
Yanai
H.
Chen
H.M.
Inuzuka
T.
Kondo
S.
Mak
T.W.
Takaoka
A.
Honda
K.
Taniguchi
T.
Role of IFN regulatory factor 5 transcription factor in antiviral immunity and tumor suppression
Proc. Natl Acad. Sci. USA
 , 
2007
, vol. 
104
 (pg. 
3402
-
3407
)
7
Sigurdsson
S.
Nordmark
G.
Goring
H.H.
Lindroos
K.
Wiman
A.C.
Sturfelt
G.
Jonsen
A.
Rantapaa-Dahlqvist
S.
Moller
B.
Kere
J.
, et al.  . 
Polymorphisms in the tyrosine kinase 2 and interferon regulatory factor 5 genes are associated with systemic lupus erythematosus
Am. J. Hum. Genet.
 , 
2005
, vol. 
76
 (pg. 
528
-
537
)
8
Graham
R.R.
Kozyrev
S.V.
Baechler
E.C.
Reddy
M.V.
Plenge
R.M.
Bauer
J.W.
Ortmann
W.A.
Koeuth
T.
Gonzalez Escribano
M.F.
Pons-Estel
B.
, et al.  . 
A common haplotype of interferon regulatory factor 5 (IRF5) regulates splicing and expression and is associated with increased risk of systemic lupus erythematosus
Nat. Genet.
 , 
2006
, vol. 
38
 (pg. 
550
-
555
)
9
Demirci
F.Y.
Manzi
S.
Ramsey-Goldman
R.
Minster
R.L.
Kenney
M.
Shaw
P.S.
Dunlop-Thomas
C.M.
Kao
A.H.
Rhew
E.
Bontempo
F.
, et al.  . 
Association of a common interferon regulatory factor 5 (IRF5) variant with increased risk of systemic lupus erythematosus (SLE)
Ann. Hum. Genet.
 , 
2007
, vol. 
71
 (pg. 
308
-
311
)
10
Shin
H.D.
Sung
Y.K.
Choi
C.B.
Lee
S.O.
Lee
H.W.
Bae
S.C.
Replication of the genetic effects of IFN regulatory factor 5 (IRF5) on systemic lupus erythematosus in a Korean population
Arthritis Res. Ther.
 , 
2007
, vol. 
9
 pg. 
R32
 
11
Reddy
M.V.
Velazquez-Cruz
R.
Baca
V.
Lima
G.
Granados
J.
Orozco
L.
Alarcon-Riquelme
M.E.
Genetic association of IRF5 with SLE in Mexicans: higher frequency of the risk haplotype and its homozygozity than Europeans
Hum. Genet.
 , 
2007
, vol. 
121
 (pg. 
721
-
727
)
12
Ferreiro-Neira
I.
Calaza
M.
Alonso-Perez
E.
Marchini
M.
Scorza
R.
Sebastiani
G.D.
Blanco
F.J.
Rego
I.
Pullmann
R.
Jr
Pullmann
R.
, et al.  . 
Opposed independent effects and epistasis in the complex association of IRF5 to SLE
Genes Immun.
 , 
2007
, vol. 
8
 (pg. 
429
-
438
)
13
Kozyrev
S.V.
Lewen
S.
Reddy
P.M.
Pons-Estel
B.
Witte
T.
Junker
P.
Laustrup
H.
Gutierrez
C.
Suarez
A.
Francisca Gonzalez-Escribano
M.
, et al.  . 
Structural insertion/deletion variation in IRF5 is associated with a risk haplotype and defines the precise IRF5 isoforms expressed in systemic lupus erythematosus
Arthritis Rheum.
 , 
2007
, vol. 
56
 (pg. 
1234
-
1241
)
14
Cheung
V.G.
Spielman
R.S.
Ewens
K.G.
Weber
T.M.
Morley
M.
Burdick
J.T.
Mapping determinants of human gene expression by regional and genome-wide association
Nature
 , 
2005
, vol. 
437
 (pg. 
1365
-
1369
)
15
Graham
R.R.
Kyogoku
C.
Sigurdsson
S.
Vlasova
I.A.
Davies
L.R.
Baechler
E.C.
Plenge
R.M.
Koeuth
T.
Ortmann
W.A.
Hom
G.
, et al.  . 
Three functional variants of IFN regulatory factor 5 (IRF5) define risk and protective haplotypes for human lupus
Proc. Natl Acad. Sci. USA
 , 
2007
, vol. 
104
 (pg. 
6758
-
6763
)
16
Graham
D.S.
Manku
H.
Wagner
S.
Reid
J.
Timms
K.
Gutin
A.
Lanchbury
J.S.
Vyse
T.J.
Association of IRF5 in UK SLE families identifies a variant involved in polyadenylation
Hum. Mol. Genet.
 , 
2007
, vol. 
16
 (pg. 
579
-
591
)
17
Fan
J.B.
Oliphant
A.
Shen
R.
Kermani
B.G.
Garcia
F.
Gunderson
K.L.
Hansen
M.
Steemers
F.
Butler
S.L.
Deloukas
P.
, et al.  . 
Highly parallel SNP genotyping
Cold Spring Harb. Symp. Quant. Biol
 , 
2003
, vol. 
68
 (pg. 
69
-
78
)
18
Blangero
J.
Goring
H.H.
Kent
J.W.
Jr
Williams
J.T.
Peterson
C.P.
Almasy
L.
Dyer
T.D.
Quantitative trait nucleotide analysis using Bayesian model selection
Hum. Biol.
 , 
2005
, vol. 
77
 (pg. 
541
-
559
)
19
Schwarz
G.
Estimating the dimension of a model
Ann. Stat.
 , 
1978
, vol. 
6
 (pg. 
461
-
464
)
20
Raftery
A.
Bayesian Model Selection in Social Research
1995
Oxford
Blackwell
21
Madigan
D.R.A.E.
Model selection and accounting for model uncertainty in graphical models using Occam’s window
J. Amer. Statist. Assoc.
 , 
1994
, vol. 
89
 (pg. 
1535
-
1546
)
22
Raftery
A.
Madigan
D.
Hoeting
J.
Bayesian Model Averaging for Linear Regression Models
J. Amer. Statist. Assoc.
 , 
1997
, vol. 
92
 (pg. 
179
-
191
)
23
Heinemeyer
T.
Wingender
E.
Reuter
I.
Hermjakob
H.
Kel
A.E.
Kel
O.V.
Ignatieva
E.V.
Ananko
E.A.
Podkolodnaya
O.A.
Kolpakov
F.A.
, et al.  . 
Databases on transcriptional regulation: TRANSFAC, TRRD and COMPEL
Nucleic Acids Res.
 , 
1998
, vol. 
26
 (pg. 
362
-
367
)
24
Suske
G.
The Sp-family of transcription factors
Gene
 , 
1999
, vol. 
238
 (pg. 
291
-
300
)
25
Das
S.
Ward
S.V.
Tacke
R.S.
Suske
G.
Samuel
C.E.
Activation of the RNA-dependent protein kinase PKR promoter in the absence of interferon is dependent upon Sp proteins
J. Biol. Chem.
 , 
2006
, vol. 
281
 (pg. 
3244
-
3253
)
26
Ward
S.V.
Samuel
C.E.
The PKR kinase promoter binds both Sp1 and Sp3, but only Sp3 functions as part of the interferon-inducible complex with ISGF-3 proteins
Virology
 , 
2003
, vol. 
313
 (pg. 
553
-
566
)
27
Mancl
M.E.
Hu
G.
Sangster-Guity
N.
Olshalsky
S.L.
Hoops
K.
Fitzgerald-Bocarsly
P.
Pitha
P.M.
Pinder
K.
Barnes
B.J.
Two discrete promoters regulate the alternatively spliced human interferon regulatory factor-5 isoforms. Multiple isoforms with distinct cell type-specific expression, localization, regulation, and function
J. Biol. Chem.
 , 
2005
, vol. 
280
 (pg. 
21078
-
21090
)
28
Dideberg
V.
Kristjansdottir
G.
Milani
L.
Libioulle
C.
Sigurdsson
S.
Louis
E.
Wiman
A.-C.
Vermeire
S.
Rutgeerts
P.
Belaiche
J.
, et al.  . 
An insertion–deletion polymorphism in the interferon regulatory factor 5 (IRF5) gene confers risk of inflammatory bowel diseases
Hum. Mol. Genet.
 , 
2007
, vol. 
16
 (pg. 
3008
-
3016
)
29
Sigurdsson
S.
Padyukov
L.
Kurreeman
F.A.
Liljedahl
U.
Wiman
A.C.
Alfredsson
L.
Toes
R.
Ronnelid
J.
Klareskog
L.
Huizinga
T.W.
, et al.  . 
Association of a haplotype in the promoter region of the interferon regulatory factor 5 gene with rheumatoid arthritis
Arthritis Rheum.
 , 
2007
, vol. 
56
 (pg. 
2202
-
2210
)
30
The_Wellcome_Trust_Case_Control_Consortium
Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls
Nature
 , 
2007
, vol. 
447
 (pg. 
661
-
678
)
31
Plenge
R.M.
Seielstad
M.
Padyukov
L.
Lee
A.T.
Remmers
E.F.
Ding
B.
Liew
A.
Khalili
H.
Chandrasekaran
A.
Davies
L.R.
, et al.  . 
TRAF1-C5 as a risk locus for rheumatoid arthritis—a genomewide study
N. Engl. J. Med.
 , 
2007
, vol. 
357
 (pg. 
977
-
986
)
32
Barnes
B.J.
Richards
J.
Mancl
M.
Hanash
S.
Beretta
L.
Pitha
P.M.
Global and distinct targets of IRF-5 and IRF-7 during innate response to viral infection
J. Biol. Chem.
 , 
2004
, vol. 
279
 (pg. 
45194
-
45207
)
33
Barnes
B.J.
Kellum
M.J.
Pinder
K.E.
Frisancho
J.A.
Pitha
P.M.
Interferon regulatory factor 5, a novel mediator of cell cycle arrest and cell death
Cancer Res.
 , 
2003
, vol. 
63
 (pg. 
6424
-
6431
)
34
Tan
E.M.
Cohen
A.S.
Fries
J.F.
Masi
A.T.
McShane
D.J.
Rothfield
N.F.
Schaller
J.G.
Talal
N.
Winchester
R.J.
The 1982 revised criteria for the classification of systemic lupus erythematosus
Arthritis Rheum.
 , 
1982
, vol. 
25
 (pg. 
1271
-
1277
)
35
Svenungsson
E.
Gunnarsson
I.
Fei
G.Z.
Lundberg
I.E.
Klareskog
L.
Frostegard
J.
Elevated triglycerides and low levels of high-density lipoprotein as markers of disease activity in association with up-regulation of the tumor necrosis factor alpha/tumor necrosis factor receptor system in systemic lupus erythematosus
Arthritis Rheum.
 , 
2003
, vol. 
48
 (pg. 
2533
-
2540
)
36
Jonsen
A.
Bengtsson
A.A.
Sturfelt
G.
Truedsson
L.
Analysis of HLA DR, HLA DQ, C4A, FcgammaRIIa, FcgammaRIIIa, MBL, and IL-1Ra allelic variants in Caucasian systemic lupus erythematosus patients suggests an effect of the combined FcgammaRIIa R/R and IL-1Ra 2/2 genotypes on disease susceptibility
Arthritis Res. Ther.
 , 
2004
, vol. 
6
 (pg. 
R557
-
R562
)
37
Purcell
S.
Neale
B.
Todd-Brown
K.
Thomas
L.
Ferreira
M.A.
Bender
D.
Maller
J.
Sklar
P.
de Bakker
P.I.
Daly
M.J.
, et al.  . 
PLINK: A tool set for whole-genome association and population-based linkage analyses
Am. J. Hum. Genet.
 , 
2007
, vol. 
81
 (pg. 
559
-
575
)
38
Almasy
L.
Blangero
J.
Multipoint quantitative-trait linkage analysis in general pedigrees
Am. J. Hum. Genet.
 , 
1998
, vol. 
62
 (pg. 
1198
-
1211
)
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/2.0/uk/) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.