Article Text

Extended report
MICA, a gene contributing strong susceptibility to ankylosing spondylitis
  1. Xiaodong Zhou1,
  2. Jiucun Wang2,
  3. Hejian Zou3,
  4. Michael M Ward4,
  5. Michael H Weisman5,
  6. Maribel G Espitia1,
  7. Xiangjun Xiao6,
  8. Effie Petersdorf7,
  9. Emmanuel Mignot8,
  10. Javier Martin9,
  11. Lianne S Gensler10,
  12. Paul Scheet6,
  13. John D Reveille1
  1. 1Department of Internal Medicine, Division of Rheumatology, The University of Texas Medical School at Houston, Houston, Texas, USA
  2. 2Ministry of Education (MOE) Key Laboratory of Contemporary Anthropology, School of Life Sciences, Fudan University, Shanghai, China
  3. 3Department of Rheumatology, Huashan Hospital, Fudan University, Shanghai, China
  4. 4NIAMS, Bethesda, Maryland, USA
  5. 5Department of Medicine, Cedars-Sinai Medical Center, Los Angeles, California, USA
  6. 6Department of Epidemiology, The University of Texas MD Anderson Cancer Center, Houston, Texas, USA
  7. 7Division of Clinical Research, Fred Hutchinson Cancer Research Center, Department of Medicine and Division of Oncology, University of Washington, Seattle, Washington, USA
  8. 8Department of Psychiatry, Stanford University, School of Medicine, Stanford, California, USA
  9. 9The Instituto de Parasitologia y Biomedicina Lopez-Neyra, CSIC, Armilla, Granada, Spain
  10. 10Division of Rheumatology, The University of California, San Francisco, USA
  1. Correspondence to Professor Xiaodong Zhou, Department of Internal Medicine, Division of Rheumatology, The University of Texas Medical School at Houston, 6431 Fannin Street, MSB5270, Houston, TX 77030, USA; Xiaodong.zhou{at}


Objective The human major histocompatibility complex class I chain-related gene A (MICA) controls the immune process by balancing activities of  natural killer cells, γδ T cells and αβ CD8 T cells, and immunosuppressive CD4 T cells. MICA is located near HLA-B on chromosome 6. Recent genomewide association studies indicate that genes most strongly linked to ankylosing spondylitis (AS) susceptibility come from the region containing HLA-B and MICA. While HLA-B27 is a well-known risk genetic marker for AS, the potential effect of linkage disequilibrium (LD) shields any associations of genes around HLA-B with AS. The aim of this study was to investigate a novel independent genetic association of MICA to AS.

Methods We examined 1543 AS patients and 1539 controls from two ethnic populations by sequencing MICA and genotyping HLA-B alleles. Initially, 1070 AS patients and 1003 controls of European ancestry were used as a discovery cohort, followed by a confirmation cohort of 473 Han Chinese AS patients and 536 controls. We performed a stratified analysis based on HLA-B27 carrier status. We also conducted logistic regression with a formal interaction term.

Results Sequencing of MICA identified that MICA*007:01 is a significant risk allele for AS in both Caucasian and Han Chinese populations, and that MICA*019 is a major risk allele in Chinese AS patients. Conditional analysis of MICA alleles on HLA-B27 that unshielded LD effect confirmed associations of the MICA alleles with AS.

Conclusions Parallel with HLA-B27, MICA confers strong susceptibility to AS in US white and Han Chinese populations.

  • Ankylosing Spondylitis
  • Gene Polymorphism
  • Inflammation

Statistics from


Ankylosing spondylitis (AS) is a chronic inflammatory disease characterised by inflammation of the axial skeleton, including the spine and sacroiliac joints, as well as extraspinal involvement such as uveitis, enthesitis and peripheral arthritis. The prevalence of AS in Western European, Chinese and North American populations ranges between 0.2% and 0.7%.1–4 Genetic factors play major roles in the pathogenesis of AS, particularly HLA-B27, which has persisted as one of the best examples of a hereditary marker with disease susceptibility. About 90% of AS patients of European ancestry have at least one HLA-B27 allele in contrast to only 7.5% of normal individuals, which confers the strongest HLA association among all human chronic diseases reported thus far.5 ,6 On the other hand, over 90% HLA-B27 positive individuals do not develop AS, which suggests that this is not the sole risk factor.

Indeed, recent genomewide association studies (GWAS) of AS have identified several novel AS-associated polymorphisms on the genes including ERAP1, IL23R, IL1R2, ANTXR2, IL12B, PTGER4, TNFR1 and TRADD, although some of them still need confirmation.7–9 However, most of the strongest genetic variants reported from the GWAS of AS come from the HLA class I region of the major histocompatibility complex, especially those polymorphisms around the HLA-B locus.7 ,8 Comparing the levels of relative risk, the non-major histocompatibility complex (MHC) genetic associations contribute only a small fraction of susceptibility to AS compared to that from the MHC.7–9 It is entirely plausible that the HLA class I region contains additional genes influencing susceptibility to AS. MICA is located next to HLA-B at the centromeric end of the HLA class I region. MICA is normally expressed on the cell membrane, and functions in immune activation in response to stimuli through binding with the natural killer (NK) cell receptor NKG2D that is expressed on the surface of NK cells, γδ T cells and αβ CD8 T cells.10 Our recent GWAS identified the single nucleotide polymorphisms (SNPs) in MICA tagged to HLA-B27 that is linked to AS susceptibility.9 This raises the further possibility that specific sequence changes of MICA may contribute to the pathogenesis of AS. However, previous confirmation studies of MICA associations with AS were hindered by linkage disequilibrium (LD) with HLA-B.

To address the confounding effect of HLA-B27 due to LD, we examined two ethnic cohorts of AS patients and non-diseased populations, including North American Caucasian and Han Chinese by sequencing MICA and typing the HLA-B locus. Sequencing MICA provides a comprehensive view of MICA alleles in association with AS, and cross-examining two ethnic groups with distinct ancestry origins and looking in HLA-B27 positive and negative individuals would capture the genetic diversity with distinct allele frequencies and LD patterns, which may facilitate the identification of true disease associations from LD effect.

Materials and methods

AS patients and controls

Two ethnic cohorts, including 1070 North American AS patients and 1003 local-matched controls of European ancestry, and 473 AS patients and 536 controls from China of Han Chinese ancestry were examined in these studies. The average ages of AS patients and controls for US Caucasian were 41 and 25 respectively, and for Han Chinese were 43 and 46, respectively. There were 70% men versus 30% women in US patients and 52% men versus 48% women in US controls, and 68% men versus 32% women in Chinese patients and 51% men versus 49% women in Chinese controls. The North American patients were examined as a discovery cohort, with associations confirmed in the Chinese cohort. White patients who were participants in the Prospective Study of Outcomes in AS,11 as well as from the North American Spondylitis Consortium,12 were examined in this study. Controls of Caucasian European ancestry were enrolled in the studies in the University of Texas Houston and Stanford School of Medicine. It is worth noting that majority of AS patients came from Texas and California. Any related patient and controls were excluded from the studies. Han Chinese patients and controls were enrolled from the clinics and hospitals in southern cities of China using a translated enrolment form of US standard that was used for North American patients. All patients met the modified New York criteria for this disease.13 All participants underwent a clinical evaluation by one of the study rheumatologists, and had pelvic and spinal radiographs to confirm their diagnosis. Unrelated controls were examined without any history of rheumatic disease. Chinese controls were obtained from a study project of Chinese population genetics in Fudan University, Shanghai, China. The ethical approvals of the studies were obtained from Institutional Review Boards in both The University of Texas Medical School at Houston and Fudan University.

Sequencing and genotyping

Genomic DNA was extracted from peripheral blood cells from subjects. The HLA-B genotyping was performed with Dynal-sequence-specific oligonucleotide (SSO) methods and confirmed by direct PCR sequencing of exon 2. MICA was examined by PCR sequencing exon 2, 3, 4 and 5 region using bidirectional Sanger sequencing methods. The primers used for PCR and sequencing were based on previous reports.14 ,15 Sequence data were analysed using SeqMan-II (DNA Star, Madison, Wisconsin, USA). Since all allelic genotypes corresponding to polymorphisms of the MICA are known from the MICA database at IMGT/HLA (, the polymorphisms of the MICA in each sample were first extracted from the sequencing data, and then were compared with specific polymorphisms in reference sequences of MICA alleles. Allelic genotypes of the MICA in each sample were obtained according to reference sequence of specific MICA alleles.

For quality control in DNA typing, all sequencing processes were performed using robotic automation system to minimise sample mislabelling and misplacing. We also had blind duplicates in each sample plate. In addition, MICA polymorphisms are in tight LD on given haplotypes or alleles that were used as an internal check for the genotyping while we performed alignment on different segments of the MICA gene.

Data analysis

Tests of association between genetic variants and AS were performed for each allele (marginally) assuming genetic dominance. Exact p values were obtained (Fisher's test) from 2×2 tables of allele counts and disease status. To explore the possibility of interactions, we considered combinations of alleles (diplotypes), which were also tabulated with 2×2 tables; we also conducted logistic regression with a formal interaction term.

Since MICA*007:01 and *019 showed a strong marginal association with AS, we sought to condition out the effect from B-27, a known risk factor. We performed a stratified analysis based on HLA-B27 carrier status, analysing separately those without any B27 alleles. Within each stratum, we created 2×2 tables based on MICA*007:01 or *019 and AS frequencies.

Since we utilised a separate (Chinese) cohort for replication, we did not consider multiple testing in the usual way, particularly given the dependent nature of the marker data. Instead, although we performed a considerable number of tests in both the European and Chinese samples, and as such, p values should be considered in this light, our recommended procedure is to identify putatively interesting results and then examine these in the replication set. In addition, among our study participants, only AS patients of US Caucasian origin were examined in the previous GWAS, 7–9 and this study was not found to have population stratification. For our US controls and Chinese cohorts, we did not have GWAS data for use in conducting an investigation of population stratification.


The discovery cohort of North American with AS showed significant associations with polymorphisms in the coding region of MICA across exon 2–5 (see online supplementary table S1). Among these, polymorphisms at the codon 6, 14, 24, 36, 64, 91, 122, 129, 173, 175, 181, 206, 210, 213, 215 and 306 represent amino acid (AA) substitutions. Some of these AA changes are distinct in side chain polarity and charges. The p values of these changes ranged from 9 × 10−7 to 10−270. It is particularly interesting that the codon 295 has a tri-nucleotide microsatellite polymorphism (GCT)n that are designated as An, and a five repetition of GCT may coexist with a guanosine insertion that is designated as A5.1. A significant increase of A4, but a decrease of A5.1 was observed in patients (table 1).

Table 1

Comparison of the presence of specific MICA variants at codon 295 in US AS patients and controls

According to IMGT/HLA database (, there are 78 identified MICA alleles. Fifteen MICA alleles were found in US AS patients (table 2). Among these alleles, MICA*007:01 was found to be only one significantly increased in the AS patients, and emerged as the major AS-associated allele in the North American cohort (86.8% in AS vs 9.9% in controls). MICA*008, particularly MICA*008:01 or *008:04, appeared to be the most likely encountered allele in the North American white controls (65%). Comparison of allelic pairs of MICA indicated that any pairs containing MICA*007:01 are increased in the AS patients, and that the genotype pair of MICA*007:01/*009:02 and MICA*018:02/*008 are observed only in the North American AS cohort (see online supplementary table S2). Age and gender adjusted results for the codon 295 and MICA alleles were shown in online supplementary tables S3 and S4, respectively.

Table 2

Comparison of the presence of MICA alleles between US AS patients and controls

The Han Chinese confirmation cohort also showed significant associations with AS at multiple MICA polymorphism sites that were consistent with North American cohort (see online supplementary table S5). Importantly, associations with A4 (increased) and A5.1 (decreased) at codon 295 are consistent between two cohorts, while A5 was increased only in Chinese AS patients (table 3).

Table 3

Comparison of the presence of specific MICA variants at codon 295 in Chinese AS patients and controls

Eleven MICA alleles were found in Chinese cohort (table 4). MICA*007:01 was consistently and significantly more common in the AS patients (25.4%) compared with that in controls (2.8%), but the major MICA allele encountered in the Chinese AS patients was MICA*019, which occurred in 64.3% of the patients versus 11% in the controls. There were 12 Chinese AS patients with both MICA*007:01 and *019 alleles, 19 with both MICA*007:01 and *009:01 or *049, and 11 with both MICA*007:01 and *027. These combinations were not found in the controls. The presence of either MICA*007:01 or *019 is 87.1% in the Chinese AS patients versus 13.8% in Chinese controls. In contrast to US controls, the major MICA alleles in Chinese controls were MICA*010:01 (40.3%), *008 (39.6%) and *002:01 (30.6%).

Table 4

Comparison of the presence of MICA alleles between Chinese AS patients and controls

HLA-B27 is a well-defined risk factor for AS, and is in strong LD with MICA. However, this LD is not complete; that is, with our sample sizes and multiple populations, we can address and confirm independent risk effects from these two loci. Therefore, several conditional analyses were performed whose results are displayed in table 5. Particularly, in North American AS patients lacking HLA-B27, 19 of 100 AS patients (19%) versus 10 of 399 controls (2.5%) were carriers of MICA*007:01, (p=4.3×10−8, OR 9.13); in HLA-B27 negative Chinese AS patients, seven of 42 (16.7%) patients versus two of 424 (0.47%) controls were positive for MICA*007:01 (p value of 9.4 × 10−7, OR 42.2), and 21 of 42 (50%) AS patients versus 42 of 424 (9.9%) controls were positive for MICA*019 (p=1.5×10−9, OR 9.1). The presence of either MICA*007:01 or MICA*019 in HLA-B27 negative Chinese patients was 64.3% vs 10.38% in controls (p=1.3×10−14, OR 15.55). A conditional analysis for HLA-B27 association with AS was performed in non-carriers of MICA*007:01 and *019. In MICA*007:01 negative North American AS patients, 39 of 120 were B27 positive compares to one of 390 controls. In the Chinese cohort, HLA-B27 was observed in 82 of 117 MICA*007:01-negative AS patients versus 26 of 448 controls, and in 48 of 69 MICA*019-negative AS patients versus 24 of 406 controls, as well as in 10 of 25 patients versus 13 of 393 controls of who lacked both MICA*007:01 and MICA*019. Of note, we have typed only a portion of US controls for HLA-B alleles. The controls were randomly selected for the HLA-B typing. We noticed that there were only 100 AS patients who were HLA-B27 negative. We obtained 399 B27-negative controls in comparing with 100 B27-negative AS patients.

Table 5

Conditional analysis on HLA-B27 negative subjects for associations of MICA alleles with AS, and on MICA*007:01 or *019 negative subjects for associations of HLA-B27 with AS

To assess statistical independence of effects from these two loci, we performed a logistic regression of AS on HLA-B27 and the MICA alleles (007:01, 019, or either). Logistic regression analysis was performed to fit a standard full model with three variables: indicator of MICA allele, indicator of B27 allele, and an interaction term. We found putatively significant interactions (p<0.05) for the following three analyses: B27*07:01 in Caucasians (β=−1.4, p=0.01); B27*07:01 in Chinese (β=−1.7, p=0.0001); and B27*019 in Chinese (β=−0.6, p=0.03). We note the coefficients for the interaction terms were consistently negative, that is, the total risk of both alleles is lower than what would be predicted based on adding their marginal effects.

In addition, the co-incidence of the MICA alleles and HLA-B27 subtypes in US Caucasian and Han Chinese cohorts were reported in online supplementary table S6.


This is the first report to suggest an association of MICA alleles with AS in both HLA-B27-positive and B27-negative AS patients, with confirmation in a different ethnic group. Genetic associations of MICA in several immune-mediated diseases have been reported previously. In particular, MICA polymorphisms have been associated with psoriatic arthritis, rheumatoid arthritis, Sjögren's syndrome, Behçet's disease and inflammatory bowel disease.16–21 However, previous studies examined a limited number of polymorphisms and small cohorts, often with lack of confirmation in other ethnic populations. The studies of single nucleotide polymorphisms in a small cohort of Algerian AS patients indicated that MICA-129 was associated with early onset disease,22 but two other reports suggested that the MICA association with AS was secondary to LD with HLA-B27.23 ,24 Tsuchiya et al examined MICA with PCR-single-strand conformation polymorphism (PCR/SSCP) method in 36 HLA-B27 positive Japanese including 18 patients with AS, one patient with Reiter’s syndrome and 17 healthy controls. MICA*007 was found in six AS patients (35.3%),23 which appeared similar in frequency in the Chinese AS patients examined here. However, the authors assumed MICA*007 was in LD with B27. Martinez-Borra et al24 examined MICA in HLA-B27 positive mixed populations including 161 AS patients and 161 controls. The mixed ethnic populations contained Spaniards, Slavic Siberians, Jews, West Africans and Asian Indians. They also reported that the HLA-B27 subtypes were associated with MICA*007. Both studies did not examine HLA-B27 negative individuals, and only studied a relatively small number of patients.

Herein, we examined polymorphisms of MICA in two large AS cohorts from different ethnic populations and identified that multiple changes in the coding region of MICA are strongly associated with AS. The association levels are similar to those seen with the HLA-B27, and largely surpass all other non-MHC genetic factors reported previously. Many of these polymorphisms encode AA substitutions with changes in side chain polarity and charges. Some of these changes may be associated with specific biological functions of MICA molecules. For instance, a previous report demonstrated that the presence of methionine versus valine at codon 129 confers stronger affinity of MICA to NKG2D.25

Comparisons between European whites and Han Chinese cohorts showed that AS patients from two different ethnic populations share many disease-associated polymorphisms, especially, at codons MICA-6, -14 and -306. On the other hand, the two ethnic AS cohorts also displayed distinct features of certain MICA polymorphisms. MICA alleles in Chinese AS patients are more likely to encode AA with more polar and positively charged side chains. In European Caucasians, the changes of specific MICA polymorphisms appeared more robust, especially at codon MICA-191 (see online supplementary table S1). An increased prevalence of the methionine encoding sequence at codon MICA-129 was observed in North American AS patients, which is in accordance with previous reports of this polymorphism in AS,24 as well as in anterior uveitis26 and psoriatic arthritis,27 two diseases that frequently occur in AS patients.

The codon MICA-295 is of particular interest. The MICA*A5.1 results in a premature stop codon, and leads to formation of soluble form of MICA.19 ,28 In contrast to the MICA protein that is expressed on the membrane, which stimulates activation of NK and T cells, soluble MICA is likely to induce immunosuppressive NKG2D CD4 T cells.29 Therefore, balance between membrane MICA and soluble MICA may control the immune process in response to stimuli. Interestingly enough, the presence of MICA*A5.1 is substantially low, but the MICA*A4 microsatellite carrying the shortest repetition of GCT is significantly increased in AS patients from both cohorts. Although, it is still unknown whether such shorter repetition of GCT affects stability and/or functional changes of MICA, both MICA*A9 and MICA*A6 microsatellites that carry relatively longer repetitions of GCT are significantly less frequent in the AS patients from these two ethnic groups.

Further analysis of MICA alleles based on combinations of all polymorphisms indicated that the MICA*007:01 is a susceptibility allele for AS in European whites. Although two ethnic populations have distinct frequencies of individual alleles, the MICA*007:01 also confers susceptibility to Chinese AS patients along with predominant MICA*019. Predominance of the MICA*019 in Chinese AS patients, but the lack of this in those of European ancestry suggests that evolution of MICA alleles differ in these two populations.

By conditioning on HLA-B27, it appears that both MICA*007:01 and *019 contribute to susceptibility to AS, independent of the LD effect of HLA-B27. Interestingly, the presence of either MICA*007:01 or *019 in Chinese AS patients is 87.1% versus 13.8% in Controls, that is similar to the occurrence rate of the MICA*007:01 in white AS patients from the US (86.8% AS vs 9.9% control), which suggests that both ethnic populations are consistent in the level of the MICA contribution to susceptibility to AS. Of note, both the MICA*007:01 and MICA*019 carry shorter repetitions of GCT, known as the microsatellite markers MICA*A4 and MICA*A5, respectively, at codon MICA-295. In contrast, the most common MICA allele in normal control is MICA*008 (table 2) that contains MICA*5.1.

In addition to possession of specific AS-associated MICA alleles, uncommon MICA allelic combinations were observed in US whites or Han Chinese, including the MICA*007:01/*009:02, *007:01/*019, *007:01/*009:01 or *049 and *007:01/*027. A common feature of these genotypes is the lack of the MICA*A5.1 microsatellite at codon MICA-295.

The MHC class I region genes comprise the most polymorphic region in the human genome. Considering the complexity of extensive LD across this region, further confirmation of a direct and independent contribution of MICA alleles to AS susceptibility may require further analysis of other genes in this region and other cohorts of AS.

It is worth noting that the increased possessions of MICA*007 and MICA*019 have been previously reported in whites with ulcerative colitis (UC)21 and Spanish patients with Behçet’s disease,30 respectively. Both diseases share clinical features with AS, which suggests that MICA*007 and *019 may be share common pathogenic features in these diseases. It is particularly interesting that MICA molecules are preferentially expressed in the epithelial cells of intestine, and chronic intestinal inflammation is commonly encountered in AS patients.31 ,32 Therefore, a gut microbe-driven inflammatory process involving the MICA*007 and *019 may be associated with the development of chronic inflammatory arthritis and spinal inflammation in AS.


This study was supported by the NIH NIAID 1U01AI09090-01 and NIH P01-052915-01 and the Intramural Research Program, NIAMS, NIH, the Major National Science and Technology Program of China, grant number 2008ZX10002-002, and the Science and Technology Committee of Shanghai Municipality (11410701800).


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Handling editor Tore K Kvien

  • Contributors All authors are involved in the following contribution to the paper: (1) conception and design, or analysis and interpretation of data; (2) drafting the article or revising it critically for important intellectual content; (3) final approval of the version to be published.

  • Funding NIH NIAID.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval The University of Texas Medical School at Houston.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement Our data will be available for sharing upon request.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.