Relationship of NKG2C Copy Number with the Distribution of Distinct Cytomegalovirus-Induced Adaptive NK Cell Subsets

Natural killer cells are innate lymphocytes capable of recognizing virus-infected and tumor cells without prior sensitization. NK cell activation results in the secretion of cytotoxic effector molecules and cytokines (e.g., IFN-γ and TNF-α), leading to target cell death and influencing the development of adaptive immune responses.

NK cells participate in the defense against herpesvirus infections. Among them, human CMV (HCMV) induces a persistent reconfiguration of the NK cell repertoire in some infected individuals, promoting the expansion and differentiation of NK cell subsets with adaptive features (1–3). These adaptive NK cells were initially identified based on the high expression of the CD94/NKG2C-activating receptor in the absence of its inhibitory counterpart CD94/NKG2A (1). In contrast to CD94/NKG2C⁺ NK cells found in HCMV-seronegative and some seropositive individuals, HCMV-induced NKG2C^bright NK cells predominantly express inhibitory killer Ig–like receptor (KIR) specific for self–HLA-C together with low NKp30, NKp46, and CD161 surface levels and include high proportions of LILRB1⁺ and CD57⁺ cells (1, 4–6). Moreover, NKG2C^bright NK cells display greater granzyme B levels and proinflammatory cytokine production (TNF-α and IFN-γ) following target cell recognition, particularly upon CD94/NKG2C or CD16 engagement (7, 8). In this regard, regulatory regions of IFNG are hypomethylated in NKG2C^bright NK cells, facilitating transcription and cytokine production (9).

Studies in congenital or perinatal HCMV infection, as well as in transplant recipients, indicate that the expansion of NKG2C^bright NK cells may be detected relatively early following HCMV infection and tends to remain fairly stable over time (10–15). The strict association between NKG2C^bright NK cells and HCMV seropositivity, even in the context of other viral infections (i.e., HIV, hepatitis C virus, EBV, Hantavirus, and Chikungunya virus) (4, 16–19), supports the existence of an HCMV-specific condition promoting the persistent modification of the NK cell compartment homeostasis.

The molecular mechanisms leading to the appearance of NKG2C^bright NK cells upon HCMV infection remain elusive. Yet, an active role for CD94/NKG2C is supported by in vitro data showing a preferential expansion of CD94/NKG2C⁺ NK cells upon stimulation with virus-infected fibroblasts in a CD94- (20), NKG2C-, and HLA-E–dependent manner (21). Moreover, ∼5% of the population is homozygous for a deletion of the NKG2C gene (officially designated KLRC2), and NKG2C zygosity has been shown to influence NKG2C⁺ NK cell numbers in steady state, as well as the receptor function in HCMV⁺ subjects (6, 14, 22). The response of NKG2C⁺ NK cells to HCMV resembles that of memory-like Ly49H⁺ NK cells, which expand following recognition of the MHC-like m157 molecule in murine CMV–infected cells and confer an efficient defense against reinfection (23–25). Yet, the role played by NKG2C⁺ NK cells in HCMV infection control remains thus far uncertain.

Besides NKG2C^bright NK cells, oligoclonal expansions of differentiated NK cells lacking NKG2C and expressing self-specific activating KIR have been also described in some HCMV⁺ healthy adults (5) as well as in patients transplanted with NKG2C^del/del hematopoietic grafts undergoing HCMV infection (26). HCMV infection has been associated with the expansion of NK cells that have epigenetically silenced the expression of particular signaling adaptor molecules (i.e., FcRγ, Syk, and Eat2) and transcription factors (PLZF, HELIOS, and IKZF2) and display a genome-wide methylation profile similar to that of memory effector T cells (27, 28). The deficiency of the FcRγ adaptor appeared the most frequently detected, and alterations in the expression of other signaling molecules were greatly confined to the FcRγ⁻ NK cell subset (27, 28). Reminiscent to observations in the NKG2C^bright NK cell subset, FcRγ-deficient NK cells displayed reduced NKp46 and NKp30 surface levels, maintaining CD16 expression and showing enhanced cytokine production and proliferation upon Ab-dependent activation (27, 29). Though FcRγ-deficient NK cells often express NKG2C and lack NKG2A, FcRγ loss has also been described in NK cells lacking NKG2C expression (27, 28, 30).

In this study, we systematically addressed the relationship between the expansion of NKG2C^bright NK cells and the existence of FcRγ-deficient NK cells in HCMV seropositive healthy adults, analyzing the influence of the NKG2C genotype and the possible participation of activating KIR in these processes.

PBMC and NK cells used in this study were obtained from peripheral blood of volunteer donors. Written informed consent was obtained from every donor, and the study protocol was approved by the local ethics committee (Comité Ético de Investigación Clínica, Parc de Salut Mar, 2010/3766/I).

Blood samples from a cohort of 81 healthy adults were fractionated to obtain a basic hemogram, DNA, serum, and PBMC. PBMCs were separated from fresh whole blood by density gradient centrifugation (Ficoll Hypaque, Lymphoprep; Axis-Shield) and were either used directly after separation or cryopreserved in 10% dimethyl sulfoxide and 90% heat-inactivated FBS for later use. The analysis of NKR and FcRγ expression in NK cells was performed in thawed PBMC samples by flow cytometry. For functional assays, fresh PBMC samples were incubated overnight with 200 U/ml IL-2 prior to NK cell isolation by negative selection using EasySep Human NK cell enrichment kit (Stemcell Technologies). Standard clinical diagnostic tests were used to analyze serum samples for circulating Ig G Abs against HCMV.

DNA was isolated from total blood using the Puregene, BloodCore kit B (Qiagen). NKG2C zygosity was assessed by PCR as in Moraru et al. (31). KIR gene content was determined by a locally designed PCR with sequence-specific primers method (32). Allele KIR3DL1*004 was identified in donor D.50 by sequence-based typing of exons 2–4 using locally designed primers (details available upon request). KIR ligands were inferred as in Moraru et al. (33) from HLA types assigned by DNA methods, following the quality standards issued by the European Federation for Immunogenetics.

Expression of NKG2A, NKG2C, and FcRγ in CD56^dimCD3⁻ cells was analyzed by five-color flow cytometry in thawed PBMC samples with anti-CD3–PerCP, anti-CD56–allophycocyanin, anti-NKG2A–PE, and anti-NKG2C indirectly labeled with a secondary goat anti-mouse-PE-Cy7. For the analysis of FcRγ expression, cells were fixed and permeabilized with fixation/permeabilization kit (BD Biosciences) followed by incubation with anti-FcRγ–FITC. Samples were analyzed in an LSR Fortessa flow cytometer (BD Biosciences), and data were analyzed with FlowJo software (Tree Star). Absolute numbers of NK cell subsets were calculated based on the hemogram counts. Donor profiles were classified as NKG2C^bright when presenting a proportion of NKG2C⁺NKG2A⁻ CD56^dimCD3⁻ cells >8% (mean + 2 SD of NKG2C⁺NKG2A⁻ NK cells in HCMV⁻ donors, excluding NKG2C^del/del individuals). Otherwise, donor profiles were classified as NKG2C^dim. Individuals with NKG2C⁻FcRγ⁻ NK-cell expansions were defined as showing >12% of NKG2C⁻FcRγ⁻ CD56^dimCD3⁻ cells (mean + 2 SD of NKG2C⁻FcRγ⁻ NK cells in HCMV⁻ donors). The coexpression of NKG2A, LILRB1, NKG2D, NKp30, NKp46, CD161, CD16, CD57, and KIR (mix of Abs recognizing KIR2DL1/S1, KIR2DL2/S2/L3, KIR3DL1/L2, and KIR2DS4) in FcRγ⁺ and FcRγ⁻ NKG2C⁺ NK cells was analyzed by multicolor flow cytometry in PBMC from HCMV-seropositive donors displaying the NKG2C^bright profile.

Anti-NKG2D (clone BAT221), anti-NKp46 (clone Bab281), anti-NKp30 (clone AZ20), and anti-KIR3DL1/S1 (z27) were kindly provided by Dr. A. Moretta (University of Genova, Genova, Italy); anti-KIR3DL1 (clone DX9) was provided by Dr. L. Lanier (University of California, San Francisco); anti-KIR2DL2/S2/L3 (clone CH-L) was provided by Dr. S. Ferrini (National Institute for Cancer Research, Genova, Italy); and anti-KIR2DL1/S1 (clone 11PB6) was provided by Dr. D. Pende (National Institute for Cancer Research). Anti-KIR3DL1/L2 (clone 5.133) was provided Dr. M. Colonna (Washington University, St. Louis, MO). Anti-LILRB1 (clone HP-F1), anti-NKG2A (clone Z199), anti-CD161 (clone HP-3G10), anti-CD57 (clone HNK1), anti-KIR2DL1 (clone DM-1), and anti-KIR2DL1/S1 (clone HP-3E4) were produced in our laboratory and used either as tissue culture supernatants or upon conjugation to FITC or Pacific Blue (PB). Anti–TNF-α (Infliximab; Remicade) was used directly labeled with PB. Anti-CD3–PerCP, anti-CD56 allophycocyanin, anti-LILRB1–FITC, and anti-CD16–FITC were from BD Biosciences; anti-NKG2A–PE was from Beckman Coulter; and anti-NKG2C–PE (clone 134591) and anti-NKG2C (clone 134522) were from R&D Systems (Minneapolis, MN). Anti-FcεRγ subunit-FITC (polyclonal) was from Merck Millipore. PBMCs were pretreated with aggregated human Ig, incubated with individual NKR-specific mAbs, washed, and further incubated with a PE-tagged F(ab')₂ rabbit anti-mouse Ig (The Jackson Laboratory) or PE-Cy7– conjugated F(ab′)₂ polyclonal goat anti-mouse IgG (BioLegend). Subsequently, samples were directly stained with anti-CD3–PerCP, anti-CD56–allophycocyanin, and anti-NKG2C–PE, fixed, and permeabilized followed by intracellular staining with anti-FcRγ–FITC. Data were acquired on an LSR Fortessa flow cytometer (BD Biosciences).

Measurements of NK cell degranulation (CD107a) and intracellular TNF-α in response to K562 and rituximab-coated 721.221 cell lines were measured in isolated NK cells from HCMV⁺ donors with NKG2C^bright NK cell profiles after 4-h coculture using standard flow cytometry protocols. Staining strategy for comparison of TNF-α responses between FcRγ⁺ and FcRγ⁻ NKG2C⁺ NK cells included anti-CD56–allophycocyanin, anti-NKG2C–PE, anti-FcRγ–FITC, and anti–TNF-α–PB. Staining strategy for comparison of degranulation responses and TNF-α production between NKG2C⁺FcRγ⁻ and NKG2C⁻FcRγ⁻ subsets in NK cells from NKG2C^+/del donors included anti–CD56-allophycocyanin–Cy7, anti-CD107a–allophycocyanin, anti-NKG2C–PE, anti-FcRγ–FITC, and anti–TNF-α–PB. Inverse redirected NK cell degranulation was measured by the CD107a mobilization in purified NK cells from NKG2C^+/del donors with concomitant NKG2C^bright and NKG2C⁻FcRγ⁻ NK cell subpopulations. NK cells were incubated with 1 μg anti-CD107a–FITC and 5 ng/ml monensin for 4 h at 37°C with 10% CO₂, either alone or in the presence of P815 murine mastocytoma cells preincubated with anti-KIR2DL2/L3/S2 mAb (clone CHL), anti-NKG2C (clone 134522; R&D Systems), anti-CD16 (clone KD1), or control IgG. After coculture, cells were surface stained with anti-CD56–allophycocyanin and anti-NKG2C–PE mAb, fixed, and permeabilized followed by intracellular staining with anti-FcRγ mAb. Data acquisition was performed with on an LSR Fortessa instrument (BD Biosciences).

Continuous variables between two groups were compared using the Mann–Whitney U test, and p values <0.05 were considered statistically significant. For comparison of matched groups, the paired Student t test or the Wilcoxon test for paired samples was used. In Fig. 4, the Fisher exact test was used. Analyses were performed using GraphPad Prism 6 software (GraphPad).

Expansions of NKG2C^bright and FcRγ⁻ NK cell subsets, associated with HCMV infection, share several phenotypic and functional features (1, 6, 7, 29). To assess in detail their relationship, NKG2C, NKG2A, and FcRγ expression were analyzed in peripheral blood NK cells from a cohort of healthy adults (n = 81; median age 32 y). According to the criteria defined in 2Materials and Methods, 45% (26 out of 58) HCMV⁺ individuals showed detectable (i.e., >8%) NKG2C^bright NK cells, and 50% (29 out of 58) displayed an NKG2C^dim profile comparable to HCMV⁻ subjects (n = 23), whereas NKG2C was undetectable in three donors homozygous for NKG2C gene deletion.

Variable proportions of CD56^dim FcRγ⁻ NK cells were present in most subjects displaying NKG2C^bright expansions, in contrast to the uniform expression of FcRγ in NK cell samples from HCMV⁻ and the majority of HCMV⁺ individuals with an NKG2C^dim profile (Fig. 1A–C), thus pointing out the relation of both phenotypic features with a common virus–host interaction pattern. Nonetheless, large proportions of CD56^dim FcRγ⁻ NK cells were found in two subjects displaying low/undetectable NKG2C expression (D.94, Fig. 1B, arrow, and D.149, homozygous for NKG2C deletion, not included in the figure), indicating that both markers occasionally dissociate.

Analysis of NKG2C⁺NKG2A⁻ (NKG2C^bright), NKG2A⁺NKG2C⁻ (NKG2A), and NKG2C⁻NKG2A⁻ (double-negative [DN]) subpopulations in NK cell samples from individuals with an NKG2C^bright NK cell profile (n = 26) showed a preferential accumulation of FcRγ⁻ cells within NKG2C^bright subset, although FcRγ⁻ NK cells were also detected in NKG2A⁺ and DN NK cell fractions in some subjects (Fig. 2A, 2B). Despite the fact that FcRγ⁻ cells were generally present among the NKG2C^bright NK cell subset, the proportions of NKG2C^bright NK cells downregulating FcRγ widely varied, being undetectable in some donors and accounting for most NKG2C^bright cells in others (Fig. 2C). Regression analysis revealed no correlation between the magnitude of NKG2C^bright NK cell expansions and the extent of FcRγ loss in NKG2C⁺ NK cells (Fig. 2D), indicating that the NKG2C^bright phenotype and FcRγ downregulation develop independently along adaptive NK cell differentiation.

Individuals showing similar proportions of FcRγ⁺ and FcRγ⁻ NKG2C^bright NK cells were selected (n = 5) to analyze the distribution of other NKR in both cell fractions in relation to NKG2C⁻FcRγ⁺ NK cells. Both FcRγ⁺ and FcRγ⁻ NKG2C^bright NK cells were KIR2D⁺, LILRB1⁺, and CD57⁺ and lacked NKG2A⁺. FcRγ loss was encompassed by a reduction in NKp46, NKp30, and CD161 surface levels. Remarkably, CD16 surface expression appeared increased in NKG2C⁺FcRγ⁺as compared with NKG2C⁺FcRγ⁻ and NKG2C⁻FcRγ⁺ NK cell fractions (Fig. 3A, 3B), and all CD56^dim NK cells expressed NKG2D at similar levels (not shown).

Both NKG2C^bright and FcRγ⁻ HCMV-induced adaptive NK cells have been described to display enhanced cytokine production upon activation (7, 30). We compared TNF-α production of the three aforementioned NK cell subpopulations in coculture experiments with K562 and rituximab-coated 721.221 cell lines. In comparison with NKG2C⁻FcRγ⁺ NK cells, both NKG2C^bright NK cell subpopulations displayed similarly higher proportions of TNF-α⁺ cells in response to Rituximab-coated 721.221 targets, regardless of FcRγ levels; this pattern of response was independent of overnight incubation of PBMCs with IL-2 (Supplemental Fig. 1). No differences in the proportions of TNF-α⁺ cells in response to K562 were noticed (Fig. 3C).

Altogether, these data support that FcRγ⁺ and FcRγ⁻NKG2C^bright NK cells represent distinct stages of a sequential differentiation process. FcRγ loss was encompassed by NCR and CD161 downregulation, whereas acquisition of enhanced competence for cytokine production in response to CD16 stimulation was already detected in NKG2C^brightFcRγ⁺KIR2D⁺ cells.

Based on the reported relation of the NKG2C genotype with the magnitude of NKG2C⁺ NK cell expansion in HCMV⁺ subjects (6, 14), the possible influence of the NKG2C copy number on FcRγ⁻ NK cell numbers was assessed. Phenotypic data from HCMV⁺ individuals with an NKG2C^bright NK cell profile (n = 26) were stratified according to their NKG2C genotype. In line with previous observations, NKG2C^+/+ individuals displayed greater relative and absolute NKG2C⁺ NK cell numbers (Fig. 4A). By contrast, FcRγ⁻ NK cell counts were similar in NKG2C^+/del and NKG2C^+/+ individuals (Fig. 4B). In a subgroup of NKG2C^+/del individuals, NKG2C^bright NK cells contained larger proportions of FcRγ⁻ NK cells as compared with those in NKG2C^+/+ subjects (Fig. 4C, D). Moreover, expansions of NKG2C⁻FcRγ⁻ NK cells (>12%, as described in 2Materials and Methods) were detected in 5 out of 10 (∼50%) NKG2C^+/del, as compared with 1 out of 16 (∼ 6%) NKG2C^+/+, individuals with an NKG2C^bright phenotype (Fig. 4C, 4E). Finally, FcRγ downregulation was also detected in one out of three NKG2C^del/del HCMV⁺ individuals, accounting for 30% of CD56^dim cells (Fig. 4F).

Thus, NKG2C copy number appeared related with distinct distribution patterns of adaptive NK cells in HCMV⁺ subjects. Larger NKG2C^bright NK cell expansions containing moderate proportions of FcRγ⁻ cells were generally found in NKG2C^+/+ individuals, in contrast to smaller NKG2C^bright NK cell pools including greater proportions of FcRγ⁻ NK cells observed in NKG2C^+/del subjects. Remarkably, NKG2C⁻FcRγ⁻ NK cells were more frequently detected in NKG2C^+/del and NKG2C^del/del individuals, further supporting the notion that NKG2C^bright NK cell expansions and FcRγ loss take place independently.

NKG2C^bright adaptive NK cells have been reported to display an oligoclonal expression pattern of self-reactive KIR, preferentially specific for HLA-C (4, 5, 7). We compared the KIR repertoires expressed by NKG2C^brightFcRγ⁻ and NKG2C⁻FcRγ⁻ NK cell subpopulations in the same individual. Based on the observations reported in the previous section, NKG2C^+/del HCMV⁺ donors were selected, genotyped for KIR and KIR ligands, and their KIR phenotypes were analyzed in freshly isolated PBMC using combinations of appropriate mAbs (Supplemental Fig. 2 for staining/gating strategy).

According to their KIR-gene profiles, three donors (#50, #77, and #98) had at least one KIR-B haplotype, whereas D.#79 was homozygous for KIR-A haplotypes with defective KIR2DS4 alleles, thus lacking activating KIR (Fig. 5A). Self-specific KIR (labeled as dark boxes in Fig. 5A) were determined based on the presence of HLA-C1, -C2, -Bw4, and -A3/A11 KIR ligands.

NKG2C^brightFcRγ⁻ NK cells predominantly expressed an inhibitory KIR specific for self–HLA-C (KIR2DL2/L3 in D.#50 and D.#98; KIR2DL1 in D.#79) and lacked NKG2A. As an exception, NKG2C^brightFcRγ⁻ NK cells of D.#77 did not express HLA-C–specific KIRs. In contrast, NKG2C⁻FcRγ⁻ NK cell subpopulations were more heterogeneous, including cells expressing HLA-C–specific KIRs (KIR2DL2/L3: 65% in D#50, 47% in D.#98; KIR2DL1: 71% in D.#79) together with significant fractions of NKG2A⁺ NK cells (29% in D.#50, 42% in D.#98, and 20% in D#79, not shown). The majority of NKG2C⁻FcRγ⁻ NK cells in D#77 were NKG2A⁺ and lacked HLA-C–specific KIR (Fig. 5B–E, first row).

NKG2C^bright and NKG2C⁻FcRγ⁻ NK cell subpopulations also differed in the expression of KIR3DL1, which was excluded from NKG2C^brightFcRγ⁻ NK cells in Bw4⁺ individuals (D.#77 with 42% KIR3DL1⁺ cells in FcRγ⁺ fraction and D.#98), being detectable in 27% of NKG2C^brightFcRγ⁻ NK cells from Bw4⁻ D.#79. As compared with NKG2C^bright NK cells, increased frequencies of KIR3DL1⁺ cells were detected in NKG2C⁻FcRγ⁻ NK cell subsets, regardless of their Bw4 phenotype (92% in D.#79, 8% in D.#77, and 27% in D#98). KIR3DL1 expression was absent in D.#50, homozygous for the KIR3DL1*004 allele, reported to be retained intracellularly (34) (Fig. 5B–E, second row).

KIR3DL2 expression was monitored with the anti-KIR3DL1/L2 mAb 5.133, which also reacts with KIR2DS4. In D.#50 and D.#79 (negative for KIR2DS4), NKG2C^bright and NKG2C⁻FcRγ⁻ NK cell subpopulations contained variable proportions of KIR3DL2⁺ cells (NKG2C^brightFcRγ⁻: 30% in D.#50, 28% in D.#79; NKG2C⁻FcRγ⁻: 68% in D.#50). In the remaining two donors (D.#77 and D.#98), expression of KIR3DL2 could only be ascertained for NKG2C^brightFcRγ⁻ NK cells (96%) in D.#77 that were devoid of KIR2DS4 (Fig. 5B–E, second and third row). Intriguingly, D.#77 lacked -A3/A11 yet was HLA-B27⁺; thus, NKG2C^brightFcRγ⁻ NK cells in this subject were apparently licensed by HLA-B27–KIR3DL2⁺.

These observations confirmed that NKG2C^brightFcRγ⁻ NK cell expansions predominantly expressed an inhibitory KIR2DL specific for self–HLA-C, lacked self-specific KIR3DL1, and could display variable proportions of KIR3DL2⁺ cells, as previously described for total NKG2C^bright NK cells (4, 5, 35). By contrast, concomitant NKG2C⁻FcRγ⁻ NK cell subpopulations found in NKG2C^+/del donors were more heterogeneous and included NKG2A⁺KIR2D⁻ together with KIR2D-restricted NK cells, with a variegated coexpression of KIR3DL2 or KIR3DL1.

Whether NKG2C⁻FcRγ⁻ NK cells would predominantly express an activating KIR was also addressed. Activating KIRs presented a variegated expression pattern in both NKG2C^bright and NKG2C⁻FcRγ⁻ NK cell subpopulations. KIR2DS4 was detected in 8 and 25% of NKG2C⁺ and in 42 and 41% of NKG2C⁻FcRγ⁻ NK cell fractions from D.#77 and D.#98, respectively (Fig. 5B–E, third row). KIR2DS1 expression was detected in 75 and 22% of NKG2C^bright and NKG2C⁻FcRγ⁻ NK cells from D.#98, who lacked its putative HLA-C2 ligand (Fig. 5B–E, fourth row).

KIR2DS2 expression could not be precisely determined because of the lack of a commercially available mAb discriminating KIR2DL2 and KIR2DS2. To indirectly address KIR2DS2 expression in NK cells from D.#50 and D.#98, we analyzed the degranulation of NKG2C^brightFcRγ⁻, NKG2C⁻FcRγ⁻, and FcRγ⁺ NK cell subsets in an Ab-redirected activation assay using the CHL mAb. Redirected activation with NKG2C- and CD16-specific mAbs was monitored in parallel, and D.#79 NK cells (lacking KIR2DS2 genes) were included as a negative control. In samples from D.#50 and D.#98, the CHL mAb triggered degranulation of small proportions of NKG2C⁻FcRγ⁻, NKG2C^bright FcRγ⁻, and FcRγ⁺ NK cells, in comparison with the broad response to anti-CD16 or anti-NKG2C mAbs. As expected, CHL stimulation did not activate NK cells from D.#79 (Supplemental Fig. 3). The results indirectly supported that KIR2DS2 may be displayed and functionally dominant in small fractions of the different NK cell subsets, though did not allow a precise definition of its distribution relative to that of KIR2DL2/L3 inhibitory receptors.

In summary, despite the fact that expression of KIR2DS1, KIR2DS4, or KIR2DS2 was detected in variable proportions of both NKG2C^brightFcRγ⁻ and NKG2C⁻FcRγ⁻ NK cells, activating KIR appeared dispensable for NKG2C⁻FcRγ⁻ NK cell differentiation.

As mentioned in the initial section, FcRγ downregulation was detected in one out of three NKG2C^del/del HCMV⁺ individuals (D.#149). Activating KIR were suggested to drive NKG2C-independent adaptive NK cell expansions in patients transplanted with NKG2C^del/del hematopoietic grafts undergoing HCMV infection (26). D.#149 was homozygous for KIR-A haplotypes with nonfunctional KIR2DS4 alleles, thus lacking activating KIR besides NKG2C (Fig. 6A). Analysis of the KIR phenotype and KIR-ligands revealed a uniform expression of self-specific KIR2DL3 associated to the absence of NKG2A and CD94 on the FcRγ⁻ NK cell subset (Fig. 6B). FcRγ⁻ NK cells in D.#149 had low levels of surface NKp30 and NKp46 and elevated proportions of LILRB1⁺ and CD57⁺ NK cells as described for adaptive NK cells (Fig. 6C). Noticeably, in contrast to NKG2C⁻FcRγ⁻ NK cell subpopulations in NKG2C^+/del individuals, FcRγ⁻ NK cells in the NKG2C^del/del donor were oligoclonal and licensed by an HLA-C–specific inhibitory KIR, reminiscent to conventional NKG2C^bright NK cell expansions.

The coexistence of NKG2C^brightFcRγ⁻ and NKG2C⁻FcRγ⁻ NK cell subsets in some individuals allowed to directly compare their effector functions. NK cells from donors #50, #98, #77, and #79 were stimulated with HLA class I–deficient K562 or rituximab-coated 721.221 cell lines, monitoring degranulation and TNF-α production by NKG2C^bright FcRγ⁻, NKG2C⁻FcRγ⁻, and FcRγ⁺ NK cell subsets. In compliance with a comparable expression of CD16 at the cell surface, degranulation in response to rituximab-coated 721.221 and K562 cells was similar, regardless of FcRγ expression (Fig. 7A–C). Production of TNF-α upon coculture with K562 cells was comparable in the different NK cell subsets analyzed. In contrast, larger proportions of NKG2C^brightFcRγ⁻ NK cells produced TNF-α upon Ab-dependent activation as compared with the other subsets (Fig. 7C, 7D), irrespectively of overnight preincubation of PBMCs with IL-2 stimulation (Supplemental Fig. 1). These results revealed relevant functional differences between NKG2C^bright and NKG2C⁻FcRγ⁻ NK cell subsets as well as the dissociation between FcRγ downregulation and enhanced TNF-α production.

HCMV infection frequently induces a reconfiguration of the NK cell repertoire by promoting the adaptive expansion and differentiation of NK cells with a mature phenotype and enhanced function (2, 3). HCMV-induced adaptive NK cells were initially identified by an elevated expression of the activating receptor CD94/NKG2C (NKG2C^bright) (1, 2). More recently, several studies have shown a relation between HCMV infection and the expansion of NK cell subpopulations that have epigenetically downregulated the FcRγ adaptor (27, 28, 30). Currently, the possible relationship between the NKG2C^bright phenotype and the downregulation of FcRγ along HCMV-induced adaptive NK cell differentiation remains unclear. In the current study, we have combined a systematic analysis of NKG2C^bright and FcRγ⁻ NK cell subsets and the NKG2C genotype in a cohort of healthy individuals, characterizing their phenotype and function at an individual level. The data indicate that the expansion of NKG2C^bright NK cells and downregulation of FcRγ are independent but closely related stages in the progressive differentiation of HCMV-induced adaptive NK cells. Moreover, a possible association among the NKG2C copy number, the accumulation of NKG2C^brightFcRγ⁻ NK cells, and the appearance of NKG2C⁻FcRγ⁻ NK cell subpopulations is suggested. Finally, a functional comparison of the different adaptive NK cell subsets supports the association between NKG2C^bright phenotype and enhanced potential for cytokine production upon CD16-dependent activation, regardless of FcRγ expression.

Reminiscent to the uneven detection of NKG2C^bright NK cells (1, 6), both markers of adaptive NK cells were concurrently detected in a subgroup of HCMV⁺ individuals, whereas others displayed NK cell profiles similar to HCMV⁻ donors. These results support that NKG2C^bright and FcRγ downregulation represent NK cell differentiation signatures linked to a common pattern of response to HCMV. The basis for the restriction of the NKR redistribution to a subset of HCMV⁺ healthy individuals remains uncertain but observations in immunocompromised patients (11, 12, 20), as well as in congenital and perinatal infection (14, 15), strongly support that the reconfiguration of the NK cell compartment may be inversely related to the T cell–mediated control of the virus along HCMV primary infection and/or reactivation.

FcRγ loss was often detected within the NKG2C^bright NK cell subset. Yet, the magnitude of NKG2C^bright NK cell expansion was unrelated to the extent of FcRγ loss, and both markers appeared occasionally dissociated, indicating their independent acquisition along adaptive NK cell differentiation. In this regard, FcRγ⁺ and FcRγ⁻ NKG2C^bright NK cells from the same individual often shared expression patterns of other NK cell markers (i.e., KIR, LILRB1, and CD57), indirectly supporting that FcRγ downregulation takes place at later stages following NKG2C^bright expansion.

It has been proposed that FcRγ downregulation favors coupling of CD16 to CD3 ζ chain adaptor, rendering the response to Ab-dependent activation more efficient (29, 30). The fact that both FcRγ⁺ and FcRγ⁻ NKG2C^bright cells showed comparably high TNF-α production upon CD16 stimulation indicated that enhanced functional competence for cytokine production is possibly acquired early after NKG2C^bright NK cell differentiation, prior to FcRγ loss. Indeed, functional comparison of coexistent NKG2C^brightFcRγ⁻ and NKG2C⁻FcRγ⁻ NK cell subpopulations, following CD16 stimulation, reasserted previous indications that FcRγ downregulation may not account for the enhanced cytokine production of adaptive NK cells. Acquisition of enhanced functional competence appeared more closely associated to the expression of self-specific inhibitory KIR by the NKG2C^bright NK cell subset, as previously recognized for conventional NK cells (36–39). This likely depends on other differentiation events, as supported by the epigenetic regulation of IFNG gene expression in NKG2C^bright NK cells (9). It is of note that NKG2C^bright NK cells developing in TAP-deficient patients express polyclonal KIR repertoires and remain hyporesponsive to CD16 activation (40), thus supporting the association between selective/preferential expression of self-specific inhibitory KIR and acquisition of competence for cytokine secretion.

In contrast, FcRγ loss was directly associated to a reduction on surface NKp46, NKp30, and CD161 levels while preserving CD16, also in NKG2C⁻FcRγ⁻ NK cells (data not shown). The fact that the modulated NKRs depend on a lysine residue in their transmembrane domain for their association with FcRγ, instead of the aspartic acid residue in CD16, could explain their different behavior upon FcRγ loss (41). Nonetheless, the contribution of alternative mechanisms (i.e., epigenetic changes) accounting for the altered surface expression of NCR should also be considered. Thus, FcRγ downregulation is closely associated to changes in the hierarchy of activating receptors governing adaptive NK cell effector function, proliferation, and survival.

In agreement with previous studies, FcRγ⁻ NK cell subpopulations lacking NKG2C were also detected (27, 28, 30). In the absence of NKG2C, activating KIR have been proposed as alternative drivers for HCMV-adaptive NK cell expansions (5, 26). Yet, we did not observe a preferential expression of activating KIRs in NKG2C⁻FcRγ⁻ cells despite the fact that KIR2DS1 and KIR2DS4 were detected among both NKG2C^bright FcRγ⁻ and NKG2C⁻FcRγ⁻ NK cell subsets. Moreover, NKG2C⁻FcRγ⁻ NK cells were present in one NKG2C^del/del and one NKG2C^+/del HCMV⁺ donor who lacked functional activating KIR (homozygous for KIR-A haplotypes with mutated KIR2DS4 alleles). Even though these results do not rule out a contribution of activating KIR to the expansion of adaptive NK cells, they suggest the participation of DAP-12–independent pathway(s) in HCMV-adaptive NK cell differentiation. In this regard, NKp46 has been previously involved in the recognition of HCMV-infected dendritic cells and macrophages (42, 43). It is conceivable that NKp46 engagement by HCMV-infected cells might provide an FcRγ-dependent trigger for initial events in NK cell proliferation/differentiation.

We proposed a role for CD94/NKG2C in the expansion and differentiation of NKG2C^bright NK cells in response to HCMV, presumably upon interaction with a still undefined ligand in infected cells (2, 3, 20). Consistent with this hypothesis, NKG2C copy number influences the surface density of the lectin-like receptor, IL-15–dependent proliferation in response to receptor engagement, and steady-state NKG2C⁺ NK cell numbers in HCMV⁺ subjects (6, 14). According to these results, CD94/NKG2C could participate in the expansion phase of the adaptive NK cell response, preferentially modulating the proliferation, long-term survival, and function of NK cell clones. In this regard, the fact that NKG2C^bright NK cells generally express inhibitory KIR specific for self–HLA-C and tend to exclude self-specific KIR3DL1 suggests a contribution of HLA class I in the selection of NKG2C-driven adaptive NK cells. Indeed, the expansion of NKG2C^bright NK cells with polyclonal KIR repertoires in TAP-deficient patients would reinforce the role of HLA class I in shaping NKG2C-driven adaptive NK cells (40).

Remarkably, NKG2C copy number and NKG2C^bright NK cell numbers appeared inversely related with the incidence of FcRγ downregulation in NKG2C^bright NK cells and the appearance of NKG2C⁻FcRγ⁻ NK cell subpopulations. The optimal response of NKG2C^bright NK cells in NKG2C^+/+ individuals could also underlie the reduced proportion of cells with late differentiation phenotype (e.g., FcRγ loss). By contrast, a suboptimal activation through the CD94/NKG2C receptor in NKG2C^+/del individuals could favor a relative increase of terminally differentiated FcRγ⁻NKG2C^bright cells, allowing the development of alternative NKG2C⁻FcRγ⁻ adaptive NK cell subpopulations. Studies in larger cohorts are warranted to confirm this association. The identification of the elusive CD94/NKG2C ligand in the context of HCMV infection remains a key challenge to precisely unravel the molecular mechanisms underlying the influence of the NKG2C genotype on the differentiation of adaptive NK cells.

Finally, the fact that NKG2C^bright and FcRγ⁻ NK cells are not detected in individuals with other chronic viral infections (e.g., HSV-1), unless coinfected with HCMV (1, 30), indicates that either the initial trigger or subsequent factors driving the differentiation of adaptive NK cells are exclusively related with HCMV infection. Nonetheless, activation of already differentiated NKG2C^bright NK cells by cytokines or via CD16 could account for secondary expansions detected along concomitant viral infections (hepatitis C virus, HIV, Hantavirus, Chikungunya virus, and EBV). The possibility that such secondary expansions contribute to the acquisition of late differentiation features (i.e., FcRγ downregulation) remains unaddressed. In this regard, recent studies have revealed that HCMV-induced adaptive NK cells gather additional epigenetic modifications (i.e., Syk and Eat2 downregulation) (27, 28). Studies on larger cohorts may allow addressing the impact of NKG2C copy number on the progressive accumulation of epigenetic changes in relation to HCMV pressure.

We thank Esther Menoyo for collaborating in obtaining blood samples, Dr. Oscar Fornas for advice in flow cytometry, and volunteer blood donors for participation.

The authors have no financial conflicts of interest.