Terminal Effector CD8 T Cells Defined by an IKZF2+IL-7R− Transcriptional Signature Express FcγRIIIA, Expand in HIV Infection, and Mediate Potent HIV-Specific Antibody-Dependent Cellular Cytotoxicity

Key Points Chronic HIV-1 is associated with increased levels of FcγRIIIA+ CD8 T cells. FcγRIIIA+ CD8 T cells display an innate transcriptomic profile akin to NK cells. ADCC is mediated by FcγRIIIA+ CD8 T cells at levels comparable with NK cells.

HIV-1 contribute to the ability of the virus to escape adaptive T cell responses (3,(12)(13)(14). Also, HIV-specific T cells become functionally impaired during chronic infection, additionally limiting their ability to control viral replication (15)(16)(17). Indeed, polyfunctional HIV-specific T cell responses are associated with better disease outcomes compared with those with a narrower functional breadth (18)(19)(20). In chronic HIV-1 infection, the replicating viral quasispecies have, to a large extent, mutated away from the originally transmitted viral sequence under T cell selection pressure, and this probably contributes to the accumulation of late-stage effector CD8 T cells with a skewed maturational phenotype (21,22). Persistent pathogen replication in chronic infections, such as untreated HIV-1 infection, engages T cell-mediated immune responses continuously with sustained antigenic challenge. Interestingly, some chronic infections have been associated with expansion of an unusual subset of CD8 T cells expressing CD16 (23)(24)(25). CD16 is the low-affinity IgG Fc receptor and exists in two isoforms, FcgRIIIA (CD16a) and FcgRIIIB (CD16b). CD16b is expressed exclusively by neutrophils and recognizes IgG-containing immune complexes, whereas CD16a is best characterized for its role in mediating Abdependent cellular cytotoxicity (ADCC) as a function of the innate immune system (26,27, and reviewed in 28). NK cells are able to mediate a strong effector function in response to signaling through CD16-mediated stimulation. Whereas Fc receptors are generally not expressed by T cells, CD16 can sometimes be expressed by subsets of TCRab T cells (29)(30)(31)(32). Growing evidence suggests the potential importance of ADCC in protection from HIV-1 infection (33,34). Additionally, nonneutralizing Abs mediate an array of effector functions through their interactions with Fc receptors that may potentiate protection from HIV-1 infection or inhibit viral replication postinfection (35)(36)(37)(38)(39)(40). Still, a better understanding of effector mechanisms, such as ADCC, involved in HIV-1 control is needed.
In this study, we hypothesized that late-stage differentiation of CD8 T cells may be associated with transcriptional changes that support innate-like effector functions in the T cell compartment. We demonstrate, in this study, that chronic, untreated HIV-1 infection is associated with the expansion of a late-stage differentiated CD8 T cell population expressing FcgRIIIA and that this population mediates HIV-specific ADCC. Furthermore, we show that the FcgRIIIA + CD8 T cells display a hybrid CD8 T cell and NK cell transcriptional profile characterized by high expression of NKp80 and the transcription factor Helios.

Patients and samples
Study participants aged 15-49 y were enrolled in a prospective community-based cohort to assess the prevalence and incidence of HIV-1 infection in Rakai District, Uganda, from 1998 to 2004 (Table I) (41)(42)(43). Infected subjects were identified between 1997 and 2002 with continued annual follow up through 2008. Blood samples from 103 randomly selected HIV-1 seropositive individuals and 40 community-matched seronegative controls were obtained. PBMCs were then isolated and cryopreserved as described previously (44). None of the patients had received antiretroviral therapy (ART). HIV-1 testing was performed as described previously (43). Positive samples were subjected to the Amplicor HIV-1 Monitor test, version 1.5 (Roche Diagnostics, Indianapolis, IN). The HIV-1-infected study participants initiating ART were from the Couples Observation Study (COS) in Kampala Uganda as previously described (45). The index partner in each HIV-1-serodiscordant couple was followed up after the initiation of ART. Samples were collected; CD4 T cell counts determined and viral load assessments made at baseline, 6 and 12 mo after initiation of ART.

Ethics statement
The study was approved by the following institutional review boards in the United States and Uganda: the institutional Review Boards of Uganda's National Council for Science and Technology and the National AIDS Research Committee, as well as Division of Human Subjects Protection at the Walter Reed Army Institute of Research. All participants gave writteninformed consent, or written-informed consent was obtained from the parent or legal guardian of those aged 17. For samples from the COS in Kampala, Uganda, all participants gave written-informed consent, and ethical approvals for the study were obtained from Uganda's National Council for Science and Technology and the National AIDS Research Committee and the University of Washington.

Flow cytometry and mAbs
Cryopreserved specimens were thawed and washed. Counts and viability were assessed on the Guava PCA (Guava Technologies, Hayward, CA), using Guava ViaCount reagent. Standard flow cytometry phenotyping was performed as previously described (46). Commercial mAbs (clone) used in flow cytometry were as follows: . For assessment of transcription factors, cells were washed, permeabilized and fixed using an optimized kit (FOXP3 transcription factor staining buffer set) before intranuclear stain. Flow cytometry data were acquired with a BD LSR II instrument or a BD FACSCanto II instrument (BD Biosciences). Sorting was performed on a four-laser BD FACSAria II SORP (BD Biosciences) contained in a biosafety cabinet. Clinical lymphocyte immunophenotyping was performed using the FACS MultiSET System and run on a FACS-Calibur using the single-platform Multitest four-color reagent in combination with Trucount tubes (BD Biosciences) (47).

Gene expression analysis
Targeted gene expression analysis was performed as previously described (48). Cells from seven donors were stained and four phenotypically distinct cell populations (CD8 T cells: CD45RA-CD57 2 , CD45RA + CD57 + FcgRIIIA 2 , CD45RA + CD57 + FcgRIIIA + , as well as CD56 dim FcgRIIIA + NK cells) (500-1000 cells per well) were sorted into wells containing 10 ml of reaction buffer (SuperScript III Reverse Transcriptase/Platinum Taq Mix, CellsDirect 23Reaction Mix; Invitrogen). Reverse transcription and specific transcript amplification were performed using a thermocycler (GeneAmp PCR System 9700; Applied Biosystems) as follows: 50˚C for 15 min, 95˚C for 2 min, then 95˚C for 15 s, 60˚C for 30 s for 18 cycles. The amplified cDNA was loaded into Biomark 96.96 Dynamic Array chips using the NanoFlex IFC controller (Fluidigm). This microfluidic platform was then used to conduct quantitative PCR in nL reaction volumes. Threshold cycle, as a measurement of relative fluorescence intensity, was extracted from the Biomark real-time PCR analysis software. A panel of 96 preselected genes related to both NK cell and CD8 T cell biology was qualified as previously described, using a script provided courtesy of Mario Roederer (49). Subsequent data analysis was performed using JMP software (version 10). Initial analyses of the transcriptome data from the Fluidigm Biomark confirmed the quality of 74 of the 96 genes, although data on 22 genes were discarded because of lack of amplification.

ADCC assays
Measurement of ADCC was performed using the PanToxiLux assay (OncoImmunin, Gaithersburg, MD) similar to the previously described assay (50). rHIV-1 BaL gp120 (catalog no. 4961; obtained through the National Institutes of Health AIDS Reagent Program, Division of AIDS, National Institute of Allergy and Infectious Diseases, National Institutes of Health) were used to coat target CEM.NKR CCR5 cells. Optimal concentration used to coat target cells was determined for each gp120 through an 11-point titration starting with 20 mg/ml and 2-fold serial dilution. After coating CEM.NKR CCR5 target cells with gp120 in 0.5% FBS-RPMI media, cells were labeled with TFL4 (OncoImmunin), a fluorescent target cell marker, for 15 min at 37˚C and 5% CO 2 . Cells were then washed twice with 13 PBS and stained with the viability dye LIVE/DEAD Fixable Aqua Dead Cell Stain (Life Technologies) for 30 min at room temperature. After washing in 0.5% FBS-RPMI media, cells were counted as above, then resuspended to reach a final concentration of 8.0 3 10 5 cells/ml. At this point, sorted effector cell populations (NK cells, CD45RA + CD57 + CD8 T cells, and CD45RA-CD57 2 CD8 T cells) were washed in 0.5% FBS-RPMI media and resuspended to a final concentration of 24 3 10 6 cells/ml for an E:T ratio of 30:1. In a 96-well polypropylene plate, 25 ml of both target and effector cell suspensions were both added to each well along with 75 ml of granzyme B substrate (OncoImmunin). After incubation for 5 min at room temperature, 25 ml of HIV-Ig (North American Biologicals, Miami, FL) at a 0.5 mg/ml dilution was added to each well, and the plate was incubated for another 15 min at room temperature. The plate was then spun at 300 3 g for 1 min and placed at 37˚C and 5% CO 2 for 1 h. Cells were washed twice with wash buffer and acquired on the LSR II (BD Biosciences) on the same day. Fluorophores were detected using a 488-nm 50-mW laser with 515/20 filters to detect granzyme B substrate, a 406-nm 100-mW laser with 525/50 filters to detect Aqua LIVE/DEAD stain, and a 640-nm 40-mW laser with 670/30 filters to detect TFL4 stain. Because of the spectral properties of the fluorescent molecules used in this panel, manual compensation of detected signals was performed to analyze the data. Data were analyzed by using FlowJo 9.7.5 (Ashland, OR).

Statistical analysis
Statistical analysis was performed using GraphPad Prism 6.0 (Version 6) for Macintosh (GraphPad Software, La Jolla, CA) or JMP software (version 10; SAS Institute, Cary, NC). Direct comparisons between two groups were performed using the nonparametric Mann-Whitney U test. Associations between groups were determined by Spearman rank correlation. To correct for multiple comparisons, the Benjamini-Hochberg false discovery rate (FDR) (51) was calculated for all observations. An FDR ,0.05 was considered statistically significant. For paired observations, a paired t test was used. A p value ,0.05 was considered statistically significant. Flow cytometry analysis and presentation of distributions were performed using SPICE version 5-1.2, downloaded from http://exon.niaid.nih.gov/spice (52). Comparison of distributions was performed using a Student t test and a partial permutation test as described previously (52).

FcgRIIIA + CD8 T cells expand in chronic untreated HIV-1 infection
HIV-1 negative (n = 40) and HIV-1 positive (n = 103) individuals from a cohort in Rakai, Uganda, were chosen for the investigation of FcgRIIIA expression in CD8 T cells (Table I). The FcgRIIIA + CD8 T cell population was identified as positive for CD3, TCRab, CD8, and FcgRIIIA and negative for CD14, CD19, and CD4 ( Fig. 1A, Supplemental Fig. 1). FcgRIIIA expression was detectable in T cells from healthy donors at a median (range) frequency of 3.8% (0.7-20.7%) of CD8 T cells (Fig. 1B). Interestingly, this population was nearly doubled in HIV-1infected donors, in which a median frequency of 5.9% (1.3-37. 9%) of CD8 T cells expressed FcgRIIIA (p , 0.001) (Fig. 1B). This expansion was positively associated with the overall CD8 T cell expansion in HIV-1-infected patients (p , 0.001, rho = 0.546) (Fig. 1C). The HIV-1-associated expansion of FcgRIIIA + CD8 T cells was not associated with the expression levels, measured as geometric mean fluorescence intensity (MFI), of FcgRIIIA on the surface of these cells (data not shown). There was no significant difference in FcgRIIIA expression levels (as measured by MFI) on FcgRIIIA + CD8 T cells between HIV-1-infected and uninfected participants (data not shown). Interestingly, the FcgRIIIA + CD8 T cells were more activated than their FcgRIIIA 2 counterparts, as assessed by CD38 expression (p , 0.001) (Fig. 1D). They also expressed less of the inhibitory receptor PD-1 (p , 0.001) (Fig. 1E). The CD38 expression levels were inversely associated with CD4 counts, albeit weakly (p = 0.02, rho = 20.367), suggesting that the FcgRIIIA + CD8 T cells become more activated as disease progresses (Fig. 1F).
To address the stability of the FcgRIIIA + CD8 T cell pool over time, we studied a second cohort of Ugandan HIV-1-infected subjects (n = 32) located in Kampala, where longitudinal samples were available from before and after initiation of ART (Table I). These patients displayed a stable population of FcgRIIIA + CD8 T cells over 12 mo of ART (Fig. 1G). However, of note, the activation levels of these FcgRIIIA + CD8 T cells declined over the course of treatment, as measured by CD38 expression (p , 0.001) (Fig. 1H). These data show that HIV-1-infected Ugandans have an expanded population of activated TCRab CD8 T cells expressing FcgRIIIA and that this population is stable over 12 mo of ART.
Next, multiplexed assays and ELISA were used to quantify a suite of 20 soluble factors in plasma in relation to the size and activation level of the FcgRIIIA + CD8 T cell population in HIV-1infected individuals. Although none of the analytes measured showed a relationship to the percentage of CD8 T cells expressing FcgRIIIA, several markers were directly associated with the activation levels of FcgRIIIA + CD8 T cells (i.e., cells coexpressing CD38) (Table II). Statistically significant correlations between the frequency of FcgRIIIA + CD8 T cells expressing CD38 and plasma   CD27, and CD45RA was significantly different between CD8 T cells positive or negative for FcgRIIIA ( Fig. 2A, Supplemental Table I) (p , 0.001). Expression of CD45RA in the absence of CCR7 and CD27 was the dominant pattern among the FcgRIIIA + CD8 T cells, consistent with a terminally differentiated status, whereas this phenotype was less common among CD8 T cells lacking FcgRIIIA (74% versus 18%, respectively) (p , 0.001). Next, the expression patterns of CD57, NKG2A, and NKG2D were evaluated, and the frequency of the subsets defined by these receptors were different in CD8 T cells expressing FcgRIIIA compared with those that did not ( To address this question further, we investigated the expression of killer Ig-like receptors (KIRs) in CD8 T cells and NK cells expressing FcgRIIIA, as well as in late-stage differentiated CD8 T cells defined by coexpression of CD45RA and CD57 and memory CD8 T cells negative for these markers (Fig. 2D). In uninfected donors, T cell populations lacking FcgRIIIA had low levels of KIR expression, whereas NK cells had high KIR levels in diverse combinations. The FcgRIIIA + CD8 T cells displayed a pattern intermediate between T cells and NK cells. Strikingly, this pattern was altered in HIV-1-infected subjects whose FcgRIIIA + CD8 T cells had adopted a KIR coexpression profile very similar to that of NK cells (p , 0.001 for FcgRIIIA + CD8 T cells in HIV-1-uninfected donors compared with HIV-1-infected donors; p = 0.250 for FcgRIIIA + CD8 T cells compared with NK cells in HIV-1-infected donors).
T cell differentiation and maturation are controlled by a set of transcription factors, including T-bet, Eomes, and Helios. PBMC from HIV-infected donors were stained intracellularly for these transcription factors, and their expression patterns were analyzed in CD8 T cells lacking or expressing FcgRIIIA, as well as in NK cells (Fig. 3A). FcgRIIIA + CD8 T cells displayed a T-bet, Eomes, and Helios expression pattern distinct from both the general CD8 T cell population and from CD56 dim NK cells, with higher levels of coexpression as compared with FcgRIIIA 2 CD8 T cells. Coexpression of all three transcription factors was common in FcgRIIIA + CD8 T cells and also relatively frequent in NK cells but uncommon in the general CD8 T cell pool. Notably, 61% of the FcgRIIIA + CD8 T cells expressed Helios, and this was significantly higher compared with the FcgRIIIA 2 CD8 T cells and NK cells (p , 0.001), in which a median of 10 and 28% expressed Helios, respectively. Characterization of T-bet and Eomes can be discriminated, based on a continuum of expression and varies on lymphocyte subsets (53). FcgRIIIA + CD8 T cells were dominated by a high T-bet expression profile with variable Eomes expression that was very similar to CD16 + NK cells (Fig. 3B, 3C). HIV-1 infection status had minimal effect on T-bet and Eomes in these populations. FcgRIIIA 2 CD8 T cells showed a much more variable expression pattern of both transcription factors, which may reflect the different states of maturation and differentiation within this compartment.
Altogether, these data indicate that the FcgRIIIA + CD8 T cell population expanded in HIV-1-infected people is characterized by Helios expression and has a late-stage differentiated effector phenotype. This population mostly retains the characteristics seen in healthy donors as it expands during HIV-1 infection, although KIR expression is significantly elevated.

The FcgRIIIA + CD8 T cell transcriptome reveals a mixed effector CD8 T cell and NK cell character
To better understand the identity of the FcgRIIIA + CD8 T cells, we next analyzed their transcriptional profile by Fluidigm Biomark. A panel of 96 genes involved in T cell function or NK cell function was selected (Supplemental Table II  analysis (PCA) was performed on the total data set of expression of these 74 genes in all four of the cell subsets (Fig. 4A).
Notably, the transcriptional profile of FcgRIIIA + CD8 T cells overlapped with both the CD45RA + CD57 + CD8 T cells lacking expression of FcgRIIIA and the CD56 dim CD16 + NK cells, whereas the CD45RA-CD57 2 memory CD8 T cell subset was A subset of genes showed expression patterns that segregated the FcgRIIIA + CD8 T cell population from the NK cells and the FcgRIIIA 2 CD8 T cell populations (Fig. 4B, Supplemental Fig. 2). Notably, the FcgRIIIA + CD8 T cell displayed significantly higher IKZF2 expression than any of the three other reference populations and lower IL-7R expression than the other T cell populations and at levels similar to CD56 dim CD16 + NK cells. Regarding a range of genes encoding NK cell-associated receptors, including KIR2DS2; KIR2DS1, KIR3DL1;KIR3DS1, KLRC2L;KLRC3, KLRD1, KLRF1, KLRK1;KLRC4-1, and NCR1, the FcgRIIIA + CD8 T cells showed a pattern intermediate between FcgRIIIA 2 CD45RA + CD57 + CD8 T cells and the CD56 dim CD16 + NK cells. In fact, KLRF1 encoding the NK cell-associated receptor NKp80, expressed at the highest levels by CD56 dim CD16 + NK cells, was expressed at significantly higher levels when compared with the FcgRIIIA 2 terminal effector CD8 T cells and effector memory CD8 T cells. Compared with their FcgRIIIA 2 counterparts, the FcgRIIIA + CD8 T cells also expressed higher levels of genes involved in regulating T cell function, including TNFSF13B. Additionally, the FcgRIIIA + CD8 T cells had significantly lower expression of TGFBR1 than the CD56 dim NK cells, but levels were above that of the other CD8 T cells populations. Altogether, the gene expression analysis indicates that FcgRIIIA + CD8 T cells have a transcriptional profile intermediate between effector CD8 T cells and CD56 dim NK cells.
Because of the distinct transcriptional signature of FcgRIIIA + CD8 T cells, we were interested in confirming expression of the IL-7R and . Transcriptome analysis reveals a mixed CD8 T cell and NK cell character in the FcgRIIIA + CD8 T cells. Supervised expression analysis of 74 genes involved in the regulation and function of innate and adaptive immune responses in seven HIV-1-infected donors using the Fluidigm Biomark system. (A) PCA of the transcriptional data from four sorted cell populations reflecting CD45RA 2 CD57 2 (blue), CD45RA + CD57 + FcgRIIIA 2 (red), CD45RA + CD57 + FcgRIIIA + (green), as well as CD56 dim FcgRIIIA + NK cells (purple). Polygons represent 95% confidence intervals in the data. (B) Expression of 10 selected genes in the same sorted subsets. (C) Successive flow cytometry gating strategy used for confirmation of IL-7Ra and KLRF1 genes at the protein level. Offset histograms showing the relative expression of IL-7Ra (CD127) and KLRF1 (NKp80) on CD8 T cells: CD45RA 2 CD57 2 (blue), CD45RA + CD57 + FcgRIIIA 2 (red), CD45RA + CD57 + FcgRIIIA + (green/filled), as well as CD56 dim FcgRIIIA + NK cells (purple). KLRF1 genes at the protein level. We further examined 10 chronically HIV-1-infected and 10 uninfected individuals for surface expression of these receptors by flow cytometry. The majority of FcgRIIIA + CD45RA + CD57 + CD8 T cells expressed NKp80 (median 68%) and lacked expression of the IL-7 receptor, CD127 (median 2%) (Fig. 4C). No differences were observed in FcgRIIIA + CD45RA + CD57 + CD8 T cells expressing NKp80 or IL-7Ra between HIV-1 positive and negative individuals, and no relationship was observed between expression and markers of HIV-1 disease progression. IL-7Ra protein expression was similar between NK cells and CD45RA + CD57 + CD8 T cells, irrespective of FcgRIIIA + expression. Interestingly, NKp80 was only found at appreciable levels in the T cells with the FcgRIIIA + CD45RA + CD57 + phenotype. Together, the FcgRIIIA + CD8 T cells have a distinct NKp80 + IL-7Ra 2 character different from other effector CD8 T cells and more akin to CD56 dim NK cells.
Potent HIV-specific ADCC activity mediated by FcgRIIIA + CD8 T cells ADCC is part of the repertoire of effector functions employed by NK cells to detect and target HIV-1-infected cells. Recent data indicating that nonneutralizing Ab-mediated effects may contribute to HIV vaccine efficacy have spurned a renewed interest in ADCC as a protective mechanism (54,55). The present observation that HIV-1 infection drives the expansion of late-stage effector CD8 T cells with a hybrid NK cell-CD8 T cell character, including FcgRIIIA and lytic protein expression, suggests that CD8 T cells might actually mediate ADCC. To test this possibility, effector cell populations from HIV-1-infected donors were sorted by flow cytometry, and the ability of these cells to mediate ADCC against HIV BaL gp120-coated CEM.NKR CCR5 target cells was evaluated by the PanToxiLux granzyme B substrate cytotoxicity assay (Fig. 5A). To avoid FcgRIIIA downregulation or blocking because of staining, CD45RA + CD57 + CD8 T cells were sorted to enrich for FcgRIIIA + cells (9-21% FcgRIIIA + ) and then compared with FcgRIIIA 2 CD45RA 2 CD57 2 memory CD8 T cells and with NK cells sorted from the same donors. In the presence of HIV-Ig, the CD45RA + CD57 + cells from three HIV + donors clearly mediated ADCC, as did the NK cells, whereas the CD45RA 2 CD57 2 CD8 T cell population did not (Fig. 5A, 5B). As such, bulk CD45RA + CD57 + CD8 T cells performed ADCC lower than the NK cells (Fig. 5B). However, after adjusting for the frequency of FcgRIIIA expression in these populations, 9-21% in CD45RA + CD57 + CD8 T cells and 69-96% in CD56 dim NK cells, ADCC capacity of FcgRIIIA + CD8 T cells was similar to that of FcgRIIIA + NK cells (Fig. 5C). Interestingly, the FcgRIIIA + MFI on FcgRIIIA + CD8 T cells was significantly lower compared with FcgRIIIA + MFI on CD56dim NK cells (p , 0.001). FcgRIIIA + CD8 T cell ability to mediate ADCC, based on normalized FcgRIIIA + MFI or the integrated MFI (frequency multiplied by the MFI), was as good as NK cells (data not shown). These data demonstrate that the FcgRIIIA + CD8 T cell population expanding during chronic HIV-1 infection can mediate HIV-specific ADCC at levels comparable to NK cells.

Discussion
CD8 T cells use a range of effector functions to combat viral infections, including cytolysis and effects mediated by cytokines and chemokines. A hallmark of these antiviral functions is that they depend on the exquisite Ag specificity of TCRs and their recognition of viral Ag in an MHC-restricted manner. In this study, we demonstrate that late-stage effector CD8 T cells acquire FcgRIIIA expression in HIV-1-infected individuals and use this Fc receptor to mediate HIV-specific ADCC in the absence of TCR recognition of Ag. Using a commercial in vitro assay, commonly used in assessing HIV-1 ADCC activity (50), we measured the effector capacity, on a per cell basis, of FcgRIIIA + CD8 T cells to mediate Ag-specific ADCC against gp120-coated targets as efficiently as NK cells from the same donors. These findings indicate that in the context of chronic uncontrolled HIV-1 infection, a significant subset of CD8 T cells acquires innate characteristics and performs a function in the immune system normally associated with NK cells. Functional diversification of adaptive CD8 T cells may be important as therapeutic strategies evolve to include Ab-mediated mechanisms to eliminate HIV-1 reservoirs (56)(57)(58).
In the Ugandan population studied in this work, expression of FcgRIIIA occurs on∼5% of CD8 T cells from healthy donors, and this frequency is doubled in patients with chronic untreated HIV-1 infection. In fact, some patients have more than 30% of their CD8 T cells expressing FcgRIIIA. The finding that the size of this population is positively associated with the global CD8 T cell expansion in these patients suggests that the FcgRIIIA + CD8 T cells expand in response to the chronic uncontrolled viral replication. These expanded cell populations have a terminally differentiated phenotype with frequent expression of CD45RA, CD57, and perforin but little expression of CD27 and CCR7, further supporting this notion. The phenotypic profile of these cells is similar between HIV-1-infected patients and healthy donors [data not shown and (23)]. However, we found one exception to this observation; the FcgRIIIA + CD8 T cells adopt a KIR expression profile similar to NK cells in HIV-1-infected subjects, an observation not seen in healthy donors (Fig. 2D). The conditions in vivo during HIV-1 infection thus seem to drive not only an expansion of these cells but also expression of surface receptors beyond FcgRIIIA normally associated with NK cells and reflective of the rise in terminally differentiated CD8 T cells in chronic viral infections (59).
These functional and, to some extent phenotypic, similarities with NK cells led us to ask how the FcgRIIIA + CD8 T cells relate to FcgRIIIA 2 T cell subsets as well as FcgRIIIA + NK cells on the transcriptional level. Based on a supervised transcriptional analysis of 74 genes in seven donors, the FcgRIIIA + CD8 T cells appear to have a transcriptional program intermediate between late-stage effector CD8 T cells lacking FcgRIIIA and CD56 dim NK cell expressing FcgRIIIA. Most interestingly, transcript and protein levels for KLRF1, encoding the activating NKp80 receptor, were expressed at high levels similar to NK cells compared with effector memory or FcgRIIIA 2 CD8 T cells. NKp80 has recently been shown to associate with the development and maturation of fully functional NK cells (60). Whereas FcgRIIIA + CD8 T cells show some features similar to CD56 dim NK cells, PCA revealed that FcgRIIIA + CD8 T cells, FcgRIIIA 2 CD8 T cells, and CD56 dim NK cells were distinct from the CD45RA 2 CD57 2 memory CD8 T cell population. Consistent with this notion, when the genes differentially expressed between the FcgRIIIA + CD8 T cell and the effector memory T cell population were entered into the Reactome pathway analysis database, the DAP12 pathway, implicated in activation of NK cells, was indicated as enriched in the FcgRIIIA + CD8 T cells (Supplemental Fig. 2, Supplemental Table II) (61,62). Furthermore, we observed the upregulation of 10 genes in FcgRIIIA + CD8 T cells compared with the effector memory CD8 T cell population that are associated with NK-like rapid effector function and the "innateness gradient" defined by Gutierrez-Arcelus et al. (63) including GZMB, PRF1, KIR3DL1, KLRK1, KLRD1, KLRF1, NCR1, KLRC2L;KLRC3, KIR2DS2, and ITGAM (Supplemental Fig. 2, Supplemental Table II). Although the overall pattern is that FcgRIIIA + CD8 T cells overlap with both FcgRIIIA 2 T cells and CD56 dim NK cells, these cells also manifest distinctive features somewhere between innate and adaptive immune cells (64). A pattern that stands out is the high expression by FcgRIIIA + CD8 T cells of the transcription factor Helios, encoded by the IKZF2 gene, both at the protein and gene levels. These cells also have very low expression of IL-7Ra. The low IL-7Ra expression level is consistent with a model in which these cells are either maintained by non-IL-7-dependent factors or, rather, short-lived in vivo. Our finding that patients initiating ART largely maintain the expanded FcgRIIIA + CD8 T cell population over 12 months suggests that these cells are not intrinsically short-lived, and thus may even be maintained by IL-7-independent mechanisms. This interpretation is supported by the recent finding of expansion of long-lived effector CD45RA + CD8 T cells that are IL-7R lo KLRG1 high in latent CMV and EBV infection, a population which phenotypically overlaps with the FcgRIIIA + CD8 T cell identified in this study (65).
The expansion of FcgRIIIA + CD8 T cells we observe in this study is reminiscent of the expansion of CD8 T cells with a similar phenotype in hepatitis C virus (HCV)-infected patients (23). Whereas HIV-1 and HCV differ in target cell tropism and mechanisms of pathogenesis, for example, they have in common establishment of chronic infections that are very difficult for the immune system to control. This is partly because of the shared features of rapid viral replication and high mutation rates. These features lead to selection of epitope immune escape variants that allow these viruses to avoid efficient recognition by clonally expanded populations of T cells. Viral quasispecies mutate away from the originally transmitted viral sequence under T cell selection pressure and some of the early responding epitope-specific T cell populations may thus lose their efficiency in targeting infected cells. Future studies are warranted to test the hypothesis that accumulation of FcgRIIIA + CD8 T cells may be a clonally driven process and this could be addressed by TCR repertoire analysis. The FcgRIIIA + CD8 T cells have a phenotype that would be expected from a T cell population expanded by Ag recognition, because they are largely negative for CD27 and CCR7, but positive for CD57, perforin and CD45RA. In the yellow fever virus vaccine model, the yellow fever vaccine-specific CD8 T cells are CD45RO + during the peak of the effector response and then revert back to CD45RA expression as the Ag is cleared and memory is established (66,67). This is consistent with a model in which CD45RA may be re-expressed when the T cells have not seen their cognate epitope for some time. This allows for the possibility that the FcgRIIIA + CD8 T cells that expand numerically after HIV-1 infection as well as in HCV infection may be driven by viral epitopes that later accumulate escape mutations. Interestingly, the expanded FcgRIIIA + CD8 T cell population described in this study displays frequent expression of inhibitory KIRs. Recent findings indicate that inhibitory KIR expression on CD8 T cells may enhance T cell survival in chronic viral infections and may facilitate the rescuing of an activated immunodominant T cell population after chronic Ag exposure (59). This population may then be viewed as a way for the immune system to repurpose Ag experienced T cells in defense against chronic viral infection (65). A recent study by Phaahla et al. (68) confirms the expansion of FcgRIIIA-expressing, ADCC-mediating CD8 T cells in an HIV positive South African cohort. Within the same cohort, FcgRIIIA expression declined on NK cells during HIV infection, which could potentially contribute to the observed decline of their capacity to mediate ADCC. These findings further support a model in which cytotoxic CD8 T cells are repurposed toward this innate-like function in the context of chronic HIV infection.
In summary, we describe a subset of late-stage differentiated CD8 T cells that acquire a distinctive hybrid NK cell and effector CD8 T cell character during untreated chronic HIV-1 infection, with expression of FcgRIIIA and potent HIV-specific ADCC activity. The development of this NK-like functionality in CD8 T cells may represent a way for the immune system to take full advantage of the cytolytic effector program of terminally differentiated cytolytic effector CD8 T cells during chronic viral infections and situations of epitope escape. In addition, the fact that expanded FcgRIIIA + CD8 T cell populations persist after initiation of suppressive ART suggests that they may be engaged and contribute to Ab-based HIV cure strategies.