Combinatorial Control of Th17 and Th1 Cell Functions by Genetic Variations in Genes Associated With the Interleukin-23 Signaling Pathway in Spondyloarthritis

Authors


Address correspondence to Lars Rogge, PhD, Institut Pasteur, Immunoregulation Unit and CNRS URA 1961, Department of Immunology, 25 Rue du Dr. Roux, 75724 Paris, France. E-mail: lars.rogge@pasteur.fr.

Abstract

Objective

Recent genome-wide association studies have revealed numerous genetic associations between specific single-nucleotide polymorphisms (SNPs) and immune-mediated inflammatory diseases. The current challenge is to identify associations of the genetic variants with effector mechanisms implicated in pathogenesis. This study was undertaken to investigate the link between genetic variation at loci associated with spondyloarthritis (SpA) and the effector function of CD4+ T lymphocyte subsets involved in chronic inflammatory disease.

Methods

Expression of Th17 and Th1 cytokines and transcription factors was measured in CD4+ T cells isolated from patients with SpA. Correlation analyses were performed to assess potential associations of these expression levels with the patient's genotype at loci genetically linked to SpA.

Results

The effector functions of Th17 and Th1 cells in patients with SpA were found to be under combinatorial control by multiple SNPs at genes associated with the interleukin-23 (IL-23)/Th17 pathway. Patients with SpA carrying risk-associated alleles of genes in the IL-23/Th17 pathway expressed the highest levels of genes involved in the differentiation and function of Th17 and Th1 cells, whereas the presence of protective alleles was associated with low-level expression of these genes. In contrast, variation at loci that were genetically linked to SpA, but not associated with the IL-23 pathway, did not affect the expression of Th17- and Th1-specific genes, suggesting that these SNPs may contribute to the pathogenesis of SpA through distinct cellular mechanisms.

Conclusion

These results show that genetic variations at genes associated with the IL-23 signaling pathway may influence the effector functions of Th17 and Th1 cells in patients with SpA. These findings provide a framework to delineate the mechanisms by which genetic variants contribute to pathology.

Spondyloarthritis (SpA) represents a family of seronegative spondyloarthritides. The various clinical forms include axial features, peripheral arthritis, enthesitis, and extraarticular features such as uveitis, psoriasis, and inflammatory bowel disease (IBD). The prototypic form of these disabling diseases is ankylosing spondylitis (AS), a highly heritable arthropathy with 80–90% of susceptibility attributable to genetic factors. The main genetic risk factor is the class I major histocompatibility complex (MHC) molecule HLA–B27, carried by 80–90% of patients. However, the presence of HLA–B27 explains only 20–40% of the genetic risk of developing AS, thus suggesting that additional genes may have an important role in the pathogenesis of AS ([1]).

To better understand the pathophysiology of SpA, several genome-wide association studies (GWAS) have been performed. These studies have uncovered genetic linkage of non-HLA genes to AS, in particular key components of the interleukin-23 (IL-23) signaling pathway, such as IL23R and IL12B, as well as genes associated with other pathways, such as ERAP1 and ANTXR2 (2–5). Nonetheless, how genetic variation at these loci affects immune function and contributes to pathology remains unknown.

For almost 2 decades, interferon-γ (IFNγ)–secreting Th1 cells have been implicated in the pathogenesis of chronic inflammatory and autoimmune diseases. Th1 cell development from naive CD4+ precursor T cells is initiated by antigenic stimulation and strongly enhanced by the heterodimeric cytokine IL-12, which activates the transcription factor STAT-4 in developing Th1 cells. STAT-4 induces expression of genes involved in the differentiation and effector functions of Th1 cells, such as TBX21, which encodes the Th1 lineage transcription factor T-bet. In addition, STAT-4 induces expression of IL12RB2 (the gene encoding the signaling subunit of the IL-12 receptor [IL-12R]) and IFNG (the gene encoding the Th1 signature cytokine IFNγ) ([6-8]).

The key role of Th1 cells in autoimmunity has been challenged by the identification of IL-23, a member of the IL-12 family of heterodimeric cytokines, and of IL-17–secreting Th17 cells as key players in mouse models of autoimmunity ([9]). IL-12 and IL-23 share a common subunit, p40, encoded by the IL12B gene, which pairs with IL-23p19 to form bioactive IL-23, and pairs with IL-12p35 to form IL-12 ([10]). Mice with a deletion of the IL-23p19 subunit, but not mice with a deletion of the IL-12p35 subunit, are protected from disease development in several experimental models of autoimmunity, such as experimental autoimmune encephalomyelitis ([11, 12]), collagen-induced arthritis ([13]), and IBD ([14, 15]). It was shown that IL-23, but not IL-12, induced the secretion of IL-17 from activated memory T cells (16), and subsequent studies revealed that IL-23 is essential for the maintenance and function of a distinct subset of effector CD4+ T cells, termed Th17 cells, which are characterized by the secretion of the proinflammatory cytokines IL-17A and IL-17F ([12, 17]). Taken together, these studies provide strong evidence to indicate that IL-23 and Th17 cells play a critical role in chronic inflammation and autoimmunity in mouse models.

A role for IL-17 and Th17 cells in human chronic inflammatory disease has been supported by studies showing increased frequencies of IL-17–producing cells in the peripheral blood of patients with SpA, as compared to the peripheral blood of healthy donors ([18-22]). Th17 cells, however, may not be the only T cell subset to promote inflammation. Several studies using experimental models of autoimmune disease have demonstrated that both Th17 and Th1 cells can induce disease development ([23, 24]), suggesting that CD4+ T cell populations with distinct functional properties may contribute to the pathogenesis of chronic inflammatory diseases.

In the present study, we investigated whether genetic polymorphisms in genes of the IL-23/Th17 pathway that have been linked to SpA have an impact on the effector functions of CD4+ T cells in patients with SpA. We demonstrate that the functions of both Th17 and Th1 cells in patients with SpA are under combinatorial control by multiple single-nucleotide polymorphisms (SNPs) at genes associated with the IL-23 signaling pathway, such as IL23R, IL12B, and CCR6. Our results provide a framework to delineate the pathways through which genetic variants contribute to pathology.

PATIENTS AND METHODS

Patients

Peripheral blood samples were obtained from patients with SpA at the Department of Rheumatology of Cochin Hospital (Paris, France). All patients met either the modified New York criteria for AS ([25]) or the criteria for SpA described by Amor et al ([26]). Age, sex, disease duration, HLA–B27 positivity, and the type of clinical presentation (axial involvement, peripheral arthritis, enthesitis, and non-rheumatologic symptoms) were recorded. Current treatment was also noted at the time of sampling (Table 1). The patients were all adults and had never received biologic antirheumatic agents. The mean age of the patients was 38 years (range 19– 81 years). Among the 49 patients, 31 had AS, 16 had psoriatic arthritis, 2 had IBD-related arthritis, and 1 had reactive arthritis associated with urethritis. Eighteen patients were female and 79% were HLA–B27 positive. To confirm the findings, a replication cohort of 24 patients with SpA, whose demographic and clinical characteristics were similar to those of the 49 patients in the primary cohort (details in Table S1, available at http://www.pasteur.fr/ip/easysite/pasteur/en/research/scientific-departments/immunology/units-and-groups/immunoregulation), was recruited.

Table 1. Demographic and clinical characteristics of the primary cohort patients recruited to the study
Patient/age/sexHLA–B27 positive or negativeType of injuryTreatmenta
AxialPeripheralEnthesitisNon-rheumatologic symptoms
  1. aNSAID = nonsteroidal antiinflammatory drug; MTX = methotrexate.
1/26/F+YesNoNoNSAID
2/47/F+YesYesYesNSAID
3/36/MYesNoNoNSAID
4/30/M+YesNoNoSulfasalazine
5/21/M+YesYesNoPsoriasis
6/26/M+YesNoYesCrohn's diseaseNSAID/MTX
7/40/M+YesYesYesNSAID/MTX
8/51/FYesNoYesNSAID
9/45/M+YesNoNoNSAID
10/46/M+YesYesNoNSAID
11/46/M+YesNoYesPsoriasisNSAID
12/41/F+YesYesYesNSAID
13/28/M+YesNoNoNSAID
14/33/MYesNoYesNSAID
15/21/M+YesNoYes
16/27/F+YesNoYesNSAID
17/30/M+YesYesNoUrethritisNSAID/MTX
18/48/F+YesNoNoCrohn's disease/psoriasisNSAID
19/28/M+YesNoNoNSAID
20/38/MYesNoYesPsoriasisNSAID
21/36/FYesYesNoPsoriasisNSAID
22/42/M+YesYesNoNSAID
23/32/F+YesNoNoNSAID
24/32/M+YesYesYesPsoriasisNSAID
25/39/MYesNoYesNSAID
26/24/F+YesNoYesNSAID
27/28/M+YesYesYesPsoriasisNSAID
28/19/M+YesNoNoNSAID
29/30/M+YesNoNoNSAID
30/44/M+YesYesNoPsoriasisMTX
31/32/F+YesNoNoNSAID
32/34/MYesNoYesNSAID
33/30/F+YesNoYesPsoriasisNSAID
34/35/MYesNoYesNSAID/sulfasalazine
35/34/F+YesNoNo
36/31/F+YesNoNo
37/28/M+YesNoNoPsoriasisNSAID
38/45/FYesNoYesPsoriasisMTX
39/60/M+YesNoYesUveitis
40/40/M+YesNoNoUveitisNSAID
41/41/M+NoYesYesPsoriasisNSAID
42/81/F+YesYesNoUveitisColchicine
43/44/F+NoYesYesNSAID
44/37/F+YesNoYesUveitisNSAID
45/63/M+YesNoYesPsoriasisNSAID
46/56/M+YesYesYesPsoriasis/uveitis
47/63/MYesYesYesPsoriasisNSAID/MTX
48/50/F+YesNoYesUveitisNSAID
49/34/M+YesNoNoPsoriasis

The study was approved by the ethics review committee, the Comité de Protection des Personnes Ile de France III. Written informed consent was obtained from all patients prior to inclusion in the study.

Cell purification and stimulation

Peripheral blood mononuclear cells were isolated by density-gradient centrifugation (Lymphoprep; Axis-Shield). Cells were stained with anti-CD3–Pacific Blue antibodies (BD Biosciences) and anti-CD4–allophycocyanin antibodies (Miltenyi Biotec), and sorted using a FACSAria II fluorescence-activated cell sorter (BD Biosciences). Sorted CD3+CD4+ T cells were cultured at a concentration of 106 cells/ml in RPMI 1640 medium (Gibco), completed with 10% fetal calf serum (Hyclone) and 1% penicillin/1% streptomycin (Gibco). Cells were stimulated with Dynabeads coated with anti-CD3 and anti-CD28 antibodies (1 bead/cell; Invitrogen) in the presence or absence of IL-12 (10 ng/ml; Roche) along with anti–IL-4 antibodies (1 μg/ml; BD Biosciences) or in the presence or absence of IL-23 (100 ng/ ml; PeproTech) along with anti–IL-4 (1 μg/ml) and anti-IFNγ antibodies (1 μg/ml) (BD Biosciences). Cells and supernatants were harvested after 48 hours of stimulation for analysis of messenger RNA expression and cytokine production, respectively.

Real-time quantitative polymerase chain reaction (PCR)

For the primary cohort (n = 49), total RNA was extracted with an RNeasy Mini kit (Qiagen). Complementary DNA (cDNA) was synthesized using a High-Capacity cDNA Archive kit (Applied Biosystems). RNA levels were assessed using custom-designed TaqMan Low-Density Arrays (Applied Biosystems) on an Applied Biosystems 7900 Real-Time PCR system, following the manufacturer's protocol. Reactions were run in triplicate, and RNA expression levels were normalized to those of 18S RNA.

For the replication cohort (n = 24), total RNA was extracted with an RNeasy Micro kit (Qiagen), and cDNA was synthesized using a High-Capacity cDNA Archive kit (Applied Biosystems). RNA levels were assessed using BioMark technology (Fluidigm), following the manufacturer's protocol (Advanced Development Protocol 17). Reactions were run in triplicate, and RNA expression levels were normalized to those of HPRT RNA.

Cytokine measurements

Cytokine production was assessed in culture supernatants using Procarta Immunoassays (Panomics), in accordance with the manufacturer's recommendations.

SNP genotyping

SNPs associated with SpA were genotyped using predeveloped TaqMan allelic discrimination assays (Applied Biosystems), following the manufacturer's protocol. Reactions were performed using an Applied Biosystems 7900HT Fast Real-Time PCR System with SDS version 2.3 software (Applied Biosystems) for allele discrimination or using a BioMark PCR System (Fluidigm) following the manufacturer's recommendations. (Complete lists of all of the assays used are given in Tables S2–S4, available at http://www. pasteur.fr/ip/easysite/pasteur/en/research/scientific-departments/immunology/units-and-groups/immunoregulation.)

Linkage disequilibrium map

The linkage disequilibrium map (r2 values) was generated using Haploview version 4.2 (see http://www.broad.mit.edu/mpg/haploview/). Scores of linkage disequilibrium were obtained from the International HapMap Project.

Statistical analysis

Mann-Whitney 2-tailed U tests were used for comparisons of gene or protein expression levels under the different stimulation conditions between CD4+ T cells carrying a minor allele and those not carrying a minor allele. Wilcoxon's 2-tailed matched pairs tests were used to analyze the induction of Th17 and Th1 marker genes by IL-23 or IL-12. Data were evaluated using GraphPad Prism software.

Univariate analysis was performed by comparing the expression level of each gene according to the presence or absence of the minor allele for each SNP. If the mean differences were significant at a P value less than 0.20, the SNP was selected for the multivariate analysis. The multivariate linear regression analysis was performed using a stepwise mode for selection of variables, with results analyzed using SAS software (version 9.3).

Genetic risk calculations were performed by following a procedure previously described (see http://www.decodeme. com/health-watch-information/risk-calculation). Briefly, we first calculated the relative risk for each allele, taking into account the odds ratio and frequency of minor and major alleles. The cumulative risk for the SNPs was then computed using a multiplicative model, assuming that the factors behave independently, as is the case for SNPs that are not in linkage disequilibrium or in only weak linkage disequilibrium. We then ranked the patients according to the cumulative risk for the 5 SNPs. The median value was used to group the patients according to whether they had a high cumulative risk or a low cumulative risk for each SNP.

RESULTS

Effect of genetic variation at the IL23R locus on expression of genes associated with development and function of Th17 and Th1 cells

To assess whether genetic variants at IL23R and other genes in the IL-23/IL-17 pathway affect CD4+ T cell functions, we isolated CD4+ T cells from the peripheral blood of 49 patients with SpA (Table 1) and stimulated the cells in the presence or absence of IL-12 or IL-23. We then analyzed the expression of genes associated with the development and function of Th17 and Th1 cells. Patients were genotyped at loci that are genetically linked to SpA ([2-5]) (listed in Table S4, available at http://www.pasteur. fr/ip/easysite/pasteur/en/research/scientific-departments/immunology/units-and-groups/immunoregulation).

We grouped patients according to the presence of either the major (G) allele or minor (A) allele at rs11209026, a SNP at the IL23R locus. The minor allele of rs11209026, which causes an R381Q amino acid exchange in the cytoplasmic portion of IL-23R, is less frequent in patients with SpA than in healthy controls. We found that SpA patients carrying the protective minor allele of rs11209026 (A) expressed substantially lower levels of IL17A transcripts, as compared to patients carrying the common allele (G) (P = 0.0361 for samples stimulated in the absence of cytokines, and P = 0.0239 for samples stimulated in the presence of IL-23) (Figure 1). Similarly, lower expression levels of transcripts for IL17F and the Th17-associated retinoic acid receptor–related orphan nuclear receptor γt (RORγt) transcription factor RORC were observed in patients with the minor allele (Figure 1).

Figure 1.

Patients with spondyloarthritis (SpA) carrying the protective IL23R rs11209026 allele express lower levels of genes associated with the differentiation and function of Th17 and Th1 cells. CD3+CD4+ T cells were sorted from the peripheral blood of SpA patients and stimulated for 48 hours with anti-CD3/anti-CD28 beads, without the addition of cytokines (−) or in the presence of interleukin-12 (IL-12) along with anti–IL-4 or IL-23 along with anti–IL-4 and anti–interferon-γ. The expression level of Th17 (upper panels) and Th1 (lower panels) marker genes was determined by real-time polymerase chain reaction in patients carrying the minor protective IL23R R381Q variant of rs11209026 (GA) (n = 8) or patients homozygous for the major allele (GG) (n = 41). Data were normalized to 18S RNA and presented as the fold change in mRNA levels relative to anti-CD3/anti-CD28–stimulated CD4+ T cells from a reference sample. Horizontal bar shows the mean.

The presence of the R381Q variant also strongly affected the function of Th1 cells, as indicated by substantial reductions in the transcript levels of IFNG, TNFA, IL12RB2, and TBX21 in CD4+ T cells from patients carrying the minor allele, both in cells stimulated in the absence of cytokines and in cells stimulated in the presence of IL-12 or IL-23 (Figure 1). Consistently, lower levels of inflammatory cytokines (IFNγ, IL-17F, and tumor necrosis factor α [TNFα]) were detected in the supernatants of CD4+ T cells from patients who were IL23R R381Q variant carriers. In contrast, levels of the homeostatic cytokine IL-2 were not altered in CD4+ T cells from patients carrying the protective IL23R R381Q variant (results in Figure S1, available at http://www.pasteur.fr/ip/easysite/pasteur/en/research/scientific-departments/immunology/units-and-groups/immunoregulation).

Reduced expression of Th17 and Th1 marker genes was not restricted to patients carrying the IL23R R381Q variant, but could also be observed in patients grouped according to either the presence or absence of the protective minor (A) allele of rs1343151, which is another SNP in the IL23R locus and is in only weak linkage disequilibrium with rs11209026 (r2 = 0.074) (results in Figure S2, available at http://www.pasteur. fr/ip/easysite/pasteur/en/research/scientific-departments/immunology/units-and-groups/immunoregulation). In particular, patients carrying the protective allele of rs1343151 had significantly lower transcript levels of IL17A, IL17F, and RORC, as well as lower levels of TNFA, IL12RB2, and TBX21.

We extended our analysis to 4 additional SNPs at the IL23R locus that have been associated with SpA ([2]). Patients carrying a protective minor allele at rs10489629 or rs11465804 showed a similar decrease in the expression of Th17 and Th1 marker genes. Moreover, an inverse pattern of gene and protein expression was consistently observed in the presence of the minor alleles at rs1004819 and rs11209032, each of which has been found to be associated with an increased risk of SpA ([2]). (Detailed results are listed in Figures S1, S3, and S4, available at http://www.pasteur.fr/ip/easysite/pasteur/en/research/scientific-departments/immunology/units-and-groups/immunoregulation.)

Effect of genetic variation in genes associated with the IL-23/Th17 pathway on Th17 and Th1 cells responses

These findings show that genetic variation at the IL23R locus affects the function of both Th17 and Th1 cells. To determine whether this effect was restricted to IL23R or could be extended to additional molecules in the IL-23/Th17 pathway, we analyzed the impact on CD4+ T cell function of genetic variation at IL12B, the gene encoding the IL-12p40 subunit shared by IL-12 and IL-23. Findings from GWAS have revealed an association of IL12B with psoriasis, Crohn's disease, and SpA ([27-29]), providing further evidence to support the notion that there is genetic overlap of these 3 diseases. We found that patients carrying the protective allele (A) of rs10045431 at IL12B showed a trend toward lower levels of Th17 and Th1 marker gene expression (Figure 2), as well as decreased production of effector cytokines (results in Figure S1, available at http://www.pasteur.fr/ip/easysite/pasteur/en/research/scientific-departments/immunology/units-and-groups/immunoregulation).

Figure 2.

Genetic variation at the IL12B locus affects effector T cell functions. CD3+CD4+ T cells were sorted from the peripheral blood of patients with spondyloarthritis and stimulated for 48 hours with anti-CD3/anti-CD28 beads, without the addition of cytokines (−) or in the presence of interleukin-12 (IL-12) along with anti–IL-4 or IL-23 along with anti–IL-4 and anti–interferon-γ. The expression level of Th17 (upper panels) and Th1 (lower panels) marker genes in stimulated CD4+ T cells was determined as described in Figure 1. Patients carrying the protective rs10045431 variant (CA/AA) (n = 22) at IL12B expressed lower levels of Th17 and Th1 markers, as compared to patients homozygous for the major allele (CC) (n = 27). Data were normalized to 18S RNA and presented as the fold change in mRNA levels relative to anti-CD3/anti-CD28–stimulated CD4+ T cells from a reference sample. Horizontal bar shows the mean.

GWAS have also uncovered a link between Crohn's disease or SpA and genetic variants at CCR6 and STAT3 ([4, 30]). CCR6 is a chemokine receptor highly expressed by Th17 cells and a subset of Th1 cells ([31, 32]). We found that CD4+ T cells from SpA patients carrying the disease-associated allele (A) of rs3093024 at CCR6 expressed significantly higher levels of IL17A, IL17F, IFNG, and TNFA (Figure 3).

Figure 3.

Genetic variation at the CCR6 locus affects effector T cell functions. CD3+CD4+ T cells were sorted from the peripheral blood of patients with spondyloarthritis and stimulated for 48 hours with anti-CD3/anti-CD28 beads, without the addition of cytokines (−) or in the presence of interleukin-12 (IL-12) along with anti–IL-4 or IL-23 along with anti–IL-4 and anti–interferon-γ. The expression level of Th17 (upper panels) and Th1 (lower panels) marker genes in stimulated CD4+ T cells was determined as described in Figure 1. Patients carrying the risk-associated minor allele of CCR6 rs3093024 (GA/AA) (n = 30) expressed higher levels of Th17 and Th1 markers, as compared to patients homozygous for the major allele (GG) (n = 19). Data were normalized to 18S RNA and presented as the fold change in mRNA levels relative to anti-CD3/anti-CD28–stimulated CD4+ T cells from a reference sample. Horizontal bar shows the mean.

STAT-3 is a transcription factor activated by multiple cytokines, including those implicated in the differentiation and function of Th17 cells, such as IL-6, IL-21, and IL-23 ([33]). However, we did not observe a significant change in the expression of Th17 and Th1 marker genes in SpA patients carrying the protective minor allele of rs6503695 (C) or the minor allele of rs744166 (G), which are 2 known SNPs at STAT3 ([4]). (Detailed results are listed in Figures S4 and S5, available at http://www.pasteur.fr/ip/easysite/pasteur/en/ research/scientific-departments/immunology/units-and-groups/immunoregulation.)

We also analyzed the effect of genetic variation at STAT4, the gene encoding the transcription factor downstream of IL-12 signaling. STAT4 has been found to be associated with increased disease risk in rheumatoid arthritis and in systemic lupus erythematosus ([34]). In contrast to the lack of effect observed for the STAT3 variants, patients carrying the minor (T) allele of rs7574865 at STAT4 expressed substantially higher levels of TBX21 and IL12RB2 (results in Figure S5, available at http://www.pasteur.fr/ip/easysite/pasteur/en/research/scientific-departments/immunology/units-and-groups/immunoregulation). Of note, both the signaling subunit of the IL-12 receptor and the Th1-specific transcription factor T-bet are direct transcriptional targets of STAT4 ([6, 7]), suggesting that genetic variation at STAT4 directly affects Th1 cell differentiation. Taken together, these findings demonstrate that the effector functions of Th17 and Th1 cells are controlled by multiple SNPs at genes associated with the IL-23/Th17 pathway.

In addition to molecules associated with the IL-23 pathway, results of GWAS have revealed an association of AS with variants at ERAP1, ANTXR2, 1q32, and intergenic regions at chromosomes 2p15 and 21q22 ([2-5]). Non-synonymous variants of ERAP1, an endoplasmic reticulum aminopeptidase, are known to have reduced enzymatic activity ([5]). Moreover, the association of ERAP1 with AS was observed only in HLA–B27–positive individuals, indicating that processing of antigenic peptides for class I MHC presentation may contribute to the pathogenesis of AS. We therefore investigated whether genetic variation at these loci, which are not directly linked to the IL-23 signaling pathway, could affect the expression of Th17 or Th1 marker genes.

We did not detect differences in the levels of inflammatory cytokine gene expression or cytokine production in patients carrying a susceptibility minor allele of rs27434 at ERAP1. Similarly, we did not detect a link between CD4+ T cell function and the protective allele of rs4333130 at ANTXR2 or genetic variants at 1q32 (rs11584383), chromosome 2p15 (rs10865331), or chromosome 21p22 (rs2242944). (Detailed results are listed in Figures S1, S4, and S6, available at http://www.pasteur. fr/ip/easysite/pasteur/en/research/scientific-departments/immunology/units-and-groups/immunoregulation.) These findings suggest that SNPs not associated with the IL-23/IL-17 axis do not specifically affect Th17/Th1 cell function, but act on pathologic development of SpA through a different cellular pathway.

We performed multivariate analysis to determine the independent contribution of each SNP, and the hierarchy of the predictive capacity of each SNP, to the expression of Th1 and Th17 marker genes. This analysis revealed that SNP IL23R rs1343151 had the highest independent predictive value for the expression of IL17A, IL17F, RORC, and IL12RB2, and also had a high predictive value for the expression of TBX21. Furthermore, IL12B rs10045431 had a high independent predictive value for the expression of IL17A, IL17F, and TBX21 (Table 2).

Table 2. Multivariate analysis for prediction of the effects of multiple single-nucleotide polymorphisms (SNPs) on the expression of genes associated with the differentiation and function of Th17 and Th1 cellsa
Gene, SNPVariables included in the modelCoefficient estimate, mean ± SDP for association
  1. aMultivariate analysis was performed by linear regression using a stepwise mode of variable selection. SNPs were selected if the mean differences in gene expression were significant at P < 0.20 in the univariate analysis.
Prediction of IL17AIL23R rs11209026, IL23R rs1004819, IL23R rs10489629, IL23R rs1343151,  
IL23R rs1343151IL23R rs11209032, IL12B rs10045431, STAT4 rs7574865, CCR6 rs3093024,−12.43 ± 4.360.007
IL12B rs1004543121q22 rs2242944−5.21 ± 2.660.056
IL23R rs10489629 7.90 ± 4.310.074
Prediction of IL17FIL23R rs11209026, IL23R rs1004819, IL23R rs10489629, IL23R rs1343151,  
IL23R rs1343151IL23R rs11209032, IL12B rs10045431, CCR6 rs3093024−14.52 ± 6.800.038
IL12B rs10045431 −12.77 ± 6.800.067
Prediction of IL23RIL23R rs11209026, IL23R rs10489629, IL23R rs1343151, IL12B rs12131065,  
IL23R rs10489629STAT4 rs7574865, CCR6 rs3093024−20.46 ± 10.600.060
Prediction of RORCIL23R rs11209026, IL23R rs1004819, IL23R rs10489629, IL23R rs1343151,  
IL23R rs1343151STAT4 rs7574865−4.01 ± 1.390.006
Prediction of IFNG CCR6 rs3093024IL23R rs11209026, IL23R rs1343151, IL23R rs11209032, STAT4 rs7574865, CCR6 rs309302453.96 ± 33.790.117
Prediction of TNFA STAT4 rs7574865IL23R rs11209026, IL23R rs1004819, IL23R rs1343151, IL12B rs10045431, STAT4 rs7574865, CCR63093024, 21q22 rs2242944, 2p15 rs108653311.97 ± 1.220.113
Prediction of IL12RB2 IL23R rs1343151IL23R rs11209026, IL23R rs1004819, IL23R rs10489629, IL23R rs1343151, IL23R rs11209032, IL12B rs10045431, STAT4 rs7574865, CCR63093024−4.65 ± 2.140.034
Prediction of TBX21IL23R rs11209026, IL23R rs1004819, IL23R rs10489629, IL23R rs1343151,  
IL12B rs10045431IL12B rs10045431, STAT4 rs7574865, CCR6 rs3093024−3.67 ± 1.830.051
STAT4 rs7574865 3.44 ± 1.930.081
IL23R rs1343151 −3.28 ± 1.880.089

To confirm the observed effects of genetic variation at genes in the IL-23/Th17 pathway on Th1 and Th17 cell functions, we analyzed gene expression in CD4+ T cells isolated from the peripheral blood of 24 patients with SpA from an independent replication cohort. Consistent with the results obtained with the primary cohort (Figure 1 and Figure S3, available at http://www.pasteur.fr/ip/easysite/pasteur/en/research/scientific-departments/immunology/units-and-groups/immunoregulation), patients in the replication cohort carrying protective minor alleles of rs11209026 or rs1343151 at IL23R expressed substantially lower levels of the Th17 marker genes IL17A, IL17F, IL23R, and RORC as well as the Th1 marker genes IFNG, TNFA, IL12RB2, and TBX21, as compared to patients carrying the common alleles. Similarly, we found that patients in the replication cohort carrying the protective allele of rs10045431 (allele A) at IL12B showed a trend toward lower levels of Th17 and Th1 marker gene expression, whereas patients carrying the risk-associated allele of rs3093024 at CCR6 expressed significantly higher levels of IL17A, IL17F, and IL23R, as well as higher levels of IFNG and IL12RB2.

Furthermore, we noted that patients carrying the protective minor allele of rs6503695 (C) at STAT3 showed a modest reduction in the expression levels of several Th17 and Th1 markers, in particular, IL17A and TNFA. We also observed a modest effect of the risk-associated allele of STAT4 rs7574865 (T) on the expression of the STAT4 target genes IFNG and IL12RB2, consistent with the findings obtained with the primary cohort.

Finally, SNPs associated with SpA, but not associated with genes in the IL-23/Th17 axis, such as ERAP1 rs27734 or ANTXR2 rs4333130, had little effect on the expression of Th1 and Th17 marker genes. Taken together, the data obtained from the replication cohort support our conclusion that genetic variation at genes associated with the IL-23/Th17 pathway controls the effector functions of Th1 and Th17 cells. (Detailed results are listed in Figures S7, S8, S9, and S10, available at http://www.pasteur.fr/ip/easysite/pasteur/en/research/scientific-departments/immunology/units-and-groups/immunoregulation.)

Combinatorial control of the effector functions of Th17 and Th1 cells

Since our data demonstrated that genetic variation at several loci affects CD4+ T cell effector functions in patients with SpA, we investigated whether there could be a combined effect of multiple SNPs in the IL-23 signaling pathway on Th17/Th1 inflammatory functions. The first analysis showed that SpA patients carrying predominantly susceptibility alleles expressed overall higher levels of IL17A, IL17F, and IFNG. In contrast, patients carrying the highest number of protective alleles expressed the lowest levels of these cytokine genes (results in Figure S11, available at http://www.pasteur.fr/ip/easysite/pasteur/en/research/scientific-departments/immunology/units-and-groups/immunoregulation).

To obtain a statistical evaluation of this effect, we tested for possible correlations of the expression levels of inflammatory cytokines with the genetic risk associated with 5 alleles targeting the IL-23 pathway (IL23R rs11209026, rs1004819, and rs1343151, IL12B rs10045431, and CCR6 rs3093024), all of which displayed no or weak linkage disequilibrium. We estimated the cumulative genetic risk for these SNPs in each patient using a multiplicative model. We then ranked the patients according to the cumulative risk for these 5 SNPs. Finally, we grouped the patients according to high versus low cumulative risk, and analyzed whether the expression of Th1 or Th17 markers differed between the 2 groups.

We found that patients having a higher cumulative risk for the 5 SNPs of the IL-23 pathway expressed significantly higher levels of Th1 and Th17 marker genes compared to patients with a lower cumulative risk (Figure 4A). To determine whether this was restricted to SNPs at loci associated with the IL-23 signaling pathway, we performed the same analysis using 5 SNPs that are associated with SpA but do not target molecules in the IL-23 pathway. This analysis revealed no differences in Th1 or Th17 marker gene expression between the 2 groups (Figure 4B), suggesting that the genetic risk associated with SNPs in the IL-23 pathway, but not SNPs in other signaling pathways, may correlate with Th1 and Th17 effector functions.

Figure 4.

Combinatorial effect on CD4 effector T cell functions of multiple single-nucleotide polymorphisms (SNPs) at genes associated with the interleukin-23 (IL-23)/Th17 pathway. A, The cumulative genetic risk for 5 alleles targeting the IL-23 pathway (IL23R rs11209026, rs1004819, and rs1343151, IL12B rs10045431, and CCR6 rs3093024) was estimated in each patient with spondyloarthritis (SpA), using a multiplicative model. Patients were ranked according to the cumulative risk (CR) of each of these SNPs, and grouped according to high cumulative risk versus low cumulative risk, using the median value as the cutoff. B, The same analysis as in A was performed using 5 SNPs associated with SpA, but not in the IL-23 pathway (ERAP1 rs27434, ANTXR2 rs4333130, 2p15 rs10865331, 1q32 rs11584383, and 21q22 rs2242944), in patients with either high or low cumulative genetic risk for these SNPs. In A and B, the expression level of Th17 (upper panels) and Th1 (lower panels) marker genes was determined by real-time polymerase chain reaction. C and D, Patients carrying the IL23R R381Q variant have reduced frequencies of CD45RO+ (memory) and CCR6+CD4+ T cells. Peripheral blood mononuclear cells were stained with anti-CD4–Pacific Blue, anti-CD45RO–fluorescein isothiocyanate, and anti-CCR6–phycoerythrin. The frequencies of CD45RO+ and CCR6+ T cells within the CD4+ T cell population were determined in patients carrying the protective minor allele (AG) compared to those homozygous for the major allele of IL23R rs11209026 (GG) (D). Immunofluorescence staining shows the distribution of expression intensity for these cell subsets in CD4+ T cells from a representative patient with each genotype (C). P values were calculated using Mann-Whitney U test. Horizontal bar shows the mean. NS = not significant.

A possible explanation for our findings is that the genotype of a patient affects the frequency, within the CD4 compartment, of memory CD4+ T cells or of CCR6+ T cells that can produce effector cytokines. We found that patients carrying the IL23R R381Q variant had lower frequencies of CD45RO+ (memory) T cells and CCR6+CD4+ T cells compared to patients carrying the common allele (Figures 4C and D), although the frequency of activated (HLA–DR+) cells in the memory compartment or in the CCR6+ compartment was similar between the 2 groups (results not shown). These data indicate that genetic variation in the IL-23 pathway may have an impact on the homeostasis of CD4 effector T cell populations and may affect the frequency of inflammatory T cell subsets.

DISCUSSION

Our investigation of the link between genetic variation at loci associated with SpA and CD4+ T cell function shows that polymorphisms at loci in the IL-23/Th17 pathway affect the expression of genes involved in the differentiation and function of Th17 and Th1 cells in patients with SpA. In particular, the presence of the IL23R R381Q variant had a strong impact on the expression of genes implicated in the differentiation and function of both effector subsets. The effect of this variant on CD4+ T cell function is still incompletely understood. One study, conducted in healthy controls, showed that memory CD4+ and CD8+ T cells from individuals carrying the IL23R R381Q variant had decreased IL-23–induced expression of IL-17A and lower frequencies of circulating IL-17–producing CD4+ and CD8+ T cells ([35]). A similar study, by Di Meglio et al, did not find differences in the frequencies of peripheral Th17 cells, but did reveal that IL-17 production in response to IL-23 stimulation of in vitro–polarized Th17 cells was reduced in healthy donors carrying the IL23R R381Q allele ([36]). Pidasheva and colleagues showed that the IL23R R381Q allele was associated with reduced IL-23–induced STAT-3 phosphorylation and decreased numbers of IL-23–responsive T cells ([37]).

The findings from these studies suggest that the IL23R R381Q variant represents a loss-of-function allele that leads to significant protection against autoimmunity ([35-37]). In contrast, another study investigated signaling downstream of IL23R, using retroviral transduction of IL23R variants of human T cell blasts. The findings of that study did not identify defects in IL23 signaling (measured as STAT-3 activation and IFNγ production) in T cells transduced with the IL23R R381Q variant ([38]).

In our study, conducted in SpA patients, we found that IL23R R381Q carriers expressed significantly lower levels of both Th17 and Th1 marker genes. However, the presence of this genetic variation was not sufficient to protect patients from development of SpA, as 8 of the 49 patients with SpA recruited to this study carried this allele. This suggests that the combinatorial action of multiple SNPs, rather than a single genetic variant, determines the pathologic outcome.

Consistently, we were able to show that the effector functions of Th17 and Th1 cells in SpA patients are under combinatorial control by multiple genetic variants in IL12B, IL23R, and CCR6. Of note, SNPs that are associated with SpA, but are not at loci related to the IL-23/Th17 pathway, did not affect CD4+ T cell activity, indicating that these SNPs may have an impact on different pathophysiologic mechanisms. Our approach, which investigated correlations of the presence of multiple disease-associated or protective polymorphisms with defined cell functions in each patient, may provide a strategy to delineate the mechanisms by which genetic variants contribute to pathology. This concept is complementary to a recent study in which individual genetic variants predisposing to inflammatory disease were clustered into groups based on the pattern of sharing of these alleles across multiple diseases ([39]). Although no functional analyses were performed, the study suggested that the resulting clusters of SNPs represent different molecular pathways.

In addition to affecting the expression of Th17 marker genes, we found that genetic variation at genes associated with the IL-23/Th17 pathway had a strong effect on genes controlling the differentiation and function of Th1 cells, such as TBX21, IL12RB2, and IFNG. We observed that IL-23 signaling enhanced the production of IFNγ and up-regulated transcripts encoding TBX21, IL12RB2, and IFNG in anti-CD3/anti-CD28–stimulated CD4+ T cells from patients with SpA, possibly as a result of the direct regulation of Th1 marker genes by IL-23, as has been indicated in a previous report ([40]). We also noted that expression of IL23R was strongly increased by stimulation with IL-23, as well as stimulation with IL-12. (Detailed results are listed in Figure S12, available at http://www.pasteur.fr/ip/easysite/pasteur/en/research/scientific-departments/immunology/units-and-groups/immunoregulation.)

Previous work from our laboratory has demonstrated an important role for IL-12 in the regulation of the IL12RB2 gene ([6]), which is only 50 kb downstream of the IL23R gene in humans. These findings point to a previously not appreciated level of coregulation of the IL23R and IL12RB2 loci. Furthermore, the finding that IL-23, which stabilizes Th17 cells, and IL-12, which induces Th1 cells, can cross-regulate the expression of the signaling subunits of their respective receptors provides evidence to indicate that some of the functions of Th17 and Th1 cells may overlap.

Taken together, our data support the notion that genetic variants in the IL12B gene (encoding the IL-12p40 subunit shared by IL-12 and IL-23) and the IL23R gene act as important rheostats in fine-tuning both Th17 and Th1 cell responses. Furthermore, the genetic link of SNPs in IL12B and IL23R to SpA, psoriasis, and IBD suggests that both Th17 and Th1 cells have a role in the pathogenesis of these human diseases. Our findings therefore provide an important link between genotype, cell function, and pathology.

In contrast to SNPs at IL23R, IL12B, and CCR6, genetic variants at STAT3 had only a minor effect on the expression of Th17 and Th1 markers. The STAT-3 transcription factor is a central player in integrating signals from multiple cytokines in various tissues. Deletion of STAT3 in the germline leads to embryonic death in mice (41). Null mutations have not been reported in humans, but patients with Job's, or hyper-IgE, syndrome carry dominant-negative mutations of STAT3 and lack Th17 cells ([42-44]), demonstrating that this transcription factor has an important role in the homeostasis of Th17 cells in vivo. Although we currently cannot explain why genetic variation at STAT3 does not seem to alter the effector function of Th17 cells in our assay system, we favor the hypothesis that the key role of this factor in multiple organ systems may restrict genetic variation that strongly alters its expression or functional activity.

Recent studies have demonstrated that the IL-23 pathway also operates in other immune cell populations, such as CD8+ T cells ([35]), γ/δ T cells ([22, 45]), innate lymphoid cells ([46]), and mast cells ([47, 48]). Kenna et al reported that γ/δ T cells expressing IL-23R and secreting IL-17, but not CD4+ Th17 cells, are enriched in the peripheral blood of patients with AS ([22]), and proposed that IL-23R–expressing γ/δ T cells may play a role in the pathogenesis of AS, providing a link between the observed genetic association of IL23R and AS. More recently, in a study by Sherlock et al ([49]) using a mouse model, the results demonstrated that IL-23 acts on a previously not identified entheseal resident cell population of CD3+CD4−CD8− T cells expressing IL-23R and the transcription factor RORγt. Stimulation of these cells with IL-23 resulted in the production of inflammatory cytokines and chemokines such as IL-6, IL-17, IL-22, and CXCL1 and led to the development of entheseal inflammation and bone remodeling, the hallmarks of SpA. It will be interesting to determine how genetic variation at molecules in the IL-23 pathway could affect these cell populations in the context of inflammatory diseases.

The introduction of biologic therapies targeting IL-1, IL-6, IL-17, and IL-23 has increased the treatment options for inflammatory diseases ([1]). However, as is the case for TNF blockers, these reagents are not effective in all patients. It is currently not possible to predict the responsiveness of patients to these treatments. We believe that understanding how genetic variation affects distinct cellular pathways may help in the design of specific treatments tailored to the genetic architecture of each individual patient.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Rogge had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Coffre, Roumier, Sechet, Dougados, Bianchi, Rogge.

Acquisition of data. Coffre, Roumier, Rybczynska, Sechet, Law, Gossec, Dougados, Rogge.

Analysis and interpretation of data. Coffre, Roumier, Rybczynska, Sechet, Gossec, Dougados, Bianchi, Rogge.

ROLE OF THE STUDY SPONSOR

Pfizer had no role in the study design or in the collection, analysis, or interpretation of the data, the writing of the manuscript, or the decision to submit the manuscript for publication. Publication of this article was not contingent upon approval by Pfizer.

ACKNOWLEDGMENTS

We thank Matthew Albert and Sergei Koralov for helpful discussions and critical reading of the manuscript, and Simon Paternotte for multivariate analysis. We also thank members of the COST-ENTIRE (BM0907) consortiums for helpful discussions.

Ancillary