Natural Killer Cells Adapt to Cytomegalovirus Along a Functionally Static Phenotypic Spectrum in Human Immunodeficiency Virus Infection

Events related to HCMV infection drive accumulation of functionally enhanced CD57posNKG2Cpos adapted NK cells. We investigated NK cell adaptation to HCMV along a proposed continuum progressing from acute activation through maturation and memory formation towards functional exhaustion. Acute exposure to conditioned medium collected 24 h after HCMV infection (HCMVsn) increased NK cell cytotoxicity for all HCMV-seronegative and seropositive donors tested, with mean 38 and 29% boosts in natural and antibody-dependent cell-mediated cytotoxicity (ADCC), respectively. Increases in NK cell cytotoxicity were completely abrogated by blocking type I interferon (IFN) receptors and equivalent responses occurred with exposure to IFN-α2 alone at the same concentration present in HCMVsn. To study longer term effects of HCMV infection, we focused on three groups of human immunodeficiency virus (HIV)-infected subjects distinguished as HCMV-seronegative or HCMV-seropositive with either high (>20%) or low (<6%) fractions of their NK cells expressing NKG2C. The NK cells of all three HIV-infected groups responded to HCMVsn and IFN-α2 in a manner similar to the NK cells of either HCMV-seronegative or seropositive controls. Neither HCMV status, nor the extent of phenotypic evidence of adaptation to HCMV infection significantly affected mean levels of ADCC or CD16-mediated NK cell degranulation and IFN-γ production compared between the HIV-infected groups. Levels of IFN-γ production correlated significantly with the fraction of NK cells lacking FcεRIγ (FcRγ), but not with the fraction of NK cells expressing NKG2C. There was negligible expression of exhaustion markers Lag-3 and PD-1 on NK cells in any of the groups and no significant difference between groups in the fraction of NK cells expressing Tim-3. The fraction of NK cells expressing Tim-3 was unaffected by CD16 stimulation. Relative to the total NK cell population, responses of Tim-3-expressing cells to CD16 stimulation were variably compromised in HCMV seronegative and seropositive groups. In general, NK cell function in response to signaling through CD16 was well preserved in HIV infection and although HCMV had a clear effect on NK cell FcRγ and NKG2C expression, there was little evidence that the level of adaptation to HCMV infection affected CD16-dependent NK cell signaling in HIV infection.

Events related to HCMV infection drive accumulation of functionally enhanced CD57 pos NKG2C pos adapted NK cells. We investigated NK cell adaptation to HCMV along a proposed continuum progressing from acute activation through maturation and memory formation towards functional exhaustion. Acute exposure to conditioned medium collected 24 h after HCMV infection (HCMVsn) increased NK cell cytotoxicity for all HCMV-seronegative and seropositive donors tested, with mean 38 and 29% boosts in natural and antibody-dependent cell-mediated cytotoxicity (ADCC), respectively. Increases in NK cell cytotoxicity were completely abrogated by blocking type I interferon (IFN) receptors and equivalent responses occurred with exposure to IFN-α2 alone at the same concentration present in HCMVsn. To study longer term effects of HCMV infection, we focused on three groups of human immunodeficiency virus (HIV)-infected subjects distinguished as HCMV-seronegative or HCMV-seropositive with either high (>20%) or low (<6%) fractions of their NK cells expressing NKG2C. The NK cells of all three HIV-infected groups responded to HCMVsn and IFN-α2 in a manner similar to the NK cells of either HCMV-seronegative or seropositive controls. Neither HCMV status, nor the extent of phenotypic evidence of adaptation to HCMV infection significantly affected mean levels of ADCC or CD16-mediated NK cell degranulation and IFN-γ production compared between the HIV-infected groups. Levels of IFN-γ production correlated significantly with the fraction of NK cells lacking FcεRIγ (FcRγ), but not with the fraction of NK cells expressing NKG2C. There was negligible expression of exhaustion markers Lag-3 and PD-1 on NK cells in any of the groups and no significant difference between groups in the fraction of NK cells expressing Tim-3. The fraction of NK cells expressing Tim-3 was unaffected by CD16 stimulation. Relative to the total NK cell population, responses of Tim-3-expressing cells to CD16 stimulation were variably compromised in INTRODUCTION Natural killer (NK) cells provide defense against malignancy and infection by recognizing certain alterations in affected cells and responding accordingly. Activation of NK cells against altered host cells is regulated by integration of positive and negative signals received through members of a relatively small set of germ-line encoded receptors. While NK cells recognize altered host cells and mediate effector functions without prior exposure to the altered cells, NK cell functional competence depends upon previous selective engagement of an inhibitory NK cell receptor with a class I human histocompatibility-linked antigen (HLA) (1). This developmental education process enables NK cells to mediate cytotoxicity and/or produce cytokines when encountering altered host cells, while ensuring appropriate levels of self-tolerance (2)(3)(4)(5). It also illustrates that NK cell function is not constitutive upon lineage determination, but dependent upon subsequent relay of specific signals. Selective NK cell education at this developmental stage raises the possibility of further maturation within select subsets under conditions associated with infections or malignancy. Such NK cell maturation clearly takes place in murine cytomegalovirus (MCMV) infection. In this setting, cytokine production and specific receptor ligand interactions between Ly49H and MCMV m157 drive a subset of NK cells expressing the Ly49H activating receptor to selectively expand, persist at elevated levels and provide protection against subsequent MCMV infection (6)(7)(8)(9)(10)(11).
The NK cell response to MCMV infection is the prototype system within which elements required to generate memory NK cells were first identified (12)(13)(14)(15)(16)(17). Infection with human (H)CMV drives selective expansion of NK cells expressing the C-type lectin-like activating receptor NKG2C, prompting speculation that NKG2C on human NK cells functions in an analogous way to Ly49H on murine NK cells (18)(19)(20)(21). In several HCMV-infected cell culture systems, selective expansion of NKG2C-expressing NK cells depends upon the presence of certain cytokines and interaction between NKG2C and HLA-E complexed with particular peptides (22,23). Peptides derived from the HCMV UL40 protein enable selective proliferation of NKG2C-expressing NK cells, including those from HCMV naïve individuals (24). The NKG2C-expressing NK cells expanded in vivo or in vitro by exposure to HCMV acquire phenotypic changes that reflect an increased capacity for effector functions (25)(26)(27). This differentiation produces CD57 pos NK cells with increased CD16 expression, lower levels of the associated FcεRIγ (FcRγ) adaptor protein, reduced natural cytotoxicity receptor (NCR) expression, and epigenetic changes increasing the accessibility of cytokine promoter regions (25,26,28,29). The CD57/NKG2C-expressing NK cells are reportedly more responsive to stimulation through CD16, at least in terms of antibody-dependent cytokine production (25)(26)(27).
Aging, and various forms of immunological stress, including congenital, iatrogenic, and HIV infection, exacerbate HCMVdriven expansion of NKG2C-expressing NK cells (21,(30)(31)(32)(33)(34). It is common for HIV/HCMV co-infected individuals to have large NK cell fractions expressing CD57 and NKG2C, within which limitations to NK cell adaptation imposed by terminal differentiation or exhaustion might be evident (34). Therefore, to assess NK cell function along a phenotypic spectrum of adaptation to HCMV infection, we studied healthy controls and HIV-infected individuals displaying varying degrees of NK cell adaptation. This included HCMV-infected and seronegative controls, an HIV-infected HCMV-seronegative group, an HIV/HCMV co-infected group with small fractions of NKG2C pos NK cells and an HIV/HCMV co-infected group with large fractions of NKG2C pos NK cells. Functional assessment began with exposure of NK cells from HCMV-seronegative controls to HCMV-related cytokines and extended across a wide range of NK cell exposure and adaption to HCMV infection, as indicated by the accumulated fractions of phenotypically adapted NK cells.

Study Subjects and Sample Collection
This study was carried out in accordance with the recommendations of the Canadian Tri-Council Policy Statement: Ethical Conduct for Research Involving Humans. The protocol was approved by the Health Research Ethics Authority of Newfoundland and Labrador, Canada. All subjects gave written informed consent in accordance with the Declaration of Helsinki. Whole blood was collected with informed consent from healthy donors and peripheral blood mononuclear cells (PBMC) isolated by Ficoll-Paque (VWR, Mississauga, ON, Canada) density gradient centrifugation were suspended in lymphocyte medium consisting of RPMI-1640 supplemented with 10% fetal calf serum (FCS), 200 IU/mL penicillin/streptomycin (P/S), 1% 1 M HEPES, 1% L-glutamine (all from Invitrogen, Carlsbad, CA, USA) and 2.0 × 10 −5 M 2-mercaptoethanol (Sigma-Aldrich, St. Louis, MO, USA). Individuals infected with HIV recruited through the Newfoundland and Labrador Provincial HIV Clinic provided informed consent for whole blood collection, immunological studies, and researcher access to medical laboratory records. Freshly isolated PBMC were resuspended in freezing medium composed of lymphocyte medium supplemented to 20% FCS with 10% dimethyl sulfoxide and cooled at 1 • C/min overnight to −80 • C. Frozen PBMC were then maintained in liquid nitrogen until analysis. Cryopreserved PBMC were recovered overnight in lymphocyte medium at 37 • C, 5% CO 2 . Humoral and CD8 pos T cell responses against CMV were measured previously as described and data included with general characteristics of the HIV-infected study subjects (34,35).  (36). Conditioned medium from uninfected (CONsn) and AD169infected (HCMVsn) MRC-5 cells was harvested and medium replaced in 24 h increments post infection for a total of 120 h. Conditioned medium from vSC8-infected MRC-5 cells (vSC8sn) was harvested and medium replaced in 24 h increments post infection for a total of 72 h. Supernatants were clarified for 10 min at 400 g to remove cell debris, frozen at −80 • C in single-use aliquots and recovered on ice. Inactivated HCMVsn was generated by collecting conditioned medium from MRC-5 cells that were exposed to the same amount of ultraviolet (UV)-irradiated AD169 (41 watts for 1 h at 30 cm).

Multiplex Array
Concentrations of cytokines and chemokines within culture medium collected from MRC-5 cells untreated or infected with AD169 or vSC8 (see above) were measured by Milliplex (Millipore, Merck KGaA, Darmstadt, Germany) according to the manufacturer instructions and analyzed on the BioPlex-200 (Bio-Rad, Mississauga, ON, Canada). Samples were run in duplicate and incubated overnight to improve the sensitivity of detection as previously described (38).

Flow Cytometry
Human PBMC were stained using directly conjugated mAb against human CD3 (

Statistical Analysis
Statistical analyses were performed using GraphPad Prism software version 5 with two-sided p < 0.05 considered significant. Normality of data distributions were assessed with Kolmogorov-Smirnov, Shapiro-Wilk and D'Agostino and Pearson tests. Student's paired or unpaired t-tests were used for group comparisons when data were normally distributed and nonparametric Wilcoxon paired signed-rank test or Mann-Whitney U-tests used otherwise as appropriate.

Soluble Factors Released Early in HCMV Infection Augment NK Cell Cytotoxicity
Natural killer cells are considered important effectors during herpesvirus infections. To examine how NK cells respond to factors released early in HCMV infection, we collected supernatant from uninfected (CONsn) and AD169-infected (HCMVsn) MRC-5 fibroblasts over 24 h intervals up to 120 h post infection. Supernatant was collected from vSC8-infected (vSC8sn) MRC-5 fibroblasts over 24 h intervals up to 72 h post infection. We then exposed NK cells to these supernatants in the presence of K562 target cells in 5 h 51 Cr release assays to assess the effects of acute exposure on NK cell cytotoxicity. The vSC8sn collected from any of the three time points had no effect on NK cell cytotoxicity against K562 cells (data not shown, n = 3) compared to CONsn. In contrast, HCMVsn collected within the first 24 h after infection increased NK cell target killing by 50% relative to cytotoxicity in the presence of CONsn ( Figure 1A). While this property of HCMVsn persisted up to 48 h post infection with conditioned media collected over a cumulative 48 h (data not shown), its production peaked during the first 24 h of infection as exposure to HCMVsn collected between 24 and 48 h post infection had little effect on NK cell activity ( Figure 1A). Increased NK cell cytotoxicity following exposure to HCMVsn collected at 24 h occurred for all 12 donors tested, with a mean 38% boost in natural cytotoxicity ( Figure 1B) and 29% increase in NK cell ADCC ( Figure 1C). An equal number of HCMVseropositive and seronegative donors, differentiated by symbol shading, are shown in Figure 1, with no apparent differences in responsiveness to stimulation with HCMVsn related to previous HCMV exposure. Factor(s) present in HCMVsn and primarily produced the first 24 h after infection cause a rapid significant increase in NK cell natural cytotoxicity and ADCC.

HCMVsn Augments Cytotoxicity Triggered Through Multiple NK Cell Activating Receptors
As HCMVsn increased both NK cell natural cytotoxicity and CD16-dependent responses, we investigated the effect of HCMVsn on cytotoxicity through other common NK cell activating receptors. We probed NKG2C, NKG2D, NKp30, NKp44, and NKp46 using a redirected lysis assay in which murine FcR-expressing P815 cells orient mAb specific for particular activating receptors to enable selective NK cell activation. A representative summary of target cell lysis by natural cytotoxicity, ADCC and redirected killing through individual activating receptors is shown in Figure 2A. Using an IgG isotype control antibody, there was little P815 cell lysis and no significant difference from control conditions when HCMVsn was present during the 5 h assay ( Figure 2B). By introducing mAb against individual NK cell activating receptors, we observed that HCMVsn significantly increased P815 cell lysis through NKG2C (Figure 2C), NKG2D (Figure 2D), and the NCRs NKp30 and NKp46, but not NKp44 (Figures 2E-G).
These data demonstrate that HCMVsn sensitizes NK cells for cytotoxicity triggered through CD16, most NCRs and through activating C-type lectin-like receptors. As with the increased killing of K562 and ADCC, there was no apparent difference in the responses of NK cells from HCMV-seropositive and seronegative subjects discriminated on the graphs by different symbols.

IFN-α Produced During HCMV Infection Enhances NK Cytotoxicity
While a cell-free component of HCMVsn increased NK cell pan cytotoxic function, it was unclear whether this effect was mediated through macromolecular interactions with different NK cell activating receptors or by cytokines. Augmented NK cell cytotoxicity was sustained following clearance of macromolecular material by HCMVsn ultracentrifugation (Figure 3A) or following selective removal of HCMV particles by magnetic beads and a mAb against HCMV glycoprotein gB ( Figure 3B). Inactivating HCMV before its addition to MRC-5 cells (UV HCMVsn) prevented the increase in NK cell cytotoxicity observed with replication competent HCMVsn ( Figure 3C), indicating that a soluble product from HCMVinfected cells increases NK cell cytotoxicity through multiple activating receptors.
To identify potential factors present in HCMVsn that increase NK cell cytotoxicity, we measured analytes using a beadbased multiplex assay. Luminex analysis of HCMVsn revealed increased concentrations of monocyte chemoattractant protein and macrophage colony stimulating factor (data not shown), IL-6, IL-8, IL-15, IFN-α2, and IFN-β compared to CONsn ( Figure 3D right and left panels). Modest concentrations of biologically active IL-15 (10 pg/mL; data not shown) were also detected. Analysis of vSC8sn revealed increased concentrations of IL-6, IL-8, and IL-15 with a marginal increase in IFN-β at the 48 h time point. Blocking stimulation through IFNAR alone with a specific mAb fully prevented any increase in NK cell activity in response to HCMVsn (Figure 3E). This blocking effect was specific for type I IFNs as increased NK cell activity in response to IL-2 was maintained in the presence of the same amount of mAb against IFNAR (data not shown). HCMV-infected fibroblasts produced both IFN-α2 and IFN-β (80 and 450 pg/mL, respectively, data not shown) during the first 24 h of infection. To determine whether the amount of IFN-α2 detected in HCMVsn during the first 24 h of infection could alone mediate the same increases Frontiers in Immunology | www.frontiersin.org in NK cell cytotoxicity as HCMVsn, we compared NK cell natural and antibody-dependent cytotoxic activity in response to HCMVsn or purified recombinant IFN-α2. Exposure to similar concentrations of IFN-α2 (20 pg/mL) as added with HCMVsn diluted 1 in 5 in 51 Cr release assays increased NK cell natural cytotoxicity ( Figure 3F) and ADCC ( Figure 3G) to an extent indistinguishable from the increases mediated by HCMVsn. Thus, stimulation through the type I IFN receptor alone rapidly sensitized NK cells to natural and antibody-dependent cytotoxicity and the acute increase in natural cytotoxicity and ADCC for NK cells from healthy controls mediated by soluble factors in HCMVsn is replicated with IFN-α2 alone at levels present in HCMVsn.
The increase in NK cell cytotoxicity following exposure to HCMVsn occurred with both HCMV pos and HCMV neg healthy donors indicating that the NK cell response to cytokines produced by HCMV-infected cells is independent of previous sensitization from in vivo HCMV infection (Figures 1, 2). To investigate such responses along a continuum of NK cell adaptation to HCMV infection, we tested effects of HCMVsn and IFN-α2 with HIV-infected subjects whose NK cells spanned a broad range of adaptation in response to HCMV infection.

HCMVsn and IFN-α2 Enhance NK Cell ADCC in HIV-Infected Individuals
After establishing that NK cells from healthy donors respond rapidly, robustly and independent of HCMV serostatus to IFN-α2 produced by HCMV-infected fibroblasts, we next studied NK cell responses from HCMV neg and HCMV pos individuals within an HIV-infected study cohort. Since adapted NK cells exhibit enhanced CD16-mediated effector responses, we measured the impact of HCMVsn and IFN-α2 on NK cell ADCC in HIV infection and compared baseline levels of ADCC between three groups reflecting a broad range of NK cell adaptation to HCMV infection. General characteristics of the individuals and comparisons between HCMV-seronegative and seropositive groups are shown in Table 1. We selected three groups of HIV-infected study subjects, two distinguished by high (>20%) vs. low (<6%) fractions of NK cells expressing NKG2C and a third distinguished by seronegative HCMV status ( Table 1). Nine of the 28 age-and sex-matched HIVinfected individuals selected for this study were HCMVseronegative. The efficacy of antiretroviral therapy was roughly equivalent between groups as indicated by undetectable HIV viral loads (<50 copies HIV RNA/mL plasma). Neither nadir, nor current CD4 pos T cell counts, indicative of past disease progression and present immunological status, respectively, differed significantly between groups. All donors had a robust fraction of mature (CD57 pos ) NK cells, independent of HCMV status or fraction of their NK cells expressing NKG2C. However, the percentage of NK cells expressing CD57 was significantly lower in the NKG2C lo group than in the NKG2C hi group (p = 0.0133, Mann-Whitney U-test), suggesting lesser overall NK cell maturation in the NKG2C lo group ( Table 1). There were no significant differences in either humoral (anti-CMV IgG levels) or cellular (% CD8 pos T cells specific for CMV pp65 and IE-1) immune responses against CMV between the HCMV pos groups with high or low fractions of NK cells expressing NKG2C.
The HCMV neg individuals and those with high or low fractions of NKG2C-expressing NK cells were pooled to assess responsiveness of NK cells from HCMV neg and HCMV pos subjects infected with HIV to HCMV-related cytokines. The mean increase in ADCC mediated by HCMV-related cytokines (24%) was similar to that seen in the HIV-naïve group and was not significantly different from the mean 14% increase elicited by recombinant IFN-α2 alone (Figure 4A). Neither HIV-infection nor adaptation to HCMV infection appeared to affect NK cell responsiveness with acute exposure to IFN-α2. Median levels of NK cell ADCC measured from HIV-infected individuals were also unrelated to either HCMV serostatus or NK cell adaptation to HCMV infection ( Figure 4B). Adaptation to HCMV infection had little effect on the capacity of NK cells from HIV-infected individuals to respond to acute IFN-α2 exposure or to mediate ADCC. anti-NKp30, (F) anti-NKp44, or (G) anti-NKp46 mAbs. Assays represented in plots (B-G) were performed in triplicate for 7-9 individuals. Subjects seropositive for HCMV are represented with filled black circles or yellow squares half-shaded black. Error bars for isotype control and anti-NKp46 plots represent median with IQR and the probability of a difference between conditions was assessed by Wilcoxon signed-rank test. All other bars represent mean ± SD with comparison between conditions by Student's paired t-test.

Long Term Adaptation of NK Cells From HIV-Infected Individuals to HCMV Infection
To study longer term effects of HCMV infection, including the possibility of progression to exhaustion, we compared features of the three groups of HIV-infected subjects distinguished by high vs. low NK cell fractions expressing NKG2C or by seronegative HCMV status ( Table 1). The flow cytometry gating strategy for a representative HIV/HCMV co-infected donor is shown in Figure 5A. Median NKG2C expression levels were similar between the HCMV neg and NKG2C lo groups (Figure 5B), however, the influence of adaptation to HCMV within the NKG2C lo group was evident from the lower mean fraction of NK cells expressing FcRγ ( Figure 5C) and strong correlation between loss of FcRγ expression and level of HCMV-specific antibodies (Spearman r = 0.5539, p = 0.0022 Table 1). Loss of the FcRγ signaling adaptor subunit in parallel with expansion of NK cells expressing NKG2C is apparent from the progressive increase in percentage of FcRγ neg NK cells across the three groups and by the significant correlation between fractions of NK cells expressing NKG2C and fractions lacking FcRγ (Figures 5C, D). Despite this general progression and significant correlation, three HIV-infected individuals within the NKG2C lo group with <6% of their NK cells expressing NKG2C and >50% lacking FcRγ illustrate that loss of FcRγ by NK cells is not inexorably linked to NKG2C expression ( Figure 5C, Table 1).

Preservation of NK Cell Antibody-Dependent Effector Functions in HIV-Infected Subjects
An association between NK cell adaptation to HCMV infection and increased capacity for antibody-dependent IFN-γ production was previously reported (39,40). Since ADCC responses against antibody-coated C1R-B27 cells by the HIV-infected subjects measured by 51 Cr release assays were similar, irrespective of HCMV status and NKG2C expression levels, we next measured and compared IFN-γ production and CD107a expression of NK cells in the three groups in response to CD16 stimulation.
Activation through CD16 induced robust NK cell IFN-γ responses (representative plots in Figure 6A), the magnitude of which directly correlated with the percentage of NK cells lacking FcRγ (Figure 6B). This is exemplified by one subject with over 80% NK cells lacking FcRγ and over 40% producing IFN-γ in response to stimulation through CD16 ( Figure 6B). In contrast, we found no significant correlation between the fraction of FcRγ neg NK cells and fraction of NK cells expressing NKG2C or between NK cell FcRγ expression and degranulation, as measured by CD107a expression following CD16 stimulation. When we compared CD16-stimulated degranulation and IFN-γ production between HCMV neg , NKG2C lo , and NKG2C hi groups, there were no significant differences in percentages of CD107a pos NK cells (Figure 6C), IFN-γ pos NK cells (Figure 6D) or polyfunctional NK cells doubly positive for CD107a and IFN-γ ( Figure 6E). All groups' NK cells responded robustly to CD16 stimulation, irrespective of NKG2C expression levels or HCMV serostatus, illustrating HCMV-independent preservation of antibody-dependent NK cell responses in chronic HIV infection.
To assess NK cell progression toward an exhausted phenotype, expression of PD-1, LAG-3, and Tim-3 was assessed on NK cells within the three defined groups. Levels of PD-1 and LAG-3 were low to undetectable on NK cells (data not shown), but substantial fractions of NK cells within each group expressed Tim-3 ( Figure 5A, Table 1).  NKG2C or (C) FcRγ. Individual donors were grouped based on HCMV serostatus (HCMV neg ) and fraction of NK cells expressing NKG2C within the HCMV pos donors (NKG2C lo or NKG2C hi ). Correlation between NK cell NKG2C and FcRγ expression was assessed in (D) with Spearman correlation coefficient (r) calculated and the probability of a significant correlation (p) shown on the graph. Error bars in (B) for the HCMV neg group represent median with IQR, while all others represent mean ± SD. Groups were compared using Mann-Whitney U-tests.  Figure 5. Error bars represent mean ± SD where data was normally distributed and median with IQR in all other cases.

Antibody-Dependent Effector Functions of Tim-3 pos NK Cells From HIV-Infected Subjects
As Tim-3 receptors were expressed at levels readily measurable by flow cytometry, we compared the extent of Tim-3 expression on NK cells within the three groups and the functional capacity of the Tim-3 pos subset. The percentage of NK cells expressing Tim-3 ranged from 0 to 10%, did not differ significantly between groups ( Table 1) and CD16-stimulation did not induce NK cell Tim-3 expression (data not shown). The percentage of Tim-3 pos NK cells responding to CD16 stimulation with IFN-γ expression, CD107a expression, or both, was compared with the percentage of total NK cells responding likewise. Tim-3-expressing NK cells from HCMV neg HIV-infected donors degranulated to a lesser extent than the general NK cell population but showed no apparent deficit in IFN-γ production ( Figure 7A). In contrast, Tim-3 pos NK cells from HIV/HCMV co-infected donors responded similarly to the general NK cell population in terms of CD107a expression, but were less likely to produce IFN-γ in response to stimulation through CD16 ( Figure 7B). The significance of Tim-3 expression on NK cells may vary with respect to the function studied and extent of adaptation to HCMV infection. There were no significant differences between HCMV neg , NKG2C lo , and NKG2C hi groups in percentage of Tim-3 pos NK cells expressing CD107a or IFN-γ in response to stimulation through CD16 Figure 7C.

DISCUSSION
Expansion of NKG2C-expressing NK cells was initially reported in the context of multiple viral infections, including hepatitis C virus, HIV, chikungunya, and hantavirus, but subsequent investigation identified HCMV infection as the critical common denominator (41,42). While only a small fraction of NK cells in healthy HCMV-seronegative individuals usually express NKG2C, HCMV infection triggers various levels of adaptation, resulting in more than 80% NKG2C pos NK cells in extreme cases (34). Therefore, we envision a continuum of NK cell adaptation to HCMV infection through which naïve NK cells respond to initial HCMV exposure, differentiate into mature effector cells and, under the influence of time, repeated exposure and other factors, progress toward terminal differentiation. In this study, we investigated NK cell function along this proposed continuum, beginning with exposure of naïve NK cells to cytokines present during acute HCMV infection and encompassing responses of NK cell populations reflecting broadly different degrees of adaptation to HCMV. To compare the quality of NK cell responses across the spectrum of adaptation to HCMV, we selected subjects from a well characterized Acute exposure to conditioned media collected from HCMVinfected fibroblasts elevated NK cell cytotoxicity through NCRs, C-type lectin receptors and CD16, irrespective of NK cell donor HIV or HCMV status. While enhanced antibodydependent effector function through CD16 is a characteristic of the NKG2C pos NK cells adapted to HCMV infection, these adapted NK cells are also reported to downregulate NCRs (18). Downregulation of NCRs is a common consequence of chronic viral infection, likely involving physical interactions and chronic cytokine stimulation. In combination with additional stimulation, pan increases in NCR-mediated cytotoxicity stimulated by HCMV-related cytokines may lend toward phenotypic alterations over periods of prolonged cytokine exposure. While IL-6, IL-8, and IL-15 were present in HCMVsn together with type I IFNs, these same cytokines were also elevated in vSC8sn, which did not increase NK cell cytotoxicity, indicating that at most, they may play some ancillary role in acute activation of NK cell cytotoxicity. Our data indicate that IFN-α2 is sufficient to effect the same increases in NK cell cytotoxicity as mediated by conditioned media from HCMV-infected fibroblasts. There were no significant differences between NK cell responses to purified recombinant IFN-α2 vs. HCMVsn in either the healthy control cohort or HIV-infected group. As type I IFNs are produced in a variety of viral infections that do not lead to expansion of NKG2C pos NK cells, it is unlikely that acute exposure to IFN-α2 alone contributes significantly to the NK cell adaptation specific to HCMV infection. Infection with HCMV in vivo may result in different patterns of type I IFN production and cellular responses dependent upon the local environment, cell type infected and virus characteristics.
By selecting three groups of individuals with comparable features of HIV infection, we were able to focus on the impact that different levels of adaptation to HCMV infection have on NK cell phenotype and function in this setting. Two HCMV-infected groups were selected for high vs. low levels of NKG2C expression, with the low NKG2C expression group indistinguishable from the HCMV neg group in this respect. Although one donor within the NKG2C lo category had a CD57 pos NK cell fraction comparable to those in the NKG2C hi group, the rest had low to moderate (10-27%) CD57 pos NK cell fractions. The influence of HCMV infection on FcRγ downregulation was the most apparent phenotypic aspect, displaying a clear hierarchy; the HCMV neg group had the lowest fraction of FcRγ neg NK cells followed sequentially by the NKG2C lo and NKG2C hi groups. A disconnect between NKG2C expression and loss of FcRγ was notable in three individuals with <6% NKG2C pos and more than 50% FcRγ neg NK cells. Although it involves only three individuals in this case, the disconnect has also been noted in studies of NKG2C null cohorts, where lack of NKG2C had little effect on HCMV-driven NK cell maturation (43)(44)(45). Despite these three outliers, there was a strong overall correlation between the fraction of NK cells expressing NKG2C and fraction of FcRγ neg NK cells. Thus, NK cells from these outliers may express features similar to those from NKG2C null subjects (45). Evidently, loss of FcRγ is a more consistent indicator of NK cell differentiation and adaptation in response to HCMV infection than NKG2C expression.
Adaptation to HCMV infection through NKG2C pos NK cell expansion and loss of FcRγ reportedly produces an NK cell population with superior CD16-mediated effector functions (25)(26)(27). In the context of our HIV-infected study cohort, there was no evidence of this in cytotoxicity assays against antibodycoated target cells. Although we saw significant correlation between the size of the IFN-γ response and fraction of NK cells lacking FcRγ, there was also no significant difference between the groups in terms of the mean fraction of NK cells producing IFN-γ in response to CD16 signaling. Despite low levels of NKG2C pos FcRγ neg NK cells, the HCMV neg group had CD16-mediated responses equally as robust as the other groups and the HCMV pos NKG2C lo group displayed no functional deficits relative to either of the other groups. This finding is somewhat unexpected in light of epigenetic remodeling of the NK cell IFNG locus following HCMV infection, with the adapted NK cell population reportedly exhibiting enhanced IFNγ responses (39,40). Whether this apparent discrepancy reflects the impact of HIV infection, different experimental methodology or peculiarities of the limited number of subjects tested in this study remains to be determined.
The extent of NK cell degranulation following CD16stimulation was also not significantly different between groups and in contrast to IFN-γ production, we noted no significant correlation between the magnitude of CD107a responses and fraction of NK cells lacking FcRγ. Polyfunctional NK cell responses, indicated by dual expression of IFN-γ and CD107a also did not differ between groups. Thus, our data suggest that in the context of HIV infection, phenotypic evidence of NK cell adaptation to HCMV infection does not equate with superior CD16-mediated effector functions. The finding here most consistent with previous studies using donors not infected with HIV was the correlation between CD16-triggered IFN-γ production and loss of FcRγ adaptor subunits (25,26). Whether loss of FcRγ is a marker for other alterations or its absence plays a direct role in enhanced signaling through CD16, this aspect of NK cell adaptation to HCMV infection appears to be maintained in HIV infection. We saw no relationship between FcRγ expression and either ADCC measured by 51 Cr release or CD16-triggered degranulation, lending credence to speculation that NK cell adaptation to HCMV affects CD16-mediated cytokine production more so than it affects cytotoxicity.
To investigate NK cell progression toward terminal differentiation, we assessed PD-1, Lag-3 and Tim-3 expression. We detected little to no PD-1 or Lag-3 in any of the three groups, yet a significant fraction of NK cells expressed Tim-3. Tim-3 expression levels did not differ between groups and were unaffected by CD16 stimulation. Although recent reports suggest enhanced function of Tim-3 pos NK cells, we observed functional deficits in degranulation and IFN-γ production in the HCMV neg and HCMV pos groups, respectively, relative to the general NK cell population (46). Although significant, these deficits were relatively slight and based on the responses of a small group of HIV-infected individuals. Further study of the role Tim-3 has on NK cell functions in different settings is warranted.
In summary, type I IFNs produced during in vitro HCMV infection of fibroblasts increased NK cell cytotoxicity through multiple receptors. This increase in NK cytotoxicity occurred with NK cells from HCMV-seronegative and seropositive healthy controls. In the HIV-infected subjects, we saw a similar increase in ADCC that was unrelated to HCMV status or extent of NK cell adaptation to HCMV infection. Cytokine and degranulation responses mediated through CD16 were well preserved in the HIV-infected individuals we studied, again unrelated to their HCMV status. Despite an inverse correlation overall between NK cell CD16-triggered IFN-γ production and FcRγ expression, the HCMV neg , HIV-infected subject group did not have a significantly lesser IFN-γ response than either of the groups with higher fractions of FcRγ neg NK cells. There was no evidence that NK cell adaptation to HCMV affects degranulation responses or cytotoxicity triggered through CD16 in HIV infection. While these findings suggest that HCMVrelated NK cell adaptation has different or lesser functional consequence in HIV infection, the clear effect of HCMV on FcRγ expression and the extreme levels of adaptation observed in terms of NKG2C expression illustrate the same selectivity operating with increased pressure. Immunological pressures associated with HIV infection may preserve or enhance NK cell function through compensatory pathways distinct from HCMV-driven adaptation.

AUTHOR CONTRIBUTIONS
The study was conceived and planned by MG and KH. Experimental work, data analysis, and data presentation was primarily performed by KH. JL carried out the luminex analysis on supernatant from CON, HCMV, and vSC8-infected cells. The manuscript was drafted and revised by KH and MG.