Abstract
Sarcoidosis is a complex granulomatous inflammatory disorder that shares several clinical and pathogenic features with inflammatory bowel disease (IBD). Postulating a common genetic basis of inflammatory diseases, we tested 106 single-nucleotide polymorphisms (SNPs) that are known or have been suggested to be associated with IBD for a potential association with sarcoidosis and its acute and chronic subphenotypes.
We genotyped 1,996 German sarcoidosis patients, comprising 648 acutely and 1,161 chronically affected individuals, 2,622 control subjects, and 342 German trios with affected offspring using SNPlex™ technology.
The nonsynonymous SNP rs11209026 (Arg381Gln) in the interleukin (IL)-23 receptor (IL23R) gene was associated with chronic sarcoidosis (OR 0.63; p = 5.58×10−5), which was supported by the result of a transmission disequilibrium test analysis in the independent family sample (OR 0.50; p = 0.031). Marker rs12035082 located at chromosome 1q24.3 was found to be associated with the acute subphenotype (OR 1.36; p = 6.80×10−7) and rs916977 (HERC2 locus; OR 1.30; p = 4.49×10−5) was associated with sarcoidosis.
Our results highlight the potential importance of the IL-23 signalling pathway for the development of chronic sarcoidosis. The finding links sarcoidosis pathogenesis to other inflammatory conditions and may contribute to new hypotheses on disease mechanisms.
Several immune-related susceptibility genes have been demonstrated as shared risk factors for chronic inflammatory disorders. These include PTPN22 and CTLA4 (various autoimmune diseases 1, 2), NOD2 (Crohn’s disease (CD) 3, 4 and atopic diseases 5, 6) and IL23R (CD 7, 8, psoriasis 9 and ankylosing spondylitis 10–12). In particular, CD and ulcerative colitis (UC), the two main phenotypes of inflammatory bowel disease (IBD), share a number of susceptibility loci (e.g. NKX2-3, CCNY, STAT3 and PTPN2) 13, 14. Presumably, this overlap of genetic risk loci, as well as familial clustering of different autoimmune diseases 15, indicates that shared biological pathways may be involved in the aetiology of clinically distinct immune-mediated diseases (reviewed in 16–18).
Sarcoidosis (Online Mendelian Inheritance in Man (OMIM) #181000) is an inflammatory disorder that can be further classified into acute or chronic sarcoidosis according to the course of the disease. It shares several clinical and immunological features with other chronic inflammatory diseases, such as general inflammatory characteristics, or more specifically, the formation of noncaseating granuloma in CD. Furthermore, sarcoidosis is a complex polygenic disorder, to which variants in numerous genes and yet unknown environmental factors are likely to contribute 19. The genetic underpinning in sarcoidosis is supported by the discovery of two susceptibility genes, namely BTNL2 20, 21 and ANXA11 22. Besides these disease loci, numerous case–control association studies of potential candidate regions have been published for sarcoidosis, but many of them with conflicting results (for review, see 23). This includes the chemokine receptor genes CCR2 and CCR5 on chromosome 3p21.3, IL18, TNFA and several HLA loci. Most of these genes seem to play key roles in the pathogenesis of other immune disorders, such as IL18 or TNFA in CD (for review, see 24). Furthermore, we recently identified a novel shared susceptibility locus on chromosome (chr)10p12.2 for both sarcoidosis and CD by combining genome-wide association datasets from the two diseases 25. This underlines the aforementioned hypothesis that a number of genetic regions exist that may act as general modifiers of the inflammatory response.
Here, we aimed to identify the putative effects of genetic variants in 74 IBD risk loci on the susceptibility to sarcoidosis, and to its acute and chronic subphenotypes. With this approach, we were able to identify HERC2 as a novel susceptibility gene for sarcoidosis, and chr12q24.3 and IL23R as subphenotype-specific risk loci for the acute and chronic subphenotype.
MATERIALS AND METHODS
Patients and control subjects
A total of 1,996 unrelated German sarcoidosis patients and 2,622 unrelated German controls were included in the study. All patients were recruited through specialised hospitals and practitioners, the German sarcoidosis patients’ organisation (Deutsche Sarkoidose-Vereinigung e.V.; www.sarkoidose.de), and via health insurance institutions. In the latter case, histological confirmation of the diagnosis was necessary for inclusion of individuals. All patients completed a questionnaire on the course of disease. Patients’ physicians were contacted to provide clinical radiology and laboratory data required to confirm the diagnosis of sarcoidosis. According to the clinical presentation of the disease, patients were classified as having either chronic or acute sarcoidosis, if possible. The acute sarcoidosis group comprised patients with sudden complaints and recovery within 2 yrs, including patients with Löfgren’s syndrome. Individuals with the chronic phenotype of sarcoidosis exhibited subtly intensifying early symptoms, followed by enduring disease activity for ≥2 yrs. In order to reduce phenotypic heterogeneity, sarcoidosis patients were separated into 648 individuals with acute sarcoidosis, including 105 patients with Löfgren’s syndrome, and 1,161 persons with the chronic form of the disease, 734 of whom suffered from enduring disease despite cortisone therapy and 427 of whom experienced recovery with or without treatment. As the remaining 187 patients could not be classified unequivocally as to the course of the disease, these individuals were excluded from the subphenotype-specific analysis. German control individuals were recruited from the PopGen biobank 26. Written informed consent was obtained from all participants. All collection protocols were approved by institutional ethics review boards and data protection authorities of the participating institutes.
The family sample comprised 342 German trios with offspring affected by sarcoidosis. Of these, 121 trios comprised offspring with the acute form of the disease and offspring of 255 trios were affected by chronic sarcoidosis. Recruitment and diagnosis was accomplished as described above for the case–control sample.
Single-nucleotide polymorphism selection, genotyping and quality control
All 106 single-nucleotide polymorphisms (SNPs) included in this study were selected from prior publications for their confirmed or suggested association with CD and/or UC. The majority of these markers (58 SNPs) were originally published by the Wellcome Trust Case–Control Consortium 27 or were identified through the meta-analysis performed by Barrett et al. 14 (see Table S1 in the online supplementary material for details). Ligation-based SNPlex™ genotyping 28 was performed as previously described 29. Genotype assignments were manually confirmed by visual inspection with the Genemapper 4.0 (Applied Biosystems, Carlsbad, CA, USA) software. All process data were written to and administered by an in-house database-driven laboratory information management system described previously 30. All samples and SNPs included in the analysis met the following quality criteria: all samples had a call rate ≥0.9; and markers had a minor allele frequency ≥0.02, a call rate ≥0.95 and showed no deviation from Hardy–Weinberg equilibrium in the control sample (p>0.01 in the exact test implemented in PLINK v1.06 31). For further quality assessment, at total of 129,840 genotypes from 55 randomly chosen SNPs was technically replicated showing a concordance rate of 99.8% between the two batches.
Statistical analysis
Statistical data analysis was carried out using PLINK v1.06 32. Allele frequency differences were assessed using a Chi-squared test (one degree of freedom (df)). Nominal p-values were adjusted for multiple testing using Bonferroni correction with a factor of 318, accounting for all 106 markers under study and the three investigated phenotypes. Power calculation for the transmission disequilibrium test (TDT) analysis was conducted using the Genetic Power Calculator 33 and for case–control analysis using the PS power and sample size program 34.
Analysis of specific expression of IL23R by RT-PCR
For investigation of tissue- and cell type-specific expression patterns of IL23R, we used standard RT-PCR procedures on a commercially available human tissue and immune panel from Clontech (Palo Alto, CA, USA). Bronchoalveolar lavage (BAL) cell samples were taken from five patients with sarcoidosis and without steroid treatment at the time of BAL and five unaffected individuals. The diagnosis of sarcoidosis was established retrospectively in accordance with previously defined criteria 35, including noncaseating granuloma identified by transbronchial biopsies. All patients and unaffected individuals gave their informed consent to participate in the study. BAL was performed as previously described 36. The BAL cell suspensions used in this study contained >95% alveolar macrophages. Briefly, the PCR program used was applied as follows: denaturation for 5 min at 95°C; 33 cycles of 20 s at 95°C, 20 s at 53°C and 60 s at 72°C; and a final extension for 5 min at 72°C. The following primers were used for amplification: hIL23R-F1 (GACACATGGAATTCTGGGCT) and hIL23R-R1 (CAAAAGCATGGTGGTTTCCT). To confirm the use of equal amounts of RNA in each experiment, all samples were checked in parallel for glyceraldehyde 3-phosphate dehydrogenase (GAPDH) expression using the following primers: GAPDH-F (CCAGCCGAGCCACATCGC) and GAPDH-R (ATGAGCCCCAGCCTTCTCCAT). All amplified DNA fragments were analysed on 1% agarose gels, and subsequently documented and quantified using a ChemiDoc XRS system (BioRad, Munich, Germany).
RESULTS
Case–control association analysis
A total of 106 SNPs from 74 different loci passed the quality control criteria (see Methods section). In the association analysis, 24 SNPs from 21 different loci reached nominal significance with nominal p-value (pnom) of <0.05 in the allelic Chi-squared test (one df) in the overall sample of 1,996 sarcoidosis cases and 2,622 controls. In the subphenotype-specific analysis, eight of these SNPs were also nominally associated (pnom<0.05) with acute sarcoidosis (648 cases) and 16 with the chronic form of the disease (1,161 cases). Three distinct SNPs were nominally significant solely for acute sarcoidosis and further four SNPs for chronic sarcoidosis only. Detailed results for all SNPs under study, including genotype counts, are shown in online supplementary Table S2.
Since every SNP was tested in the overall, acute and chronic samples, nominal p-values were Bonferroni-corrected for 3×106 = 318 independent tests. Three association signals, namely from rs916977, rs11209026 and rs12035082, remained significant after correction for multiple testing (table 1). SNP rs916977, which is located in intron 12 of the HECT domain and RLD 2 (HERC2) gene on chr15q13.1, yielded a corrected p-value (pcorr) of 0.014 in the overall sarcoidosis sample (OR 1.30, 95% CI 1.15–1.48), with both subphenotypes contributing to the signal to a similar extent (ORacute 1.34, 95% CI 1.12–1.61; ORchronic 1.29, 95% CI 1.11–1.50). These odd ratios were in the same range as for CD (OR 1.26, 95% CI 1.07–1.47), but reduced compared to UC (OR 1.46, 95% CI 1.23–1.75), as reported previously for a German sample 37. Furthermore, two subphenotype-specific associations were observed: the nonsynonymous SNP rs11209026 in the interleukin (IL)-23 receptor (IL23R) gene, which is a well established risk factor for CD 7, 14, 37, 38, was solely associated with chronic sarcoidosis (ORchronic 0.63, 95% CI 0.50–0.79; pcorr,chronic = 0.018). The effect of this SNP only slightly differed between patients with enduring disease despite cortisone treatment and patients that experienced recovery (OR 0.66 (95% CI 0.50–0.68) and OR 0.57 (95% CI 0.40–0.82), respectively). As the IL-23 receptor (IL-23R) directly interacts with Janus kinase 2 (JAK2) at the molecular level 39, it is interesting to note that the two SNPs, rs10758669 and rs10974944, located in the JAK2 gene, are associated with chronic sarcoidosis with nominal significance (ORchronic 1.14 (95% CI 1.03–1.26; pchronic = 0.013) and ORchronic 1.16 (95% CI 1.04–1.30; pchronic = 0.0061), respectively). However, no epistatic effects were detected between IL23R and JAK2 SNPs (data not shown). Finally, an exclusive association with acute sarcoidosis was observed for SNP rs12035082 (ORacute 1.36, 95% CI 1.21–1.54; pcorr,acute = 2.34×10−4), which is located downstream of the tumour necrosis factor (ligand) superfamily member 18 gene (TNFSF18 or GITRL) on chr1q24.3 and was reported to modestly increase the risk for CD (OR 1.20, 95% CI 1.05–1.37) 37. In the Löfgren’s disease subsample, this SNP was not significantly associated, probably due to the small number of patients (n = 105; power = 0.34).
Additional evaluation in a family-based design
In order to substantiate these novel associations, we tested the three markers in an independent family sample comprising 342 German trios with affected offspring, including 121 “acute” trios and 255 “chronic” trios, using the TDT (table 2). SNP rs916977 located in the HERC2 gene, which was found to be associated with sarcoidosis in the population-based design (OR 1.30), yielded a borderline, but not significant, result (OR 0.73, 95% CI 0.52–1.03; p = 0.069; 29% power). Marker rs11209026, located in the IL23R gene, showed a nominal significant association with the chronic subphenotype (OR 0.50, 95% CI 0.26–0.95; pchronic = 0.031) and a similar effect size as in the case–control analysis (OR 0.63). Finally, SNP rs12035082 was found to have yielded a slightly higher odds ratio (OR 1.50, 95% CI 0.96–2.35) in the acute trios than in the population-based design (OR 1.36), but also only a borderline, nonsignificant p-value (pacute = 0.074; 27% power).
Expression analysis of IL23R
To gain first insights on a possible effect of IL23R variants on sarcoidosis, the transcript levels of IL23R were assessed by RT-PCR in a panel of different healthy human tissues and resting and activated immune cells (fig. 1). The expression analysis in this panel showed elevated IL23R expression in testis and activated CD4+ T-cells and moderate expression in healthy lung tissue. Next, we analysed the expression levels of IL23R in cells derived from BAL cells using quantitative real-time PCR, and cDNA from sarcoidosis patients and unaffected controls (n = 5 per group). Statistical analysis revealed no significant differences in relative expression levels between patients and controls (data not shown).
DISCUSSION
Under the hypothesis of a common genetic basis of inflammatory diseases, we selected SNPs that are known to be associated with CD or UC as likely candidates for also playing a role in sarcoidosis aetiology. We identified three significant associations with sarcoidosis, with two being specific to the acute and chronic subphenotypes.
The overall sarcoidosis phenotype was associated with marker rs916977, which is a nonsynonymous SNP in the HERC2 gene. Our study sample originated from Germany and, therefore, we do not expect a significant population structure for this sample 40. However, allele frequencies of markers located at the HERC2 locus have recently been shown to exhibit a north–south gradient across Europe 41, possibly even within Germany. HERC2 was previously reported to be associated with CD and UC in German and Dutch–Belgian samples 37, 38. However, neither the association with CD and UC nor with sarcoidosis was investigated in a well-powered, family-based study and, to date, there are no reports of a possible functional role of HERC2 in the inflammatory process. Therefore, the association signal for HERC2 has to be considered with caution and requires further investigation.
The association of IL23R variants with IBD and with a number of chronic inflammatory diseases is well established 7, 9–13, 42–46. Our study is the first report on an association of an IL23R variant with chronic sarcoidosis (rs11209026; OR 0.63, 95% CI 0.50–0.79) and provides further support for this signal by the nominal significant association in an independent family sample (OR 0.50, 95% CI 0.26–0.95). The associated SNP rs11209026 (Arg381Gln) causes an amino acid exchange in the intracellular domain of IL-23R and may thus affect downstream signalling of the receptor. Preliminary expression studies showed elevated IL23R expression in activated CD4+ T-cells and low expression in monocytes, which is consistent with previous findings 39. In addition, we found moderate expression in healthy lung tissue. Differential IL23R expression has been reported for CD4+ T-cells from UC and CD patients 47. However, the lack of differential expression of IL23R in BAL cells from sarcoidosis patients compared to healthy individuals (data not shown) does not contradict our genetic finding, mainly because the causative SNP(s) may influence IL-23R function rather than mRNA expression.
Signal transduction through IL-23R and the downstream elements of the IL-23 pathway is well characterised. IL-23 signalling is crucial for the induction of T-helper (Th)17 cells 48, and there is increasing evidence that this is a key pathway in the development of chronic inflammation in the course of various skin, intestinal and lung inflammatory diseases (reviewed in 49–52). However, to date, the particular role of Th17 cells in the pathogenesis of sarcoidosis has not been investigated 53. The tyrosine kinase JAK2 is directly involved in signal transduction from the functional receptor complex consisting of IL-23R and IL-12 receptor β1. The nominally significant genetic association of chronic sarcoidosis with rs10974944, which is located in the JAK2 gene (OR 1.17, 95% CI 1.05–1.30) could, therefore, provide an additional hint to the potential importance of IL-23 signaling for the development of this phenotype, although no formal genetic SNP–SNP interaction was detected. Future experiments will clarify the role of associated SNP rs11209026 and, possibly, other causative variants in the development of chronic sarcoidosis, and how the genetic variation influences IL-23R function regarding sarcoidosis pathogenesis.
SNP rs12035082 located on chromosome 1q24.3 was significantly associated with acute sarcoidosis in the population-based design. Replication in the independent German family sample most likely failed due to low power (27%), while the odds ratio in the TDT analysis points to a true finding. The association signal cannot easily be attributed to a specific gene with potential functional relevance, since the SNP is located approximately 112 Mb from the next gene. No conclusions could be drawn from in-house genome-wide association data 22, whether the signal extends further to the TNFSF18 (GITRL) or even further to the TNFSF4 (OX40L) gene (data not shown). Both have been shown to regulate T-cell responses 54, 55 (reviewed in 56, 57); they could both be plausible candidates for genes with functional relevance in sarcoidosis aetiology. In order to assess an association of this region and the suggested candidate genes, validation in an independent population and fine-mapping experiments are necessary.
In general, only limited conclusions can be drawn from negative findings in this variant-centered, candidate gene approach, where we tested previously reported IBD risk variants. Disregarding the power issue, there are two major reasons why a negative finding does not exclude the specific locus as a sarcoidosis risk gene. First, for distinct diseases, “private” causative variants may exist at a common risk locus, which results in the observed overlapping pathology (e.g. granuloma formation). Secondly, it must be assumed that the majority of the markers investigated here do not represent causative SNPs, but are merely in allelic association with them. Therefore, different linkage disequilibrium structures between study populations can lead to false negative results. For example, SNP rs2066844, located in the CD risk locus NOD2, was very recently reported to be associated with a severe course of sarcoidosis 58. Two CD risk variants (rs10521209 and rs2076756) in the NOD2 gene region were included in our study, but were not associated with either sarcoidosis or the clinical subphenotypes in this study.
In summary, this study identified the association of IL23R variant Arg381Gln with chronic sarcoidosis and provides support for this novel finding by TDT analysis in an independent family sample. We further report on an association of rs916977 (HERC2) with overall sarcoidosis and of rs12035082 (chr1q24.3 locus) with the acute subphenotype. This genetic result may point to possible aetiological similarities between sarcoidosis and other chronic inflammatory diseases, such as IBD, psoriasis and ankylosing spondylitis 7, 10, 18, 43. Our study adds further evidence to the postulated common genetic basis of inflammatory disorders. The finding may lead to new hypotheses on disease pathogenesis and encourages further functional investigation of potential causative IL23R variants and their effect on the IL23 signaling pathway in the context of sarcoidosis.
Acknowledgments
The authors wish to thank all patients, families and physicians involved in the study for their cooperation. The support of the German Sarcoidosis patients association (Deutsche Sarkoidose-Vereinigung e.V.), the PopGen biobank and of the contributing pulmonologists is gratefully acknowledged. Finally, we want to acknowledge M. Davis, S. Ehlers, C. Fürstenau, T. Henke, I. Urbach, T. Wesse, R. Vogler and Michael Wittig at the Institute for Clinical Molecular Biology (Christian-Albrechts University, Kiel, Germany) for expert technical help.
Footnotes
This article has supplementary material available from www.erj.ersjournals.com
For editorial comments see page 489.
Statement of Interest
None declared.
- Received March 28, 2010.
- Accepted June 28, 2010.
- ©ERS 2011