NK Cell and Ig Interplay in Defense against Herpes Simplex Virus Type 1: Epistatic Interaction of CD16A and IgG1 Allotypes of Variable Affinities Modulates Antibody-Dependent Cellular Cytotoxicity and Susceptibility to Clinical Reactivation

Herpes simplex virus type 1 infects most individuals, establishing life-long latency with periodic reactivations that result in clinical relapses or asymptomatic virus shedding. Although exposure to HSV-1 is virtually universal in adults, the clinical course of the infection varies greatly and depends, in part, on host genetic factors, including polymorphisms at the interface of innate and adaptive immunity, like those of KIR and their HLA ligands and the IgG receptor CD16A (1–3). Viral replication is controlled by both branches of the immune response, of which key components are human NK cells. This lymphocyte subset mediates both cytokine production and cellular toxicity, and its behavior in the presence of a target cell depends on the balance of activating and inhibitory signals transduced by multiple NK cell receptors (4). HSV-1–specific Abs are detectable ∼1 mo after primary infection and last throughout one’s life. The great majority of seropositive individuals carry Abs of IgG1 subclass, whereas HSV-specific IgG3 is detected in only approximately half of them; HSV-IgG4 and HSV-IgG2 are rare (5). These specific Abs have a demonstrated neutralizing activity and readily mediate complement-dependent and Ab-dependent cellular cytotoxicity (ADCC) (5). Yet, their presence in virtually all infected individuals does not prevent sporadic or recurrent episodes of symptomatic reactivation in up to half of them.

Several mechanisms help HSV-1 to evade the host immune system (6). Among these, HSV-1 glycoprotein E (gE) and glycoprotein I (gI) form a viral receptor for the crystallizable fragment of Igs (vFcγR), which binds human IgG1, IgG2, and IgG4 with different affinities and is expressed on both the viral envelope and the infected cell membrane (7–11). The gE/gI decoy vFcγR does not transmit any activating signal, but it can hamper host cell FcγR and C1q binding to immune complexes, thus protecting the infected cells from ADCC, complement-dependent neutralization, and phagocytosis.

ADCC, a major pathway in the clearance of pathogens, depends on simultaneous binding of IgG molecules to infected cells and to FcγR expressed on a variety of effector cells, most importantly NK cells (12). Low-intermediate affinity receptors for the Fc of IgG are essential for the cellular immune response to Ag-bound IgG. Among these, CD16A, CD16B, CD32A, CD32B, and CD32C are encoded by five genes (FCGR3A–B and FCGR2A–C, respectively) clustered in ∼150,000 bp on the long arm of chromosome 1, the FCGR locus. Several of these genes display functional polymorphisms, including both sequence and copy number variation (13, 14). Potentially relevant for NK cell–mediated ADCC are genetic variations involving three of the FCGR genes (FCGR2B, FCGR2C, and FCGR3A, coding for CD32B, CD32C, and CD16A, respectively) that can be expressed on these cells, whereas CD32A (FCGR2A) is expressed primarily on myeloid leukocytes, and CD16B (FCGR3B) is expressed on polymorphonuclear cells. CD16A, the major FcγR on NK cells (and, hence, in ADCC), is expressed on the great majority of CD3⁻CD56^dim lymphocytes in all healthy individuals. In contrast, only a minority of humans have NK cells expressing the activating FcγR CD32C, resulting from a single nucleotide polymorphism in exon 3 or a splice-site mutation in intron 7 determining premature termination of the FCGR2C reading frame on both chromosomes of most individuals (15, 16). Furthermore, a recently described ∼70-kbp deletion in the FCGR cluster, involving FCGR2C, HSPA7, and FCGR3B, was shown to determine ectopic NK cell expression of inhibitory CD32B (normally a B cell and myeloid lineage receptor) (15, 17). Finally, a functional allelic dimorphism in FCGR3A, which encodes CD16A, generates allotypes with different affinity for IgG (18): a valine change for phenylalanine 158, a residue that interacts with the lower hinge region of IgG (19, 20), increases the receptor affinity for IgG1 and IgG3 and allows binding of IgG4.

Both the changes in the affinity for IgG and NK cell expression of additional FcγR besides CD16A might influence ADCC efficacy. Indeed, previous studies related the development of autoimmune processes with functional polymorphisms in both FCGR3A and FCGR2C (14, 21), and with FCGR3B deletion (22, 23) that determines ectopic CD32B expression on NK cells. Paradoxically, the possible influence of the FcγR polymorphism on the defense against viral infection has not been explored sufficiently; in this regard, we showed previously that homozygosis for the higher-affinity allele CD16A-158V is associated with an asymptomatic course of HSV-1 infection, whereas the CD32A-131H/R dimorphism is not (2).

IgG allotypes, determined by polymorphic residues on the constant region of H and L chains, also modulate avidity of the FcγR–IgG interaction and could influence, through additional mechanisms, the efficacy of a humoral response (24). Indeed, several of these genetic markers were associated with autoimmunity, malignant diseases, and susceptibility to infection (24-26) and could modulate the immune response to HSV-1. Twenty-six Ig allotypes have been identified in humans; these include 20 Gm allotypes, determined by polymorphisms in the gamma H chain locus (IGHG), and another 3 found on the kappa L chain (Km1, Km2, and Km3) (27). Polymorphic residues on the gamma1 H chain determine six IgG1 allotypes (G1m1, G1m2, G1m3, G1m, 17, G1m27, and G1m28). Because linkage disequilibrium between G1m allotypes within a racial group is almost absolute, these genetic markers are inherited in fixed haplotypes. G1m common haplotypes in whites carry G1m3 (arginine 214, normally linked to glutamate 356 and methionine 358) or G1m17,1 (lysine 214, aspartate 356, leucine 358) (27). Of note, G1m3 binds to the HSV-1–encoded FcγR with lower affinity than does G1m17,1 (7), leading investigators to hypothesize a differential risk for developing HSV-1–driven disease determined by these allotypes (28). Furthermore, affinity-binding assays found slightly, but consistently, higher affinity of CD16A-158V and CD16A-158F for the G1m3 allotype compared with the G1m17,1 allotype (29).

In the current study, we attempted to extend our knowledge on the influence of the FCGR cluster genetic variation and other immunogenetic factors on immunity to HSV-1. To that end, we analyzed the contribution of the major IgG1 and kappa chain allotypes, as well as their interaction with FcγR allelic variants, to the clinical course of HSV-1 infection. Furthermore, using an in vitro setting of HSV-1 infection, we studied the functional influence of IgG1 and CD16A allotypes, as well as CD32B and CD32C NK cell expression, on NK cell–mediated ADCC.

A total of 226 healthy individuals was selected from a previously studied cohort (2), based on the presence, or lack thereof, of specific IgG Abs against HSV-1 and clinical HSV-1 manifestation. A group of 65 seronegative subjects was used as controls for several genetic markers, and their genotype distributions did not differ from those of seropositive individuals (Supplemental Table I). Frozen DNA, serum, and PBMC samples previously extracted from venous blood after informed consent were used for genotyping and functional studies, respectively. When required, new blood samples were obtained, and PBMCs were isolated by density gradient separation using Lymphoprep (Axis-Shield, Oslo, Norway).

For ADCC experiments, we used PBMCs from 10 donors carrying a CD16A-158V/V genotype and another 10 with a CD16A-158F/F genotype. Additionally, sera were obtained from 10 donors: 2 seronegative and 8 seropositive for HSV-1. Of these, 4 each had the G1m3/3 or G1m17/17 genotype. Pairs of G1m3/3 and 17/17 sera were made, ensuring that both sera of each pair had similar specific anti–HSV-1 IgG titers on ELISA tests (Vircell, Granada, Spain). Serum isolated from these subjects was heat inactivated and frozen in individual aliquots for each ADCC experiment. For the assessment of CD32B and CD32C influence on NK cell–mediated ADCC against HSV-1–infected targets, we selected 12 individuals with different CD32B/C genotypes and expression patterns (4 donors with each of the following genotypes: CD32B⁻/CD32C⁻, CD32B⁺/CD32C⁻, and CD32B⁻/CD32C⁺, Supplemental Fig. 1). For phenotype analysis of degranulating NK cells, we selected six individuals displaying clearly identifiable expansions of NKG2C^bright NK cells.

For the determination of IgG1 markers G1m3 and Glm17 (a G-to-A substitution determining an arginine-to-lysine change at residue 214 of the CH1 region), we used a TaqMan genotyping assay from Applied Biosystems (Foster City, CA). In brief, the assay includes a PCR with primers 5′-CCCAGACCTACATCTGCAACGTGA-3′ (forward) and 5′-CTGCCCTGGACTGGGACTGCAT-3′ (reverse), which specifically amplify a 161-bp fragment of the IGHG1 gene, as well as probes that discriminate the single nucleotide polymorphism VIC-CTCTCACCAACTTTCTTGT-NFQ (G1m17 specific) and FAM-CTCTCACCAACTCTCTTGT-NFQ (G1m3 specific). A PCR-restriction fragment-length polymorphism-based approach (30) was used to identify Km1 and Km3 allotypes. Three alleles, Km1, Km1,2, and Km3, segregate at the Km locus in IGKC. The Km1 allele, without Km2, is extremely rare; >98% of the individuals positive for Km1 are also positive for Km2. In this article, as well as in most other publications, positivity for Km1 includes the Km1 and Km1,2 alleles.

FCGR cluster polymorphism was analyzed using a multiplex ligation-dependent probe amplification (MLPA) method (MRC-Holland, Amsterdam, The Netherlands). The FCGR-specific MLPA assay includes gene-specific probes designed to determine the copy number variation in FCGR2A, FCGR2B, FCGR2C, FCGR3A, and FCGR3B, as well as HSPA6 and HSPA7, which are found in the same gene cluster, thus allowing for a highly accurate prediction of ectopic CD32B expression on NK cells. Furthermore, it contains probes for multiple single nucleotide polymorphisms in FCGR2A, FCGR2B, FCGR2C, FCGR3A, and FCGR3B, including those encoding CD16A-158V/F and CD32A-131H/R dimorphisms and determining a premature stop codon in FCGR2C. Therefore, this method further validated our original results on CD16A-158V/F and CD32A-131H/R genotyping using a PCR method with confronting two-pair primers, as previously described (2, 31).

Human telomerase reverse-transcriptase–immortalized (hTERT) HCA2 dermal fibroblasts (HCA2 hTERT; HLA-A*01, B*08,*41, C*07,*17, kindly provided by Dr. Gavin Wilkinson, University of Wales College of Medicine, Cardiff, U.K.) and BHK21 cells (a gift from Dr. S. Efstathiou, Department of Pathology, University of Cambridge, Cambridge, U.K.) were grown in DMEM (Lonza, Verviers, Belgium) supplemented with 10% FBS, penicillin, and streptomycin. The 49-44 cell line was grown in DMEM supplemented with 10% FBS, 0.6 mg/ml G418, and 4 mg/ml puromycin.

To ensure high viability of infected fibroblasts as stimulators of ADCC, we used a glycoprotein H–null HSV-1 mutant, dh1a, derived from HSV-1 strain 17, unless stated otherwise (32). Dh1a stocks were prepared and titered in 49-44 cells, a complementing cell line expressing glycoprotein H. In selected experiments we used sΔUS8, a gE⁻ mutant derived from the SC16 HSV-1 strain (33) (kindly provided by Dr. Helena Browne, Department of Pathology, University of Cambridge, Cambridge, U.K.); this virus was titered on both HCA2 hTERT and BHK21 cell lines. Viral infections were performed with a multiplicity of infection of 10, conditions under which >95% of fibroblasts were infected, as confirmed by flow cytometry assays using anti-gE Ab (clone H600; Santa Cruz Biotechnology, Santa Cruz, CA) and anti-ICP4 [hybridoma 58S (34); American Type Culture Collection] (data not shown).

Degranulation of NK cells was assessed with anti-CD107a–PE–Cy5 (clone H4A3), anti-CD56–ECD (clone N901/NKH-1), and anti-CD3–FITC (clone UCHT1) (Beckman Coulter, Fullerton, CA). Surface expression of FcγRs and other NK cell receptors was analyzed using anti-CD16–FITC (clone NKP15; BD Biosciences, San Diego, CA), anti-CD32B/C (clone 2B6; purchased from LGC Standards, Middlesex, U.K. and used as culture supernatant), anti-CD57–PE (clone hNK-1; BD Biosciences) or anti-CD57–VioBlue (clone hNK-1; Miltenyi Biotec, Bergisch Gladbach, Germany), anti-NKG2C–PE (clone 134591; R&D Systems, Minneapolis, MN), anti-NKG2A–FITC (clone REA110; Miltenyi Biotec), anti-CD3–PE–Cy7 (clone SK7; BioLegend, San Diego, CA), anti-CD56–allophycocyanin (clone CMSSB; eBioscience, San Diego, CA), anti-CD107a–allophycocyanin–Cy7 (clone 1D4B; BioLegend), anti-FcεRIγ subunit (Millipore, Temecula, CA), PE-conjugated F(ab′)₂ goat anti-mouse Ig, PE-conjugated F(ab′)₂ goat anti-rabbit Ig, and PerCP-conjugated F(ab′)₂ goat anti-rabbit Ig (Jackson ImmunoResearch, West Grove, PA). For vFcγR-blocking experiments, we used two monoclonal anti-gE Abs (clone H600; Santa Cruz Biotechnology and clone M612452; Abcam, Cambridge, U.K.) and purified rabbit IgG (Jackson ImmunoResearch) aggregated at 63°C for 20 min; blocking was assessed by comparing the binding of human serum obtained from two HSV-1–seronegative individuals with HSV-1–infected fibroblasts in the presence or absence of anti-gE Abs.

Cryopreserved PBMCs were thawed and rested overnight in RPMI 1640 supplemented with 10% FBS, glutamine, penicillin, and streptomycin. HCA2 hTERT cells were infected with dh1a HSV-1 strain for 2 h, washed with PBS to remove unadsorbed virus, and cultured overnight in DMEM supplemented with 10% FBS, glutamine, penicillin, and streptomycin. The next day, fibroblasts were harvested using a 5-mM EDTA solution in PBS. Infected or uninfected HCA2 hTERT cells were cultured with PBMCs and 0.5% heat-inactivated human serum in 96-well round-bottom plates for 2 h at 37°C. PBMC numbers were adjusted to achieve an E:T (NK cell/fibroblast) ratio of 1:1. Where indicated, a 30-min incubation with blocking Abs preceded coculture of HCA2 hTERT–infected fibroblasts with PBMCs. Each experiment was performed with PBMCs from two age-matched individuals with CD16A-158V/V and CD16A-158F/F genotypes in parallel, as well as with serum from at least one seronegative and two seropositive individuals with G1m3/3 and G1m17/17 allotypes, respectively.

Because the presence of CD32B or CD32C on NK cells from some individuals might influence ADCC potency, we selected paired samples for the functional assays, ensuring that both individuals in each pair either lacked or expressed CD32C at similar levels on their NK cells and that no individual carrying a CD16A-158F/F allotype expressed the inhibitory FcR CD32B. In line with this, all selected donors had two copies of FCGR3A and comparable baseline expression levels of CD16A.

NK cell degranulation was assessed by four-color flow cytometry, using a COULTER EPICS XL-MCL Flow Cytometer, and analyzed with Expo32 ADC software (both from Beckman Coulter). The memory-like phenotype of degranulating NK cells was assessed by eight-color flow cytometry using a MACSQuant Analyzer and MACSQuantify software (Miltenyi Biotec). Percentages of degranulating NK cells were obtained by selecting CD107a⁺ from CD56^dimCD3⁻ lymphocytes or from specific subsets of these cells (gating details are shown in Supplemental Fig. 1). Background degranulation levels, attributable to natural cytotoxicity against infected fibroblasts, were assessed in all ADCC experiments in the presence of serum lacking HSV-1–specific IgG (mean, 8.7%; range: 2.5%–19.6%); all reported percentages of degranulating NK cells were calculated by subtracting this background level from the degranulation levels obtained in the presence of HSV-1–immune sera.

A possible confounding effect derived from interference of KIR2DL2 and its HLA-C1 ligands (associated with HSV-1 disease) in ADCC assays was doubly controlled. First, by normalization of degranulation levels by subtracting natural cytotoxicity, as described above; in addition, we used multiple pairs of NK cell donors with divergent CD16A-158 genotypes such that both, neither, or one of the donors could have NK cells educated by KIR2DL2 and C1 (as ascertained by the presence of both factors in their genomes). In no experiment did these KIR/ligand phenotype combinations modify the predominance of the ADCC response exerted by CD16A-158V/V over CD16A-158F/F NK cells (see 8Results).

Gene frequencies were estimated by direct counting, and they were compared using the Fisher exact test. The nonparametric Wilcoxon matched-pairs signed-rank test was used to compare NK cell degranulation between different subgroups. A significance level of 5% was chosen. A two-sided p value was always chosen, unless otherwise stated.

Forward stepwise multiple logistic regression was used to measure the independent effect on HSV-1 disease of the joint presence of a CD16A-158V/V and G1m3/3 allotypes, and of another five previously identified susceptibility and protection genetic markers (SPSS 14.0 package; SPSS, Chicago, IL).

Ig allelic polymorphisms could, by themselves or through interaction with other genes, modify the defense against infection. To assess their possible influence on the HSV-1–driven host response and to extend our previous analysis on immunogenetic factors affecting susceptibility to the pathogen, we genotyped the two main H chain allotypes of IgG1, a major component of the humoral response against HSV-1, in symptomatic and asymptomatic seropositive individuals. The frequencies of G1m3 and G1m17 allotypes in our study group (Supplemental Table I) were similar to those reported in whites (35, 36), and G1m polymorphism alone did not associate significantly with the clinical course of HSV-1 infection (G1m3/3 frequency in asymptomatic and symptomatic, seropositive individuals was 55.88 and 51.27%, respectively, odds ratio [OR] = 0.83, not significant) (Fig. 1A, Supplemental Table I).

We also explored the association of the major L chain allotypes with HSV-1 disease and possible interactions with G1m allotypes. Assessment of the distribution of Km allotypes showed that 17.25% of individuals with HSV-1 Abs carried Km1 (Supplemental Table I), in line with the published frequencies for this allotype in whites (35, 36). Again, neither Km allotypes alone nor their combination with G1m significantly influenced susceptibility to HSV-1 (Supplemental Table I).

Previous analysis of CD16A-158V/F dimorphism in a cohort that included the 226 donors selected for this study showed that homozygosity for the higher-affinity CD16A-158V allele was more common in subjects without clinically manifest HSV-1 reactivation than in symptomatic patients (2). To evaluate whether the apparently protective effect of CD16A-158V (higher affinity for IgG) might be modified by the G1m3 allotype (lower affinity for vFcγR and higher affinity for CD16A), we analyzed the joint effect of both polymorphisms in our study population. Indeed, homozygosis for CD16A-158V and G1m3 associated negatively with a symptomatic course of HSV-1 infection (OR = 0.19 versus all other genotypes, p = 0.003) more strongly than did CD16A-158V/V alone (OR = 0.38, p = 0.017) (Fig. 1A). Furthermore, stratification for IgG1 allotypes revealed that the CD16A-158V/V genotype was protective against clinical HSV-1 reactivations in the presence of a G1m3/3 genotype (OR = 0.19, p = 0.003, Fig. 1B) but not in its absence (OR = 0.97, not significant). Reciprocally, the G1m3/3 genotype was also underrepresented among symptomatically infected individuals when its distribution was analyzed in the subgroup of 28 individuals carrying a CD16A-158V/V genotype; however, the small sample size of CD16A-158V/V donors precluded a statistically valid conclusion (OR = 0.22, not significant, Fig. 1C); in contrast, G1m3/3 was equally frequent among symptomatic and asymptomatic donors with other CD16A-158 genotypes (OR = 1.04, not significant).

Our previously published study showed that, in contrast with CD16A-158V/F, CD32A-131H/R dimorphism did not modify HSV-1 susceptibility (2). In line with that result, the combined presence of G1m3/3 (lower vFcγR affinity) and CD32A-131H/H (higher IgG2 and IgG3 affinity) genotypes had no appreciable influence on the risk for clinical HSV-1 recurrence (Supplemental Table I). Similarly, stratification for Km allotypes revealed no new, unsuspected interactions with the genes encoding CD16A and CD32A (Supplemental Table I).

In contrast with CD16A, which is expressed on most NK cells of healthy humans, only a minority of individuals express the low-affinity FcγRs CD32B and/or CD32C on NK cells, which is controlled by genetic polymorphisms. Therefore, these are potential candidates to influence immunity to HSV-1. To test this hypothesis, we used MLPA to genotype the FCGR cluster. The frequencies of CD32C^ORF (determined by the absence of two point mutations in the FCGR2C gene, c169C>T and c798+1A>G, at the donor splice site in intron 7) and the FCGR ∼ 70-kbp deletion conditioning CD32B ectopic expression on NK cells (CD32B^NK hereafter) are shown in Supplemental Table I. FCGR2C genotype frequencies approached those observed previously in whites (21, 37), whereas, to our knowledge, the distribution of CD32B^NK-related genotypes had never been reported. Neither CD32C^ORF nor CD32B^NK showed any appreciable association with herpetic susceptibility (OR =1.11 and 0.79, respectively; not significant) (Fig. 2). In agreement with previously published data (37), we found strong linkage disequilibrium between CD32C^ORF and CD16A-158V (D = 0.025, D′ = 0.61, p < 0.001). To rule out that association of the latter dimorphism with HSV-1 protection might reflect one of a CD16A-CD32C haplotype, we analyzed CD32C^ORF and CD16A-158V/V frequencies, each in the presence and absence of the other; again, we found no new significant association with the risk for HSV-1 recurrence (data not shown).

We showed previously that a series of polymorphic key regulators of the cytotoxic lymphocyte response modifies the clinical course of HSV-1 infection: the CD16A-158V/V genotype, the NK cell receptor–ligand pair KIR2DL2:HLA-C1, and certain HLA class I allotypes (2). Moreover, a multiple logistic regression model performed in that study supported an independent and cumulative effect of those genetic markers on the clinical course of HSV-1 infection (2). To confirm the association found in univariate tests in this study, we performed a new logistic regression analysis. Its results validated the observed protective effect of the joint presence of CD16A-158V/V and G1m3/3 allotypes (rather than CD16A-158V/V alone) on the risk for symptomatic HSV-1 reactivation, as well as its independence from the other variables previously found to influence susceptibility to HSV-1 (Table I).

The protection from clinical HSV-1 reactivation apparently conferred by the joint presence of the CD16A-158V/V and G1m3/3 genotypes could be determined by enhanced NK cell–mediated ADCC against HSV-1–infected cells in those subjects carrying this combined genotype; however, to our knowledge, such a functional effect of IGHG1 and FCGR polymorphisms has never been demonstrated in infection. To test this hypothesis, we evaluated degranulation of NK cells from individuals with a CD16A-158V/V or CD16A-158F/F genotype after in vitro stimulation with HSV-1–infected fibroblasts and HSV-1–immune sera with different IgG1 allotypes. In all experiments, irrespective of CD16A and IgG1 allotypes, NK cell degranulation in the presence of sera with HSV-1–specific IgG largely exceeded the results obtained with sera lacking those Abs, representing natural cytotoxicity levels against infected targets.

In line with the observed association of the CD16A-158V/V genotype with a better course of HSV-1 infection, significantly more NK cells from CD16A-158V/V donors degranulated against opsonized HSV-1⁺ cells compared with NK cells from CD16A-158F/F individuals cultured in similar conditions (p < 0.001) (Fig. 3A). Of note, this was the case in the presence of G1m3/3 and G1m17/17 sera (p < 0.001), despite CD16-158V/V association with lack of HSV-1 disease only in the presence of the former G1m genotype. Furthermore, although our association study found no significant influence of the G1m genotype on resistance to HSV-1, in all cases G1m3/3 sera triggered significantly more NK cell degranulation in ADCC than did G1m17/17 sera (p < 0.001) (Fig. 3B). Again, stratification for CD16A revealed that such an effect of the IgG1 genotype was exerted on NK cells expressing either a high- or a low-affinity allele of this FcγR (p < 0.001).

The previous experiments identified CD16A and IgG1 genotypes associated with more potent ADCC and/or a milder course of HSV-1 infection. The interaction of both polymorphisms allows the identification of combined genotypes of theoretical maximal protection (CD16A-158V/V+G1m3/3) and maximal susceptibility (CD16A-158F/F+G1m17/17). ADCC assays consistently revealed a greater NK cell response when PBMCs from individuals carrying the CD16A-158V/V genotype were cultured with G1m3/3 serum compared with NK cells expressing lower-affinity CD16A-158F/F cultured with serum containing G1m17/17 IgG1 (p < 0.001) (Fig. 3C). On average, the presence of a combined susceptibility genotype reduced the percentage of CD107a⁺ NK cells by 65.3% (range: 31.2–100%) relative to the combined protective genotype. Of note, differences between degranulation levels of NK cells with each CD16A-158 genotype were seen consistently with all donor combinations, regardless of whether cells were derived from donors having or lacking a KIR2DL2/HLA-C1 receptor–ligand pair, previously associated with HSV-1 disease (data not shown).

The observed differences in ADCC intensity determined by IgG1 major allotypes might be due to G1m3/3 IgG having a lower binding affinity for vFcγR, a higher affinity for CD16A, or both. To assess the contribution of the decoy vFcγR, we analyzed the effect of IgG1 allotypes on NK cell degranulation in its absence. vFcγR-blocking experiments were unsuccessful, because the capacity of infected cells to bind human IgG was not efficaciously abolished by aggregated rabbit IgG [capable of binding to the HSV-1 FcγR (10)] or two commercially available mAbs against HSV-1 gE (data not shown). As an alternative approach, we used HSV-1 gE⁻ mutant strain sΔUS8 to infect target fibroblasts. These were then exposed to NK cells and immune sera of different genotypes, and the results were compared with those obtained using gE⁺ HSV-1. As seen in Fig. 4, degranulation induced by G1m3/3 and 17/17 sera showed greater variability with the gE⁻ virus than with the gE⁺ virus. Still, sera of the former genotype triggered significantly greater ADCC than did sera of the latter (Fig. 4), suggesting that the effect of G1m polymorphism on ADCC is at least partially independent of vFcγR. Along the same lines, comparison of the fold ratios of CD107a⁺ NK cells induced by G1m3/3 or 17/17 sera in each experiment revealed no significant difference between the results obtained with the gE⁻ (0.9–3.5) and the gE⁺ (1.2–2.9) HSV-1 strains (Fig. 4).

Although our association studies showed that polymorphisms linked to CD32B or CD32C expression on the surface of NK cells had no significant influence on susceptibility to HSV-1, those results do not exclude a possible influence on ADCC potency. Hence, we performed ADCC assays to compare individuals carrying the same CD16A-158 genotype, but different CD32B/CD32C genotypes and expression patterns (CD32B⁻/CD32C⁻, CD32B⁺/CD32C⁻, and CD32B⁻/CD32C⁺, as determined using DNA typing and flow cytometry). These assays revealed that NK cell expression of CD32B or CD32C did not significantly or consistently influence ADCC potency in our in vitro setting of HSV-1 infection (Fig. 2).

Mature CD56^dim NK cells can undergo clonal expansion and further differentiation, processes that are characterized by phenotypic and functional changes and eventually generate memory-like, long-lived NK cells of unique phenotypes (38–42). The best-known example of this is an NKG2C^bright memory-like NK cell subset with a defined phenotype (NKG2A⁻KIR⁺NCR^low) that expands in many, but not all, healthy individuals infected with human CMV (HCMV) (40, 42). Moreover, an overlapping memory-like, long-lived, mature CD57⁺NKG2A⁻NCR^low NK cell subset, observed in HCMV-exposed individuals, frequently downregulates the common signaling adaptor FcRγ (also known as FcεRIγ); those cells were proposed to be highly responsive to HCMV- and HSV-1–infected cells in the presence of specific Abs (43). To address whether and to what extent NK cells with memory-like phenotypes exert ADCC against HSV-1–infected fibroblasts, we measured degranulation of NK cells expressing or lacking particular differentiation markers in that setting. For these experiments, we selected previously immunophenotyped donors displaying measurable proportions of the aforementioned subpopulations. As in previous experiments, comparisons of ADCC activity were made after subtracting natural degranulation against infected targets (i.e., that seen in the presence of serum lacking HSV-1–specific IgG).

Not surprisingly, ADCC was not restricted to a defined NK subset but was exerted by cells with diverse phenotypes. However, and in every individual, higher proportions of NKG2C^brightNKG2A⁻ NK cells compared with NKG2C^dim/− NK cells degranulated in response to HSV-1–infected, opsonized fibroblasts; the same was true for CD57⁺ versus CD57⁻ NK cells (Fig. 5A, both p = 0.031). In contrast, the behavior of the FcRγ⁻ and FcRγ⁺ NK cell subsets varied in different donors (not significant). In line with these results, NKG2C^brightNKG2A⁻CD57⁺FcRγ⁻ memory-like NK cells degranulated consistently more than did NKG2C^dim/−CD57⁺FcRγ⁻ cells, as well as more than did NKG2C^bright cells with other CD57/FcRγ phenotypes (Fig. 5B, p = 0.031). Taken together, our data indicate that ADCC in response to HSV-1–infected fibroblasts opsonized with human immune serum is exerted by NK cells of heterogeneous phenotypes, and it is enhanced in the highly differentiated NKG2C^bright subset that expands in response to HCMV.

Previous studies support the hypothesis that polygenic inheritance controls common variations in the clinical course of HSV-1 infection (6). We reported that a series of polymorphic key regulators at the interface of innate and adaptive immunity (i.e., HLA class I molecules, KIR and CD16A) are associated with this clinical variability, strengthening the essential role of cytotoxic lymphocytes in the immune control of HSV-1 (2). Polymorphisms of other genes, including APOE and CSSG-1, were related to susceptibility to recurrences, but their exact roles in HSV-1 control are unknown (1, 44, 45). In turn, inborn mutations of TLR3-IFN signaling pathway genes predisposing to herpetic encephalitis support a crucial role for TLR3 in the control of HSV-1 primary infection in the CNS (3), but those mutations are apparently not connected with common forms of HSV-1 infection. In this study, we extend our previous findings, assessing the contribution of the interaction between Ig and FcγR genetic polymorphisms to HSV-1 susceptibility.

Genetic variations in both FcγR and IgG H chain are attractive candidates to influence the immune responses against HSV-1, because they potentially modulate ADCC intensity, an important NK cell effector mechanism. We showed previously that a CD16A-158V/V genotype confers protection from symptomatic herpes (2). Furthermore, major G1m allotypes in whites were recently discovered as potential candidates to influence susceptibility to HSV-1, based on their different binding affinity to HSV-1 gE/gI (7, 29), hence, the variable likeliness to be scavenged by the vFcγR decoy (28). The G1m1 allotype was reported as a possible susceptibility factor for HSV-2 infection (46). According to our present results, G1m3/17 allotypes are not associated, by themselves, with the clinical course of HSV-1 infection; however, they interact epistatically with a FCGR3A (CD16A) functional dimorphism, thus conditioning susceptibility to symptomatic HSV-1 recurrences. Specifically, the previously reported protective effect conferred by a CD16A-158V/V genotype (2) was enhanced in, or limited to, G1m3/3 individuals. Conversely, herpetic disease was not associated with Km1/3 allotypes (with or without stratification for CD16A genotype), and we did not detect any interaction between G1m or Km allotypes or between them and CD32A-131H/R dimorphism, revealing a dominant effect of IgG1–CD16A interaction among the analyzed polymorphisms of Igs and their cellular receptors.

Notwithstanding the demonstrated effect of CD16A-158V/F polymorphism on IgG affinity, and vice versa, few studies addressed the influence of those polymorphisms on ADCC potency (47–49). In the current study, we observed a consistently enhanced NK cell degranulation against opsonized, HSV-1–infected targets in CD16A-158V/V carriers compared with CD16A-158F/F carriers. Similarly, G1m3/3 Abs induced stronger NK cell degranulation in this context compared with G1m17/17 Abs, despite our observation that G1m allotypes do not appear to associate on their own with HSV-1 immunity. Hence, the combined genotype CD16A-158V/V+G1m3/3 was associated with the most potent NK cell–mediated ADCC compared with any other combination of CD16A+IgG1 genotypes. Thus, we expanded our knowledge about FcγR–IgG interactions and provide further evidence on the implication of IGHG polymorphism in the NK cell response by demonstrating that CD16A-158V/F and G1m allotypes condition ADCC intensity against HSV-1–infected cells.

The strength of the FcR–Ab interaction was shown to condition NK cell activation and ADCC potency (47, 50). In an infection setting, HSV-1 vFcγR was proposed to hamper ADCC efficacy (8, 51); therefore, a higher-affinity host FcγR could better compete with the vFcγR decoy and compensate for this subversion mechanism. Indeed, the estimated K_D for vFcγR–G1m17,1 binding is similar to the K_D for the CD16A-158F–IgG1 interaction (7, 29, 47, 52). Thus, lower-affinity G1m3-vFcγR, together with higher-affinity G1m3-CD16A-158V/V interactions could suffice to counteract the viral immune-evasion mechanism, explaining all of our functional and clinical observations. However, we observed that Abs of the G1m3 allotype induced stronger NK cell degranulation, even against targets infected with an HSV-1 strain lacking vFcγR. This reveals a direct effect of G1m polymorphism on ADCC intensity, independent of the virus-subversion mechanism, and possibly mediated by the binding affinity of CD16A to different IgG1 allotypes.

Although our in vitro setting might not accurately mirror the overall response to HSV-1 in vivo, murine models showed that human anti–HSV-1 Abs confer significant protection to subsequent HSV-1 infection, irrespective of the vFcγR⁺ (53–56). Likewise, our findings point to a limited capacity of vFcγR to subvert FcγR-driven NK cell activation, and they are consistent with previous reports indicating that G1m polymorphism also affects IgG1-CD16A binding. IgG1 allotype G1m17 differs from G1m3 at position 214 (CH1, arginine for lysine), and it associates almost invariably in whites with two other substitutions (Glu356Asp and Met358Leu, CH3) that determine the G1m1 allotype (35, 36, 57). Site-directed mutagenesis assays demonstrated the importance of IgG1 H chain residue 214 (and to a lesser extent 356 and 358) for HSV-1 FcγR binding (7, 29), but none of these interacts directly with human FcγR (19, 50). Yet, these substitutions might allosterically influence IgG1 avidity for cellular FcγR. Supporting this hypothesis are affinity-binding assays showing higher affinity of G1m3 compared with G1m17,1 for both CD16A-158V and CD16A-158F (29). Likewise, ADCC-competition assays showed that IgG aggregates carrying the G1m17,1 allotype (but not the G1m3 one) bind with lower affinity to CD16A-158F/F NK cells than to CD16A-158V/V NK cells (58). All of those results, like ours, point collectively to a G1m3 allotype directly enhancing ADCC, regardless of the presence or absence of a pathogen FcγR.

The inhibitory receptor CD32B, classically considered to be expressed only on B lymphocytes or myeloid cells, and the activating receptor CD32C, which is generally not expressed, can both be detected on the NK cells of some individuals. Their expression on this lymphocyte subset was reported to modulate ADCC potency in some, but not all, assays (15, 16, 59, 60). In our in vitro setting of HSV-1 infection, significant CD32B and CD32C contributions to NK cell–mediated ADCC were undetectable. Among FcγRs, CD32B and CD32C bind to IgG1 with the lowest affinity (52, 61), and the CD32C staining pattern in flow cytometry is consistent with low expression on NK cells (Supplemental Fig. 1); therefore, inefficient competition of CD32B/C with CD16A could explain the latter FcγR dominating the overall NK cell–mediated ADCC. In addition, lower-affinity CD32B/C binding to IgG could be competed with more efficiently by vFcγR than that of CD16A. A more precise assessment of the CD32C⁺/CD32B⁺ NK cell role in immunity falls outside the scope of this work; therefore, we cannot exclude a marginal contribution of these receptors to Ab-dependent cytotoxic or noncytotoxic NK cell responses to HSV-1. Nevertheless, our association study showing that genetic variations controlling such expression do not significantly influence the clinical course of the infection point against a major role for NK cell expression of CD32B and CD32C in the immune response to HSV-1.

A recent study showed increased Ab-driven IFN-γ production by FcRγ⁻ NK cells in response to both HSV-1– and HCMV-infected cells (43). Likewise, enhanced degranulation and IFN-γ production in response to HIV⁺ ADCC targets were observed for differentiated CD57⁺/KIR3DL1⁺ NK cells (62, 63). Furthermore, a major role in cytokine production and specific degranulation triggered by opsonized HCMV-infected targets was established for NKG2C^bright NK cells that expand in response to this herpesvirus, many of which express CD57 and downregulate FcRγ as memory-like differentiation markers (64, 65). Of note, expansion of NKG2C^bright cells is specifically triggered, through unknown mechanisms, by HCMV but not by HSV-1 or other human viruses (40, 41). In our in vitro setting of NK cell–mediated ADCC against HSV-1–infected targets, phenotype analysis showed that HSV-1 can readily trigger degranulation in NK cells of different phenotypes and maturation stages. Nonetheless, an enhanced capacity to degranulate in response to opsonized, HSV-1–infected fibroblasts was seen among NKG2C^brightNKG2A⁻ and CD57⁺ NK cells and, in particular, among those with the combined NKG2C^brightNKG2A⁻FcRγ⁻CD57⁺ phenotype. These data, in line with those obtained with HCMV, identify a phenotypic variable that seems to modify efficacy of the NK cell response. Because these phenotypic traits are determined, in part, by the environment (i.e., HCMV infection), and current genetic tests cannot predict them, further specific studies are warranted to determine whether, how, and to what extent an expanded NKG2C^brightNKG2A⁻FcRγ⁻CD57⁺ NK subset might modify immunity to HSV-1.

In summary, our results show that both CD16A-158V and G1m3 allotypes enhance NK cell–mediated cytotoxicity against HSV-1–infected opsonized targets and that those polymorphisms interact epistatically, protecting from clinical HSV-1 recurrence. These data identify NK cell–mediated ADCC as an important antiviral mechanism controlling HSV-1 reactivation and combined FCGR3A (CD16A) and IGHG1 (IgG1) genetic variability as a key contributor to susceptibility to infection.

We thank Drs. S. Efstathiou and H. Browne for the gifts of viruses and cell lines, Dr. Elvira Ramil and Dr. Aránzazu García Grande (DNA sequencing and flow cytometry facilities, Instituto de Investigación Sanitaria Puerta de Hierro) for expert support, and all of the individuals who kindly donated blood samples.

The authors have no financial conflicts of interest.