MICA polymorphism: biology and importance in cancer

The major histocompatibility complex class i polypeptide-related sequence A gene ( MICA ) encodes a membrane-bound protein acting as a ligand to stimulate an activating receptor, NKG2D, expressed on the surface of essentially all human natural killer (NK), γδ T and CD8 + αβ T cells. MiCA protein is absent from most cells but can be induced by infections and oncogenic trans- formation and is frequently expressed in epithelial tumors. Upon binding to MiCA, NKG2D activates cytolytic responses of NK and γδ T cells against infected and tumor cells expressing MiCA. Therefore, membrane-bound MiCA acts as a signal during the early immune response against infection or spontaneously aris-ing tumors. On the other hand, human tumor cells spontaneously release a soluble form of MiCA, causing the downregulation of NKG2D and in turn severe impairment of the antitumor immune response of NK and CD8 + T cells. This is considered to promote tumor immune evasion and also to compromise host resist- ance to infections. MICA is the most polymorphic non-classi-cal class i gene. A possible association of MICA polymorphism with genetic predisposition to different cancer types has been investigated in candidate gene-based studies. Two genome-wide association studies have identified loci in MICA that influence susceptibility to cervical neoplasia and hepatitis C virus-induced hepatocellular carcinoma, respectively. Given the current level of interest in the field of MICA gene, we discuss the genetics and biology of the MICA gene and the role of its polymorphism in cancer. Gaps in our understanding and future research needs are also discussed.

The major histocompatibility complex class i polypeptide-related sequence A gene (MICA) encodes a membrane-bound protein acting as a ligand to stimulate an activating receptor, NKG2D, expressed on the surface of essentially all human natural killer (NK), γδ T and CD8 + αβ T cells. MiCA protein is absent from most cells but can be induced by infections and oncogenic transformation and is frequently expressed in epithelial tumors. Upon binding to MiCA, NKG2D activates cytolytic responses of NK and γδ T cells against infected and tumor cells expressing MiCA. Therefore, membrane-bound MiCA acts as a signal during the early immune response against infection or spontaneously arising tumors. On the other hand, human tumor cells spontaneously release a soluble form of MiCA, causing the downregulation of NKG2D and in turn severe impairment of the antitumor immune response of NK and CD8 + T cells. This is considered to promote tumor immune evasion and also to compromise host resistance to infections. MICA is the most polymorphic non-classical class i gene. A possible association of MICA polymorphism with genetic predisposition to different cancer types has been investigated in candidate gene-based studies. Two genome-wide association studies have identified loci in MICA that influence susceptibility to cervical neoplasia and hepatitis C virus-induced hepatocellular carcinoma, respectively. Given the current level of interest in the field of MICA gene, we discuss the genetics and biology of the MICA gene and the role of its polymorphism in cancer. Gaps in our understanding and future research needs are also discussed.

introduction
The major histocompatibility complex (MHC) region on chromosome 6 contains numerous polymorphic and multicopy genes that play important roles in the immune system that provide protection against pathogens and tumorigenesis. The first MHC products were discovered on the surface of leucocytes and the human MHC was initially referred to as the human leucocyte antigen (HLA) complex. The MHC gene family is divided into three subgroups: MHC class I, II and III. MHC class I (HLA-A, -B and -C) and class II (HLA-DR, -DQ and -DP) genes encode cell surface antigen-presenting molecules that bind short peptides derived from non-self and self proteins, including infections and autoantigens. The presentation of these HLA-anchored peptides to T lymphocytes triggers a cascade of responses in immuneassociated genes that leads to adaptive immunity. MHC class III genes encode diverse group of molecules that perform various immune function in the body such as complement proteins involved in the antibody response and inflammatory cytokines (1). In 1994, a new set of loci related to MHC class I genes called MHC class I chainrelated genes (MIC) or Perth beta block transcript 11 (PERB11) were identified independently by Bahram et al. (2) and Leelayuwat et al. (3). A member of this family, MICA, has been found to be the most polymorphic non-classical class I gene (4).
Unlike the classical class I molecules, MICA does not bind β2microglobulin (5). Instead, it encodes a membrane-bound protein which acts as a ligand to stimulate an activating receptor, NKG2D, expressed on the surface of essentially all human natural killer (NK), γδ T and CD8 + αβ T cells (5)(6)(7). Normally, MICA is constitutively expressed in low levels on epithelial cells in the gut and thymus, endothelial cells, fibroblasts and monocytes (8)(9)(10). However, it is upregulated or expressed de novo in stressed conditions, such as during viral and bacterial infections (7,11,12), heat shock (9), DNA damage response (13), oncogenic transformation (5,6) and in autoimmune conditions (14). MICA serves as signal of cellular stress, and engagement of NKG2D by MICA triggers NK cells, and costimulates some γδ T cells and antigen-specific CD8 + αβ T cells, resulting in a range of immune effector functions, such as cytotoxicity and cytokine production (8,15). The recognition of the MICA molecule by the NKG2D receptor enables immune cells to identify and attack infected or transformed cells without the need of MHC class I expression or antigen recognition (16). Thus, the MICA/ NKG2D interaction is an effective mechanism for immunosurveillance. Nevertheless, tumor cells have evolved mechanisms to minimize or avoid the response mediated by NKG2D by shedding MICA from the cell surface (17,18). The shedding of MICA has been reported to be mediated by matrix metalloproteinases through proteolytic cleavage of the extracellular parts and palmitoylation of two cysteine residues in the cytoplasmic tail of MICA was found to be necessary for efficient cleavage (18,19). The shedding of soluble MICA (sMICA) by human tumors not only hinders recognition of the MICA-expressing tumor cells but also results in systemic downregulation of NKG2D on NK and CD8 + T cells and evasion of NKG2D-mediated immune recognition (17,18). This is considered to promote tumor evasion and also to compromise host resistance to infections. High levels of the soluble form of MICA have been detected in sera of patients with malignancy and precancerous condition, but not in healthy subjects (20)(21)(22)(23)(24)(25)(26)(27)(28)(29), which may reduce immunogenicity of cancer cells and promote tumor progression. In cancer patients, elevated sMICA levels correlated significantly with cancer stage and metastasis (28,29). Given the important role of MICA in immune activation and surveillance against infection and tumorigenesis, the association between MICA polymorphism and susceptibility to cancer has attracted considerable attention. In this report, we review the functional relevance and implications of MICA polymorphism in different cancer types. Gaps in our understanding and future research needs are also discussed.

MICA gene and polymorphism
The human MICA gene is ~15.5 kb in size, located at the MHC region on chromosome 6p21.33, ~46.4 kb centromeric to HLA-B gene. Complex linkage disequilibrium (LD) pattern extends across MICA and multiple HLA genes within the MHC region (1). MICA gene has six exons separated by five introns. The domain structure of MICA, similar to those of classical class I molecules, consists of three extracellular domains, namely α1 (encoded by exon 2), α2 (encoded by exon 3) and α3 (encoded by exon 4), a transmembrane domain (encoded by exon 5) and a hydrophobic cytoplasmic tail (encoded by exon 6) ( Figure 1). The top surface of the MICA α1-α2 domains has been found to interact directly with NKG2D (30,31 According to the nomenclature of international ImMunoGeneTics information system (IMGT)/HLA Database (32), 100 sequence alleles have been so far identified for MICA as defined by the combination of the sequence polymorphism of the coding region of the mature protein (http://hla.alleles.org/alleles/classo.html).
Several variants in MICA gene have gained more interest since their discovery. The transmembrane domain of MICA encoded by exon 5 harbors a variable number of short tandem repeat (STR) polymorphism consisting of four, five, six or nine GCT repeats, which encode four, five, six or nine alanine (Ala) residues (alleles designated A4, A5, A6 or A9, respectively). Additionally, the A5.1 allele (rs67841474) contains an extra guanine (G) insertion after two GCT triplets that causes a frameshift mutation resulting in a premature stop codon that, in turn, truncates 10 amino acids of the transmembrane domain as well as the hydrophobic cytoplasmic tail (33). A5.1 is most commonly found in the MICA*008 allele. Suemizu et al. found that the cytoplasmic tail-deleted MICA-A5.1 gene product was aberrantly transported to the apical surface of human intestinal epithelial cells instead of the basolateral surface where the interaction with intraepithelial T and NK lymphocytes takes place (34). Thus, MICA-A5.1 carriers may have an aberrant immunological surveillance by NK and T cells. Meanwhile, in contrast to other MICA alleles that are shed as truncated soluble species after proteolysis by metalloproteinases, the protein translated from the MICA-A5.1 allele is released from cells as a membrane-anchored full-length molecule in exosomes due to the lack of the two cysteines required for proteolytic shedding. Incubation of NK cells with the MICA-A5.1 (MICA*008) containing supernatant triggers significantly more NKG2D downregulation than the MICA*019 culture supernatant. Strikingly, incubation with exosomes containing MICA-A5.1 (MICA*008) also impairs NK cell cytotoxicity (35). MICA-A4 is in high LD with the amino acid substitution of glycine (Gly) by tryptophan (Trp) at position 14 in the α1 domain and MICA-A5 is in high LD with the amino acid substitution of Gly by serine (Ser) at position 175 in the α2 domain of MICA in the IMGT/ HLA database. Further studies are warranted to determine whether these amino acid changes could affect the binding affinity of MICA to the NKG2D receptor. Another interesting variant is a substitution of valine (Val) by methionine (Met) at position 129 in the α2 domain (rs1051792) encoded by exon 3. MICA carrying 129Met (A allele) was found to have much higher binding affinity to NKG2D compared with 129Val (G allele) (31). Varying affinities of MICA alleles for NKG2D may affect thresholds of NK cell triggering and T-cell modulation.

MICA polymorphism and cervical neoplasia
Cervical cancer and its precursor lesions, cervical intraepithelial neoplasia (CIN), are caused by persistent infection with high-risk human papillomavirus, where CIN III is considered the same as carcinoma in situ (36). Cervical cancer was among the first malignant diseases investigated with regard to association with MICA polymorphism. As shown in Table I, no significant association was found for MICA STR alleles in the transmembrane domain and amino acid substitutions in the extracellular domains with regard to risk of CIN or cervical cancer in early candidate gene studies (37)(38)(39)(40). However, all these studies are limited by the small sample size. In a recent genome-wide association study (GWAS) of cervical neoplasia in the Swedish population, rs2516448 was identified to be significantly associated with risk of cervical neoplasia, which is in perfect LD with MICA-A5.1 (D′ = 1.0, r 2 = 1.0), independently of previously known associations with classical HLA alleles (41). The effect of this variant was also replicated in two independent case-control series (41,42). The functional analysis shows that cervical neoplasia patients carrying the A5.1 allele have less membrane-bound MICA in their lesions (41), which may compromise their ability to alert the immune system of human papillomavirus infection or neoplastic change, leading to impaired immune activation and increased risk of tumor development. In a subsequent study, protective effects of the MICA-A4 and MICA-A5 alleles were also identified in another independent study of the Swedish population (42). The associations with these variants are unlikely to be driven by the nearby HLA alleles. Nevertheless, these findings need to be validated in other ethnic populations. Association between MICA allele 184 and cervical squamous cell carcinoma (SCC) was also reported in the population of the northern Netherlands, but detailed information of this allele has been unknown (43).

MICA polymorphism and hepatocellular carcinoma
Chronic infection with hepatitis B virus (HBV) and/or hepatitis C virus (HCV) are the main causes of hepatocellular carcinoma (HCC). As shown in Table II, Jiang et al. first showed that MICA-A5.1 was significantly associated with increased risk of HCC in the South Han Chinese population. They also identified that MICA-A5.1 polymorphism was associated with higher serum levels of sMICA in 141 HCC patients and may contribute to the development of HCC by releasing sMICA to evade immunosurveillance (44). In the same year, a GWAS identified A allele of SNP rs2596542 in the 5′ flanking region of MICA to be strongly associated with increased risk of       HCV-induced HCC in the Japanese population, independently of HLA-tagging SNPs (Table II). Subsequent analyses indicated that this SNP is not associated with chronic hepatitis C susceptibility but is significantly associated with progression from chronic hepatitis C to HCC. They also found that the protective allele G of rs2596542 was associated with higher sMICA protein levels in individuals with HCV-induced HCC. Based on an earlier finding that the level of sMICA was proportional to the level of membrane-bound MICA, they speculated that the individuals who carry rs2596542 A allele would express low levels of membrane-bound MICA in response to HCV infection, which thus leads to poor or no activation of NK cells and CD8 + T cells against virus-infected cells. They further genotyped 673 cases with HCV-induced HCC and 890 non-HCV controls for the MICA STR locus and found that alleles A9 and A6 were associated with higher risk of HCC, whereas the A5 and A5.1 alleles had a protective effect. However, no significant association was found between the STR alleles and sMICA levels among 665 HCV-induced HCC patients (22). In a subsequent study by the same group, a functional SNP rs2596538 located at 2.8 kb upstream of the MICA gene which is in high LD with rs2596542 (D′ = 0.953, r 2 = 0.832) was identified to regulate the MICA expression. The protective G allele exhibited a higher binding affinity to the transcription factor Specificity Protein 1 (SP1) and a higher transcriptional activity compared with the risk A allele. Moreover, the functional SNP rs2596538 showed stronger association with HCV-induced HCC than the previously identified SNP rs2596542. They also found significantly higher serum level of sMICA in HCV-induced HCC patients carrying the G allele than those carrying the A allele (45).
Two following studies investigated the relationship between SNP rs2596542 and risk of HBV-induced HCC, with opposite results (Table II). One study observed that the G allele was associated with increased risk of HBV-induced HCC in a Japanese population. Concordant with their previous report for HCV-induced HCC patients, no significant association was found between the MICA STR locus and sMICA level in 111 HBV-induced HCC patients (23). The other study found A allele to be the risk allele in a Vietnamese population (24). Significant difference in allele frequency of rs2596542 is observed between these two populations (Table II). These conflicting results warrant further validation in larger and multiethnic populations. However, concordant with earlier report for HCV-induced HCC patients, both studies found that the G allele of rs2596542 was associated with higher serum sMICA levels in HBV-induced HCC patients. Furthermore, MICA-129Met and MICA-251Gln allele were observed at a significantly higher rate in the HBV-induced HCC group compared with liver cirrhosis and non-HCC cases (patients with chronic hepatitis B, liver cirrhosis or asymptomatic HBV carriers) in the Vietnamese population (24).

MICA polymorphism and oral SCC
Oral squamous cell carcinoma (OSCC) is a solid tumor originating from epithelial cells. HLA has been implicated in the development of OSCC (46)(47)(48). The association between the MICA STR polymorphism and risk of OSCC has been investigated in three populations, with conflicting results. As shown in Table III, the first case-control study found that MICA-A6 in subjects with OSCC was significantly higher than that in controls in a Taiwanese population (49). In another study of the Dutch population, MICA-A9 was found to be significantly associated with decreased risk of OSCC but not with SCC in the hypopharynx, larynx or oropharynx (50). On the other hand, Tamaki et al. reported that MICA-A5.1 allele was significantly higher in patients with OSCC compared with normal controls in two studies of Japanese population (25,51). However, the subjects in these two studies are overlapped and no correction for multiple testing was performed. Nevertheless, OSCC patients with the homozygous A5.1 genotype were found to higher levels of sMICA and lower survival rate in the Japanese study. Because of the small sample size in each study and conflicting results, no firm conclusions can be drawn for the role of MICA STR polymorphism in the susceptibility for OSCC. In addition, no association was found between MICA sequence alleles and risk of OSCC in a small study of the Dutch population (48).

MICA polymorphism and breast cancer
Two studies examined the MICA STR polymorphism in relation to risk of breast cancer. One study reported that MICA-A5 allele was reduced in breast cancer patients compared with healthy controls in a Spanish population. Given the association between the HLA-B7 allele and the susceptibility to breast cancer, they also found that the MICA-A5 and MICA-A5.1 alleles significantly decrease and increase the risk of breast cancer in HLA-B7 patients, respectively (52). The other study in the Iranian population showed that MICA-A4 and A5.1 play a role in reducing the risk of breast cancer, whereas MICA-A6 increased risk for developing breast cancer (53). However, no correction for multiple testing was performed in the latter study (Table  III).

MICA polymorphism and nasopharyngeal carcinoma
Nasopharyngeal carcinoma (NPC) is an Epstein-Barr virus-associated epithelial malignancy. In patients with NPC from Morocco and a few Asian countries, associations with several HLA alleles and/or haplotypes have been reported (54)(55)(56)(57). In a case-control study, Tian et al. found that MICA-A9 and A5.1 were associated with increased and decreased risk of male but not female NPC in southern Chinese population, respectively (58). In a subsequent study, they enlarged the sample size and found similar results (59). This constitutes the first demonstration of a gender-specific association between MICA STR polymorphism and NPC and the authors suggested that this could be largely attributable to the underlying gender-related mechanisms that modulate MICA gene expression. In another study of a Tunisian population, MICA-129Val allele was found more frequent in the NPC patient group than in the healthy controls, although it failed to reach significance after correction for multiple testing. Nevertheless, homozygosity for MICA-129Val was significantly more prevalent in NPC patients as compared with healthy controls, suggesting that MICA-129Val behaves as a recessive allele for NPC susceptibility (Table I) (60).

MICA polymorphism and colorectal cancer
As shown in Table III, the association of MICA STR polymorphism with colorectal cancer has been addressed in studies of both the German and Chinese populations, respectively, with negative results (29,61). In contrast, the frequency of the MICA-129Met was significantly decreased in the patient group in the Chinese population. But this result needs to be interpreted with caution as no correction was performed for multiple testing in this study.

MICA polymorphism and skin cancer
HLA molecules have been implicated in the development of both melanoma and non-melanoma skin cancer (62,63). The relationship between MICA STR polymorphism and risk of melanoma and nonmelanoma skin cancer has been studied in the Dutch population, but no significant associations were observed (64,65) (Table III).

MICA polymorphism and other cancer types
HLA allele associations with leukemia and gastric cancer have been reported previously (66)(67)(68)(69)(70)(71). Only one study has assessed the potential contribution of MICA polymorphism toward the pathogenesis of leukemia in a Southern Han Chinese population. They found that MICA-A5 and MICA*019 were associated with increased risk of leukemia, whereas MICA-A5.1 and MICA*008, which are in perfect LD with each other, were associated with decreased risk of leukemia after correction for multiple testing (72). On the other hand, MICA-A9 was  found to be associated with increased risk of gastric cancer in a casecontrol study of Taiwanese population (73) (Table III).

Concluding remarks and perspective
GWASs have emerged as an important tool for discovering regions of the genome that harbor genetic variants that confer risk for different types of cancers. Compared with the older candidate gene approach, the GWAS approach investigates not only the region around candidate genes with known or predicted role in disease but across the entire genome using a SNP array, which simultaneously genotypes hundreds of thousands to millions of marker SNPs. This approach has provided the opportunity to scan across the genome in a sufficiently large set of cases and controls without a set of prior hypotheses in search of susceptibility alleles with low effect sizes. The association in the GWAS and the independent replication as well as functional analysis provided firm evidence for the frameshift mutation A5.1 of MICA as a causal variant for cervical neoplasia in the Swedish population. As the number of invasive cervical cancer cases included in these two studies is small, the role of MICA-A5.1 in invasive cervical cancer warrants further investigation in both European and other ethnic groups. On the other hand, another GWAS provided strong evidence for the role of rs2596542 in HCV-induced HCC and a functional variant rs2596538 in strong LD with it was found to have a stronger effect in the Japanese population.
Based on the hypothesis that the mechanism of action for causal variants is shared across human populations (74), cross-population studies are warranted to confirm the causal variant at this locus in MICA for HCVinduced HCC. Furthermore, given the opposite effects of membranebound MICA and sMICA in immune surveillance against infection and tumorigenesis, it is important in future studies to directly evaluate the correlation between MICA alleles and level of both membrane-bound MICA and sMICA in patients of HCV-related HCC in order to better understand the underlying mechanism. It is worth noting that the current commercial arrays used for GWA scans are designed based on HapMap which contains only ~30% of the common SNPs that are present in the genome and these arrays also have limited potential to capture rare and low frequency variants (i.e. those with a minor allele frequency <5%) (75). Once there is evidence that a particular variant in a given gene is associated with risk of certain cancer, the probability that other functional variants in the same gene also modify cancer risk is markedly enhanced. Therefore, deep sequencing in the MICA gene and flanking region will be required to allow for the discovery of rare causal variants in MICA affecting risk of cervical cancer and HCV-induced HCC. At present, the evidence is less convincing for a primary MICA allele association with other cancer types. No GWASs have identified MICA as a susceptibility gene for other cancer types. However, we cannot exclude the possibility of involvement of MICA in these cancers as causal variants with moderate effects could be missed due to incomplete coverage of the SNP arrays and the stringent significance threshold of GWASs. On the other hand, previous candidate gene studies of MICA polymorphism in relation to susceptibility for these cancer types showed lack of consistence. Several observations of these studies need to be noted. First, the sample size of majority of the original candidate gene studies is small, and they do not have sufficient power to detect the moderate effects of loci involved in cancer. Second, the hypothesis-driven candidate gene studies performed generally had poor coverage of genetic variants. Most case-control studies have focused on polymorphisms of the STR in exon 5. Far fewer cancer-association studies have attempted to analyze the polymorphisms in the regulatory region and those in the more complex exons 2-4, which encode the external domains of the molecule containing the recognition motifs for NKD2D-bearing cells. The low statistical power and non-uniform SNP coverage both contribute to false negatives (i.e. true susceptibility loci with non-significant candidate gene findings). Third, the majority of the variants in one cancer have only been examined in one case-control study from one population, which makes it impossible to elucidate the level of false positives or false negatives finding. When more than one ethnic group has been examined, conflicting results were observed. Given the small effect size of most genetic variants involved in cancer and the distorting potential of a publication bias in favor of positive associations, a single positive finding cannot be viewed as sufficient evidence. Fourth, HLA alleles have been implicated in several cancer types, but most studies reported association analyses without controlling for LD of MICA with HLA alleles. This makes a conclusion about an independent association of MICA difficult. Furthermore, for the association of low penetrance genetic variants and cancer risk, heterogeneity could also come from variations in study design, differences in biological effect between populations due to modification of environmental factors, lack of adjustment for known risk factors and issues related to multiple significance testing.
Therefore, future studies require larger sample size, a more careful study design, study execution and data analysis. First, it is beneficial to form multicenter research consortia for not only increasing study sample sizes, but also for ensuring data quality and generalizing the findings and conclusions. Second, given the biological plausibility of MICA polymorphism affecting cancer risk and limited coverage of variants in GWAS, association studies based on deep sequencing of the MICA region may be more appropriate than GWAS. Third, given the LD between MICA and multiple HLA genes within the MHC region, more refined cancer studies on MICA polymorphism, independent of HLA association, need to be undertaken to confirm its role in pathogenesis, in particular, tumorigenesis. Fourth, most of the cancers are highly affected by environmental factors, including infections. The interaction between MICA polymorphism and environmental factors could be important for the development of different cancers. Future studies investigating the interaction between MICA polymorphism and environmental factors in different cancer types are needed in order to address this issue in more detail. Last but not the least, in order to further our understanding of the role of MICA polymorphisms in different cancers, it is imperative that association studies, whether based on a hypothesis-free or candidate gene design, include a replication in an independent case-control cohort from the same population and a characterization of the biological function for candidate variants.