Background
An array of host genetic factors have been reported to alter HIV acquisition or markers of HIV disease progression [
1,
2]. Identifying and understanding these genetic correlates may accelerate preventive and therapeutic efforts.
The most widely replicated correlates of HIV acquisition include homozygosity in the 32-base pair deletion in
CCR5 that reduces acquisition risk. Although, in comparison, greater effort has been focused on identifying correlates of HIV disease progression there are similarly only a handful of widely replicated findings. Natural Killer (NK) cell and CD8+ T-cell responses stand out as being reproducibly implicated in HIV control [
3]. Concordantly, variation in
the Human Leukocyte Antigen (HLA) and
Killer-cell Immunoglobulin-like Receptor (KIR) loci, the two most polymorphic regions of the human genome that encode receptors involved in NK and CD8+ T-cell function, is associated with rates of HIV disease progression across several studies [
2]. Genome-wide association studies identify
HLA-B*57 alleles, and variants that affect HLA-C expression [
4‐
6] as important modifiers of HIV viraemia, the former of which had been identified in many earlier candidate gene studies.
Natural Killer cells are amongst the earliest responders to viral infection and mediate protective responses by secreting pro-inflammatory cytokines and by direct cytolysis of infected cells. Their function is governed, at least in part, by the combinatorial array of inhibitory and activating receptors including the KIR, Leukocyte immunoglobulin-like receptors (LILR), the C-type lectin receptors-NKG2A-F, and the natural cytotoxic receptors (NCRs) -NKp30, NKp44 and NKp46. The dominant regulators of NK cell recognition of virus-infected cells are thought to be KIR, because these are the natural receptors for HLA class I [
7]. Alteration in HLA expression and presentation of pathogen-derived or self-peptides is a common feature of many viral infections [
8]. Diversity in
KIR gene content, polymorphism and structural variation within the 14
KIR genes, and variation in expression confer additional variation in the ability of NK cells to identify and respond to virus-infected cells. KIRs show partial specificity in their recognition of HLA ligands: KIR3DL1 and KIR3DS1 recognise HLA-A and HLA-B molecules with the Bw4 epitope, KIR2DL1 recognises HLA-C molecules of the C2 group exclusively, KIR2DL3 recognises HLA-C of the C1 group exclusively and KIR2DL2 recognizes HLA-C of both C1 and C2 groups. The KIR genes segregate in two groups of haplotypes. Group A haplotypes consist of nine genes that encode predominantly inhibitory receptors, whereas group B haplotypes represent a more diverse collection of haplotypes based on gene content and contain more activating
KIRs compared with haplotype A.
Several combinations of
KIR and
HLA have been associated with protection from HIV acquisition [
9‐
13], though these findings are based on small sample sizes and none have been replicated in the literature. In contrast, a large study reported in 2002, found that
KIR3DS1 together with
HLA Bw4 alleles encoding Isoleucine at position 80 (80I) are associated with slower disease progression [
5]. Several additional lines of evidence from subsequent observational and functional studies support a role for KIR in HIV control. NK cells are expanded in primary HIV infection [
14], the expansion is modified by specific KIR and ligand repertoires [
15], the degree of HLA C ligand expression associates with protection [
4,
16] and viruses sequenced from individuals with specific
KIR show evidence of ‘escape’ mutation at sites that appear to alter recognition by KIR [
17]. As in infection with EBV [
18], CMV [
19], HCV or HTLV-1 [
20], KIR have also been shown to modulate T-cell responses to HIV [
21].
We studied the role of
HLA and
KIR on risk of HIV acquisition and course of HIV viraemia and CD4+ T-cell counts through the first 5–10 years of infection in South African women infected with HIV-1 clade C in a nested case–control and prospective cohort study respectively. In prior studies in this cohort we have reported that a) innate immune cell activation is associated with enhanced HIV acquisition [
22]; b) HIV-directed NK cells secreting IFN-y are associated with reduced HIV acquisition risk [
23]; c) HIV acquisition results in profound alteration in NK cell function [
24] and d) the rate of HIV disease progression is more rapid than similar cohorts elsewhere [
25,
26]. This current study strongly implicates
KIR2DL2, belonging to the group B haplotype with HIV control.
Discussion
KIR and HLA haplotypes have been associated with HIV disease outome, but little is known about their role in HIV acquisition. Using a large cohort of prospectively enrolled women at risk for HIV, we did not observe evidence of association between KIR profiles and HIV-1 acquisition. This finding is consistent with recent studies that suggest that genetic variation does not explain a substantial proportion in liability to acquire HIV-1 [
33,
34]. In contrast, we found that KIR genotype BB, which encodes KIR2DL2, was associated with lower HIV viraemia and higher CD4+ T-cell counts sustained over more than 5 years in a cohort of more than 130 prospectively followed South African women from a homogenous ethnic background.
Lack of association between KIR haplotype and HIV-1 acquisition observed here are in contrast to previous smaller studies linking higher activating:inhibitory KIR receptor repertoires, and the presence of
KIR3DS1 in particular, and genotype AB, with reduced HIV acquisition [
10]. The
KIR3DS1 gene is infrequent in the population under study, which is typical for populations of African descent [
35]. Our study, despite being larger than previous studies, was powered to identify covariates with unadjusted odds ratio <0.63 or >1.6, hence we may commit type 2 error if true effect sizes exist and are smaller. Although not feasible here, a more robust strategy would have been a prospective cohort analysis.
The association between KIR and HIV disease course is in agreement with immunogenetic [
5] and functional studies [
36] as well as recent evidence that NK cell function is a feature of HIV-1 control in patients with poor CD8+ T-cell responses [
37]. Similarly, in HCV and HTLV-1 infection KIR2DL2 modifies HLA mediated effects on disease and is thereby implicated in protective responses [
20]. However, our data are in contrast with observations reported by Khakoo et al. on Hepatitis C infection, where KIR2DL3 homozygous individuals had superior resolution [
38]. These different outcomes may be explained by differences in pathogenesis between HCV and HIV-1. Our findings are also in contrast to a previous report of lower CD4+ T-cell counts in HIV infected KIR haplotype B carriers [
39]. However, this apparent discrepancy may be due to cross-sectional sampling in Jennes et al. [
39] leading to a frailty bias in selection of participants and consequent reversal of direction of effect as was recently observed in a separate study [
33]. Gaudieri et al. also reported that carriage of either of the haplotype B genes KIR2DS2 or KIR2DL2 was associated with more rapid CD4+ T-cell decline, but they did not specifically assess the combined haplotype [
40]. Nevertheless, these findings suggest the potential for population heterogeneity in effect and highlight the challenges in confidently delineating which KIR gene contributes to the haplotype effect.
Several underlying mechanisms may be involved in KIR2DL2 mediated enhancement of HIV control, or reciprocally of KIR2DL3 impairment. Firstly, the presence of KIR2DL2 associated footprints in virus sequenced from KIR2DL2+ donors supports a model in which KIR2DL2 may bind HIV-derived viral peptides presented by HLA-C [
17]. In vitro studies using viral peptide variants suggest that selected viral peptides enhance KIR2DL2 binding, resulting in NK cell inhibition and diminished degranulation, hence affording the virus a selection advantage [
41]. This may explain why the beneficial effect of KIR2DL2 may be lost later in infection. Whilst differing by only a few amino acids, KIR2DL2 has been reported to have higher affinity for HLA-CI than KIR2DL3, and KIR2DL2/2DL3 differ in their sensitivity to peptide bound in the HLA C groove offering a further potential explanation for how subtle differences may explain the divergent effects of KIR2DL2/L3 [
42]. Although Korner et al. [
15], show that KIR2DL2+ NK cells are functionally more potent in the presence of HLA-C1/C1, because KIR2DL2 mediates inhibitory signals following binding to HLA-C molecules of both HLA-C1 or HLA-C2, the absence of an interaction between KIR2DL2 and HLA-C1/C2 group in our study is compatible with this model [
43]. The linkage disequilibrium that we found between KIR2DL2 and KIR2DS2 was consistent with previous studies [
44] and implies that a role for KIR2DS2 cannot be excluded. An alternative mechanism is suggested by Schonberg et al. [
43] who report that due to differences in timing of KIR2DL2 and KIR2DL1 expression, the presence of KIR2DL2 may affect ligand-instructed NK cell education during development. Finally, KIR2DL2 may act indirectly to alter T-cell recognition of virally infected cells as described in HTLV-1 and HCV infection [
20].
Blockade of inhibitory KIR interaction with HLA-C has been pursued as therapeutic strategy in malignancy and viral infection [
45]. An antibody that blocks interaction of KIR2DL1/L2/L3/S1/S2 (1-7 F9) with HLA-C, was shown to enhance degranulation of NK cells from HIV infected donors cocultured with target cells in vitro [
46]. The enhancement in NK cell degranulation observed with 1-7 F9 was higher amongst NK cells from donors with KIR haplotype B in congruence with our observations. However, precise delineation of the role of KIR2DL2, KIR2DL3 and KIR2DS2 is difficult. This approach has been further developed in acute myeloid leukaemia (AML), multiple myeloma (MM) and lymphoma models [
47‐
49]. A humanised version (lirilumab) was shown to be safe [
50] in humans and is currently in clinical trials for treatment of AML (NCT01687387), MM (NCT02252263, NCT01592370), lymphoma (NCT01592370) and some solid organ tumours (NCT01750580, NCT01714739). We speculate that similar approaches may be beneficial in HIV.
In spite of having a pre-determined data analysis plan and consistent viral load and CD4+ T-cell measures across both sub-cohorts, we cannot exclude that this finding is a false positive due to multiple comparisons. Larger studies, that limit analyses to replication of this finding, are required. In addition, this study has other noteworthy limitations. Firstly, it is limited by the lack of resolution of specific KIR alleles and copy number. Secondly, in the case control analysis we were unable to account for the transmitting partners KIR status constraining our ability to confirm previous reports of KIR2DS4*001 carriage in the transmitting partner being linked to enhanced transmission risk [
51]. Thirdly, the limited sample size prohibited subgroup analyses, for example, of HLA C1/C2 ligand interactions. Finally, we were unable to link these clinical observations with previously reported flow-cytometry based measures of NK cell function [
23] due to the absence of KIR2DL2/L3 specific antibodies used in our previous work.
Acknowledgements
We thank the participants of the CAPRISA study cohorts; women who are dedicated and committed to improving their and their peers’ health and who donate samples to make this research possible. We thank Mary Carrington for helpful advice in preparation of this manuscript. This work was supported by the South African HIV/AIDS Research Platform (SHARP), and US National Institutes for Health FIC K01-TW007793. The parent trial (CAPRISA004) was supported by the United States Agency for International Development (USAID), Family Health International (FHI) co-operative agreement # GPO-A-00-05-00022-00, contract # 132119, and LIFElab, a biotechnology centre of the South African Department of Science & Technology. These studies were also supported by the TRAPS (Tenofovir gel Research for AIDS Prevention Science) Program, which is funded by CONRAD co-operative grant # GP00-08-00005-00, subproject agreement # PPA-09-046. We thank the US National Institutes for Health’s Comprehensive International Program of Research on AIDS (CIPRA grant # AI51794) for the research infrastructure. V.N. was supported by LIFElab, the Columbia University-South Africa Fogarty AIDS International Training and Research Program (AITRP #D43 TW000231) and the Rhodes Trust. M.A. is a Distinguished Clinical Scientist of the Doris Duke Charitable Foundation. W.H.C was supported by a Massachusetts General Hospital Physician Scientist Development Award. T.N. holds the South African Research Chair in Systems Biology of HIV/AIDS supported by the South African Department of Science and Technology through the National Research Foundation. T.N. received additional funding from the Victor Daitz Foundation and is a Howard Hughes Medical Institute International Early Career Scientist. VN and AVSH were partially supported through a grant from the Wellcome Trust which supports core facilities (090532/Z/09/Z).
Competing interests
The authors have no relevant conflict of interests to report.
Authors’ contributions
SSAK, QAK, AK, NG, SS and KM were involved in cohort accrual and clinical follow up. VN, DdAR, HH, CG, and RM performed the KIR and HLA typing. VN, LW, DC, MA, AVSH, WHC, SSAK, and TN led and performed analyses and all authors read and commented on the manuscript. All authors read and approved the final manuscript.