MICA and NKG2D variants as risk factors in spondyloarthritis: a case–control study

The major histocompatibility complex class I polypeptide-related sequence A (MICA) glycoprotein mediates the activation of the natural killer group 2D receptor (NKG2D) expressed on NK and CD8+ T cells. A methionine or valine at position 129 in exon 3 results in strong (MICA129 met) or weak (MICA129 val) binding to NKG2D. The MICA A5.1 allele causes a premature stop codon. Various NKG2D polymorphisms are associated with low (NKC3 C/C and NKC4 C/C) or high (NKC3 G/G and NKC4 T/T) levels of NK cell cytotoxic activity. In 162 patients with spondyloarthritis (115 with ankylosing spondyloarthritis, 46 with psoriatic arthritis and 1 with reactive arthritis) compared to 124 healthy controls, MICA-129 with methionine allele was more frequent in patients with spondyloarthritis (odds ratio (OR) (95% confidence interval) = 4.84 (2.75‒8.67)), whereas MICA-129 val/val, MICA A5.1 and NKC3 C/C variants were less frequent (OR = 0.20 (0.11‒0.37), 0.15 (0.06‒0.36) and 0.24 (0.13‒0.44), respectively). After adjustment for HLA-B*27 status, only NKC3 C/C remained linked to spondyloarthritis (adjusted OR = 0.14 (0.06‒0.33)). Homozygosity for MICA A5.1 is linked to ankylosing spondyloarthritis, and NKC3 C/C and MICA-129 val/val to psoriatic arthritis. MICA and NKC3 polymorphisms (related to a low NK cell cytotoxic activity) constituted a genetic association with spondyloarthritis.


Introduction
Spondyloarthritis (SpA) refers to a group of chronic inflammatory diseases with a number of similar features, including axial manifestations (such as sacroiliitis or spondylitis), peripheral manifestations (such as arthritis, enthesitis and dactylitis) and extra-articular manifestations (such as psoriasis, uveitis and inflammatory bowel disease (IBD)) [1]. Ankylosing spondyloarthritis (AS) and psoriatic arthritis (PsA) are the two most prevalent SpA diseases. In young adults, AS and PsA cause irreversible tissue damage and worsen the quality of life [2]. The prevalence of SpA is 0.43% in France, with a sex ratio of 1 [3]. The presence of the well-known genetic marker HLA-B*27 allele in patients with symptoms increases the risk of developing SpA (risk ratio (95% confidence interval (CI)) = 39 (17-86)), and is one of the Assessment of Spondyloarthritis International Society (ASAS) criteria for diagnosis of this disease [4]. In the general population, the HLA-B*27 allele is found in 1 to 5% of symptom-free individuals. However, this allele accounts for only 20 to 40% of the genetic susceptibility to SpA, suggesting that other genes are significantly involved in the physiopathology of these diseases [5].
On chromosome 6, the gene closest to the HLA-B locus is major histocompatibility complex class I chain A related (MICA). It codes for a cell stress-inducible glycoprotein [6] that mediates the activation of natural killer group 2D receptor (NKG2D) expression and the associated proinflammatory pathway [7][8][9]. Exons 2-4 of MICA correspond to three extracellular domains, and exon 5 corresponds to a transmembrane domain. The MICA-129 polymorphism (resulting in the replacement of methionine by valine) in exon 3 has been shown to influence NKG2D signaling [10]. Indeed, the presence of methionine vs. valine at position 129 of the α2-heavy chain domain results in strong (MICA-129 Met/Met) or weak (MICA-129 Val/Val) binding to the NKG2D receptor. Furthermore, seven alleles have been described in exon 5 of MICA (from the wild type A4 to A9). The alleles differ with regard to GCT triplet repeat microsatellite polymorphisms. Moreover, the MICA A5.1 allele has an additional G nucleotide insertion, which creates a premature stop codon and results in the production of truncated transmembrane MICA protein (i.e., lacking the cytoplasmic domain) [11].
NKG2D is an activating receptor from the C-type lectinlike family of transmembrane proteins. It is expressed on natural killer (NK) cells, CD8+ T cells and NK T cells. Eight single-nucleotide polymorphisms (SNPs) are known to be closely associated with differences in the level of cytotoxic activity. In particular, NKC3 C/C and NKC4 C/C are associated with a low level of NK cell cytotoxicity, whereas NKC3 G/G and NKC4 T/T are associated with a high level [12].
Contrasting results have been reported for the association between SpA and MICA polymorphisms [13][14][15][16][17]. To the best of our knowledge, NKG2D receptor polymorphisms have been studied in some inflammatory diseases (such as rheumatoid arthritis) but not yet in SpA [18].
The primary objective of this study was to determine whether MICA and NKG2D polymorphisms (linked to the cytotoxic activity of NK cells) are associated with a predisposition to SpA (i.e., AS or PsA) and the clinical and radiological features of the disease.

Demographic characteristics of the study groups
Patients with SpA (n = 162, mean (range) age at diagnosis of SpA: 40 (9-72), 55% male) were compared with healthy control participants (n = 90, 63.7% male) ( Table 1). The mean (range) disease duration was 8.6 (1-46) years. The HLA-B*27 allele was observed in 99 patients (61%) and 4 controls (3.2%, p = 1.8 × 10 −18 ). At diagnosis, 80, 44, 23 and 11% of the patients presented with back pain, arthritis, enthesitis and dactylitis respectively. All patients met ASAS criteria. One hundred and thirty-one patients were ASASpositive axial SpA: 110 patients had structural damage or sacroiliitis (SI X-ray, computed tomography (CT) scan or magnetic resonance imaging (MRI)) and 21 patients were HLA-B*27 positive associated with more than 2 others symptoms. Thirty-one patients were only ASAS-positive peripheral SpA. In 51 patients who had a SI X-ray, 21 patients (13%) met the modified New York (mNY) criteria. Other patients had SI CT scan or MRI or did not have SI explorations because they had only peripheral symptoms. One hundred and fifteen patients (71%) had AS, 46 (28%) had PsA and 1 patient had reactive arthritis.

Frequencies of MICA/NKG2D polymorphisms in the SpA population
The overall observed frequencies of MICA/NKG2D polymorphisms are listed in Table 2. The homozygote MICA 129 Val/Val was found to protect against the development of SpA (odds ratio (OR) (95% CI) = 0.20 (0.11-0.37)), whereas the MICA 129 Met allele was more prevalent in patients with SpA than in controls and thus constituted a genetic association for SpA (OR = 4.84 (2.75-8.67)).
In multivariate analysis, before adjustment for the HLA-B*27 status, heterozygous and homozygous MICA A5.1 mutations, the presence of a G mutation in NKC3 constituted a genetic association with SpA, with ORs of 0.60 (0.37-0.95), 0.20 (0.09-0.45) and 4.51 (2.35-8.93), respectively, whereas the NKC3 C/C polymorphism was   Any association was found between MICA 129 Met/Val, MICA A5.1, NKC3 C/G, NKC4 C/T polymorphisms and clinical data (Table 3). Indeed, these polymorphisms are not linked to uveitis, psoriasis, inflammatory bowel disease and sacroiliac joint damage in our cohort.

Discussion
Associations between MICA/NKG2D polymorphisms and several immune-mediated diseases have been reported previously. In particular, MICA polymorphisms are associated with inflammatory rheumatic diseases [19], Behçet's disease [20], IBD1 [21], systemic lupus [22] and ankylosing spondylitis [23,24]. NKG2D variants have also been linked to rheumatoid arthritis susceptibility and severity [18]. Given the complex gene sequences and protein expression profiles of MICA/NKG2D, studies of the pathogenesis of immune-mediated SpA are essential. To address this issue, we genotyped MICA129 and MICA A5.1 polymorphisms, and two NKG2D SNPs (rs1049174 for NKC3 and rs2255336 for NKC4) in patients with SpA and in healthy controls.
Our study population of mainly Caucasian patients was similar to the literature data with regard to age at diagnosis, sex ratio, the frequency of AS vs. PsA and the percentage of patients bearing HLA-B*27 alleles [2,3]. We had a small number of patients meeting the mNY criteria because the majority of our patient did not have the result of SI X-ray in their data (the result was not available at the time of collecting data and/or the used of MRI is more common than the SI X-ray to diagnose a SpA). However, all patients met ASAS criteria.
We found that MICA 129 Val/Val constituted a genetic association with SpA (i.e., both AS and PsA). In the study of Zhou et al. [23] of 1070 American patients with AS (all of whom met the New York modified criteria), the same result was found both before and after adjustment for HLA-B*27 status. We found that MICA-129 Val/Val protected against PsA. We also found that this polymorphism constituted a genetic association with X-ray-confirmed sacroiliitis. In view of the link between MICA129 Val/Val and low affinity for its receptor, one pathophysiological hypothesis holds that patients with SpA and MICA129 Val/Val polymorphism have lower levels of inflammation near their sacroiliac joint and thus suffer less damage from sacroiliitis.
To the best of our knowledge, the association between NKC3 polymorphisms and SpA has not previously been studied. Here, we found that the presence of a G mutation in NKC3 was a risk factor for SpA both before and after adjustment for HLA-B*27 status. NKC3 G/G status could be used to refine the diagnosis of SpA, particularly when a patient is negative for HLA-B*27 status. Given the association between NKC3 G/G and high levels of cytotoxicity [12], it would be interesting to understand the relationship between this polymorphism and severe, aggressive forms of the disease.
It is well established that the MICA gene confers susceptibility to SpA [23]; indeed, MICA*007:01 is a significant risk allele for SpA in Caucasian and Han Chinese populations, and MICA*019 is a major risk allele in Chinese patients. Moreover, the MICA129 met/met genotype was linked to juvenile SpA (independently of HLA-B*27 status) in Algerian patients [24], rheumatic disease-associated IBD in Spanish patients [25] and systemic lupus erythematosus in Japanese patients [19]. NKG2D gene variants have also been linked to rheumatoid arthritis susceptibility and severity [18]. Indeed, the frequencies of the NKG2D9 A allele and the NKG2D10 T allele were significantly higher in patients with deformities, whereas the frequencies of the NKG2D9 G allele and the NKG2D10 A allele were higher in patients without deformities. The mechanism by which MICA/NKG2D polymorphisms protect against inflammatory rheumatic diseases has not been fully elucidated. However, Nielsen et al. [26] showed that fibroblast-like synoviocytes from patients with rheumatoid arthritis (RA-FLS) express many ligands of activating and inhibitory NK cell receptors. Blockade of the interaction between CD94/ NGA (an inhibitory receptor) and its ligand HLA-E expressed on RA-FLSs enhanced degranulation of the Nishi human NK cell line [26]. Moreover, blockade of NKG2D in a murine model of collagen-induced arthritis was associated with significant joint protection, relative to control animals [27]. Patients with psoriatic arthritis have elevated expression levels of interleukin-15 and MIC in their affected synovial tissues, and this inflammatory environment enabled NK cell activation and tissue destruction through NKG2D [28].
The present study was performed in a large, representative cohort of patients with SpA with the use of asymptomatic control patient. It constituted a multidisciplinary collaboration between geneticists and rheumatologists. In view of the retrospective, single-center design, our study had some limitations. Furthermore, the PsA subgroup was relatively small (n = 46), even though the proportion of patients with PsA was much the same as in other cohorts [3].
Studies of the association between MICA/NKG2D polymorphisms and inflammatory diseases have generated conflicting results. Our present results show that MICA A5.1, MICA129 val/val and NKC3 C/C polymorphisms (related to low levels of cytotoxic activity in NK cells) protected against SpA. In contrast, MICA129 met, NKC3 G/ G and G/C polymorphisms (related to elevated levels of cytotoxic activity and an inflammatory environment) were risk factors for SpA. Further research must focus on the involvement of MICA/NKG2D signaling in rheumatic disease.

Study participants
We performed a retrospective, single-center study of Caucasian patients with SpA being treated in the Rheumatology Department at Amiens University Hospital (Amiens, France). The procedures were approved by the Nord Ouest II ethics committee, Amiens, France. In accordance with French legislation and the Declaration of Helsinki, all participants gave their prior, written, informed consent to genetic testing. Spondyloarthritis had been diagnosed by an experienced rheumatologist in all cases.
A total of 214 patients with SpA were screened between August 2014 and May 2015. We excluded 52 patients who were lost to follow-up or had not provided written, informed consent. Hence, 162 patients were included in the study. Demographic, clinical, laboratory, radiological and HLA-B*27 genotyping (performed by PCR-SSP: Single Specific Primer-Polymerase Chain Reaction, Bionobis, Paris, France) data were extracted from the patients' medical records. All the radiological data had been analyzed by the same specialist radiologist at the time of diagnosis. Sacroiliac joint damages were diagnosed on an X-ray, CT scan or MRI of the sacroiliac joint [29,30]. Erosions were diagnosed on hands and foot X-rays. Retrospectively, patients were classified with mNY and ASAS criteria [29,30].

Controls
One hundred and twenty healthy control Caucasian participants were selected from adult hematopoietic stem cell donor registries. In line with the World Marrow Donor Association's guidelines, none had a history of autoimmune disease, neoplastic disease or thromboembolic events.

NKG2D and MICA polymorphisms
NKG2D genotyping was performed as described previously [31], using a TaqMan® allelic discrimination method on a 9700-HT real-time polymerase chain reaction (PCR) system (Applied Biosystems, Foster City, CA, USA). An allelespecific fluorogenic oligonucleotide (Applied Biosystems) was used to discriminate between genotypes for each studied pair of alleles. The following two SNPs were probed: (i) rs1049174, featuring a G-C substitution that distinguishes between the high-activity-related HNK1 NKG2D haplotype (G) and the low-activity-related LNK1 (C) haplotype (NKC3); and (ii) rs2255336, featuring a C-T substitution that distinguishes between the HNK2 (T) and LNK2 (C) haplotypes (NKC4). All participants underwent sequence-based typing for MICA. PCR primers were designed to amplify MICA exons 3 and 5, as described previously [31]. Sequencing was performed on an automatic 24-capillary ABI Prism system (Applied Biosystems), and the electropherogram was interpreted using SeqPilot® software (JSI Medical Systems GmbH, Ettenheim, Germany).

Statistical analysis
Data were quoted as the mean and range (for continuous variables) or the number and percentage (for categorical variables). Univariate logistic regression was used to compare the frequency of polymorphism in patients vs. controls. The odds ratio was adjusted (OR adj ) for HLA-B*27 status by including this variable in the logistic regression. Each genotype was compared with the others as the reference. Variables with a p value < 0.1 in a univariate analysis were included in a multivariate model, where the threshold for statistical significance was set to p < 0.05. Associations between clinical and radiological features and polymorphisms were also probed. All analyses were performed with R software (version 3.1).
Funding This work was funded by Amiens University Hospital as part of the French government's hospital-based clinical research program.

Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/.