Report

HLA and NK Cell Inhibitory Receptor Genes in Resolving Hepatitis C Virus Infection

See allHide authors and affiliations

Science  06 Aug 2004:
Vol. 305, Issue 5685, pp. 872-874
DOI: 10.1126/science.1097670

Abstract

Natural killer (NK) cells provide a central defense against viral infection by using inhibitory and activation receptors for major histocompatibility complex class I molecules as a means of controlling their activity. We show that genes encoding the inhibitory NK cell receptor KIR2DL3 and its human leukocyte antigen C group1 (HLA-C1) ligand directly influence resolution of hepatitis C virus (HCV) infection. This effect was observed in Caucasians and African Americans with expected low infectious doses of HCV but not in those with high-dose exposure, in whom the innate immune response is likely overwhelmed. The data strongly suggest that inhibitory NK cell interactions are important in determining antiviral immunity and that diminished inhibitory responses confer protection against HCV.

Natural killer (NK) cells are key components of the innate antiviral immune response. In vivo, they are under the constitutively dominant influence of inhibitory receptors for self-MHC class I ligands (1, 2), such that effector functions occur only when activating signals overcome inhibitory signals (3, 4). The killer cell immunoglobulin-like receptors (KIR) represent a diverse family of activating and inhibitory receptors that are integral in this model. As with their MHC class I ligands, the population diversity and rapid evolution of the KIR genes strongly suggests that they are under pathogen-mediated selection (57).

KIR haplotypes vary in number and type of genes present, and because HLA and KIR map to separate chromosomes, some individuals lack specific KIR-HLA receptor-ligand pairings. To date, only activating KIR have been associated with disease outcome (810), whereas the influence of inhibitory KIR on disease is undetermined.

Hepatitis C virus (HCV) is a common infection worldwide, causing cirrhosis and hepatocellular carcinoma. About 20% of individuals resolve acute infection, an outcome associated with specific components of the adaptive immune system (11), including HLA class I (12). Because resolution of HCV infection may also involve the innate immune system, including NK cells (13, 14), we examined the possible synergistic influence that corresponding KIR-HLA combinations might have on the outcome of HCV infection.

Individuals who were exposed to HCV (685 with persistent and 352 with resolved infection) (table S1, A to C) were categorized according to their KIR-binding motifs based on HLA-B and -C genotyping data (15). Group 1 HLA-C (HLA-C1) allotypes have asparagine at residue 80 and are ligands for the inhibitory receptors KIR2DL2 and KIR2DL3, which segregate as alleles of a single locus (Table 1). The remaining HLA-C allotypes (group 2, HLA-C2) have lysine at position 80 and are ligands for KIR2DL1 (an inhibitory receptor) and KIR2DS1 (the homologous activating receptor). HLA-B Bw4 allotypes serve as ligands for KIR3DL1.

Table 1.

Frequency of KIR and HLA receptor-ligand pairings in the population studied, stratified by race, study site, and route of infection.

NKIR2DL1-HLA-C2 N (%) KIR2DL2-HLA-C1 N (%) KIR2DL3-HLA-C1 N (%) KIR2DS1-HLA-C2 N (%) KIR2DS2-HLA-C1View inlineN (%) KIR3DL1-HLA-Bw4 N (%) KIR3DS1-HLA-Bw4View inlineN (%)
All 1037 689 (66.4) 591 (57.0) 754 (72.7) RaceView inline 231 (22.3) 441 (42.5) 635 (61.2) 216 (20.8)
UK Caucasian 340 219 (64.4) 144 (42.4) 271 (79.7) 78 (22.9) 144 (42.3) 205 (60.1) 83 (24.4)
USA Caucasian 355 220 (62.0) 163 (45.9) 265 (74.6) 88 (24.8) 167 (47.0) 205 (57.7) 89 (25.1)
USA African-American 271 205 (75.6) 108 (39.9) 166 (61.3) 47 (17.3) 99 (36.5) 188 (69.4) 31 (11.4)
USA other 69 44 (63.8) 30 (43.5) 52 (75.4) Route 17 (24.6) 30 (43.5) 35 (50.7) 12 (17.4)
No blood products 543 372 (68.5) 229 (42.2) 382 (70.3) 115 (21.2) 222 (40.9) 344 (63.3) 105 (19.3)
Blood products 494 317 (64.2) 217 (43.9) 372 (75.3) 116 (23.5) 219 (44.3) 291 (59.0) 111 (22.5)
  • View inline* Receptor ligand pairing inferred by protein sequence but not directly demonstrated.

  • View inline Excludes two non-Caucasian individuals from the United Kingdom.

  • The frequency of individuals with two copies of HLA-C1 alleles (HLA-C1C1) was higher in the group that had resolved infection (37.5%) relative to those with persistent infection (29.9%) (P = 0.01; OR = 1.40). The reciprocal association of two HLA-C2 alleles (HLA-C2C2) with viral persistence (14.5% in resolved versus 20.2% in persistent infection) was also observed (P = 0.02; OR = 0.67) (Table 2). The frequency of HLA-B Bw4 alleles did not differ significantly between the groups.

    Table 2.

    HLA-C and KIR-HLA-C interactions are associated with resolution of HCV infection. Frequencies of HLA-C and KIR-HLA-C combinations among individuals with resolved and persistent HCV infection from all individuals combined are shown. HLA-C1C1 indicates two group 1 HLA-C alleles, HLA-C2C2 indicates two group 2 HLA-C alleles, and HLA-C1C2 indicates one of each. P values were calculated by using the chi-square test from two-by-two contingency tables; a positive odds ratio indicates a protective association with resolution of infection.

    Genetic factor Frequency resolved N (%) N = 348-352 Frequency persistent N (%) N = 681-685 OR 95% CI P
    HLA-C1C1 132 (37.5) 205 (29.9) 1.40 1.07-1.84 0.01
    HLA-C1C2 169 (48.0) 342 (49.9) 0.93 0.72-1.20 0.6
    HLA-C2C2 51 (14.5) 138 (20.2) 0.67 0.47-0.95 0.02
    2DL2 + HLA-C1C1 64 (18.2) 121 (17.7) 1.04 0.74-1.45 0.9
    2DL3 + HLA-C1C1 119 (33.9) 182 (26.7) 1.41 1.06-1.86 0.02
    2DS2 + HLA-C1C1 64 (18.2) 120 (17.5) 1.05 0.75-1.46 0.9
    2DL1 + HLA-C2C2 50 (14.2) 135 (19.7) 0.68 0.47-0.96 0.03
    2DS1 + HLA-C2C2 22 (6.3) 32 (4.7) 1.36 0.78-2.37 0.3
    3DS1 + HLA-Bw4 86 (24.7) 130 (19.1) 1.39 1.02-1.90 0.04
    2DL2/2DL2 + HLA-C1C1 11 (3.1) 23 (3.4) 0.93 0.45-1.92 1.00
    2DL2/2DL3 + HLA-C1C1 52 (14.8) 98 (14.4) 1.03 0.72-1.49 0.9
    2DL3/2DL3 + HLA-C1C1 68 (19.4) 84 (12.3) 1.71 1.20-2.42 0.003
    2DL3/2DL3 + HLA-C1C2 82 (23.3) 165 (24.2) 0.95 0.70-1.29 0.8
    2DL3/2DL3 + HLA-C2C2 21 (6.0) 73 (10.7) 0.53 0.32-0.88 0.01

    Although both KIR2DL3 and KIR2DL2 bind HLA-C1 allotypes, KIR2DL2 binds HLA-C1 with greater affinity than does KIR2DL3 (16). We hypothesized that a weaker inhibitory receptor-ligand (KIR2DL3-HLA-C1) interaction would be protective, because it should be more easily overridden by activating signals than a stronger inhibitory interaction such as KIR2DL2-HLA-C1 or KIR2DL1-HLA-C2. Consistent with this model, the protective association of HLA-C1C1 was significant only among individuals homozygous for KIR2DL3 (P = 0.003; OR = 1.71) and not among KIR2DL2/KIR2DL3 heterozygotes or KIR2DL2 homozygotes (Table 2). Thus, the presence of KIR2DL2 appears to counteract KIR2DL3-HLA-C1C1 protection. Further, KIR2DL3 did not associate with HCV resolution in individuals who were lacking HLA-C1C1, which indicates a synergistic protective effect between HLA-C1C1 and KIR2DL3/KIR2DL3, as opposed to additive, independent effects of each. All individuals in this study have at least one copy of KIR2DL1, so it was not possible to determine whether the susceptible HLA-C2C2 effect is independent of KIR2DL1. Neither KIR2DS2 nor KIR2DS1, activating receptors with high sequence similarity to KIR2DL2/KIR2DL3 and KIR2DL1, respectively, were associated with HCV resolution, but KIR3DS1 displayed a weak protective effect in combination with HLA-B Bw4+ alleles (P = 0.04; OR = 1.39).

    Resistance to murine cytomegalovirus infection is dependent on the NK cell receptor Ly49H (17) and can be overcome by increasing the size of the infecting inoculum (18). To investigate whether a similar dose-response relation could be detected with HCV infection, individuals were stratified by the expected inoculum size, assuming that individuals who contract HCV by transfusion of either blood or concentrated blood products (N = 494) receive larger inocula than those infected by injection drug use and needle-stick injuries (nontransfused) (N = 543) (table S1B) (19, 20).

    Among nontransfused individuals, 20.4% of those resolving infection had the compound genotype KIR2DL3/KIR2DL3-HLA-C1C1, as compared with 9.9% with persistent infection (P = 0.001; OR = 2.33) (Table 3). Further, homozygosity for HLA-C1 was protective only among individuals who were homozygous for KIR2DL3 (P = 0.0001; OR = 3.01) but not in those with one or no KIR2DL3 genes (P = 0.6 and P = 0.3, respectively) (Table 4). Protection conferred by KIR2DL3/KIR2DL3-HLA-C1C1 was stronger than any other KIR-HLA combination tested, indicating a direct, primary effect of this genotype on HCV clearance (tables S2 and S3). Alternatively, KIR2DL3/KIR2DL3-HLA-C1C1 showed no protection among transfused individuals.

    Table 3.

    Comparison of the frequencies of HLA-C and KIR-HLA-C combinations in individuals, stratified by history of transfusion of blood or plasma products. Definitions and calculations as for Table 2.

    Genetic factor Frequency resolved N (%) Frequency persistent N (%) OR 95% CI P
    Nontransfusion N = 185-187 N = 353-356
        HLA-C1C1 72 (38.5) 93 (26.1) 1.77 1.21-2.58 0.003
        HLA-C1C2 85 (45.4) 180 (50.6) 0.81 0.57-1.16 0.3
        HLA-C2C2 30 (16.0) 83 (23.3) 0.63 0.40-1.00 0.06
        2DL2/2DL2 + HLA-C1C1 3 (1.6) 14 (4.0) 0.40 0.11-1.40 0.2
        2DL2/2DL3 + HLA-C1C1 30 (16.1) 44 (12.5) 1.35 0.82-2.23 0.2
        2DL3/2DL3 + HLA-C1C1 38 (20.4) 35 (9.9) 2.33 1.42-3.85 0.001
    Transfusion N = 165 N = 328-329
        HLA-C1C1 60 (36.3) 112 (34.0) 1.11 0.75-1.63 0.6
        HLA-C1C2 84 (50.9) 162 (49.2) 1.07 0.74-1.55 0.8
        HLA-C2C2 21 (12.7) 55 (16.7) 0.73 0.42-1.25 0.3
        2DL2/2DL2 + HLA-C1C1 8 (4.85) 9 (2.74) 1.81 0.68-4.77 0.3
        2DL2/2DL3 + HLA-C1C1 22 (13.3) 54 (16.5) 0.78 0.46-1.33 0.4
        2DL3/2DL3 + HLA-C1C1 30 (18.2) 49 (14.9) 1.27 0.77-2.08 0.4
    Table 4.

    Frequency of HLA-C1C1 among nontransfused individuals stratified according to KIR2DL2 and KIR2DL3 genotype. Definitions and calculations as for Table 2.

    Frequency HLA-C1C1
    KIR genotype Resolved N (%) Persistent N (%) OR 95% CI P
    2DL2/2DL2 (N = 57) 3 (18.6) 14 (34.5) 0.45 0.11-1.83 0.3
    2DL2/2DL3 (N = 223) 30 (36.1) 44 (31.3) 1.24 0.70-2.19 0.6
    2DL3/2DL3 (N = 258) 38 (43.7) 35 (20.5) 3.01 1.72-5.29 0.0001

    KIR2DL3/KIR2DL3-HLA-C1C1 protection was observed in both nontransfused Caucasians and African Americans. Among Caucasians homozygous for KIR2DL3 (N = 145), 21.7% of persistently infected versus 49.1% of resolved individuals were HLA-C1C1 (P = 0.0009; OR = 3.47), as compared with 20.6% versus 36.4%, respectively, in African Americans (N = 106; P = 0.096; OR = 2.21). Although the protective effect did not reach significance in African Americans, the consistent trends across racial groups further suggest that a synergistic interaction between KIR2DL3 and HLA-C1 directly confers protection against HCV, rather than indirectly through linkage disequilibrium with neighboring loci.

    Multiple variable logistic regression analyses of variables that were significant (p < 0.05) in univariate analysis supported a protective effect of HLA-C1C1 only in the context of KIR2DL3 homozygosity and only among nontransfused individuals (P = 0.001; OR = 2.24) (Table 5). Conversely, the adverse effect of HLA-C2C2 alone and in combination with KIR2DL1 was no longer significant, suggesting that its effect in univariate analysis derives from the absence of protective HLA-C1 alleles. The weak effect of KIR3DS1-Bw4 persisted, appearing to be independent of KIR2DL3/KIR2DL3-HLA-C1C1 (Table 5). The KIR-HLA associations detected in HCV resolution were not affected by hepatitus B virus (HBV) infection or human immunodeficiency virus (HIV) infection [both previously associated with differential HCV recovery (21)], age, or sex; correspondingly, the associations were more significant in the UK cohort, which contained a lower proportion of transfused individuals.

    Table 5.

    HLA-C1C1 protection is present only in the context of KIR2DL3 homozygosity. Multiple variable logistic regression analyses of the effect of KIR and HLA effects in the resolution of HCV infection demonstrate that the protective effect of HLA-C1C1 is due to its epistatic interaction with KIR2DL3 and that this effect is present only among nontransfused individuals. Analysis was performed by stepwise logistic regression with the PROC LOGISTIC procedure (15), with the variables KIR2DL3/KIR2DL3-HLA-C1C1, HLA-C1C1 (without KIR2DL3/KIR2DL3), HLA-C2C2, and KIR3DS1-Bw4.

    Group Genotype OR 95% CI P
    All HLA-C1C1 1.05 0.73-1.50 0.80
    (N = 1023) HLA-C2C2 0.75 0.51-1.08 0.12
    2DL3/2DL3+ HLA-C1C1 1.75 1.21-2.55 0.003
    KIR3DS1-Bw4 1.49 1.09-2.04 0.01
    No transfusion HLA-C1C1 1.2 0.73-1.98 0.48
    (N = 533) HLA-C2C2 0.78 0.48-1.28 0.33
    2DL3/2DL3+ HLA-C1C1 2.42 1.42-4.13 0.001
    KIR3DS1-Bw4 1.55 0.99-2.41 0.06
    Transfusion HLA-C1C1 0.92 0.55-1.53 0.24
    (N = 490) HLA-C2C2 0.71 0.40-1.26 0.75
    2DL3/2DL3+ HLA-C1C1 1.28 0.75-2.18 0.36
    KIR3DS1-Bw4 1.45 0.93-2.26 0.11

    KIR are clonally expressed on NK cells in a stochastic manner such that each NK cell clone expresses only a portion of the genes within the genetic profile (2, 22). Thus, homozygotes for KIR2DL3 will have more NK cells solely under the inhibitory control of KIR2DL3 than will KIR2DL2/KIR2DL3 heterozygotes. Similarly, individuals who have the HLA-C1C1 genotype (and are therefore missing the HLA-C2 ligand for KIR2DL1) will have more NK cells under the inhibitory control of KIR2DL3 than individuals who have the HLA-C1C2 genotype, in whom a proportion of NK cells will be inhibited by KIR2DL1 (an inhibitory receptor that is present in virtually all individuals). Consistent with this thesis, we observed a linear trend between the number of KIR2DL3-HLA-C1 interactions and the odds of resolving infection (χ2 for trend = 12.33; P = 0.0004) (Fig. 1).

    Fig. 1.

    Progressive effect of KIR2DL3-HLA-C1 on the outcome of HCV infection in nontransfused individuals. Individuals were divided according to KIR2DL3 and HLA-C genotype. 2DL3-C1 homozygous individuals have the genotype KIR2DL3/KIR2DL3-HLA-C1C1; 2DL3-C1 heterozygous have the genotypes 2DL3/2DL3-C1C2, 2DL3/2DL2-C1C1, or 2DL3/2DL2-C1C2; 2DL3-C1 null comprise the remainder and are missing KIR2DL3, HLA-C1, or both. The odds ratios, the 95% confidence intervals for resolution of HCV as calculated from two-by-two contingency tables, and the results of a chi-square test for trend based on the number of these interactions are shown.

    In this quantitative model, protective NK cell activity may be mediated through weak inhibitory KIR2DL3-HLA-C1 interactions (i.e., the lack of strong NK cell inhibition), perhaps in combination with one of the many nonvariable NK cell activating receptors (23). The protection was observed only among individuals presumably receiving low-dose HCV inocula, which suggests that the difference in the ability of distinct KIR-HLA genotypes to regulate NK cell activity is great enough to alter the outcome when faced with low-dose, but not high-dose, infection. The beneficial effect of lower inhibitory signals in HCV infection is consistent with other disease models in which activating interactions are advantageous against HIV disease (9) but disadvantageous in autoimmune disease (8, 10). In light of the protection conferred by KIR2DL3-HLA-C1 against HCV, the known conservation of the MHC-C1 motif across primate species (24) may indicate a selective advantage of this genotype against viral disease in general.

    Supporting Online Material

    www.sciencemag.org/cgi/content/full/305/5685/872/DC1

    Materials and Methods

    Figs. S1 and S2

    Tables S1 to S3

    References

    References and Notes

    View Abstract

    Navigate This Article