Imprint of human cytomegalovirus infection on the NK cell receptor repertoire

Killer immunoglobulin-like receptors (KIRs), CD94/NKG2 killer lectin-like receptors (KLRs), and CD85j (immunoglobulin-like transcript 2 [ILT2], leukocyte immunoglobulin-like receptor 1 (LIR1), leukocyte immunoglobulin-like receptor B1 [LILRB1]) specifically recognize HLA class I molecules and are expressed by natural killer and T-cell subsets.^1-4  Single cells²  bear variable combinations of these natural killer cell receptors (NKRs), presumably resulting from stochastic gene activation/silencing events that take place during their maturation.⁵  The diversity of NKRs observed in different individuals is in part genetically determined, because distinct KIR haplotypes include variable sets of genes.¹  On the other hand, there is evidence that microbial infections may also influence the NKR repertoire. In this regard, murine cytomegalovirus (MCMV) promotes an expansion of NK cells bearing the Ly49H receptor specific for the m157 viral glycoprotein,⁶  which plays a crucial role in the immune response to infection.^7-9  Moreover, increased proportions of CD8⁺ T cells with an effector/memory phenotype bearing inhibitory NKRs (ie, CD94/NKG2A) have been observed in mice infected by different viruses^10-12  as well as in human immunodeficiency virus (HIV)–infected patients.¹³ 

CD94, NKG2A, and NKG2C are C-type lectins encoded at the NK gene complex (NKC) in human chromosome 12.¹⁴  Surface expression of NKG2A/C molecules requires their covalent assembly with CD94.^4,15,16  The CD94/NKG2A heterodimer constitutes an inhibitory receptor that recruits the protein tyrosine phosphatase containing SH2 domain-1 (SHP-1) through the immunoreceptor tyrosine-based inhibitory motif (ITIM)–bearing NKG2A subunit, whereas CD94/NKG2C is coupled to a tyrosine kinase activation pathway through the DAP12 adapter.^17,18  In human beings, both NKRs specifically recognize HLA-E, which presents peptides derived from the signal sequences of other HLA class I molecules^19-22 ; HLA-E allotypes contain either an Arg (HLA-E^R107) or a Gly (HLA-E^G107) at position 107.²³  The biologic relevance of such structural dimorphism remains unclear, but it may affect surface expression levels of the class Ib molecule^24,25  and its interaction with CD94/NKG2 receptors.²⁶ 

A number of studies have addressed the characterization of CD94/NKG2A in NK and T cells, whereas little is known about the biologic role of the CD94/NKG2C dimer. According to the current hypotheses proposed to interpret the function of activating NKRs, either a selective down-modulation of the ligands for inhibitory receptors expressed by CD94/NKG2C⁺ cells or/and an increased avidity of CD94/NKG2C interaction with infected cells might favor NK cell activation through that pathway.⁴  In support of the first possibility, HLA-E appears selectively preserved from the action of some human cytomegalovirus (HCMV) proteins that interfere with HLA class Ia expression.^27,28  Moreover, the class Ib molecule can be stabilized by a peptide derived from the UL40 HCMV protein,^29,30  potentially allowing the pathogen to evade the response mediated by CD94/NKG2A⁺ cells.

In the present study we analyzed CD94/NKG2C expression in peripheral blood lymphocytes (PBLs) from healthy adult blood donors; moreover, other NKRs (ie, CD94/NKG2A, KIR, CD85j, CD161, NKG2D) and triggering NKp46 and NKp30 natural cytotoxicity receptors (NCRs)³¹  were also studied. HCMV infection as well as the HLA-E and KIR genotypes were assessed as potentially relevant variables related to CD94/NKG2C expression. Our observations support that HCMV, but not other herpesviruses (Epstein-Barr virus [EBV]; and herpes simplex virus, [HSV]), shapes the NKR repertoire of healthy carrier individuals, promoting an expansion of CD94/NKG2C⁺ NK and T cells.

Blood samples derived from a cohort of 70 healthy adult individuals, including 51 men and 19 women (age range, 20 to 56 years; median, 27 years) were analyzed. Written informed consent was obtained from every donor, and the study protocol was fully approved by the Institut Municipal d'Investigació Medica Ethics Committee. Standard clinical diagnostic tests were used to analyze serum samples from blood donors for circulating immunoglobulin G (IgG) antibodies against CMV (Abbot Laboratories, Abbot Park, IL), HSV, and EBV (Trinity Biotech, Jamestown, NY).

HP-3B1 anti-CD94, HP-1F7 anti–HLA-class I, HP-MA3 anti-KIR2DL1/S1, HP-3G10 anti-CD161, and HP-F1 anti-CD85j monoclonal antibodies (mAbs) were generated in our laboratory and have been previously reported.^32,33  Z199 anti-CD94/NKG2A,¹⁵  p25 anti-CD94/NKG2A/NKG2C,³⁴  C218 anti-CD56, KD1 anti-CD16, AZ20 anti-NKp30, Bab281 anti-NKp46, and OM72 anti-NKG2D mAbs were generously provided by Dr A. Moretta (University of Genova, Italy).³¹  Dx9 anti-KIR3DL1 mAb was provided by Dr L. Lanier (University of California, San Francisco). CH-L anti-KIR2DL2/S2/L3 was provided by Dr S. Ferrini (University of Genova). Anti-NKG2C mAb (MAB1381) was from R&D Systems (Minneapolis, MN); biotin labeling of MAB1381 was carried out using EZ-Link Sulfo-NHS-Biotin (Pierce, Rockford, IL) according to the manufacturer's instructions. Z199 (anti-NKG2A) mAb was conjugated to fluorescein isothiocyanate (FITC) (Sigma, St Louis, MO). Indirect immunofluorescence analysis was carried out with phycoerythrin (PE)– or FITC-tagged F(ab′)₂ rabbit antimouse Ig antibodies (Dako, Glostrup, Denmark). Anti-CD3–peridin chlorophyll protein (Per CP), CD56-PE, CD4-PE, CD8-PE, TcRαβ-FITC, TcRγδ-FITC, and streptavidin-FITC were from BD Biosciences Pharmingen (San Jose, CA).

CMV-specific T cells were detected with an R-phycoerythrin–labeled HLA-A^*0201 tetramer bound to the HLA-A^*0201–restricted NLVPMVATV peptide (amino acids 495 to 503 of the lower matrix protein, pp65) (Proimmune, Oxford, United Kingdom).

Heparinized peripheral blood was obtained by venous puncture. Peripheral blood mononuclear cells (PBMCs) were isolated using a Ficoll-Hypaque (Axis-Shield, Oslo, Norway) density gradient centrifugation. For indirect immunofluorescence staining, cells were pretreated with saturating concentrations of human aggregated Ig to block Fc and subsequently labeled with saturating concentrations of the different mAbs. Samples were subsequently analyzed by flow cytometry (FACScan; Becton Dickinson, Mountain View, CA). For multicolor staining the following procedures were used. Protocol 1: Cells were incubated with anti-NKG2A/C (p25) mAb followed by washing and labeling with PE-tagged F(ab′)₂ rabbit antimouse Ig antibody (Dako); subsequently, samples were incubated with the Z199 anti-NKG2A–FITC and anti-CD3–Per CP (BD Biosciences Pharmingen). Protocol 2: Cells were incubated with individual anti-NKR or NCR mAbs followed by washing and labeling with FITC-tagged F(ab′)₂ rabbit antimouse Ig antibody (Dako); subsequently, samples were incubated with anti-CD56–PE and anti-CD3–Pcep (BD Biosciences Pharmingen). Protocol 3: Cells were incubated with either anti-NKp46, NKp30, CD85j, NKG2D, or a mixture of anti-KIR (2DL1/S1, 2DL2/S2/L3, and 3DL1) mAbs, followed by washing and labeling with PE-tagged F(ab′)₂ rabbit antimouse Ig antibody (Dako); subsequently, sample cells were incubated with either anti-NKG2A–FITC or anti-NKG2C–biotin with streptavidin-FITC (BD Biosciences Pharmingen) and with anti-CD3–Per CP (BD Biosciences Pharmingen).

DNA was isolated from total blood using the Genomic DNA Purification Kit (Gentra Systems, Minneapolis, MN). An HLA-E gene sequence between exons 2 and 3 was amplified by the polymerase chain reaction (PCR) using the following primers: 2FHLAE, 5′-CGCACAGATTTTCCGAGTGAA-3′; and 382-ALL-AS, 5′-CCGCCTCAGAGGCATCATTTG-3′.²³  PCR reactions were run at 94°C for 1 minute, at 64°C for 30 seconds, and 72°C for 30 seconds for 35 cycles, with a final 10-minute extension at 72°C. DNA sequencing was performed using BigDye Terminator v3.1 Cycle Sequencing kit and a 3100 ABI automatic sequencer (Applied Biosystems, Foster City, CA).

The NKG2C gene sequence between intron 1 and exon 3 was amplified using the following primers: FNKG2C, 5′–GGCATTGTTCAACTGTAATCTGCG-3′; and RNKG2C, 5′-ACCTTTCTGCGTTCTTGTATTCGG-3′. PCR amplifications were run at 94°C for 1 minute, 61°C for 30 seconds, and 72°C for 1 minute and 30 seconds for 35 cycles, with a final 10-minute extension at 72°C.

KIR gene typing was performed using the PCR-SSP method (PCR with sequence-specific primers) as previously described.³⁵  Mutant KIR2DS4 alleles bearing a 22–base pair (bp) deletion in exon 5 were distinguished from normal ones by PCR using primers that produce amplicons of different lengths for each of these allotypes (C.V., unpublished data, 2003). The mutant KIR2DS4 alleles were not taken into account for estimating the number of activating KIRs of each donor. The HLA-A alleles were analyzed using the PCR-SSO method (Dynal Biotech, Wirral, United Kingdom).

The Kolmogorov-Smirnov nonparametric test was applied to check the normal distribution of the continuous variables, and either Pearson or Spearman correlation coefficients were computed to compare levels of continuous variables. To assess the relationship between a categoric variable with 2 levels and normally or nonnormally distributed quantitative variables, Student and Mann-Whitney U tests were applied, respectively. To determine the independence of HLA-E and CD94/NKG2C expression and the serologic status for HCMV, the χ² tests and Kruskall-Wallis were applied, respectively. Multivariate lineal regression was chosen to assess the relationship between CD94/NKG2C expression and HCMV status considering age as potential confounding variable. The model was then stratified by HLA-E genotypes. Analyses were performed with the SPSS 9.0 (SPSS, Chicago, IL) statistical package. Results were considered significant at the 2-sided P level of .05.

CD94/NKG2C expression was studied in PBLs from a cohort of 70 healthy adult blood donors (51 men and 19 women; age range, 20 to 56 years; median, 27 years). To circumvent the unavailability of NKG2C-specific reagents, we initially used a combination of anti-NKG2A mAb (Z199) and NKG2A/C (p25) mAbs; in that way the relative expression of the inhibitory and triggering CD94/NKG2 receptors was simultaneously assessed. Because the NKG2C gene is deleted in some individuals,³⁶  genomic analysis was required to interpret the phenotypic data; NKG2C appeared undetectable by PCR in 2 of 70 donors that were thus separately considered.

NKG2C⁺ cells, indirectly defined as p25⁺Z199^-, were observed to vary within a wide range (<0.1% to 22.1%; mean ± SD = 1.9% ± 3.5%), being virtually undetectable in approximately 40% of the donors. Three-color immunofluorescence analysis confirmed that the lectin-like dimer was expressed by both CD3⁺ (0.1% or less to 5%; 0.6% ± 0.8%) and CD3^- (0.1% or less to 45%; 4.5% ± 8.2%) lymphocyte subsets, the latter including NK cells. Figure 1A shows the distribution of p25⁺Z199⁺ and p25⁺Z199^- lymphocytes in 2 representative donors, illustrating the marked variability of CD94/NKG2C expression.

A commercial NKG2C-specific mAb (MAB1381) became recently available allowing comparison of its staining pattern with that obtained with the Z199/p25 mAb pair in samples from 21 individuals. MAB1381 brightly stained a cell population that did not coexpress NKG2A (Figure 1B), and no significant differences were substantiated between the proportions of NKG2C⁺ cells, defined either as p25⁺Z199^- or MAB1381^bright. NKG2C⁺ NK cells, identified as CD3^-CD56⁺ MAB1381^bright, represented 25.3% ± 25.1% of total NK cells (range, 2.5% to 80%). This approach revealed that most NKG2A⁺ PBLs do not detectably coexpress the NKG2C protein, an important question that could not be solved using the p25/Z199 mAb combination. A minor fraction of dull-stained MAB1381⁺ cells was also observed even in samples in which p25⁺Z199^- lymphocytes were undetectable (Figure 1B); yet, the significance of this observation is unclear and requires further characterization of the commercial mAb. Interestingly, a significant correlation (r = 0.83, P < .001) between the proportions of NKG2C⁺CD3⁺ and either NKG2C⁺CD3^- or NKG2C⁺CD3^-CD56⁺ cells was observed, suggesting that the regulatory event(s) controlling the expression of the activating KLR acts concomitantly on both NK and T-cell lineages; conversely, such a correlation was not substantiated for NKG2A⁺ cells. Three-color analysis of samples from 10 donors revealed that most NKG2C⁺ T cells displayed a TcRαβCD8⁺ phenotype; yet, in some individuals NKG2C was also detected in a subpopulation of TcRγδ lymphocytes and a minor proportion of CD4⁺ cells (data not shown).

Blood donors were classified according to the detection of HCMV-specific circulating IgG into 2 groups: HCMV-positive (n = 34) and HCMV-negative (n = 34); notably, according to previous studies, some carriers may appear seronegative.^37,38  A striking association between the detection of NKG2C⁺ cells and HCMV-specific antibodies was noticed (P < .001). As shown in Table 1 and Figure 2, the proportions of NKG2C⁺ cells, including CD3^- and CD3⁺ subsets, were significantly increased in HCMV-positive donors; by contrast, no relationship was observed between the serologic status for HCMV and the percentage of NKG2A⁺ or CD94⁺ cells (data not shown). Consequently, the NKG2A/NKG2C ratio also varied within a broad range and appeared significantly reduced among HCMV-positive individuals (Table 1). Despite the strong association between the serologic status for HCMV and expression of the activating KLR, NKG2C⁺ cells were less than 0.5% in 9 HCMV-positive and 0.5% or more in 5 HCMV-negative donors, the latter eventually corresponding to seronegative carriers. Thus, a clear-cut threshold in the proportions of NKG2C⁺ cells discriminating HCMV-positive and HCMV-negative subjects could not be precisely established. A significant correlation was observed between donor age and the percentage of NKG2C⁺ cells (P = .026), although it disappeared when the HCMV variable was jointly considered in the model (P = .232).

The clear-cut relationship between HCMV serology and increased proportions of circulating NKG2C⁺ NK and T cells indirectly supported that the viral infection may shape the NKR repertoire in healthy carriers. Remarkably, analyses performed in PBLs from 31 individuals, including HCMV-positive (n = 17) and HCMV-negative (n = 14), indicated that the numbers of circulating NKG2C⁺ cells were unrelated to the serologic status for EBV or HSV (Table 2) and, in fact, all HCMV-negative cases were either EBV-positive and/or HSV-positive. These results point out that the driving force(s) that underlie the expansion of NKG2C⁺ cells do not operate as a general response to all latent herpesvirus infections.

HLA-E allotypes contain either arginine (HLA-E^R107) or glycine (HLA-E^G107) at position 107. It has been reported that these allotypes differ in their expression levels, their sensitivity to the action of the US6 HCMV protein, as well as in their ability to interact with CD94/NKG2 receptors.^24-26  Thus, the possibility that the HLA-E genotype might influence the expression of CD94/NKG2C was also addressed. The HLA-E allotypes were defined by sequencing the specific PCR products amplified from genomic DNA. Although HLA-E genotypes were not associated either with the percentage of NKG2C⁺ cells or with HCMV serology (Table 3), the correlation between NKG2C expression and HCMV serology among HLA-E^G107 donors was not as strong as among the other HLA-E genotypes: β = 1.3 (P = .097) versus β = 3.1 (P = .010) for HLA-E^R107, and β = 4.4 (P = .020) for HLA-E^G107 HLA-E^R107. Yet, the small sample size of the HLA-E^G107 group impairs the interpretation of this result, further studies in a larger population being required to precisely evaluate this effect.

The number and identity of KIRs encoded in the human genome vary greatly in different individuals, the variation being greatest for activating KIRs.¹  While the combination of KIRs expressed by NK cell clones appears to be completely stochastic, a bias toward increased expression of CD94/NKG2A has been reported in individuals having lower numbers of functional inhibitory KIRs (ie, lower numbers of self-HLA ligands for the inhibitory KIRs encoded in their genomes).³⁹  To test the possibility of a reciprocal increase of NKG2C expression in individuals having lower numbers of activating KIRs, we determined the KIR genotypes of 31 individuals (16 HCMV-negative and 15 HCMV-positive) having diverse proportions of NKG2C⁺ cells (range, 0.1% or less to 22.1%). Neither a correlation between the number of activating KIR genes of each individual and the proportions of NKG2C⁺ cells nor appreciable deviations between the KIR gene frequencies of HCMV-positive and HCMV-negative subjects were observed (data not shown). KIR2DS3 was underrepresented among donors bearing less than 0.5% NKG2C⁺ cells in comparison with those having higher numbers of these cells, but this deviation was not statistically significant.

To explore whether HCMV infection was associated with additional features of the NKR expression pattern, 3-color fluorescence-activated cell sorter (FACS) analysis was performed employing anti-CD3 and CD56 mAbs in combination with a panel of reagents specific for NKRs (ie, CD85j, KIR2DL1/S1, KIR2DL2/S2/L3, KIR3DL1, NKG2D, CD161) and NCRs (ie, NKp46, NKp30). The impossibility to discriminate by flow cytometry between homologous KIR2DL and KIR2DS molecules, due to the cross-reactivity of available mAbs, somehow limits the interpretation of the phenotypic data. The proportions of NK and T cells bearing KIR3DL1 (Figure 3), NKG2D (Figure 4), and CD161 (data not shown) were comparable in HCMV-negative and HCMV-positive subjects. By contrast, a significant increase in the minor fraction of T cells expressing CD85j or KIR2D was observed in samples from HCMV-positive individuals (Figure 3B) with a tendency to display as well higher proportions of CD85j⁺ and KIR2D⁺ NK cells (Figure 3A).

NKp46 and NKp30 have been reported to be expressed on virtually all NK cells, but some NK cell subsets have been shown to bear low levels of these NCRs and to mediate inefficient cytotoxicity against tumor cells.⁴⁰  Thus, the staining intensity with NCR-specific mAbs was considered; Figure 4A shows the different staining patterns observed with NKp46 and NKp30 mAbs, corresponding to samples from 3 representative donors (D1, D2, and D3). Notably, NK cells from all HCMV-negative donors expressed pattern D1, whereas samples from HCMV-positive individuals displayed either patterns D1, D2, or D3. Altogether, HCMV-positive donors tended to display significantly lower proportions of NKp46^bright and NKp30^bright NK cells (Figure 4B).

These data raised the question as to whether the features observed in the NKR repertoire from HCMV-positive donors might constitute independent events or rather reflect a coordinated distribution of the other receptors with NKG2C. To address this issue, 3-color flow cytometry analysis was carried out in PBLs from 5 HCMV-positive donors using the biotin-labeled anti-NKG2C mAb in combination with anti-CD3 and either NCRs, CD85j, or KIR-specific mAbs. Remarkably, as compared with NKG2A⁺ lymphocytes from the same individuals, NKG2C⁺ cells displayed lower levels of NKp30 and NKp46 NCRs (Figure 5) and included higher proportions of KIR⁺ and CD85j⁺ lymphocytes (Figure 6), thus supporting the second possibility.

The detection of CD94/NKG2C in T lymphocytes from HCMV-positive donors raised the question as to whether such cells might correspond to an expansion of cytotoxic T lymphocytes (CTLs) specific for viral antigens. To address this issue, we selected 10 HLA-A^*0201 HCMV-positive donors and analyzed by flow cytometry the expression of NKG2C in T cells stained by HLA-A^*0201 tetramers refolded with an immunodominant peptide epitope from the pp65 HCMV lower matrix protein. In every case, most T lymphocytes binding the tetramer did not coexpress NKG2C (data not shown). These data do not entirely exclude that the CD94/NKG2C might be preferentially coexpressed by CTL subsets specific for other viral antigens but strongly suggest that the NKG2C⁺ cell expansion may occur independently of the T-cell receptor (TcR) specificity.

HCMV infection is quite prevalent (50% to 100%) in most populations, and it generally follows an indolent course. Yet, HCMV may cause a severe congenital disease and important disorders in immunocompromised individuals; moreover, the virus has been proposed to constitute a cofactor in the development of artheriosclerosis.³⁸  HCMV infects different cell types and tends to remain latent in immunocompetent hosts, where occasional reactivation allows its dissemination. An effective defense against CMV requires the participation of both NK cells and T cells.^41,42  HCMV-specific antibodies and circulating CTLs specific for viral peptides reflect the adaptive immune response to the virus, thus constituting conventional parameters to assess exposure and response to the pathogen.^38,43,44  Our data provide a first evidence indicating that HCMV infection may selectively shape the NKR repertoire of healthy individuals and, moreover, that the driving force(s) leading to the expansion of NKG2C⁺ cells may act coordinately on NK and T-cell lineages. Beyond the association between the serologic status for HCMV and the increase of NKG2C⁺ cells, studies in progress strongly support a causal role for the virus (M.G. and M.L.-B., unpublished data, 2004). A key open issue is whether NKG2C⁺ NK and T cells indeed play a role in the defense against HCMV or, alternatively, whether their expansion merely constitutes an epiphenomenon of the infection.

The increased proportions of NKG2C⁺ cells likely reflect the challenge exerted by HCMV on the innate immune system and thus may become another useful parameter to explore the complex host-pathogen relation during the course of infection and latency. The selective imprint of HCMV on the NKR repertoire is reminiscent of the expansion of CTLs specific for viral antigens during the adaptive immune response. Preliminary longitudinal analyses carried out in some donors pointed out that the distribution pattern of NKG2C⁺ and NKG2A⁺ subsets tended to remain rather stable along time (data not shown). Yet, an increase of the KLR expression should predictably follow primary infection, and oscillations in the proportions of circulating NKG2C⁺ cells may occur in HCMV-positive donors. The basis for the wide variability in the numbers of NKG2C⁺ cells observed among HCMV-positive individuals needs to be addressed. Although the proportions of NKG2C⁺ cells did not significantly correlate with serum levels of HCMV-specific IgG in HCMV-positive donors, it is conceivable that their increase may be related to the incidence of reactivation and/or reinfection episodes. Furthermore, a prospective follow-up of the NKR repertoire should be carried out in the context of different clinical settings in which HCMV is involved. Although HCMV and MCMV are genetically disparate pathogens, CD94/NKG2 receptors are conserved in mice and specifically recognize Qa1^b,⁴⁵  a functional homolog of HLA-E. Thus, studies are required to evaluate whether MCMV infection may have any impact on the expression of NKG2C.

There is limited information on the function played by activating NKRs (ie, CD94/NKG2C and KIR) in CTLs, and it has been proposed that they might play a costimulatory role rather than directly triggering T-cell effector functions.⁴⁶  The expansion of NKG2C⁺ T cells in HCMV-positive donors raised the question as to whether they might be CTL-specific for viral antigens. Arguing against that possibility, we observed in HCMV-positive HLA-A^*0201 donors that most T cells specifically stained by HLA-A^*0201 tetramers bound to an immunodominant pp65-derived peptide did not coexpress CD94/NKG2C; consistent with this observation others have reported that HCMV-specific CTLs were CD94^-.⁴³  As an alternative, we also considered the possibility that CD94/NKG2C⁺ cells might correspond to HLA-E–specific T cells. We previously described a CD94/NKG2C⁺ T-cell clone that recognized HLA-E via the TcR.⁴⁷  Moreover, Mingari and colleagues also identified HLA-E–specific T cells bearing the CD94/NKG2A inhibitory receptor and provided evidence supporting that they recognize peptides derived from the UL40 HCMV protein.^48,49  CD94/NKG2C⁺ T-cell clones were tested for their ability to kill the HLA-E–transfected 721.221 cells (.221-AEH).²²  In every case, the enhanced lysis against .221-AEH cells was completely prevented by an anti-CD94 mAb, consistent with an involvement of the activating KLR rather than of the TcR (M.G. and M.L.-B., unpublished results, 2004); yet, further studies are required to precisely assess the frequency of CD94/NKG2C⁺ T cells bearing an HLA-E–specific TcR.

Several mechanisms may account for the variable increase of CD94/NKG2C⁺ cells in HCMV-positive individuals. First, changes in the NKR distribution might result from alterations in the cytokine network secondary to the viral infection. In this regard, CD94/NKG2A has been reported to be inducible in T cells by transforming growth factor-β (TGF-β) and IL-15.⁵⁰  Recently, IL-21 was shown to promote the differentiation of CD34⁺ precursors to NK cells and their sequential acquisition of NCRs and NKRs⁵¹ ; yet, there is no evidence that cytokines may regulate differential NKR expression during maturation. Alternatively, signaling by NKRs may control not only lymphocyte effector functions but also the proliferation and/or survival of NK cells subset(s) that participate in ligand recognition, as shown for the expansion of Ly49H⁺ cells in MCMV-infected mice.⁶  Thus, NKG2C-mediated recognition of HCMV-induced alterations in infected cells could lead to the expansion of the corresponding NK and T-cell subsets. Based on the current view of NKR biology, this might happen either upon an increased avidity of the KLR-ligand interaction and/or secondarily to a selective loss of the ligands for inhibitory receptors (ie, KIR, ILT2) expressed by CD94/NKG2C⁺ cells. Thus far, there is no evidence supporting that CD94/NKG2C may recognize HLA-E–bound microbial peptides or viral molecules, as shown for Ly49H.^7,8  Although HLA-E bound to an HSP60-derived peptide interacts with CD94/NKG2A,⁵²  there is no information as to whether this complex may efficiently engage CD94/NKG2C. Regarding the second possibility, a nonamer derived from the UL40 HCMV protein binds to HLA-E and favors its transporter associated with antigen processing (TAP)–independent and US6-resistant expression.^29,30  Moreover, the US11 and US2 HCMV proteins reduced the surface levels of class Ia molecules but did not affect HLA-E.²⁸  Thus, preservation of the class Ib molecule to maintain HCMV-infected cells resistant against CD94/NKG2A⁺ subsets might promote their recognition by NKG2C⁺ cells.

Differences between the surface levels of HLA-E allotypes and their interaction with CD94/NKG2 receptors have been reported. Moreover, the HLA-E^G107 transfected in the HLA-negative K562 cell line was shown to be insensitive to the US6 HCMV protein.^24-26  When individually analyzed, the HLA-E allotypes/genotypes were not significantly associated either with the proportions of circulating NKG2C⁺ cells or the serologic status for HCMV. Yet, the possibility that the HLA-E genotype may influence the impact of HCMV infection on the proportions of NKG2C⁺ cells cannot be entirely excluded, requiring further studies in a larger population.

The effect of HCMV infection on the expression of other NKRs and NCRs was also addressed. Human NKG2D is coupled to phosphatidylinositol-3 kinase (PI-3K) signaling pathways through the DAP10 adapter and interacts with stress-inducible class I–related molecules, including MHC class I–related A/B (MICA/B)– and “UL16 binding proteins” (ULBPs) or RAET1 (Rae-1 like transcripts).^53,54  Expression of NKG2D ligands in CMV-infected cells has been shown to costimulate virus-specific CTLs.⁵⁵  Moreover, the HCMV UL16 glycoprotein impairs the expression of MICB, ULBP1, and ULBP2, presumably constituting a mechanism to escape from the NKG2D-mediated response.^56-58  According to our data, the NKG2D expression was comparable in HCMV-positive and HCMV-negative donors as well as among the NKG2C⁺ and NKG2A⁺ cell subsets.

A putative involvement of natural cytotoxicity receptors (NCRs) of the Ig superfamily (IgSF) (ie, NKp46, NKp44, and NKp30)³¹  in the response to HCMV is uncertain. Although their cellular ligands have not yet been defined, there is indirect evidence that they are widely distributed in different tissues and, thus, NCRs might contribute to the NK cell response against some virus-infected cells in which the expression of HLA class I molecules has been down-regulated. Paradoxically, lower proportions of NKp30⁺ and NKp46⁺ “bright” NK cells were detected in HCMV-positive individuals; consistent with this observation, NKG2C⁺ NK cells displayed significantly lower levels of these NCRs than NKG2A⁺ cells from the same donors. This is in line with a previous study showing that the NKG2A⁺ cells were confined to the NKp46^bright subset.⁴⁰  A low expression of NCRs has been recently reported in NK cells from viremic HIV-positive patients, and the putative influence of HCMV infection on the NKR repertoire of HIV-positive individuals should be addressed.⁵⁹ 

CD85j (ILT2/LIR-1) is an inhibitory receptor of the IgSF expressed by different leukocyte lineages that interacts with a broad spectrum of HLA class I molecules and binds with high affinity to the class I–like UL18 HCMV glycoprotein.^60-62  The hypothesis that UL18 may interfere with NK cell activity during HCMV infection has not yet received convincing experimental support.²⁷  However, CD85j expression was reported to increase in lymphocytes from patients undergoing HCMV infection after lung transplantation.⁶³  In line with this finding, we observed that the proportions of CD85j⁺ and KIR2D⁺ cells tended to be raised in HCMV-positive individuals. This association reflected the higher coexpression of CD85j and KIR in NKG2C⁺ cells as compared with the NKG2A⁺ subset. Such coordinated distribution of NKRs suggests that CD94/NKG2C⁺ lymphocytes displaying low levels of NCRs that coexpress inhibitory NKRs (ie, CD85j and KIR) may be preferentially selected during maturation, preventing their potential autoreactivity against normal HLA-E–positive cells. Further studies are required to assess whether NKG2C indeed preferentially associates to inhibitory KIRs. No relation could be established between the KIR genotype and either HCMV infection or CD94/NKG2C expression, ruling out that the activating KLR may be preferentially associated to haplotypes containing fewer activating KIR genes.

Altogether, our results support that HCMV shapes the NKR repertoire, expanding NKG2C⁺ NK and T-cell subsets that coexpress CD85j and KIR and display low levels of NCRs. The driving force(s) that determine the HCMV imprint on the NKR repertoire are being currently explored at the cellular and molecular levels, with special attention to the role played by CD94/NKG2C itself.

We are grateful to Raquel Flores for technical assistance, Dr Oscar Fornas for advice in flow cytometry analysis, Drs Alessandro Moretta and Daniela Pende (University of Genova) for generously sharing reagents, and Dr Magí Farré and Esther Menoyo (IMIM) for kindly collaborating in obtaining blood samples.