Association of killer cell immunoglobulin-like receptor genotypes with microscopic polyangiitis




Genetic background and infection have been implicated in the etiology of microscopic polyangiitis (MPA). Killer cell immunoglobulin-like receptors (KIRs) are a diverse family of activating and inhibitory receptors expressed on natural killer (NK) cells and T cells, the genes of which show extreme polymorphism. Some KIRs bind to HLA class I subgroups, and genetic interactions between KIR genes and their ligand HLA have been shown to be associated with several autoimmune and viral diseases. In this study, we examined possible associations of the presence or absence of KIR loci with a genetic predisposition to MPA.


The presence or absence of 14 KIR loci was determined in 57 myeloperoxidase antineutrophil cytoplasmic antibody–positive Japanese subjects (43 patients with MPA and 239 healthy controls).


The carrier frequency of activating KIR2DS3 was significantly decreased among patients with MPA compared with healthy controls (4.7% versus 16.7%; P = 0.038, odds ratio [OR] 0.24, 95% confidence interval [95% CI] 0.06–0.94). When KIRs were analyzed in combination with their HLA ligands, the proportion of individuals carrying inhibitory KIR3DL1 and HLA–Bw4 but not activating receptor KIR3DS1, a combination presumed to be the most inhibitory of all KIR3DS1/3DL1/HLA–B combinations, was significantly increased in the MPA group compared with the control group (46.5% versus 27.0%; P = 0.014, OR 2.35, 95% CI 1.18–4.70). Furthermore, when subjects were classified according to KIR3DL1/3DS1/HLA–B and KIR2DL1/ HLA–C combinations, an increasing trend toward susceptibility was observed as combinations became more inhibitory.


The decreased activation potential of NK and/or T cells associated with KIR/HLA genotypes may predispose to MPA, possibly through insufficient resistance against infections.

Although the etiology of antineutrophil cytoplasmic antibody (ANCA)–associated vasculitis (AAV) is poorly understood, differences in the prevalence or incidence among different populations and sporadic reports of multiplex families suggest a role of genetic predisposition. Indeed, association of several genes with AAV has been previously reported (1). Bacterial and viral infections have also been implicated as environmental factors, based on clinical observations and animal models (1–4).

We previously conducted a multicenter study of the genetic background of AAV in Japan and demonstrated significant association of the HLA–DRB1*0901;DQB1*0303 haplotype with microscopic polyangiitis (MPA) (5, 6). A higher incidence of MPA among Japanese patients with AAV may partly be ascribed to such a genetic background, because this haplotype is common in Asians but is rare in other populations.

Killer cell immunoglobulin-like receptors (KIRs) are a diverse family of activating and inhibitory receptors expressed on human natural killer (NK) cells and a subset of T cells. The KIR gene cluster is located on human chromosome 19q13.4 and shows extreme polymorphism in gene content, namely, the presence or absence of each locus (7, 8). In fact, no fewer than 100 KIR genotype profiles (the presence or absence of each gene in a given individual) have been reported to date (7, 8).

HLA class I molecules have been identified to be ligands for some of the inhibitory KIR molecules (7, 8). KIR2DL1 binds to HLA–C group 2 molecules, which have amino acids Asn77 and Lys80, KIR2DL2/2DL3 binds to HLA–C group 1 molecules, which have amino acids Ser77 and Asn80, and KIR3DL1 binds to HLA–B allele products, which contain the Bw4 epitope determined by amino acid positions 77–83. The ligands for activating KIRs are less well defined, although it has been suggested that they recognize the same HLA–B or HLA–C molecules as recognized by their related inhibitory KIRs. Because both HLA and KIR genes are polymorphic and are encoded on different chromosomes, an individual may have both KIRs and the corresponding HLA ligands, may have only KIRs and no corresponding HLA ligands, or may have only the HLA ligand but no KIR for a certain HLA–KIR interaction.

Such diversity of KIR and HLA genes among individuals might be related to the heterogeneity of the immune response to infectious agents or susceptibility to autoimmune or inflammatory diseases (for review, see ref. 8). Indeed, a higher frequency of KIR2DS2 was observed in patients who had both rheumatoid arthritis and vasculitis. Psoriatic arthritis (KIR2DS1 and/or 2DS2), type 1 diabetes mellitus (KIR2DS2), psoriasis vulgaris (KIR2DS1/2DL5, haplotype B), and scleroderma (KIR2DS2+/2DL2−) have also been shown to be associated with activating KIRs. With respect to viral infections, KIR3DS1 was associated with delayed progression of human immunodeficiency virus (HIV) infection in the presence of HLA–Bw4, and KIR2DL3 was associated with the clearance of hepatitis C virus (HCV) in the presence of HLA–C group 1, both of which are presumed to be less-inhibitory combinations.

In view of the association of KIRs with autoimmune, inflammatory, and infectious diseases, we considered that KIR polymorphisms might be associated with susceptibility to AAV. In this study, we analyzed Japanese patients with AAV for the association of KIR genotypes in combination with HLA–B and HLA–C ligands. Due to the rarity of proteinase 3 ANCA–positive AAV in Japan (5), the analysis was limited to myeloperoxidase (MPO) ANCA–positive patients.


Patients and controls.

Patients with vasculitis who were positive for MPO ANCA were recruited from 15 clinical centers in central Japan. The classification of vasculitides was based on the Chapel Hill Consensus Conference definitions (9). Because those definitions do not provide diagnostic or classification criteria, and because the American College of Rheumatology (ACR) classification criteria do not include MPA, the diagnosis of MPA was made according to the criteria proposed by the Research Committee on Intractable Vasculitides (Ministry of Health, Labor, and Welfare of Japan) in 1998 (5). The diagnoses of Wegener's granulomatosis (WG), Churg-Strauss syndrome (CSS), and classic polyarteritis nodosa (PAN) were based on the ACR criteria (10–12). In most cases, the diagnosis was confirmed by biopsy. MPO ANCA was measured by an enzyme-linked immunosorbent assay (Nissho, Osaka, Japan), using MPO extracted from human neutrophil cytoplasmic α granules (Wieslab, Lund, Sweden).

The study group comprised 57 patients positive for MPO ANCA, including 43 with MPA (19 men and 24 women, mean ± SD age 66.0 ± 11.5 years), 8 with CSS (6 men and 2 women, mean ± SD age 62.1 ± 10.1 years), 3 with WG (1 man and 2 women, mean ± SD age 43.7 ± 27.1 years), and 3 male patients with classic PAN (mean ± SD age 62.0 ± 18.5 years), as well as 239 healthy controls (139 men and 100 women, mean ± SD age 32.6 ± 9.8 years) recruited from the Tokyo area. These patients are part of the same group described in our previous reports (5, 6) and were consecutively recruited at each of the participating institutions, usually within several months after the first presentation of symptoms. All patients, except 3 patients with MPA and 1 with classic PAN, were receiving treatment with corticosteroids. Twenty patients with MPA, 6 with CSS, and 2 with WG were being treated with cyclophosphamide, and 1 patient with classic PAN was being treated with methotrexate. This study was reviewed and approved by the ethics committees of the participating institutions.


Genotyping was performed using genomic DNA isolated from peripheral blood leukocytes. KIR locus typing was performed to detect the presence or absence of a total of 14 KIR loci (KIR2DL1, 2DL2, 2DL3, 2DL4, 2DL5, 2DS1, 2DS2, 2DS3, 2DS4, 2DS5, 3DL1, 3DL2, 3DL3, and 3DS1), using sequence-specific primer–polymerase chain reaction (PCR) (13) with minor modifications. Briefly, PCR was performed in 10-μl reaction mixtures containing 100 ng of genomic DNA, 0.1–2.5 μM of each primer, 0.2 mM dNTPs, 1.5 mM MgCl2, and 0.5 units of AmpliTaq DNA polymerase (Takara, Otsu, Japan), using a GeneAmp PCR System 9700 Thermocycler (Applied Biosystems, Foster City, CA). The PCR conditions were as follows: initial denaturation at 95°C for 2 minutes, 10 cycles of denaturation at 94°C for 20 seconds, annealing at 64°C for 10 seconds, and extension at 72°C for 90 seconds, followed by 25 cycles (for KIR2DL1, 2DL3, 2DS3, and 3DL3) or 20 cycles (for other loci) of denaturation at 94°C for 20 seconds, annealing at 61°C for 45 seconds, and extension at 72°C for 90 seconds, then a final extension step at 72°C for 5 minutes. HLA–B and HLA–C genotypes were determined by reverse sequence-specific oligonucleotide probing (LABType SSO; One Lambda, Canoga Park, CA).

Statistical analysis.

The chi-square test or Fisher's exact test was performed to analyze disease association, using StatView-J5.0 software (SAS Institute, Cary, NC). Fisher's exact test was applied when 1 or more variables in the contingency table were <5. The test for trend was performed as previously described (14).


Carrier frequency of each KIR locus.

The carrier frequency of each KIR locus among patients and controls is shown in Table 1. The carrier frequency of activating KIR2DS3 was significantly decreased in patients with MPA as compared with healthy controls (4.7% versus 16.7%; P = 0.038, odds ratio [OR] 0.24, 95% confidence interval [95% CI] 0.06–0.94). The difference in the carrier frequency of KIR2DS3 between MPO ANCA–positive patients as a whole and controls did not reach statistical significance. Significant differences for other loci were not observed.

Table 1. Carrier frequencies of KIR loci in Japanese patients with vasculitis and healthy controls*
KIRMPA (n = 43)MPO ANCA+ (n = 57)Controls (n = 239)
  • *

    Values are the number (%). KIR = killer cell immunoglobulin-like receptor; MPA = microscopic polyangiitis; MPO = myeloperoxidase; ANCA = antinuclear cytoplasmic antibody.

  • P = 0.038 versus controls (odds ratio 0.24, 95% confidence interval 0.06–0.94).

2DL141 (95.3)54 (94.7)238 (99.6)
2DL28 (18.6)10 (17.5)33 (13.8)
2DL342 (97.7)56 (98.2)239 (100.0)
2DL443 (100.0)57 (100.0)239 (100.0)
2DL516 (37.2)21 (36.8)116 (48.5)
2DS115 (34.9)20 (35.1)109 (45.6)
2DS28 (18.6)10 (17.5)35 (14.6)
2DS32 (4.7)4 (7.0)40 (16.7)
2DS436 (83.7)48 (84.2)208 (87.0)
2DS514 (32.6)17 (29.8)77 (32.0)
3DL143 (100.0)57 (100.0)227 (95.0)
3DL242 (97.7)55 (96.5)237 (99.2)
3DL343 (100.0)57 (100.0)239 (100.0)
3DS115 (34.9)20 (35.1)110 (46.0)

KIR genotype profile.

We next compared KIR genotype profiles determined by the presence or absence of each KIR locus in a given individual. A total of 34 profiles were detected (Table 2). Profiles observed in only 1 individual (among either patients or controls) were combined and listed as “others.” Although the statistical significance of the difference was marginal, profile 3, which contained activating KIR2DS3 and KIR3DS1, was absent in patients with MPA (P = 0.053).

Table 2. Killer cell immunoglobulin-like receptor (KIR) profile frequencies in patients and controls*
KIR profile2DL12DL22DL32DL42DL52DS12DS22DS32DS42DS53DL13DL23DL33DS1MPA (n = 43)MPO ANCA+ (n = 57)Controls (n = 239)
  • *

    Values are the number (%). The myeloperoxidase antinuclear cytoplasmic antibody–positive (MPO ANCA+) group included all patients with microscopic polyangiitis (MPA) and 14 other patients positive for MPO ANCA. Profiles observed in only 1 subject were combined (Others).

  • P = 0.053 versus controls.

1+++++++20 (46.5)26 (45.6)94 (39.3)
2+++++++++++9 (20.9)12 (21.1)43 (18.0)
3+++++++++++0 (0)1 (1.8)21 (8.8)
4+++++++++5 (11.6)7 (12.3)13 (5.4)
5++++++++++2 (4.7)2 (3.5)8 (3.3)
6++++++1 (2.3)1 (1.8)7 (2.9)
7+++++++++0 (0)0 (0)6 (2.5)
8+++++++++++++0 (0)0 (0)5 (2.1)
9++++++++++++0 (0)0 (0)5 (2.1)
10++++++++0 (0)0 (0)4 (1.7)
11+++++++++++++0 (0)0 (0)3 (1.3)
12++++++++++0 (0)0 (0)3 (1.3)
13+++++++++++1 (2.3)1 (1.8)2 (0.8)
14++++++++++++0 (0)0 (0)2 (0.8)
15++++++++++1 (2.3)2 (3.5)2 (0.8)
16++++++++++0 (0)0 (0)2 (0.8)
17+++++++++0 (0)0 (0)2 (0.8)
Others              4 (9.3)5 (8.8)17 (7.1)

KIR/HLA combinations.

We next analyzed combinations of activating/inhibitory KIRs and their HLA ligands for possible association with vasculitis. HLA–B and HLA–C genotypes were determined for all patients. Among the controls, HLA–B and HLA–C genotypes previously determined for 153 and 142 individuals, respectively, were used. Three combinations, KIR2DL1/2DS1/HLA–C2, KIR2DL2/2DL3/2DS2/HLA–C1, and KIR3DL1/3DS1/HLA–Bw4, were analyzed. Activating KIRs that showed high homology with the inhibitory counterparts were included in the analyses.

The frequency of subjects who carried the inhibitory KIR3DL1 receptor and its ligand HLA–Bw4 but who did not possess activating KIR3DS1 was significantly increased among patients with MPA compared with controls (P = 0.014, OR 2.35, 95% CI 1.18–4.70) (Table 3). This association did not reach statistical significance when all of the MPO ANCA–positive patients were compared with controls. This genotype combination is presumed to have the highest inhibitory effect of all 6 combinations that involve HLA–Bw4/KIR interactions. In contrast, analyses for HLA–C/KIR combinations did not reveal statistically significant association (Tables 4 and 5).

Table 3. KIR3DL1, KIR3DS1, HLA–Bw4 genotype combinations in patients and controls*
HLA–Bw4KIRMPA (n = 43)MPO ANCA+ (n = 55)Controls (n = 163)
  • *

    Values are the number (%) of subjects. KIR = killer cell immunoglobulin-like receptor; MPA = microscopic polyangiitis; MPO = myeloperoxidase; ANCA = antinuclear cytoplasmic antibody.

  • HLA–B genotyping was unsuccessful in 2 MPO ANCA–positive patients who did not have MPA.

  • P = 0.014 versus controls, by chi-square test (odds ratio 2.35, 95% confidence interval 1.18–4.70).

+3DL1+/3DS1−20 (46.5)21 (38.2)44 (27.0)
3DL1+/3DS1−8 (18.6)14 (25.5)39 (23.9)
+3DL1+/3DS1+6 (14.0)8 (14.5)38 (23.3)
3DL1+/3DS1+9 (20.9)12 (21.8)33 (20.2)
+3DL1−/3DS1+0 (0.0)0 (0.0)7 (4.3)
3DL1−/3DS1+0 (0.0)0 (0.0)2 (1.2)
Table 4. KIR2DL1, KIR2DS1, HLA–C group 2 genotype combinations in patients and controls*
HLA–C group 2KIRMPA (n = 43)MPO ANCA+ (n = 57)Controls (n = 142)
  • *

    Values are the number (%). Statistically significant differences were not observed. See Table 3 for definitions.

+2DL1+/2DS1−5 (11.6)7 (12.3)13 (9.2)
2DL1+/2DS1−22 (51.2)29 (50.9)64 (45.1)
+2DL1+/2DS1+2 (4.7)3 (5.3)9 (6.3)
2DL1+/2DS1+12 (27.9)16 (28.1)56 (39.4)
+2DL1−/2DS1+0 (0.0)0 (0.0)0 (0.0)
2DL1−/2DS1+1 (2.3)1 (1.8)0 (0.0)
2DL1−/2DS1−1 (2.3)1 (1.8)0 (0.0)
Table 5. KIR2DL2, KIR2DL3, KIR2DS2, and HLA–C group 1 genotype combinations in patients and controls*
HLA–C group 1KIRMPA (n = 43)MPO ANCA+ (n = 57)Controls (n = 142)
  • *

    Values are the number (%). Statistically significant differences were not observed. See Table 3 for definitions.

+2DL2−/2DL3+/2DS2−35 (81.4)47 (82.5)123 (86.6)
2DL2−/2DL3+/2DS2−0 (0.0)0 (0.0)1 (0.7)
+2DL2+/2DL3+/2DS2+7 (16.3)9 (15.8)18 (12.7)
2DL2+/2DL3+/2DS2+0 (0.0)0 (0.0)0 (0.0)
+2DL2−/2DL3+/2DS2+0 (0.0)0 (0.0)0 (0.0)
2DL2−/2DL3+/2DS2+0 (0.0)0 (0.0)0 (0.0)
+2DL2+/2DL3−/2DS2+1 (2.3)1 (1.8)0 (0.0)

We further addressed this issue based on a model proposed by Carrington et al (15), in which a gradient of activation/inhibitory potential derived from the combination of KIR and HLA ligand genes is taken into account. Thus, patients and controls were classified into 8 subgroups (a–h in Figure 1) according to combinations of the presence or absence of KIR3DS1/KIR3DL1 and its ligand HLA–Bw4, as well as KIR2DL1 and its ligand HLA–C group 2. Next, the subgroups were categorized into 3 large groups (groups I, II, and III) for the purpose of a test for trend. Groups I and III represented the most activating and inhibitory groups, respectively. As shown in Figure 1, an increasing trend for susceptibility was observed as combinations became more inhibitory, although the statistical significance of these differences was marginal (P for trend = 0.056).

Figure 1.

Effect of combined killer cell immunoglobulin-like receptor (KIR) and HLA ligand genotypes on susceptibility to microscopic polyangiitis (MPA). Subjects were classified into 8 subgroups (a–h) according to the presence of activating KIR3DS1 and KIR3DL1/HLA–Bw4 or KIR2DL1/HLA–C group 2 combinations that transmit inhibitory signals. The subgroups were then grouped into 3 large groups according to the presumed strength of activating or inhibitory signals. The trend for susceptibility increased as combinations became more inhibitory.


To our knowledge, studies on an association between KIR genotypes and susceptibility to AAV have not previously been reported. In the present study, we demonstrated a decrease in the carrier frequency of KIR2DS3 and an increase in the KIR3DL1+/KIR3DS1− genotype in the presence of the HLA–Bw4 ligand among patients with MPA. The latter combination is presumed to be the most inhibitory combination of these genes. Both of the detected associations indicate that KIR and HLA genotypes that favor NK and/or T cell inhibition are associated with MPA. An increased risk for MPA in association with the inhibitory genotype combinations was also supported by more global analyses (Tables 1 and 2 and Figure 1), although the associations were marginal.

Several studies on autoimmune or inflammatory diseases, including psoriasis, psoriatic arthritis, type 1 diabetes mellitus, rheumatoid arthritis accompanied by vasculitis, and scleroderma, unequivocally demonstrated association of “less inhibitory” KIR2DS and 2DL profiles, either alone or in combination, with HLA–C groups (for review, see ref. 8). In contrast, in the setting of HIV and HCV infection, less-inhibitory combinations have been shown to be associated with resistance (8). In the case of human papillomavirus, more-inhibitory combinations are associated with protection against progression to cervical cancer (15). The latter unexpected observation may be explained by the possibility that a chronic inflammatory environment may facilitate neoplastic transformation.

Taken together with these previous observations on KIR genotypes and anecdotal reports that suggested a role of viral infection on AAV (2–4), our present observations seem to imply that defective clearance of viral infections by NK cells and/or T cells might be involved in the pathogenesis of MPA. Because MPA generally affects older individuals, it is hypothesized that NK cell and/or T cell functions in subjects with more-inhibitory KIR and HLA genotype combinations become further compromised with aging, to a point that renders them susceptible to MPA. Such a hypothesis should be examined by studies on the expression or function of NK and T cells, as well as the contribution of viral infections in patients with MPA.

In contrast to the association of HLA–DRB1*0901 (6), KIR/HLA class I genotypes appeared to be more strongly associated with MPA than with MPO ANCA positivity as a whole. This raises the possibility that KIR/HLA combinations, possibly through ineffective clearance of infectious agents, may represent one of the pathways, more relevant to MPA than to other forms of vasculitis, that triggers the production of pathogenic MPO ANCA, which could be facilitated by the HLA–DRB1*0901;DQB1*0303 haplotype (5, 6). However, this issue needs to be addressed by future studies with a larger number of MPO ANCA–positive patients.

Due to the rarity of MPA, the number of the patients analyzed in this study was small. Indeed, this study and our previous studies (5, 6) examined the largest number of patients with AAV among all genetics studies thus far reported in Japan. The marginal P values observed in this study could be at least partly ascribed to the small number of subjects. Nevertheless, all analyses in this study point to the contribution of an inhibitory potential of KIR, which is unlikely to be a mere coincidence. Further independent studies are required to validate the role of KIR in MPA.


We are indebted to the patients and healthy donors who participated in this study, to the doctors who recruited the patients, and to Aya Kawasaki (Department of Human Genetics, University of Tokyo) for technical assistance.