KIR3DL1 is a highly polymorphic killer cell Ig-like receptor gene with at least 23 alleles described, including its activating counterpart, KIR3DS1. Recently, the KIR3DS1 allele has been shown to slow progression to AIDS in individuals expressing HLA-Bw4 with isoleucine at position 80. However, due to the lack of a specific Ab, KIR3DS1 expression and function is not well characterized. In this study, we demonstrate KIR3DS1 expression on a substantial subset of peripheral natural killer cells through its recognition by the mAb Z27. The fidelity of this detection method was confirmed by analysis of KIR3DS1 transfectants and the identification of a novel KIR3DS1 null allele. Interestingly, KIR3DS1 is also expressed by a small proportion of CD56+ T cells. We show that ligation of KIR3DS1 by Z27 leads to NK cell IFN-γ production and degranulation as assessed by expression of CD107a. Furthermore, we document the persistence of KIR3DS1+ NK cells in HIV-1 viremic patients. The high frequency of KIR3DS1 expression, along with its ability to activate NK cells, and its maintenance during HIV-1 viremia are consistent with the epidemiological data suggesting a critical role for this receptor in controlling HIV-1 pathogenesis.
Natural killer cell activity is controlled in part by receptors belonging to the killer cell Ig-like receptor (KIR)4 family, which includes inhibitory KIR that specifically recognize HLA class I molecules (1). The stochastic expression of a subset of the KIR repertoire on individual NK cells permits efficient detection of alterations in MHC class I expression resulting from viral invasion, thus allowing peripheral NK cells to eliminate infected cells (2). Although KIR-meditated regulation of NK cells in vitro has been extensively investigated, our understanding of the in vivo role of KIR is largely limited to analysis of genetic association across disease cohorts due to the lack of specific anti-KIR Abs. These studies have now implicated several KIR in the pathophysiology of a variety of human diseases including cancer, autoimmune disease, and infectious diseases such as AIDS and hepatitis C (3).
The KIR gene cluster resides within the leukocyte receptor complex on chromosome 19q13.4 (4). Genotypic analysis has identified >30 distinct haplotypes comprised of 4–14 KIR genes suggesting active, rapid evolution of the KIR locus (5, 6). KIR genes encode receptors with either two or three extracellular Ig-like domains, known as KIR2D or KIR3D, respectively. The inhibitory receptors possess long cytoplasmic tails (i.e., KIR2DL or KIR3DL) containing a canonical ITIM that becomes tyrosine phosphorylated upon KIR engagement of their HLA class I ligands. The phosphorylated ITIM recruits the Src homology 2 domain containing protein tyrosine phosphatases Src homology 2 domain-containing phosphatase 1 and/or Src homology 2 domain-containing phosphatase 2, which dephosphorylate cellular substrates, thus aborting the NK activation signal, i.e., sparing target cells with appropriate self-MHC class I expression (7, 8). In addition to the inhibitory receptors, the KIR cluster encodes receptors with short (i.e., KIR2DS or KIR3DS) cytoplasmic tails that lack ITIMs. These activating KIR contain a charged residue within their transmembrane domain facilitating interaction with the signaling chain KARAP/DAP12. Engagement of the KIR2DS family of receptors has been shown to lead to a cascade of KARAP/DAP12-mediated signaling events culminating in increased NK cell cytolytic activity and the production of proinflammatory cytokines such as IFN-γ (8, 9). Although several inhibitory KIR have been described to bind HLA class I molecules, and Stewart et al. (10) recently demonstrated the binding of KIR2DS1 to HLA-C, the high-affinity ligands of most activating KIR remain poorly defined.
Individual peripheral blood NK cells express a complex KIR repertoire that is determined by multiple factors, including gene content and interactions with host MHC class I molecules during NK cell development (11). Although each intact KIR gene of a given individual is thought to be represented on the surface of some NK cells, not every NK cell will express all of the KIR present within the haplotype. The molecular and/or biochemical mechanisms underlying this stochastic expression are still unclear. However, Saleh et al. (12) recently described a bidirectional promoter element that functions as a probabilistic switch controlling the stochastic expression pattern of the murine Ly49 genes (12) and bidirectional KIR promoters have been identified (13, 14).
KIR3DL1 is a highly polymorphic inhibitory gene with 23 known alleles including an apparent activating counterpart, KIR3DS1, making the KIR3DL1/S1 locus unique within the KIR cluster (http://www.ebi.ac.uk/ipd/kir/align.html). Previous studies correlated KIR3DL1 polymorphisms with the binding pattern of two anti-KIR3DL1 mAb clones, DX9 and Z27. These data allowed the characterization of three distinct levels of mean channel fluorescence intensities (MFI) referred as low (KIR3DL1*005, *006, *007), high (KIR3DL1*001, *002, *008, *015, *020), and null (KIR3DL1*004) KIR3DL1-binding patterns (15, 16). Further analysis suggested that the distinct MFI patterns were indeed a reflection of receptor density on the cell surface and not different affinities of the Ab for the various allotypes (16). KIR3DL1 is known to bind multiple HLA-B allotypes that possess the public serological epitope Bw4 determined by residues 77–83 of the HLA α1 domain (17). However, KIR3DL1-expressing NK cell cytolysis of target cells bearing those HLA molecules is most efficiently inhibited with HLA-Bw4 allotypes presenting an isoleucine at position 80 (Bw4*80I) (18). Recently, Thananchai et al. (19) expanded the list of known KIR3DL1 ligands by directly demonstrating its ability to recognize specific HLA-A molecules of the serologically defined epitope Bw4. Moreover, this study showed differential binding of various KIR3DL1 allotypes to HLA-Bw4 molecules folded with various peptides, suggesting a high degree of specificity in this KIR-HLA interaction.
In contrast to KIR3DL1, very little is known regarding the activating allele of the locus KIR3DS1. Despite sequence similarity >95% in the extracellular domains of KIR3DL1 and KIR3DS1, recent attempts at demonstrating KIR3DS1 interactions with HLA-Bw4 tetramers containing various peptides failed to show any binding (20). However, genetic association studies have suggested a role for KIR3DS1 in the outcome of multiple diseases associated with viral infection. For example, the absence of KIR3DS1 is associated with the development of cervical cancer in human papillomavirus infection (21, 22). In addition, KIR3DS1 is beneficial in the resolution of acute hepatitis C virus infection (23) and protective against the development of hepatocellular carcinoma in chronically infected HLA-Bw4*80I patients (24). One of the most striking effects of KIR3DS1 is seen in HIV-1 infection. KIR3DS1 is correlated with significantly slower progression to AIDS, but this effect is only seen in patients with both KIR3DS1 and HLA-Bw4*80I (25).
Taken together, these findings suggest that interactions between KIR3DS1 and HLA-Bw4 play a role in human disease pathophysiology. Moreover, the dual requirement for HLA-Bw4*80I and KIR3DS1 in these studies suggests that this activating KIR may also interact with a subset of HLA-Bw4 molecules. Despite the mounting evidence suggesting a role for KIR3DS1 in infectious disease, no direct interactions between KIR3DS1 and HLA-Bw4 have been documented. In fact, due to the lack of specific anti-KIR3DS1 Abs, expression and function is not well characterized. In this study, we characterize, in a large-scale study, KIR3DS1 expression on the surface of a substantial subset of peripheral blood NK cells and a small proportion of CD56+ T cells. Moreover, we describe a novel KIR3DS1 allele that is not expressed on NK cells. In addition, we demonstrate that KIR3DS1 ligation activates NK cells to degranulate and produce IFN-γ. Finally, we report that KIR3DS1 can be detected on a subset of NK cells from HIV-1 viremic patients. Our findings are consistent with the proposed role of KIR3DS1 in the pathophysiology of HIV-1 infection.
Materials and Methods
Donors and HIV-infected patients
Healthy volunteers were recruited through the National Cancer Institute-Frederick Research Donor Program. HIV-1-infected patients were included as described previously (26) after approval by either the Massachusetts General Hospital Institutional Review Board or a National Institute of Allergy and Infectious Diseases Institutional Review Board as applicable. Each subject gave informed consent for participation in the study.
The mAb used in this study were: purified or PE-conjugated anti-CD158e1 (KIR3DL1, ZIN276, IgG1); FITC-conjugated anti-CD3 (IgG1, UCHT1); allophycocyanin- or PC5-conjugated anti-CD56 (IgG1, NKH.1); PE-conjugated anti-IFN-γ (IgG1, 45.15; Beckman Coulter); purified, FITC-, or PE-conjugated anti-CD158e1 (KIR3DL1, DX9, IgG1; Biolegend); purified anti-CD16 (3G8, IgG1; BD Pharmingen); PE-conjugated anti-CD16 (B73.1, IgG1); PE-conjugated anti-CD107a (lysosomal-associated membrane protein 1, H4A3, IgG1); PE-Cy5- or allophycocyanin-Cy7-conjugated anti-CD16 (IgG1, 3G8); allophycocyanin-Cy7-conjugated anti-CD3 (SK7, IgG1); PE-Cy5-conjugated anti-CD56 (B159, IgG1; BD Biosciences); PE-Alexa Fluor 700-conjugated anti-CD3 (S4.1, IgG2a; Invitrogen Caltag); and Pacific Blue-conjugated anti-CD3 (IgG2a, OKT3; eBioscience). Appropriately labeled mouse isotype control mAb were purchased from the respective companies.
Flow cytometric analysis of CD158e1 (KIR3DL1) expression on peripheral NK cells
The proportion of CD158e1 (KIR3DL1) expressing NK cells in control donors was assessed on whole blood by three- or four-color flow cytometry. Briefly, 100 μl of EDTA-treated blood was incubated with the appropriate mixture of mAb. Erythrocytes were lysed with an ammonium chloride solution and the pelleted cells were analyzed with a FACSort flow cytometer (BD Biosciences). Events were collected in the lymphocyte gate and analyzed. NK cells were defined as CD3−CD56+ lymphocytes. KIR3DS1 expression on NK cells was investigated by using DX9 and ZIN276 (Z27) Abs. Results were expressed as percentages of NK cells positive for KIR3DS1. Because the expression of CD56 but not CD16 has been described to decrease on NK cells from HIV-1-infected patients, KIR3DS1 expression was assessed on CD3−CD16+CD56+/− NK cells by five-color flow cytometry. Frozen PBMC samples from chronic or acute HIV-1-infected patients were thawed, stained, then fixed with a 1% paraformaldehyde solution before analysis with a LSRII flow cytometer (BD Biosciences). The direct comparative analysis of KIR3DS1 expression on the NK cells of healthy donors and HIV-1-infected individuals was performed by four-color flow cytometry using a FACSAria Special Order System flow cytometer (BD Biosciences).
KIR and HLA class I typing and KIR3DS1 sequencing
KIR3DL1/S1 and HLA-Bw4/Bw6 typing, as well as KIR3DS1 genomic sequencing were performed as previously described (http://www.ihwg.org/components/ssopr.htm).
The mouse mastocytoma P815 and the human embryonic kidney HEK293 cell lines were cultured in RPMI 1640 and DMEM, respectively, both supplemented with 10% FBS, 2 mM l-glutamine, 100 IU/ml penicillin, and 100 μg/ml streptomycin.
Generation of a KIR3DS1-TCRζ (KIR3DS1ζ) chimeric fusion protein
Expression of KIR3DS1 on HEK293 cells in the absence of the adaptor molecule KARAP/DAP12 was achieved by transfecting a chimeric construct containing the extracellular domain and stalk region of KIR3DS1 fused to the transmembrane and cytoplasmic domain of a CD8/TCRζ chimera, as described in the study by Irving et al. (27). A FLAG epitope was added to the carboxyl terminus of the KIR3DS1-TCRζ (KIR3DS1ζ) construct as a control for cell surface expression. HEK293 cells were transfected by using FuGene according to the manufacturer’s instructions (Roche Diagnostic) (28). KIR expression was assessed 24 h posttransfection by flow cytometry using both DX9 and Z27. The KIR3DL1*015 expression construct is described in detail elsewhere (19).
IL-2-activated NK cell preparation
PBMC from healthy donors were isolated from whole blood diluted 1/2 in PBS after centrifugation in lympho separation medium (ICN Pharmaceuticals). PBMC were further cultured in RPMI 1640 medium supplemented with 10% FBS, 2 mM l-glutamine, 100 IU/ml penicillin, 100 μg/ml streptomycin, and 10,000 IU/ml IL-2 for 5 days. Nonadherent cells were removed at day 3 of cell culture. NK cell functional assays were performed on day 6.
CD107a degranulation assay
The frequency of NK cell degranulation was analyzed through the expression of CD107a as described in the study by Fischer et al. (29). Briefly, IL-2-activated NK cells were incubated with P815 cells at an E:T ratio of 1:2 in the presence of anti-CD107a or the isotype-matched control mAb. NK cell degranulation was triggered by adding 1 μg of anti-CD16, Z27 or DX9. Treatment with PMA (10 ng/ml) and ionomycin (1 μg/ml) was used as a positive control. After 1 h, GolgiStop (BD Biosciences) was added and cells were incubated for another 2 h at 37°C. Induction of degranulation by KIR3DS1-expressing NK cells was explored by three-color flow cytometry.
The production of IFN-γ by NK cells after plate-bound mAb stimulation was investigated by intracellular flow cytometry using the BD Pharmingen Cytofix/Cytoperm with GolgiStop kit according to the manufacturer’s instructions. In brief, 5 × 105 IL-2-activated NK cells were plated for 6 h at 37°C in a 24-well flat-bottom plate, previously coated for 4 h at 37°C with 10 μg of anti-CD16, Z27, DX9, or isotope control. Treatment with PMA (10 ng/ml) and ionomycin (1 μg/ml) was used as a positive control. Results are expressed as percentage of IFN-γ+CD3−CD56+ NK cells.
Determination of viral load in HIV-infected patients
Viral loads were determined using the Food and Drug Administration-approved Amplicor HIV-1 Monitor test, version 1.5 (Roche Molecular Diagnostics) according to the manufacturer’s instructions. For samples that fell below the lower detection limit of 400 RNA copies/ml, the ultrasensitive assay with a detection limit of 50 RNA copies/ml was applied where sufficient material was available.
Graphic representation and descriptive statistics were obtained by using GraphPad Prism software. Comparisons of distributions were done using a Kruskal-Wallis test (one-way ANOVA) or a Mann-Whitney U test. Correlation was tested using a Spearman test. A p < 0.05 was considered to be significant.
Identification of the KIR3DS1+ NK cell subset
A previous study correlated KIR3DL1 polymorphisms with characteristic allele-based binding patterns of the anti-KIR3DL1 Abs DX9 and Z27 (15, 16). The authors defined three distinct levels of MFI identified as low (KIR3DL1*005, *006, *007), high (KIR3DL1*001, *002, *008, *015, *020), and null (KIR3DL1*004) KIR3DL1-binding patterns. Considering the similarity between KIR3DL1 and KIR3DS1, we tested whether DX9 or Z27 might also reveal KIR3DS1 expression on NK cells. Fig. 1 shows different patterns observed for Z27 and DX9 binding on peripheral blood NK cells from four healthy donors. Consistent with previous reports, both Z27 and DX9 react in a similar fashion with unimodal high (Fig. 1,A), bimodal high/low (Fig. 1,B), and low (data not shown) KIR3DL1 allotypes. However, we could detect in some donors a subset of NK cells that produce a dim peak with Z27 but not with DX9 (Fig. 1, C and D). This Z27dim subset was clearly distinct from isotype control staining and resulted in lower MFI than expected even for low KIR3DL1 allotypes (Fig. 1,D). In addition, this Z27dim binding appeared to be independent of coexpressed, previously defined null (Fig. 1C), high (data not shown), or low (Fig. 1,D) KIR3DL1 allotypes. Genotyping demonstrated that the Z27dimDX9− NK subset was detected only in KIR3DS1-positive donors (Fig. 1,E). These data suggested that Z27, but not DX9, might recognize KIR3DS1 on the surface of NK cells. We therefore transfected HEK293 cells with expression constructs encoding a KIR3DS1ζ chimera or KIR3DL1 to directly test the KIR3DS1 reactivity of Z27 (Fig. 1 F). As expected, staining with Z27 showed clear reactivity with cells expressing either the KIR3DS1ζ chimera or KIR3DL1, whereas DX9 only reacted with KIR3DL1 transfectants.
Expansion of our phenotypic analysis to a large cohort of donors with known KIR genotypes (n = 67) confirmed that KIR3DL1 homozygotes (n = 11) lacked Z27dimDX9− NK cells (Fig. 2,A). In contrast, we detected a Z27dimDX9− NK subset in only 53 of 56 individuals with one or two KIR3DS1 alleles. Despite our analysis of multiple donors and the demonstration that Z27 directly reacts with KIR3DS1 transfectants, the possibility remained that Z27dimDX9− NK cells might represent a subset expressing a novel KIR variant or other unknown, cross-reactive receptor in linkage disequilibrium with KIR3DS1. This possibility was suggested by our identification of three individuals (subjects 643, 091, and 702) who, although positive for a KIR3DS1 allele, lack Z27dimDX9− NK cells (Fig. 2,A, arrow). Fig. 2,B shows the DX9- and Z27-binding patterns of NK cells from these donors compared with what is typically observed for their respective KIR genotypes (KIR3DS1/L1*004 for subjects 643 and 091 and KIR3DS1/L1*002 for subject 702). Remarkably, genomic sequencing of KIR3DS1 demonstrated that these three unrelated individuals carry an identical single bp deletion along with a G to A substitution within exon 4 (Fig. 2 C). The shift in reading frame caused by this deletion results in a premature termination codon 66 bp downstream of the deletion. Translation of this KIR3DS1null allele would, therefore, be predicted to result in a protein product lacking the Ig-D2, stalk, transmembrane, and cytoplasmic domains. Along with our flow cytometric analysis, these genetic findings prove that KIR3DS1 is expressed on the surface of NK cells. Thus, from this point forward, we will refer to the Z27dimDX9− NK subset as KIR3DS1+ NK cells.
The discovery that Z27 can detect KIR3DS1 cell surface expression allowed, for the first time, a complete characterization of KIR3DS1 distribution on PBL from healthy donors. As shown in Fig. 2,A, the frequency of KIR3DS1+ NK cells was nearly 2-fold greater (p = 0.0015) in KIR3DS1 homozygous donors (61 ± 7%, 41–81%, n = 5) than in donors heterozygous for KIR3DS1/L1 (34 ± 1%, 15–58%, n = 48), consistent with independent transcriptional regulation of the two KIR3DS1 alleles. As expected, the lowest frequency of KIR3DS1+ NK cells observed in the KIR3DS1/S1 group (41%) was in a KIR3DS1/KIR3DS1null heterozygous individual. Interestingly, we could also detect a small proportion of Z27dimDX9−CD3+ T cells in 11 of 53 donors positive for at least one KIR3DS1 allele. An example of KIR3DS1 expression on CD3+ T cells is presented for two donors, one homozygous (Fig. 3,A, top) and one heterozygous (Fig. 3,A, bottom) for KIR3DS1. These cells were found to be largely CD56+CD3+ T cells (data not shown). Further analysis showed that KIR3DS1 is expressed on 3.7 ± 0.5% (1–6%, n = 10) of the CD56+CD3+ T cell subset (Fig. 3,A, right panels), except for one KIR3DS1 homozygous donor, where KIR3DS1 expression was detected on 26% of this subset, thus representing 6% of this individual’s total CD3+ T cell population (Fig. 3,B). Moreover, analysis of this donor’s KIR3DS1+ T cells revealed very high levels of CD8 expression, suggesting they may be CTLs (Fig. 3 B).
Exploration of KIR3DS1 expression on NK cells from KIR3DS1/L1 heterozygotes allowed us to assess the frequency of expression on the NK cell population in the presence of various KIR3DL1 subtypes (Fig. 4 A). This analysis indicated a statistically significant variability of expression on the NK cell population with respect to the KIR3DL1 allele present on the opposite haplotype (ANOVA, p = 0.0269). We found the largest KIR3DS1+ NK cell subsets in donors possessing KIR3DL1*004 as the opposite haplotype (39 ± 2%, 28–52%, n = 11), whereas those carrying KIR3DL1*001 (25 ± 4%, 15–41%, n = 6, p = 0.0067), KIR3DL1*005 (25 ± 4%, 15–41%, n = 6, p = 0.0207), KIR3DL1*002 (32 ± 2%, 19–41%, n = 11, p = 0.0568), and KIR3DL1*015 (32 ± 6%, 16–43%, n = 4, p = 0.4333) alleles had a lower frequency of KIR3DS1 expression relative to KIR3DL1*004 individuals. Taken together, these data demonstrate that even though KIR3DS1 is highly expressed in KIR3DS1/L1 heterozygotes, the observed Z27dimDX9− NK cell subset is an underestimate of the total percentage of KIR3DS1-expressing NK cells, since KIR3DS1 expression is masked by the coexpression of brightly staining KIR3DL1 allotypes. In support of this idea, even though coexpression of KIR3DL1 and KIR3DS1 is not detectable by FACS analysis, RT-PCR of sorted DX9+Z27bright NK cells from a KIR3DS1/L1 heterozygote individual revealed the presence of KIR3DS1 mRNA (data not shown). The total percentage of NK cells expressing KIR3DS1 from a single allele is therefore best represented in KIR3DL1*004/KIR3DS1 donors, because KIR3DL1*004 is not expressed on the cell surface (30).
Genetic analysis has revealed a significant effect of KIR3DS1 on the progression to AIDS only in patients with HLA-Bw4*80I, suggesting KIR3DS1/HLA-Bw4*80I as a ligand/receptor pair. Therefore, we tested whether the presence of HLA-Bw4, as well as HLA-Bw4*80I, might affect the frequency of KIR3DS1+ NK cells in our cohort. In contrast to the associations reported for KIR3DL1, we did not find any statistically significant relationship between the presence of HLA-Bw4 or HLA-Bw4*80I and the frequency of KIR3DS1+ NK cells (Fig. 4 B and data not shown).
NK cell activation by KIR3DS1
The predicted protein sequence of KIR3DS1 suggests the ability to couple to DAP12 and transmit activating signals to NK cells. Along with the importance of KIR3DS1 in HIV-1 pathogenesis, this observation prompted us to investigate its functional capacity by examining the effects of receptor cross-linking on granule exocytosis (CD107a expression) and IFN-γ production. Potential suppression of Z27-mediated signals due to cross-reactivity with KIR3DL1 was prevented by performing functional assays exclusively on PBMC derived from KIR3DS1 or KIR3DL1 homozygous donors. KIR3DS1 expression was confirmed on IL-2-activated NK cells before each experiment (Fig. 5, A and B). We found that KIR3DS1 cross-linking with plate-bound Z27 Ab or through coculture with Z27-coated P815 target cells produced only a modest effect on freshly isolated PBMC or negatively selected NK cells (data not shown). In contrast, Z27 but not DX9 cross-linking triggered substantial IFN-γ secretion (Fig. 5,C) and degranulation (Fig. 5 D) in IL-2-activated KIR3DS1+ NK cells. As expected, NK cell activation by Z27 was only observed for individuals carrying the KIR3DS1 allele. Moreover, analysis of a donor expressing the KIR3DS1null allele did not show any activation via Z27 as compared with DX9 cross-linking (data not shown).
KIR3DS1 and HIV-1 pathogenesis
HIV-1 infection has been shown to cause dramatic changes in the NK cell compartment (31, 32, 33, 34, 35, 36, 37), including effects on the expression levels of a variety of NK receptors (31, 32, 35, 36, 38, 39, 40, 41). In this respect, the continued presence of KIR3DS1 on NK cells during HIV-1 viremia would be supportive of the hypothesis that this receptor plays an active role in the host response to infection. Therefore, cryopreserved PBMC of chronically HIV-1-infected, KIR3DS1/L1 individuals were analyzed for KIR3DS1 expression on NK cells. Due to the substantial down-regulation of the prototypic NK cell marker CD56 during HIV-1 infection, PBMC were stained for CD56, CD16, CD3, DX9, and Z27 and gated as described in Materials and Methods. Although KIR3DS1 expression was more difficult to appreciate in cryopreserved HIV-1-infected samples, we could still detect the presence of a Z27dimDX9− NK cell subset (Fig. 6,A). Fig. 6,B shows the results obtained from 12 viremic HIV-1-infected individuals plotted against their HIV-1 viral titers at the time of study. These data confirmed that KIR3DS1 can be detected on peripheral NK cells from HIV-1 patients, even with viral titers as high as 750,000 copies/ml. Although it appears that KIR3DS1 expression might change as a function of viral load (Spearman correlation test, r2 = 0.39860, p = 0.1993) or CD4 count (Fig. 6 C, Spearman correlation test, r2 = 0.22378, p = 0.4845), analysis of many more patients will be required before such a relationship with either viral load or CD4 count can be confirmed.
Because our initial data on HIV-1 patients suggested levels of KIR3DS1 expression on NK cells much lower than predicted by our screen of healthy donors, we performed a comparative analysis of KIR3DS1 expression on NK cells, where EDTA-treated peripheral blood samples of healthy donors and HIV-1-infected individuals were processed in parallel (Fig. 6,D). Our phenotypic analysis of HIV-1-infected individuals identified KIR3DS1 expression in 9 (25%) of 36 patients. Compared with the data from eight healthy donors known to carry at least one copy of KIR3DS1, our phenotypic analysis confirmed that the measures required to detect KIR3DS1 in HIV-1-infected individuals underrepresents actual receptor expression. Whereas, our large-scale analysis demonstrated 36 ± 2% (Fig. 2,A, 15–81%, n = 53) of NK cells to be KIR3DS1 positive in healthy donors, analysis under the conditions required for our clinical samples demonstrated only 10 ± 2% KIR3DS1-expressing NK cells (3–24%, n = 8). On average, HIV-1-infected individuals presented 13 ± 3% (3–35%, n = 9) KIR3DS1+ NK cells. These data clearly demonstrate that KIR3DS1 expression is maintained in HIV-1-positive individuals and suggest that the KIR3DS1+ NK cell subset is not significantly expanding or contracting in these patients (p = 0.8148). Only two patients included in this phenotypic comparative analysis stained for KIR3DS1 and suffered viral loads >50 copies/ml (79,010 and 130,714 copies/ml). The KIR3DS1-expressing NK cell subset of these patients represented 35 and 12% of the NK cell compartment, respectively, confirming the persistent expression of KIR3DS1 during HIV-1 viremia shown in Fig. 6. To further address possible regulation of KIR3DS1 during the course of HIV infection, we are continuing our analysis using this comparative flow cytometry analysis.
The genetic association of KIR3DS1 with slower progression to AIDS (25), more efficient clearance of hepatitis C virus (23), and reduced development of hepatocellular carcinoma (24) and cervical cancers (21, 22) has driven considerable interest in its function. Although this activating receptor has been known for quite some time, no ligand has been described, and expression on NK cells is only now being appreciated. The findings reported here demonstrate that KIR3DS1 is expressed on a considerable subset of peripheral blood NK cells as well as some CD56+ T cells, and cross-linking of the receptor on NK cells leads to activation. There are three lines of evidence that conclusively demonstrate that the Z27dimDX9− NK cell subset represents KIR3DS1-expressing cells. First, we have analyzed significant numbers of individuals with known KIR haplotypes and found Z27dimDX9− NK cells only in those with at least one copy of KIR3DS1 (Fig. 2,A). Second, Z27 readily detects HEK293 transfectants expressing either a KIR3DS1ζ chimeric protein or a KIR3DS1 protein modified so as not to require coexpression of KARAP/DAP12 (Fig. 1,F and data not shown). Third, the only donors found to carry the KIR3DS1 gene that lacked Z27dimDX9− NK cells all possess a novel KIR3DS1 allele with a frame-shift mutation resulting in a premature stop codon that prevents cell surface expression (Fig. 2, B and C).
Analysis of KIR3DS1 distribution on PBL revealed remarkable results in both frequency and pattern of distribution. KIR3DS1 was expressed on a surprisingly large subpopulation of NK cells. Indeed, >30% of NK cells from KIR3DS1/L1 heterozygotes expressed the receptor and nearly two-thirds when donors are homozygous for KIR3DS1. The expression of KIR3DS1 is thus higher than most of the KIR described to date, with the exception of KIR2DL4, which is expressed on all NK cells (42). It remains unclear whether the low MFI of Z27 detected here reflects low KIR3DS1 levels on the cell surface or inefficient detection due to reduced affinity of Z27 for KIR3DS1 relative to KIR3DL1. Although the KIR3DS1ζ chimeric receptor was detected on nearly 100% of HEK293 transfectants with anti-FLAG Ab, a much smaller proportion of these cells reacted with Z27 (data not shown and Fig. 1 F). This suggests that Z27 has a low affinity for KIR3DS1 and the receptor density on NK cells could be higher than indicated by Z27. Development of specific anti-KIR3DS1 Abs could address this possibility.
In donors heterozygous for KIR3DS1, the modulation of KIR3DS1 expression frequency with the various KIR3DL1 alleles present on the opposite haplotype suggests that some NK cells coexpress both the activating and inhibitory allotypes encoded by this locus. We also confirmed this coexpression by RT-PCR on sorted NK cells. Based on the product rule of stochastic KIR expression, we predict that ∼10% of NK cells coexpress KIR3DL1 and KIR3DS1. Extensive analysis of the expression of KIR3DL1 allotypes has suggested that both their frequency and density on NK cells is regulated by the presence of cognate MHC class I ligands (16). Yawata et al. demonstrated that donors with two high KIR3DL1 alleles showed an increase in KIR3DL1-expressing NK cell frequency in the presence of HLA-Bw4. This effect was greatest in donors heterozygous for HLA-Bw4. In this study, we did not detect any difference between KIR3DS1 expression in donors who are HLA-Bw4 positive relative to those with HLA-Bw6 allotypes. This observation may reflect an inability of KIR3DS1 to be regulated by its ligand in vivo. Alternatively, KIR3DS1 may only bind HLA-Bw4*80I as suggested by the genetic studies, but our current cohort does not contain enough of these donors to draw firm conclusions on this issue. Regardless, in light of our demonstration of KIR3DS1 expression and function on a significant percentage of NK cells, the elucidation of KIR3DS1 ligands is of primary importance.
Expression of inhibitory KIR on T cells has been reported on activated, memory, effector, or senescent T cells. Similarly, murine NK1.1+CD3+ T cells express inhibitory Ly49 molecules but not their activating counterparts. The definitive detection of KIR3DS1 on resting CD56+ T cells is intriguing given that, except for KIR2DS4 (43), evaluation of activating KIR cell surface expression has been prevented by Ab cross-reactivity between inhibitory and activating KIR isoforms. Indeed, expression of activating KIR2DS1 and KIR2DS2 on T cells has only been shown indirectly by using flow cytometry with Abs known to react with both activators and inhibitors in conjunction with either RT-PCR to demonstrate the presence of mRNA for an activating KIR, or redirected killing assays to show an increase in cytolytic activity. Along these lines, KIR2DS1 and/or KIR2DS2 expression has been suggested on CD8+TCRγδ+ T cells (44), EBV-selected synovial TCRαβ+ T cell clones derived from rheumatoid arthritis patients (45), and expanded CD8+CD28null T cells in melanoma patients (46). In addition, KIR2DS2 has been shown to be expressed by the CD4+CD28null T cells that expand in some patients with rheumatoid arthritis (47). In these T cells, cross-linking of KIR2DS2 can synergize with TCR signals, promoting full activation even in the absence of CD28 ligation (48). The signaling partner for the activating KIR in NK cells, KARAP/DAP12, is not expressed in all of these CD4+CD28null T cells, and its expression has not been described in CD56+ T cells. We speculate, therefore, that like KIR2DS2 in CD4+CD28null T cells, KIR3DS1 might be paired with an alternative signaling molecule in CD56+ T cells. Alternatively, some KIR3DS1+ T cells may possess KARAP/DAP12 in a manner analogous to the TCRαβ+CD8+CD94+NKG2C+ T cells that develop in patients with celiac disease (49). The ability of KIR3DS1 to trigger CD56+ T cell activation is currently under investigation as is the nature of the associated signaling chain used by these cells. Regardless, in light of our data demonstrating KIR3DS1 expression on CD8+CD56+ T cells, one can no longer restrict the interpretation of the genetic association between KIR3DS1 expression and HIV-1 progression as solely a NK cell effect.
In this study, we clearly demonstrate the ability of KIR3DS1 to mediate NK cell activation. This result is consistent with the presence of a positively charged residue within the receptor’s transmembrane domain and its proposed role in controlling viral infection. It is worth noting that stimulation of fresh, resting NK cells demonstrated little if any KIR3DS1-mediated IFN-γ production. In contrast, we could readily detect both IFN-γ production and CD107a expression in response to Z27 stimulation of IL-2-activated NK cells. These data suggest that preactivation is required for substantial KIR3DS1-mediated activation to occur. Whether this is inherent to KIR3DS1 or simply a result of the relatively low levels of Z27 binding to KIR3DS1 on NK cells is currently unclear.
Interestingly, stimulation of IL-2-activated NK cells from KIR3DS1 homozygous donors with plate-bound DX9 reproducibly induced low levels of IFN-γ, whereas this treatment had little effect on NK cells from KIR3DL1 homozygous individuals. This observation is most likely the consequence of low-level, DX9-induced NK activation via FcγRIIIα that is quenched by the concurrent ligation of KIR3DL1 in donors homozygous for this receptor. However, in some experiments, intentional ligation of FcγRIIIα with anti-CD16 Ab appears to induce greater levels of IFN-γ and CD107a expression in NK cells from KIR3DS1 homozygous donors as compared with those from KIR3DL1 homozygotes (Fig. 5). We currently have no explanation for this enhanced FcγRIIIα function in NK cells from some KIR3DS1 homozygous individuals, but it should be noted that we did not detect any difference between the levels of FcγRIIIα expression observed in KIR3DS1/S1 and KIR3DL1/L1 donors (data not shown).
Several KIR3DS1 alleles are recorded in databases but direct sequencing of multiple KIR haplotypes has identified only one, KIR3DS1*013 (http://www.ebi.ac.uk/ipd/kir/align.html). Similarly, our extensive KIR analysis has failed to detect any KIR3DS1 alleles other than *013 and the null allele we describe here. A detailed analysis of the distribution of this novel allele is underway and a re-evaluation of the studies of disease cohorts implicating KIR3DS1 may even be necessary, since the presence of this null allele could have obscured positive associations.
Previous studies have reported various effects of acute HIV-1 infection on NK cell expression of KIR (31, 32, 35, 36, 38, 39, 40, 41). The demonstration of KIR3DS1 expression on NK cells from HIV-1 viremic patients reported here lends further support to the notion that this activating KIR plays a role in the regulation of HIV-1 infection. Extensive analysis of the effect of KIR3DS1 expression on viral load in different stages of HIV-1 disease is now clearly warranted. It is important to note that we have found the detection of KIR3DS1 to be highly dependent on the methodology and instrumentation used. Indeed, analysis of known KIR3DS1+ healthy donors using the method required for proper identification of CD56dim and/or negative NK cells in clinical samples, dramatically underestimates KIR3DS1 detection. Thus, one must exercise caution when evaluating potential alterations in the KIR3DS1-expressing NK cell subset in HIV-1-infected patients. Finally, it will be of significant interest to investigate the expression of KIR3DS1 over the course of infection in patients stratified by HLA allotype.
In conclusion, we have demonstrated in a large-scale study that: 1) KIR3DS1+ NK cells can be detected as a Z27dimDX9− subset; 2) KIR3DS1 is expressed by a large percentage of peripheral blood NK cells (>70% in some donors); 3) a subset of NK cells coexpress KIR3DS1 and its inhibitory counterpart KIR3DL1; 4) a null allele of KIR3DS1 exists and is not expressed on the cell surface; 5) KIR3DS1 engagement by Z27 activates NK cells; and 6) KIR3DS1 expression is maintained in HIV-1-positive individuals. While this manuscript was in process, three other laboratories have similarly shown recognition of biologically active KIR3DS1 on NK cells by Z27 (50, 51, 52). These manuscripts also report the lack of KIR3DS1 recognition of HLA Bw4*80I when expressed on 721.221 cells, furthering our recent data showing that Bw4*80I tetramers loaded with various HIV-derived peptides fail to bind KIR3DS1. Taken together, all of these data are consistent with the involvement of KIR3DS1 in human disease, as suggested by genetic association, but demonstrate that the recognition of Bw4*80I by the receptor may require peptides other than those presented by 721.221 or derived from HIV-1-immunodominant epitopes (20). Clearly, additional study of this receptor in both healthy and virally infected individuals and vigorous pursuit of its putative ligands is now warranted.
We acknowledge the outstanding technical assistance provided by Anna T. Mason and Earl W. Bere, Jr.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This project has been funded in whole or in part with federal funds from the National Cancer Institute, the Intramural Research Program of the National Institute of Allergy and Infectious Diseases and the National Institutes of Health, under Contract DHHS N01-C0-12400 and 1 R01 AI067031-01A2. This research was supported in part by the Intramural Research Program of the National Institutes of Health, National Cancer Institute, Center for Cancer Research.
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. By acceptance of this article, the publisher or recipient acknowledges the right of the U.S. government to retain a nonexclusive, royalty-free license in and to any copyright covering the article.
Abbreviations used in this paper: KIR, killer Ig-like receptor; MFI, mean channel fluorescence intensity.