Killer cell immunoglobulin-like receptors (KIRs) influence both innate and adaptive immunity. But while the role of KIRs in NK-mediated innate immunity is well-documented, the impact of KIRs on the T cell response in human disease is not known. Here we test the hypothesis that an individual's KIR genotype affects the efficiency of their HLA class I-mediated antiviral immune response and the outcome of viral infection. We show that, in two unrelated viral infections, hepatitis C virus and human T lymphotropic virus type 1, possession of the KIR2DL2 gene enhanced both protective and detrimental HLA class I-restricted anti-viral immunity. These results reveal a novel role for inhibitory KIRs. We conclude that inhibitory KIRs, in synergy with T cells, are a major determinant of the outcome of persistent viral infection.
Hepatitis C virus (HCV) and human T-cell leukemia virus (HTLV-I) infect millions of people worldwide. Some HCV-infected individuals spontaneously clear the virus and many HTLV-1-infected people remain asymptomatic; however, in both cases the infection can lead to serious illness such as cancer. The factors which determine outcome are still elusive. We have found that a gene that encodes a receptor (KIR2DL2) enhances both protective and detrimental HLA class I-mediated immunity to HCV and HTLV-1. Strikingly, although KIRs are primarily associated with innate immunity, our observations suggest that they also have a major impact on the efficiency of the adaptive immune response. This work helps to explain why one individual infected with a virus remains healthy but another, infected with the same virus develops disease; it also helps to explain why particular HLA class I molecules do not always protect or cause susceptibility as expected. Interestingly, the impact of the KIR is entirely context dependent: if an HLA class I molecule is protective then protection is enhanced, but in the context of a detrimental HLA then susceptibility is enhanced. This study reveals a novel role for inhibitory KIRs in adaptive immunity.
Citation: Seich al Basatena N-K, MacNamara A, Vine AM, Thio CL, Astemborski J, Usuku K, et al. (2011) KIR2DL2 Enhances Protective and Detrimental HLA Class I-Mediated Immunity in Chronic Viral Infection. PLoS Pathog 7(10): e1002270. doi:10.1371/journal.ppat.1002270
Editor: Daniel C. Douek, NIH/NIAID, United States of America
Received: May 27, 2011; Accepted: August 1, 2011; Published: October 13, 2011
This is an open-access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication.
Funding: This work is supported by the Wellcome Trust, the Medical Research Council and the National Institutes of Health. Specifically, BA was supported by the MRC and NKS was supported by the Wellcome Trust. This project has been funded in whole or in part with federal funds from the National Cancer Institute, National Institutes of Health, under Contract No. HHSN261200800001E. This Research was supported in part by the Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research. The HGDS was funded by the National Institutes of Health, National Institute of Child Health and Human Development, 1 R01 HD41224. ALIVE was supported by the National Institute on Drug Abuse, grants R01-DA-04334, R01-DA-12568 and R01-DA-16078. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Killer cell immunoglobulin-like receptors (KIRs) are a family of transmembrane proteins that are expressed on natural killer (NK) cells and subsets of T cells –. They bind HLA class I molecules and have activatory and inhibitory isoforms . KIRs contribute directly and indirectly to antiviral immunity. Directly, KIRs on NK cells sense the loss of HLA class I molecules from the cell surface and trigger NK-mediated cytolysis. Indirectly, NK cells regulate adaptive immunity via crosstalk with dendritic cells and by the production of chemokines and cytokines –.
HLA class I molecules can be grouped into allotypes with similar KIR binding properties . For example, KIR2DL2 binds group C1 HLA-C molecules which have asparagine at residue 80, and, with a weaker affinity, group C2 molecules which have a lysine at position 80 .
Early research on KIRs investigated NK-mediated protection by studying disease associations with KIRs in the context of their HLA class I ligands –. There is now compelling evidence that KIRs also regulate adaptive immunity –, but it is not known whether this has a significant impact on the response to infection in vivo. Differences between human KIRs and their mouse functional homologues (the Ly49 receptors) and the paucity of KIR allele-specific antibodies have hindered work on the role of KIRs in controlling adaptive immune responses. Here we used immunogenetics to investigate whether KIR genotype modulates HLA-mediated anti-viral protection in vivo. We focussed on HLA class I alleles which have previously been associated with disease outcome and investigated whether these effects were altered by the KIR background. We studied 4 well-documented HLA class I allele-disease associations in two viral infections: human T lymphotropic virus type 1 (HTLV-1) and hepatitis C virus (HCV).
HTLV-1 is a persistent retrovirus that infects 10–20 million people worldwide. Most infected individuals remain lifelong asymptomatic carriers (ACs). However, approximately 10% of infected individuals develop associated diseases including HTLV-1-associated myelopathy/tropical spastic paraparesis (HAM/TSP), an inflammatory disease of the central nervous system that results in progressive paralysis. It is poorly understood why some individuals remain asymptomatic whereas others develop disease, but one strong correlate of disease is the proviral load, which is significantly higher in HAM/TSP patients than in ACs . We have previously shown that HLA-A*02 and C*08 are each associated with both a reduced risk of HAM/TSP and a reduced proviral load in ACs and that HLA-B*54 is associated with an increased prevalence of HAM/TSP and an increased proviral load in HAM/TSP patients –.
HCV is among the most widespread viral infections, with 170 million infected people worldwide. As in HTLV-1 infection, the outcome of HCV infection is heterogeneous: the virus persists in approximately 70% of infected individuals while the rest clear the infection spontaneously. Chronic HCV infection can cause serious liver damage including cirrhosis and hepatocellular carcinoma . The origins of this heterogeneity are not completely understood but several genetic determinants have been identified, including HLA-B*57 which is associated with spontaneous clearance in several cohorts –.
The aim of this study was to test the hypothesis that KIR genotype determines the efficiency of HLA class I-mediated anti-viral immunity. We tested this hypothesis for 4 HLA class I associations: HLA-C*08, A*02 and B*54 in HTLV-1 infection and B*57 in HCV infection. We show, using multiple independent measures, that for both HCV and HTLV-1, possession of the KIR2DL2 gene enhanced HLA class I-restricted immunity.
Two independent cohorts were studied: HCV-infected individuals (N = 782) and HTLV-1-infected individuals (N = 402). Ten KIR genes were typed; of these, 4 were present at an informative frequency in the HTLV-1 cohort and 9 in the HCV cohort (defined in Methods). The modulation of HLA class I associations by the informative KIRs was analysed; the models included all other known determinants of outcome in the cohorts as covariates.
HTLV-1 and disease status
In the cohort from Southern Japan, HLA-C*08 was associated with a significantly reduced odds of developing HAM/TSP (OR = 0.47, p = 0.03, OR<1 indicates a protective effect while OR>1 indicates a detrimental effect). We investigated the impact of KIRs on this protective effect by stratifying the cohort by KIR genotype. Of the KIR studied, one particular KIR, KIR2DL2, had a noticeable interaction with C*08 (Table 1 and Figure 1). We found that the C*08 protective effect was weakened and no longer statistically significant in the subset of individuals who were KIR2DL2- (OR = 0.67, p = 0.4) but enhanced in KIR2DL2+ individuals (OR = 0.16, p = 0.02). There were more KIR2DL2- individuals than KIR2DL2+ individuals so the absence of significance in the KIR2DL2- individuals was not simply due to reduced cohort size. Similarly, HLA-B*54, which is associated with a significantly increased risk of HAM/TSP (OR = 3.11, p = 0.0009), had a weakened impact on disease risk in the absence of KIR2DL2 (OR = 1.70, p = 0.2) but an enhanced impact in the presence of KIR2DL2 (OR = 12.05, p = 0.004). Again, the absence of a significant effect of B*54 in KIR2DL2- individuals was not attributable to a loss of power. In contrast, although HLA-A*02 was associated with a reduced risk of HAM/TSP, there was no significant additional impact of KIR genotype which could not be attributed to power.
HTLV-1 and proviral load
As an independent test of the observation that KIR2DL2 enhanced the effect of both protective and detrimental HLA class I alleles in HTLV-1 infection, we investigated the interaction between HLA class I alleles, KIR2DL2 and HTLV-1 proviral load (pvl). We investigated pvl in ACs and HAM/TSP patients separately, so any observed impact on pvl is independent of the impact on disease status. C*08 has previously been associated with a low pvl in ACs (difference in log10 pvl between C*08+ and C*08- Δ = −0.33, p = 0.05); again, this effect was weakened in KIR2DL2- individuals (Δ = −0.29 p = 0.2) but enhanced in KIR2DL2+ individuals (Δ = −0.66 p = 0.07); Table 1. Similarly, HLA-B*54, which is associated with a high pvl in HAM/TSP patients (Δ = +0.24 p = 0.01) showed a weakened effect in the absence of KIR2DL2 (Δ = +0.22, p = 0.05) but an enhanced effect in the presence of KIR2DL2 (Δ = +0.42, p = 0.01).
Two previous observations on HTLV-1 immunogenetics have, until now, remained unexplained. Firstly, although C*08 has been associated with a low pvl in ACs it has no detectable impact on pvl in HAM/TSP patients; similarly, B*54, which was associated with a high pvl in HAM/TSP patients, had no impact on pvl in ACs –. Why some HLA class I alleles apparently “cease working” in some populations was unknown. We hypothesised that the lack of the expected C*08 and B*54 effects in HAM/TSP patients and ACs respectively was due to a low frequency of KIR2DL2 in these groups and that the decrease or increase in pvl due to C*08 or B*54 respectively would be manifest only in KIR2DL2+ individuals. Consistent with this hypothesis we found that the frequency of KIR2DL2 carriage in the groups that did not show the expected effect of HLA genotype on pvl was approximately half that of the groups in which HLA-associated effects were observed (prevalence of KIR2DL2+ amongst B*54+ individuals: 12.5% in ACs vs 29.5% in HAM/TSPs; prevalence of KIR2DL2+ amongst C*08+ individuals: 18.2% in HAM/TSPs vs 27.8% in ACs; Table 1). The small numbers of individuals in the stratified cohorts (HAM/TSP KIR2DL2+: C*08+ N = 4, C*08- N = 50. AC KIR2DL2+: B*54+ N = 3, B*54- N = 45) precluded a reliable test for an impact of HLA on pvl in KIR2DL2+ individuals. However, in the larger of these groups there was a significant impact; i.e C*08 was associated with a significant reduction in pvl in KIR2DL2+ individuals (Δ = -0.86, p = 0.005). This provides, for the first time, a plausible explanation for the reported observation  that the B*54 effect on pvl was not manifest in ACs and the C*08 effect on pvl was not manifest in HAM/TSP patients.
HTLV-1 and HLA class I specificity
We recently reported that in HTLV-1 infection, HLA class I molecules that bind peptides from the virus protein HBZ are associated with a reduced risk of HAM/TSP and, independently, a reduced pvl . In the same study we showed, using IFNg ELISpot, chromium release and CD107 staining, that HBZ-specific CD8+ T cells were present and functional in fresh PBMC from infected individuals. We therefore investigated the interaction between KIR2DL2 and the protective effect of binding HBZ. We used epitope prediction software  to predict the strength of binding of HBZ peptides to an individual's HLA-A and B molecules. We found that ACs had HLA-A and -B molecules that are predicted to bind HBZ significantly more strongly than those in HAM/TSP patients (median difference 12%, p = 0.00005)  and that this effect was stronger in KIR2DL2+ individuals (median difference 25%, p = 0.00006) than in KIR2DL2- individuals (median difference 7%, p = 0.06); Figure 2a. We reasoned that this difference in HBZ binding between ACs and HAM/TSP patients was due to HLA-A*02 and B*54, which differ in their HBZ peptide-binding affinities  and are associated with different outcomes in HTLV-1 infection. We therefore removed all individuals with A*02 or B*54 from the cohort and repeated the analysis. Surprisingly, we still found the same pattern: possession of HLA molecules that bind HBZ strongly was significantly associated with remaining asymptomatic (median difference 10%, p = 0.04) and this effect was strengthened in KIR2DL2+ individuals (median difference 23%, p = 0.02) but not in KIR2DL2- individuals (median difference 3%, p = 0.2); Figure 2b. This demonstrates that the protective effect of binding HBZ peptides by multiple HLA class I molecules, both A and B, is enhanced by KIR2DL2.
HCV and viral clearance
As previously reported, HLA-B*57 was associated with significantly decreased odds of chronic infection (OR = 0.571, p = 0.02). This protective effect was enhanced in the presence of KIR2DL2 (OR = 0.40, p = 0.007) but weakened in the absence of KIR2DL2 (OR = 0.83, p = 0.6) (Table 2). Furthermore, the impact of B*57 was strongest in KIR2DL2 homozygote individuals (OR = 0.21, p = 0.02), weaker in KIR2DL2 heterozygote individuals (OR = 0.48, p = 0.07) and absent in KIR2DL2-negative individuals (OR = 0.83, p = 0.6). KIR2DL2 enhanced the association between B*57 and spontaneous clearance independently and with similar strength in both African Americans and Caucasians (Table S1 in Text S1).
HCV and viral load
Next we investigated the impact of B*57 and KIR2DL2 on HCV viral load. We only considered patients with chronic infection, so any observed impact on viral load is independent of the impact on viral clearance. This analysis was possible in two cohorts: MHCS and ALIVE. We found that B*57 was associated with reduced chronic HCV viral load, particularly in the MHCS cohort (MHCS: difference in log10 VL Δ = −3.1 p = 0.0003; ALIVE: Δ = −0.18 p = 0.3. Combined p = 0.0006). Consistent with our observations in HTLV-1 infection, this reduction was enhanced in the presence of KIR2DL2 (MHCS: Δ = −4.5 p<0.0001, ALIVE: Δ = −0.46 p = 0.05. Combined p = 0.00003) but weakened in the absence of KIR2DL2 (MHCS Δ = −1.64 p = 0.2, ALIVE Δ = +0.32 p = 0.3. Combined p = 0.7); Table 2 and Figure 3. In MHCS (but not ALIVE) we also observed a progressive effect with KIR2DL2 copy number (2 copies: Δ = −6.5, p = 0.0005. 1 copy: Δ = −4.1 p = 0.001. 0 copies Δ = −1.6 p = 0.2); the number of homozygous individuals are too small to draw firm conclusions but this progressive effect is consistent with our other observations.
These data show that KIR2DL2 enhances both protective and detrimental HLA class I associations. Therefore, KIR2DL2 would not be predicted to have a significant net impact across all HLA class I molecules. That is, possession of KIR2DL2 (alone or with its C1 ligand) without a particular protective or detrimental HLA allele, would not be expected to be significantly protective or detrimental. This prediction was verified (Table S2 in Text S1).
There is strong linkage disequilibrium among the KIR genes and among the HLA class I genes. Analysis of the linked genes indicates that the primary genes driving the observed associations are most likely to be KIR2DL2 in combination with HLA-B*54, C*08 and B*57 rather than individual linked KIR, multiple stimulatory linked KIRs or linked HLA class I genes (sections 3 and 4 in Text S1). We cannot rule out an effect of linkage between KIR2DL2 and neighbouring loci outside the KIR genes. However, there is little evidence of significant linkage between KIRs and even the next closest gene cluster, the LILR . Furthermore, we observed the same effect of KIR2DL2 in three different populations (Japanese, African-American and Caucasian) so a putative linked locus driving the effect would have to be linked to KIR2DL2 in all three populations.
Canonical KIR-HLA binding
Although HLA-C*08, as a group C1 molecule, is expected to bind KIR2DL2, the most frequent subtype in our cohort (Cw*0801, 88%) binds KIR2DL2 very weakly (comparable to background ), furthermore HLA-B*54 and HLA-B*57 are not expected to bind KIR2DL2 and the most frequent subtypes in our cohorts (B*5401 and B*5701) have been shown not to bind KIR2DL2 , . Finally, KIR2DL2 enhanced the protective effect of binding HBZ peptides by multiple HLA-A and –B molecules. With the exception of B*4601 and B*7301 (which were not responsible for the enhancement, data not shown) KIR2DL2 is not thought to bind HLA-A and B molecules and has been shown not to bind 29 HLA-A and 54 HLA-B allotypes . We therefore hypothesised that the effect of KIR2DL2 on HLA class I-mediated immunity we have observed is not attributable to KIR2DL2 directly binding the HLA molecule whose effect is enhanced. To test this hypothesis we first investigated whether the other group C1 alleles had the same effect as C*08 in HTLV-1 infection. Grouping all the C1 alleles we found no significant association between C1 and decreased risk of HAM/TSP either in the whole cohort or in KIR2DL2+ individuals. Similarly, there was no relation between pvl and C1 in either ACs or HAM/TSP patients. Analysis of the individual C1 alleles confirmed the hypothesis that the C*08 effect we observed was not exhibited by other group C1 alleles (Tables S4 and S5).
HLA-B*54, a group Bw6 HLA allele, is not known to bind any KIR molecule. We therefore tested whether the observed B*54 effect was attributable to C*01, which is in linkage disequilibrium with B*54 and which encodes molecules that bind KIR2DL2. This analysis suggested that B*54, not C*01, was the gene driving the observed detrimental effect on HTLV-1 outcome (section 5.2 in Text S1). This result, and the observation that no other C1 allele shows “B*54-like” behaviour, indicate that, as postulated, the interaction between B*54 and KIR2DL2 cannot be explained by direct KIR-HLA binding.
The most frequent B*57 allele in our cohort is B*5701, which does not bind KIR2DL2 . There are therefore two ways in which the observed interaction between KIR2DL2 and HLA-B*57 could be attributed to “classical” KIR-HLA binding: either KIR2DL2 might bind a class I HLA molecule whose encoding gene is linked to HLA-B*57, or the effect might be due to KIR3DL1/S1, which does bind B*57. Analysis of both these possibilities indicated that they did not explain the KIR2DL2-B*57 effect (section 5.3 and 5.4 in Text S1).
Therefore, as hypothesised, the enhancement of C*08, B*54 and B*57–restricted immunity by KIR2DL2 is not explained by direct binding between the respective HLA molecules and KIR2DL2. Instead, we suggest that KIR2DL2 binds its HLA-C ligands and indirectly modulates C*08, B*54 and B*57–restricted T cells. Consistent with this, we found some evidence that KIR2DL2 enhanced HLA Class I effects more strongly when it's stronger C1 ligands are present (section 6 in Text S1).
Other inhibitory KIR
It seems unlikely that KIR2DL2 behaves fundamentally differently to other inhibitory KIRs. The effect of KIR2DL2 may be most apparent because KIR2DL2 is present at informative frequencies and its C1 and C2 ligands are ubiquitous. We addressed the role of other inhibitory KIRs in 3 ways. We studied the effect of individual inhibitory KIRs (section 4.1 in Text S1), we investigated whether the number of inhibitory KIR:ligands had a cumulative effect (section 7 in Text S1) and we examined the role of the group A KIR haplotypes which are dominated by inhibitory KIRs but do not contain KIR2DL2 (section 8 in Text S1). We found little evidence that the other inhibitory KIRs enhanced HLA class I-mediated immunity but this may be due to small cohort sizes and masking by the dominant KIR2DL2 effect.
We found no evidence that activating KIR were enhancing HLA class I-restricted immunity. Haplotype B, the more activatory KIR haplotype, enhanced HLA class I associations but this was only true if the haplotype contained KIR2DL2 (section 8 in Text S1). We found no evidence that the cumulative presence of activating KIR enhanced HLA class I restricted immunity (section 4.3 in Text S1). And, as far as it was possible to separate KIR2DL2 and KIR2DS2, which are in tight linkage disequilibrium, the enhancement of HLA class I restricted immunity appeared to be attributable to KIR2DL2 rather than KIR2DS2 (section 4.2 in Text S1).
We show that KIR2DL2 enhanced several independent HLA class I-mediated effects in two unrelated viral infections. In HTLV-1 infection, KIR2DL2 enhanced the protective and detrimental effects of HLA-C*08 and B*54 respectively on disease status. KIR2DL2 also enhanced the association between C*08 and low proviral load in ACs and between B*54 and high proviral load in HAM/TSP patients. Additionally, KIR2DL2 enhanced the protective effect of HBZ binding by multiple HLA molecules. Strikingly, on stratifying by KIR2DL2, we observed, for the first time, a protective effect of C*08 on pvl in HAM/TSP patients and explained the lack of impact of B*54 on pvl in ACs. In HCV infection, KIR2DL2 enhanced the protective effect of B*57 on spontaneous clearance and the association between B*57 and low viral load in chronic carriers; for both clearance and viral load a progressive effect with KIR2DL2 copy number was observed. This progressive effect is consistent with reports of an association between KIR gene copy number and the frequency of cell-surface expression of the respective KIR molecule –.
There are two mechanisms by which KIR2DL2 could act: it could enhance either NK-mediated or T cell-mediated immunity. That is, NK cell killing of virus-infected cells could be altered by KIR2DL2 expression or, alternatively, the virus-specific CD8+ T cell response could be modified by KIR2DL2 expression (on NK cells or T cells). Two observations indicate that it is the T cell response that is more likely to be enhanced. First, strong binding of HBZ viral peptides via multiple different HLA-A and B molecules was associated with asymptomatic status  and this protective effect was enhanced by KIR2DL2. KIR2DL2 is not known to bind HLA-A or–B molecules (with the exception of B*4601 and B*7301) , ,  so it is unlikely that the enhancement of the protective effect of binding HBZ by KIR2DL2 is due to direct binding between KIR2DL2 and HBZ peptide in the context of HLA-A and –B molecules. Furthermore, although NK cells exhibit peptide dependence , it is hard to reconcile protein-specificity via multiple HLA molecules with an NK cell-mediated mechanism. Second, the KIR2DL2 enhancement could not be explained by binding between KIR2DL2 and any of the 3 HLA class I molecules investigated. One further observation also suggests a T cell-mediated mechanism. Two protective genotypes in HCV infection that are postulated to operate via innate immune mechanisms  (namely a SNP upstream of IL28B and KIR2DL3-HLA-C1) had no impact on viral load in chronic infection , . The authors hypothesised that this was because innate barriers offer little protection once overcome. In contrast, the KIR2DL2/B*57 effect that we report here had a significant impact on viral load: again, this is perhaps more consistent with adaptive immunity.
Our results indicate that KIR2DL2 enhances HLA class 1-restricted CD8+ T cell-mediated adaptive immunity. KIRs on both NK cells and CD8+ T cells have been reported to shape adaptive immunity -. Of particular interest are reports – that inhibitory KIRs on CD8+ T cells promote the survival of a subset of memory phenotype CD8+ αβ T cells with enhanced cytolytic potential (Tm1 ) by reducing activation-induced cell death. Ugolini et al suggested that this phenomenon helps maintain specific CD8+ T lymphocytes during chronic viral infection . Tm1 cells have been described in both HTLV-1 and HCV infections, where they constitute a minority of virus-specific CD8+ T cells but the majority of perforin-bright cells –. Consistent with our findings, these studies showed that the HLA molecule that restricts the T cell whose survival is promoted was independent of the HLA-C molecules that ligated the KIR , . We postulate that, in the face of chronic antigen stimulation, protective T cells survive longer if they carry KIR2DL2 and therefore exert stronger protection. Likewise, T cells restricted by HLA alleles associated with increased disease susceptibility also survive for longer in the presence of KIR2DL2 and so are more detrimental (Figure 4). Hence, KIR2DL2 enhances both protective and detrimental HLA class I associations.
Alternatively, it is known that NK cells kill activated T cells and that this killing is reduced by inhibitory KIR –. So again, T cells restricted by protective and detrimental HLA class I molecules may survive longer in the presence of inhibitory KIR and thus the protective and detrimental associations would be enhanced.
Ugolini et al proposed that inhibitory KIRs promote T cell survival by increasing the activation threshold of T cells. This may explain why the HLA-A*02 protective effect in HTLV-1 is not significantly enhanced by KIR2DL2. A*02 molecules bind peptides significantly more strongly than other alleles (section 9 in Text S1) and the immunodominant HTLV-1 peptide Tax 11-19 is bound exceptionally strongly. Therefore, even if the T cell activation threshold were increased, the strength of signalling may remain above the threshold and consequently the A*02 protective effect cannot be enhanced.
Why does KIR2DL2 enhance T cell responses whereas the other inhibitory KIRs apparently do not? The effect of KIR2DL2 may be most apparent because KIR2DL2 is present at informative frequencies and its C1 and C2 ligands are ubiquitous; i.e. unlike the other KIR every individual carries a KIR2DL2 ligand.
It will be important to determine whether inhibitory KIRs play a similar role in enhancing CD8+ and possibly CD4+ T cell-mediated immunity to other pathogens and in autoimmune disease. KIR-expressing virus-specific CD8+ T cells have been reported in other chronic infections including HIV-1, CMV and EBV -. Furthermore, in HIV-1 infection, high expression alleles of an inhibitory KIR, KIR3DL1, in the context of HLA-Bw4I have been associated with slow progression to AIDS . In order to explain protection by an inhibitory KIR the authors proposed a model based on NK cell development. Our results suggest an alternative explanation, i.e. that KIR3DL1 enhances protective HLA-B-restricted responses to HIV-1.
In contrast to previous studies of KIR genotype, which investigated the antiviral action of NK cells, we investigated the impact of KIRs on HLA class I-mediated antiviral immunity. We find a clear and consistent effect of KIR2DL2. The effect sizes are striking: KIR2DL2 homozygotes with B*57 are almost 5 times more likely to clear HCV infection spontaneously; if they fail to clear the virus they have a viral load that is reduced by 6.5 logs. Until now, the advantages offered by inhibitory KIRs in virus infections have been unclear. Our data support an alternative role in which inhibitory KIRs enhance both beneficial and detrimental T cell-mediated immunity in persistent viral infection.
The HTLV-1 cohort has been approved by the following committees: 1) St. Mary's Local Research Ethics Committee, 1995: title “The immunology and virology of the treatment of HTLV-1-associated inflammatory disease”. Approval reference number: EC3108. 2) Kagoshima University Hospital Clinical Research Ethics Committee: 27th May 1999. Title: “Investigation of HAM pathomechanism: relationship between host genetic background and clinical status of HTLV-1 infection”. Approval reference number: 22. All samples were taken under written informed consent.
The HCV cohort consisted of four sub-cohorts: AIDS Link to Intravenous Experience (ALIVE, N = 262) , Multicenter Hemophilia Cohort Study (MHCS, N = 320) , Hemophilia Growth and Development Study (HGDS, N = 110)  and a UK cohort (N = 341) . 251 individuals were excluded due to incomplete information. The cohort had 257 resolved and 525 chronic patients. HLA class I associations in three of these 4 cohorts (ALIVE, MHCS and HGDS) have previously been reported .
The HTLV-1 cohort (N = 431)  consists of individuals recruited in Kagoshima Japan. All individuals were of Japanese ethnic origin and resided in Kagoshima Prefecture, Japan. The cohort had 229 HAM/TSP patients and 202 asymptomatic carriers.
Genomic DNA was amplified using locus specific primers as described , . The resulting PCR products were blotted onto nitrocellulose membranes and hybridized with sequence specific oligonucleotide probes (SSO). US: alleles were assigned according to the reaction patterns of the SSO probes, ambiguities were resolved by sequencing analysis. UK: PCR products were typed by direct sequencing. HLA types that were not resolved by sequencing or which gave unusual results were also tested by SSO typing using commercial kits (Dynal, RELI SSO, Wirral, UK).
PCR–sequence-specific primer reactions were performed as described .
KIR genotyping of the HCV cohort (but not the HTLV-1 cohort) was performed previously .
The presence or absence of ten KIR genes (KIR2DL1, KIR2DL2, KIR2DL3, KIR3DL1, KIR2DS1, KIR2DS2, KIR2DS3, KIR2DS4, KIR2DS5 and KIR3DS1) was determined by PCR using sequence specific primers as described in .
The presence or absence of ten KIR genes (KIR2DL1, KIR2DL2, KIR2DL3, KIR2DL4, KIR3DL1, KIR2DS2, KIR2DS3, KIR2DS4, KIR2DS5 and KIR3DS1) was determined by PCR using sequence specific primers . 30 of the 431 individuals were not KIR genotyped due to insufficient DNA.
This method of KIR typing does not allow the direct determination of KIR2DL2 copy number. Instead we use the allelic nature of KIR2DL2 and KIR2DL3 at the 2DL2/2DL3 locus to infer the number of copies of KIR2DL2.
HCV Cohorts were analysed separately as viral load was measured using different assays. Two cohorts (MHCS and ALIVE) had sufficient numbers of individuals with measurements of all variables for this analysis. For MHCS we used the median of multiple measurements of viral load; for ALIVE, a single measurement was available for each subject. We only analysed viral load in patients with chronic infection, so any observed impact on viral load is independent of an observed impact on viral clearance.
HCV RNA was assessed by branched DNA (Quantiplex HCV RNA 2.0 assay; Chiron Corporation) [ALIVE] or the HCV COBAS AMPLICOR system (COBAS AMPLICOR HCV; Roche Diagnostics) [MHCS].
The HTLV-1 provirus load in peripheral blood mononuclear cells (PBMC) was measured as described in . Quantitative PCR was performed using an ABI 7700 sequence detector (Perkin–Elmer Applied Biosystems). All DNA standards and samples were amplified in triplicate. A standard curve was generated by using the β-actin gene from HTLV-1-negative PBMC and the Tax gene from TARL-2, a cell line containing a single copy of HTLV-1 proviral DNA. We analysed viral load separately in ACs and HAM/TSP patients, so any observed impact on viral load is independent of any observed impact on disease status.
HLA class I.
HLA class I alleles which were significantly associated with outcome in our cohorts and that were independently verified (either in a large independent cohort or on an independent outcome) were studied. For HTLV-1 there are three such associations: HLA-A*02 and C*08 which are associated with reduced proviral load in ACs and reduced risk of HAM/TSP and HLA-B*54 which is associated with increased proviral load in HAM/TSP patients and increased risk of HAM/TSP -. For HCV there are two such associations: HLA-B*57 and C*01 which have both been associated with increased odds of viral clearance in two independent cohorts , . The C*01-protective effect in HCV infection has been shown to be NK-mediated ,  and therefore we did not consider it here.
We investigated KIR genes which were present at an informative frequency. We define an informative frequency as sufficient to detect a “moderate” effect size (δ = 0.5, ); this is equivalent to at least 32 people carrying the gene and 32 not carrying the gene. Of the KIR genes typed, 4 were present at an informative frequency in the HTLV-I cohort (KIR2DL2, KIR2DS2, KIR2DS3 and KIR3DS1) and 9 in the HCV cohort (KIR2DL2, KIR2DL3, KIR3DL1, KIR2DS1, KIR2DS2, KIR2DS3, KIR2DS4, KIR2DS5 and KIR3DS1).
The effect of individual HLA class I alleles in different KIR genetic background was investigated by stratifying the cohorts for the absence and presence of the informative KIRs and the effect of the already described HLA class I associations were re-evaluated in each stratum. The impact of the HLA alleles on two response variables was studied: 1) status (AC vs. HAM/TSP for HTLV-1 infection and resolved vs. chronic for HCV infection) and 2) viral burden (log10[proviral load] for HTLV-1 infection and log10[viral load] for HCV infection). Multiple logistic regression was used to study the variation in status and multiple linear regression was used for the variation in viral burden. All statistically significant confounding variables were included in the models (HTLV-1: age and gender; HCV: HBV status, mode of infection, SNP rs12979860 and cohort). We focused on the difference between KIR+ and KIR- individuals in the size of the protective/detrimental effects associated with the individual HLA class I molecules (i.e. odds ratios and differences in viral load) rather than p values as the latter comparison is confounded by differences in strata sizes.
We quantified the false discovery rate for our analysis using Monte Carlo methods. For each of the HLA associations studied, we performed 10,000 random stratifications of the relevant (HCV or HTLV-1) cohort with the size of the two strata being the number of KIR2DL2+ and KIR2DL2- individuals in that cohort and asked how many times we would see odds ratios equal to or more extreme than we observed in the actual cohorts. The probability of making our observations by chance is less than p = 2×10−11.
Linkage disequilibrium (LD) between HLA class I alleles was calculated using the tool available at www.hiv.lanl.gov. The LD between KIRs was calculated based on the Chi-squared test on a 2x2 contingency table. All the reported p values are two-tailed. Where applicable, independent p values were combined using Fisher's combined test. Statistical analysis was performed using R v2.9.2.
The binding strength of HLA class I molecules to viral proteins was assessed using epitope prediction software. Prediction of T cell class I epitopes is now highly accurate and algorithms can achieve accuracy of up to 94% . In this study we use the epitope prediction software Metaserver  (http://web.bioinformatics.ic.ac.uk/metaserver).
Rank measure for predicted epitopes
We used the rank measure technique  in which the strength of an allele's preference for a particular protein is quantified by ranking the strength of binding of the top binding peptide from the protein of interest amongst the strength of binding of peptides from the entire proteome to that allele. Specifically, we split the HTLV-1 proteome into overlapping nonamers offset by a single amino acid and predicted a binding affinity score for each nonomer. For each allele we rank all nonamers from the proteome from the weakest to strongest predicted binding scores. This produces a list of rank values for each protein to that particular allele that quantifies the binding relationship between that allele and the protein. We then invert the rank so the bigger 1/rank, the stronger the preference of the allele for the protein; the logarithm of this measure is plotted on the y axis of Figure 2 as “HBZ binding score”. Each individual therefore contributes up to 4 values (alleles for which no predictive algorithms were available were excluded from the analysis). Binding scores were compared between ACs and HAM/TSP patients using the Wilcoxon rank sum test and reported both as separate p values for HLA-A and B molecules and combined (since we found no evidence to reject the null hypothesis that the HBZ binding score of an individual's A and B molecules was independent, spearman correlation = 0.05 p = 0.5). The median difference in binding score is the median of the difference of average HBZ binding between ACs and HAM/TSP patients expressed as a percent of the AC binding score for HLA-A and –B molecules.
Additional analysis and results. Supporting information on impact of race, KIR and HLA linkage disequilibrium, canonical KIR-HLA binding, KIR haplotypes, other inhibitory and activatory KIRs and other additional results.
We thank J. Trowsdale, J. Traherne, C. Johnson, W. Jiang, K Gould and U. Kadolsky for comments and discussion.
Conceived and designed the experiments: BA. Performed the experiments: NKS AM BA. Analyzed the data: NKS BA. Contributed reagents/materials/analysis tools: AMV CLT JA KU MO GDK SMD JJG MC SIK CRMB. Wrote the paper: BA NKS. Contributed to interpretation: BA NKS CRMB MC SIK.
- 1. Mingari MC,Moretta A,Moretta L (1998) Regulation of KIR expression in human T cells: a safety mechanism that may impair protective T-cell responses. Immunol Today 19: 153–157.
- 2. Speiser DE,Pittet MJ,Valmori D,Dunbar R,Rimoldi D,et al. (1999) In vivo expression of natural killer cell inhibitory receptors by human melanoma-specific cytolytic T lymphocytes. J Exp Med 190: 775–782.
- 3. Speiser DE,Valmori D,Rimoldi D,Pittet MJ,Lienard D,et al. (1999) CD28-negative cytolytic effector T cells frequently express NK receptors and are present at variable proportions in circulating lymphocytes from healthy donors and melanoma patients. Eur J Immunol 29: 1990–1999.
- 4. Long EO,Rajagopalan S (2000) HLA class I recognition by killer cell Ig-like receptors. Semin Immunol 12: 101–108.
- 5. Gerosa F,Baldani-Guerra B,Nisii C,Marchesini V,Carra G,et al. (2002) Reciprocal Activating Interaction between Natural Killer Cells and Dendritic Cells. J Exp Med 195: 327–333.
- 6. Piccioli D,Sbrana S,Melandri E,Valiante NM (2002) Contact-dependent Stimulation and Inhibition of Dendritic Cells by Natural Killer Cells. J Exp Med 195: 335–341.
- 7. Raulet DH (2004) Interplay of natural killer cells and their receptors with the adaptive immune response. Nat Immunol 5: 996–1002.
- 8. Wagtmann N,Rajagopalan S,Winter CC,Peruzzi M,Long EO (1995) Killer cell inhibitory receptors specific for HLA-C and HLA-B identified by direct binding and by functional transfer. Immunity 3: 801–809.
- 9. Moesta AK,Norman PJ,Yawata M,Yawata N,Gleimer M,et al. (2008) Synergistic polymorphism at two positions distal to the ligand-binding site makes KIR2DL2 a stronger receptor for HLA-C than KIR2DL3. J Immunol 180: 3969–3979.
- 10. Khakoo SI,Thio CL,Martin MP,Brooks CR,Gao X,et al. (2004) HLA and NK cell inhibitory receptor genes in resolving hepatitis C virus infection. Science 305: 872–874.
- 11. Martin MP,Qi Y,Gao X,Yamada E,Martin JN,et al. (2007) Innate partnership of HLA-B and KIR3DL1 subtypes against HIV-1. Nat Genet 39: 733–740.
- 12. Nagai M,Usuku K,Matsumoto W,Kodama D,Takenouchi N,et al. (1998) Analysis of HTLV-I proviral load in 202 HAM/TSP patients and 243 asymptomatic HTLV-I carriers: high proviral load strongly predisposes to HAM/TSP. J Neurovirol 4: 586–593.
- 13. Jeffery KJ,Siddiqui AA,Bunce M,Lloyd AL,Vine AM,et al. (2000) The influence of HLA class I alleles and heterozygosity on the outcome of human T cell lymphotropic virus type I infection. J Immunol 165: 7278–7284.
- 14. Jeffery KJ,Usuku K,Hall SE,Matsumoto W,Taylor GP,et al. (1999) HLA alleles determine human T-lymphotropic virus-I (HTLV-I) proviral load and the risk of HTLV-I-associated myelopathy. Proc Natl Acad Sci U S A 96: 3848–3853.
- 15. Dustin LB,Rice CM (2007) Flying Under the Radar: The Immunobiology of Hepatitis C. Annu Rev Immunol 25: 71–99.
- 16. Kim AY,Kuntzen T,Timm J,Nolan BE,Baca MA,et al. (2011) Spontaneous control of HCV is associated with the expression of HLA-B*57 and preservation of targeted epitopes. Gastroenterology 140: 686–696.
- 17. Kuniholm MH,Kovacs A,Gao X,Xue X,Marti D,et al. (2010) Specific human leukocyte antigen class I and II alleles associated with hepatitis C virus viremia. Hepatology 51: 1514–1522.
- 18. Thio CL,Gao X,Goedert JJ,Vlahov D,Nelson KE,et al. (2002) HLA-Cw*04 and Hepatitis C Virus Persistence. J Virol 76: 4792–4797.
- 19. Chuang WC-M,Sarkodie F,Brown CJ,Owusu-Ofori S,Brown J,et al. (2007) Protective effect of HLA-B57 on HCV genotype 2 infection in a West African population. J Med Virol 79: 724–733.
- 20. MacNamara ,Rowan ,Hilburn ,Kadolsky ,Fujiwara ,et al. (2010) HLA class I binding of HBZ determines outcome in HTLV-1 infection. PLoS Pathog 6: e1001117.
- 21. MacNamara A,Kadolsky U,Bangham CR,Asquith B (2009) T-cell epitope prediction: rescaling can mask biological variation between MHC molecules. PLoS Comput Biol 5: e1000327.
- 22. Norman PJ,Cook MA,Carey BS,Carrington CV,Verity DH,et al. (2004) SNP haplotypes and allele frequencies show evidence for disruptive and balancing selection in the human leukocyte receptor complex. Immunogenetics 56: 225–237.
- 23. Winter CC,Gumperz JE,Parham P,Long EO,Wagtmann N (1998) Direct binding and functional transfer of NK cell inhibitory receptors reveal novel patterns of HLA-C allotype recognition. J Immunol 161: 571–577.
- 24. O'Connor GM,Guinan KJ,Cunningham RT,Middleton D,Parham P,et al. (2007) Functional polymorphism of the KIR3DL1/S1 receptor on human NK cells. J Immunol 178: 235–241.
- 25. Yawata M,Yawata N,Draghi M,Little AM,Partheniou F,et al. (2006) Roles for HLA and KIR polymorphisms in natural killer cell repertoire selection and modulation of effector function. J Exp Med 203: 633–645.
- 26. Vilches C,Parham P (2002) KIR: diverse, rapidly evolving receptors of innate and adaptive immunity. Annu Rev Immunol 20: 217–251.
- 27. Fadda L,Borhis G,Ahmed P,Cheent K,Pageon SV,et al. Peptide antagonism as a mechanism for NK cell activation. Proc Natl Acad Sci U S A 107: 10160–10165.
- 28. Ahlenstiel G,Martin MP,Gao X,Carrington M,Rehermann B (2008) Distinct KIR/HLA compound genotypes affect the kinetics of human antiviral natural killer cell responses. J Clin Invest 118: 1017–1026.
- 29. Thomas DL,Thio CL,Martin MP,Qi Y,Ge D,et al. (2009) Genetic variation in IL28B and spontaneous clearance of hepatitis C virus. Nature 461: 798–801.
- 30. Ugolini S,Arpin C,Anfossi N,Walzer T,Cambiaggi A,et al. (2001) Involvement of inhibitory NKRs in the survival of a subset of memory-phenotype CD8+ T cells. Nat Immunol 2: 430–435.
- 31. Young NT,Uhrberg M,Phillips JH,Lanier LL,Parham P (2001) Differential expression of leukocyte receptor complex-encoded Ig-like receptors correlates with the transition from effector to memory CTL. J Immunol 166: 3933–3941.
- 32. Gati A,Guerra N,Gaudin C,Da Rocha S,Escudier B,et al. (2003) CD158 Receptor Controls Cytotoxic T-Lymphocyte Susceptibility to Tumor-Mediated Activation-Induced Cell Death by Interfering with Fas Signaling. Cancer Res 63: 7475–7482.
- 33. Young NT,Uhrberg M (2002) KIR expression shapes cytotoxic repertoires: a developmental program of survival. Trends Immunol 23: 71–75.
- 34. Anfossi N,Pascal V,Vivier E,Ugolini S (2001) Biology of T memory type 1 cells. Immunol Rev 181: 269–278.
- 35. Bieganowska K,Hollsberg P,Buckle GJ,Lim DG,Greten TF,et al. (1999) Direct analysis of viral-specific CD8+ T cells with soluble HLA-A2/Tax11-19 tetramer complexes in patients with human T cell lymphotropic virus-associated myelopathy. J Immunol 162: 1765–1771.
- 36. Bonorino P,Leroy V,Dufeu-Duchesne T,Tongiani-Dashan S,Sturm N,et al. (2007) Features and distribution of CD8 T cells with human leukocyte antigen class I-specific receptor expression in chronic hepatitis C. Hepatology 46: 1375–1386.
- 37. Soderquest K,Walzer T,Zafirova B,Klavinskis LS,Polic B,et al. Cutting edge: CD8+ T cell priming in the absence of NK cells leads to enhanced memory responses. J Immunol 186: 3304–3308.
- 38. Takao S,Ishikawa T,Yamashita K,Uchiyama TThe rapid induction of HLA-E is essential for the survival of antigen-activated naive CD4 T cells from attack by NK cells. J Immunol 185: 6031–6040.
- 39. Alter G,Rihn S,Streeck H,Teigen N,Piechocka-Trocha A,et al. (2008) Ligand-independent exhaustion of killer immunoglobulin-like receptor-positive CD8+ T cells in human immunodeficiency virus type 1 infection. J Virol 82: 9668–9677.
- 40. Poon K,Montamat-Sicotte D,Cumberbatch N,McMichael AJ,Callan MF (2005) Expression of leukocyte immunoglobulin-like receptors and natural killer receptors on virus-specific CD8+ T cells during the evolution of Epstein-Barr virus-specific immune responses in vivo. Viral Immunol 18: 513–522.
- 41. van der Veken LT,Campelo MD,van der Hoorn MA,Hagedoorn RS,van Egmond HM,et al. (2009) Functional analysis of killer Ig-like receptor-expressing cytomegalovirus-specific CD8+ T cells. J Immunol 182: 92–101.
- 42. Vlahov D,Anthony JC,Munoz A,Margolick J,Nelson KE,et al. (1991) The ALIVE study, a longitudinal study of HIV-1 infection in intravenous drug users: description of methods and characteristics of participants. NIDA Res Monogr 109: 75–100.
- 43. Goedert JJ,Kessler CM,Aledort LM,Biggar RJ,Andes WA,et al. (1989) A prospective study of human immunodeficiency virus type 1 infection and the development of AIDS in subjects with hemophilia. N Engl J Med 321: 1141–1148.
- 44. Hilgartner MW,Donfield SM,Willoughby A,Contant CF,Evatt BL,et al. (1993) Hemophilia growth and development study. Design, methods, and entry data. Am J Pediatr Hematol Oncol 15: 208–218.
- 45. Bunce M,O'Neill CM,Barnardo MCNM,Krausa P,Browning MJ,et al. (1995) Phototyping: comprehensive DNA typing for HLA-A, B, C, DRB1, DRB3, DRB4, DRB5 & DQB1 by PCR with 144 primer mixes utilizing sequence-specific primers (PCR-SSP). Tissue Antigens 46: 355–367.
- 46. Martin MP,Gao X,Lee J-H,Nelson GW,Detels R,et al. (2002) Epistatic interaction between KIR3DS1 and HLA-B delays the progression to AIDS. Nat Genet 31: 429–434.
- 47. Uhrberg M,Valiante NM,Shum BP,Shilling HG,Lienert-Weidenbach K,et al. (1997) Human diversity in killer cell inhibitory receptor genes. Immunity 7: 753–763.
- 48. Nagai M,Usuku K,Matsumoto W,Kodama D,Takenouchi N,et al. (1998) Analysis of HTLV-I proviral load in 202 HAM/TSP patients and 243 asymptomatic HTLV-I carriers: high proviral load strongly predisposes to HAM/TSP. J Virol 4: 586–593.
- 49. Thio CL,Gao X,Goedert JJ,Vlahov D,Nelson KE,et al. (2002) HLA-Cw*04 and hepatitis C virus persistence. J Virol 76: 4792–4797.
- 50. Cohen J (1988) Statistical Power Analysis for the Behavioral Sciences.
- 51. Yang X,Yu X (2009) An introduction to epitope prediction methods and software. Rev Med Virol 19: 77–96.
- 52. Borghans JAM,Mølgaard A,de Boer RJ,Keşmir C (2007) HLA Alleles Associated with Slow Progression to AIDS Truly Prefer to Present HIV-1 p24. PLoS ONE 2: e920–e920.