Immune exhaustion is an important feature of chronic infections, such as HIV, and a barrier to effective immunity against cancer. This dysfunction is in part controlled by inhibitory immune checkpoints. Blockade of the PD-1 or IL-10 pathways can reinvigorate HIV-specific CD4 T cell function in vitro, as measured by cytokine secretion and proliferative responses upon Ag stimulation. However, whether this restoration of HIV-specific CD4 T cells can improve help to other cell subsets impaired in HIV infection remains to be determined. In this study, we examine a cohort of chronically infected subjects prior to initiation of antiretroviral therapy (ART) and individuals with suppressed viral load on ART. We show that IFN-γ induction in NK cells upon PBMC stimulation by HIV Ag varies inversely with viremia and depends on HIV-specific CD4 T cell help. We demonstrate in both untreated and ART-suppressed individuals that dual PD-1 and IL-10 blockade enhances cytokine secretion of NK cells via restored HIV-specific CD4 T cell function, that soluble factors contribute to these immunotherapeutic effects, and that they depend on IL-2 and IL-12 signaling. Importantly, we show that inhibition of the PD-1 and IL-10 pathways also increases NK degranulation and killing of target cells. This study demonstrates a previously underappreciated relationship between CD4 T cell impairment and NK cell exhaustion in HIV infection, provides a proof of principle that reversal of adaptive immunity exhaustion can improve the innate immune response, and suggests that immune checkpoint modulation that improves CD4/NK cell cooperation can be used as adjuvant therapy in HIV infection.
The progressive loss of immune functions occurring during chronic Ag exposure, termed exhaustion, is an important feature of chronic infections such as HIV (1) and is also a barrier to effective immunity in cancer, along with tolerance (2). A key advance in the field has been the demonstration that T cell exhaustion is in part under the active control of inhibitory immune checkpoints, such as PD-1, whose selective blockade can reinvigorate Ag-specific CD8 T cell function, including in HIV infection (3–5). The relevance of this reversibility is illustrated by the recent dramatic progresses made in clinical oncology by targeting coinhibitory receptors such as PD-1 (6, 7). These strategies hold promise for treating HIV infection as well, in which modulation of immune checkpoints may be relevant to enhance immune function, as HIV-specific immunity is not restored by antiretroviral therapy (ART) alone (8, 9), and to reactivate latent reservoirs for targeting (10, 11). Importantly, exhaustion mechanisms are not restricted to CD8 T cells, but also affect other cell subsets that are critical for control of chronic infection, such as CD4 T cells, B cells, and NK cells. Virus-specific CD4 T cell dysregulation results from the combination of an exhaustion program and skewing in Th cell lineage differentiation that impacts function (12–14), presenting both similarities and differences with their CD8 counterparts. Although PD-1 and IL-10 mediate HIV-specific CD4 T cell exhaustion (15–17), consistent with their impact on CD8 T cell responses, some coinhibitory receptors are differentially expressed between subsets, with a preferential upregulation of CTLA-4 on CD4 T cells (18, 19). Combined blockade of PD-1 and IL-10 has an additive effect on HIV-specific CD4 T cell function compared with blockade of either pathway alone (20). A positive feedback loop between IFN-γ produced by HIV-specific CD4 T cells and IL-12 produced by APCs contributes to this additive effect. As Th cells act mostly by modulating function of other immune cell types, it is important to define whether reversion of HIV-specific CD4 T cell exhaustion can improve the efficacy of effector arms of the immune system.
NK cells are another important component of the immune response to HIV (21) and other viral infections, providing a link between the adaptive and innate immune system. NK cells can directly kill infected cells (22, 23), provide pressure on viral evolution (24), and mediate Ab-dependent cellular cytotoxicity against infected cells (25). Studies have shown that NK cells can develop an exhausted phenotype akin to T cell exhaustion in cancer and chronic viral diseases, including in HIV and SIV infections (26). Exhausted NK cells can present a range of defects (27), including lower proliferative capacity, decreased expression of activating receptors (28–31), and increased expression of coinhibitory receptors (30), as well as lost ability to degranulate, secrete cytokines, and promote Ab-dependent cellular cytotoxicity (32).
Several studies demonstrate an important cross-talk between CD4 T cells and NK cells. Murine models have shown that CD4 T cell help contributes to optimal NK cell function (33). In SIV infection of nonhuman primates, IL-2–secreting virus-specific CD4 T cells activate NK cells, and this cooperation is impaired in progressive infection (34). In humans, CD4 T cell help enhances responsiveness of NK cells to influenza and to CMV-infected cells in vitro (35, 36), and results in Plasmodium falciparum malaria suggest a similar link in vivo (37). Notably, restoration of virus-specific CD4 T cell help by therapeutic vaccine candidates improved NK cell responses in both HIV-uninfected human donors (38) and SIV-infected rhesus macaques (39).
However, whether immune checkpoint blockade in HIV infection can lead to NK cell functional restoration via reinvigorated CD4 T cell help remains to be determined. In this study, we show that IFN-γ expression by NK cells upon PBMC stimulation by HIV Ag varies inversely with viral load and depends on HIV-specific CD4 T cell help. We demonstrate both in untreated and ART-suppressed individuals that PD-1 and IL-10 blockade enhances cytokine secretion, degranulation, and killing capacity of NK cells via restored HIV-specific CD4 T cell function and that soluble factors contribute to these immunotherapeutic effects, which depend on IL-2 and IL-12 signaling. This study demonstrates a previously underappreciated relationship between CD4 T cell impairment and NK cell exhaustion in HIV infection, provides a proof of principle in vitro that reversal of adaptive immunity exhaustion can improve an important arm of the innate immune response, and suggests that immune checkpoint modulation that improves CD4/NK cell cooperation can be used as adjuvant therapy in HIV infection.
Materials and Methods
Peripheral blood was obtained from HIV-infected individuals at the Massachusetts General Hospital in Boston, the Centre Hospitalier de l’Université de Montréal, and the McGill University Health Centre in Montreal. The study was approved by the respective institutional review boards and written informed consent was obtained from all study participants prior to enrollment in the study. All participants were adults (18 y old or older). All clinical investigations were conducted according to the Declaration of Helsinki principles. PBMCs from chronically HIV-infected individuals with a broad range of viral loads prior to initiation of ART and individuals treated for 0.6–28 y with undetectable levels of viral RNA (˂50 copies/ml) were isolated from blood samples by Ficoll density centrifugation. Freshly isolated PBMCs were cultured in RPMI 1640 containing 10% heat-inactivated FBS (Sigma-Aldrich) supplemented with 50 IU penicillin, 50 μg/ml streptomycin, 2 mM l-glutamine, and 10 mM HEPES (Mediatech) (R10 medium).
Phenotypic analysis of cytokine secretion
To investigate the impact of combined blockade on cytokine secretion, CD8 T cell–depleted PBMCs (RosetteSep CD8 depletion reagent; STEMCELL Technologies) were incubated at 37°C in 5% CO2 for 48 h with an HIV-1 Gag peptide pool (66 overlapping peptides spanning the Clade B consensus sequence; 14–18 aa long and overlapping by 11 aa; 1 μg/ml/peptide) or left unstimulated in the presence of blocking Abs against PD-L1 (clone 29E.2A3 [10 μg/ml]) and IL-10Rα (clone 37607/MAB274 [10 μg/ml]; R&D Systems) or the corresponding isotype control Abs (IgG2b [10 μg/ml] plus IgG1 [10 μg/ml]). For selected control experiments, total T cells were depleted (RosetteSep CD3 depletion reagents; STEMCELL Technologies; or Dynabeads CD8 positive isolation kit; Invitrogen). For all samples, brefeldin A (5 μg/ml; Sigma-Aldrich), GolgiStop (containing monensin) (0.3 μl/ml; BD Biosciences, La Jolla, CA), and anti-CD107α (clone H4A3, PE-Cy5; BD Biosciences, or BV786; BD Biosciences) were added for the last 12 h of stimulation. After 48 h, cells were stained with viability dye (LIVE/DEAD fixable dead cell dye; Invitrogen/Thermo Fisher Scientific) for 20 min at room temperature and subsequently stained for fluorescent Abs against CD3 (clone SK7, APC-Cy7; BD Biosciences, or PerCP-eFluor710; eBioscience), CD4 (clone RPA-T4, V450 or BV605; BD Biosciences), CD8 (clone 3B5, Qdot 605; Invitrogen/Thermo Fisher Scientific, or clone RPA-T8, V500; BD Biosciences), CD19 (clone HIB19, V500; BD Biosciences, or APCeFluor780; eBioscience), CD14 (clone M5E2, V500 or BUV737; BD Biosciences), and CD56 (clone NCAM16.2, APC or BV421; BD Biosciences). Intracellular cytokine staining (ICS) for IFN-γ (clone B27, PE-Cy7; BD Biosciences), TNF-α (clone MAb11, Alexa 700 or APC; BD Biosciences), and IL-2 (clone 5344.111, FITC, or clone MQ1-17H12, AF488; BD Biosciences) was performed using BD Cytofix/Cytoperm Fixation/Permeabilization solution according to the manufacturer’s instructions. Cells were acquired on a BD LSRFortessa (BD Biosciences).
To evaluate IL-10 and PD-L1 expression, CD8-depleted PBMCs were stimulated with an HIV Gag peptide pool or left unstimulated. For all samples, brefeldin A (5 μg/ml; BD Biosciences) was added for the last 12 h of stimulation. After 18 h, cells were stained with viability dye (LIVE/DEAD fixable dead cell dye; Invitrogen/Thermo Fisher Scientific) for 20 min at room temperature and subsequently stained for fluorescent Abs against CD3 (clone UCHT1 APC; BD Biosciences), CD4 (clone RPA-T4 BV605; BD Biosciences), CD8 (clone RPA-T8 BUV395; BD Biosciences), CD19 (clone H1B19 APCeFluor780; eBioscience), CD14 (clone 61D3 PerCPCy5.5; BD Biosciences), and CD56 (clone SK1 BUV737; BD Biosciences). ICS for IL-10 (clone JES3-19F1 PE; BD Biosciences) was performed using BD Cytofix/Cytoperm Fixation/Permeabilization solution according to the manufacturer’s instructions. Cells were acquired on a BD LSRFortessa (BD Biosciences).
Analysis of NK cell function after HIV-peptide stimulation
CD8-depleted PBMCs were stimulated with an HIV-1 Gag peptide pool or left unstimulated in the presence of isotype control or PD-L1/IL-10Rα blocking Abs as described above. To investigate mechanisms of action of combined blockade on NK cells, we also used anti–IL-12 (clone no. 24910, MAB219; R&D Systems) or anti–IL-2Rα (clone no. 22722, MAB223; R&D Systems) neutralizing Abs or their respective isotype controls. For transwell experiments, we used 24-well plates with 0.4-μm polycarbonate membrane inserts (E&K Scientific). Two million five hundred thousand CD8-depleted PBMCs were stimulated for 48 h in the bottom well in the presence of isotype control or PD-L1/IL-10Rα blocking Abs. Five hundred thousand purified NK cells (STEMCELL Technologies kit) were placed in the top chamber, corresponding to a 1:5 cell number ratio between the two compartments of the transwells. Brefeldin A, monensin, and anti-CD107α/LAMP1 (clone H4A3 BV786; BD Biosciences) were added for the last 12 h of incubation. Subsequent staining was performed using viability dye (LIVE/DEAD fixable dead cell dye; Invitrogen/Thermo Fisher Scientific) for 20 min at room temperature and fluorescent Abs against CD3 (clone SK7 PerCP-eFluor710; eBioscience), CD4 (clone RPA-T4 BV605; BD Biosciences), CD8 (clone RPA-T8 V500; BD Biosciences), CD19 (clone HIB19 APC-eFluor780; eBioscience), CD14 (clone M5E2 BUV737; BD Biosciences), and CD56 (clone NCAM16.2 BV421; BD Biosciences). ICS for IFN-γ (clone B27 PE-Cy7; BD Biosciences), TNF-α (clone MAb11 APC; BD Biosciences), and IL-2 (clone MQ1-17H12 AF488; BD Biosciences) was performed using BD Cytofix/Cytoperm Fixation/Permeabilization solution according to the manufacturer’s instructions. Cells were acquired on a BD LSRFortessa (BD Biosciences).
NK cell killing assay
CD8-depleted PBMCs were stimulated overnight (16 h) with an HIV-1 Gag peptide pool (1 μg/ml/peptide) or staphylococcal enterotoxin B (SEB) (1 μg/ml) or were left unstimulated in the presence of isotype control or combined anti–PD-L1/anti–IL-10Rα Abs as described above. After 16 h, equal numbers of CFSE-stained RAJI cells (American Type Culture Collection) and CellTrace Violet (Thermo Fisher Scientific)–stained K562 target cells (American Type Culture Collection) were added to the CD8-depleted PBMCs at a total E:T ratio of 10:1 and incubated for an additional 16 h. Cells were collected and stained with LIVE/DEAD viability dye (Thermo Fisher Scientific) and fluorescent Abs against CD3, CD4, CD8, CD56, CD14, and CD19 (BD Biosciences/Pharmingen) before fixation and acquisition by flow cytometry as described above. Lysis of K562 cells was quantified by the frequency of LIVE/DEAD+ cells among CellTrace Violet+ cells. NK-resistant RAJI cells were included as controls to determine the frequency of background, spontaneous cell death. Specific lysis was calculated as (percent sample lysis − percent background lysis) / (100 − percent background lysis).
Flow cytometry data were analyzed with FlowJo version 10 (Tree Star). Statistical analyses were performed using Prism 6 (GraphPad Software). Pairwise comparisons for cytokine secretion were performed using the Wilcoxon matched-pair test. We used a Friedman test with Dunn posttest for comparison of more than three groups with paired data. Correlation coefficients were calculated using the Spearman test.
Stimulation of PBMCs by HIV Gag induces a CD4 T cell–dependent production of cytokines by NK cells
To define the impact of CD4 T cell help on NK cell function in HIV-infected subjects, we first measured the modulation of NK cells by virus-specific CD4 T cells stimulated by an HIV Gag peptide pool (Fig. 1) in PBMCs of 21 chronically infected individuals prior to initiation of ART (see Supplemental Table I for characteristics of study participants). We reasoned that a delayed ICS assay, in which cell subsets would be allowed to freely interact and secrete cytokines in the extracellular medium for several hours before addition of protein transport inhibitors, would be suitable to measure a secondary activation of NK cells by autologous CD4 T cells. We thus performed kinetic experiments, in which PBMCs were stimulated for 12 or 36 h before addition of brefeldin A for 12 h prior to ICS. We depleted CD8 T cells prior to the assays to avoid a confounding impact of this subset. We observed that although IFN-γ production by HIV-specific CD4 T cells was clearly detectable at all time points, IFN-γ expression by NK cells was only apparent at the later 48-h time point (Fig. 1A), which we thus selected for subsequent experiments. Stimulation of CD8-depleted PBMCs with HIV Gag peptides resulted in a significant increase in the fraction of IFN-γ–producing NK cells compared with the unstimulated condition (Fig. 1B, median fold increase: 2.0; p < 0.0001). As previous studies have shown that progressive HIV disease is associated with NK dysfunction (22), we next examined the association of the inducible IFN-γ secretion by NK cells with markers of disease progression. We observed a strong inverse correlation between the frequency of IFN-γ+ NK cells and viral load (Fig. 1C; r = −0.59, p = 0.005), whereas no significant correlation was found with CD4 T cell count (data not shown). Notably, we found a direct correlation between frequency of IFN-γ+ Gag–specific CD4 T cells and net increase in the IFN-γ+ NK cells in the stimulated CD8-depleted PBMCs (Fig. 1D; r = 0.54, p = 0.01), suggesting a functional link between these responses. To confirm this hypothesis, we compared a subgroup of seven subjects for IFN-γ secretion by NK cells in PBMCs depleted of CD8 T cells alone versus PBMCs depleted of both CD4 and CD8 T cells by CD3+ cell removal (Fig. 1E, 1F). CD4 T cell depletion in addition to CD8 T cell removal resulted in abrogation of the NK response measured after HIV Ag addition, contrasting with the enhancement observed after CD8 T cell depletion alone. Thus, these data show that HIV Ags can stimulate NK cells via HIV-specific CD4 T cell–dependent mechanisms and that this response decreases with higher viral loads.
Combined blockade of the PD-1 and IL-10 pathways leads to a CD4 T cell–dependent enhancement of NK cell function
We previously showed that blockade of either the PD-1 or the IL-10 pathways reinvigorated HIV-specific CD4 T cell function in vitro and that combined blockade of these immune checkpoints resulted in additive effects compared with inhibition of a single pathway (20). We thus investigated whether combined blockade with anti–PD-L1 and anti–IL-10Rα Abs could improve NK cell function via augmented CD4 T cell help. We first examined PD-L1 and IL-10 expression by four major PBMC subsets: monocytes, CD4 T cells, NK cells, and B cells (Fig. 2A–C). In line with previous findings (16, 17), monocytes expressed the highest levels of PD-L1 (Fig. 2A) and/or IL-10 (Fig. 2B), whether or not the cells were stimulated with the Gag peptide pool. IL-10 was frequently coexpressed with PD-L1 by monocytes (Fig. 2C). We verified that the delayed ICS assay described above gave results consistent with our previous data using Luminex bead arrays on cell culture supernatants (40), confirming the additive effect of dual blockade on IFN-γ secretion by Gag-specific CD4 T cells (Fig. 2D). We next compared the respective impact of blockade on NK and CD4 T cell function. Compared with isotype control Ab conditions, we observed a significant increase of several NK functions upon simultaneous addition of PD-L1/IL-10Rα Abs and Gag Ag, including IFN-γ production (Fig. 2E, 2G; p < 0.0001), degranulation as measured by cell-surface CD107a coexpression with intracellular IFN-γ (Fig. 2H; p = 0.013), and TNF-α expression (Fig. 2I; p = 0002). IFN-γ expression by Gag-specific CD4 T cells increased concurrently (Fig. 2F, 2J; p < 0.0001). However, at this later 48-h time point after Ag stimulation, production of other cytokines, such as IL-2, was not measurable by ICS in most subjects investigated (data not shown). We noted a direct significant correlation between the gain in CD4 and NK function upon combined PD-L1 and IL-10 blockade, as measured by the fold increase in immune function observed by combined blockade compared with isotypic control Abs (Fig. 2K; r = 47; p = 0.03). Thus, blockade of inhibitory immune checkpoints that restore HIV-specific CD4 T cell function also improves NK cell activity in chronically infected individuals in vitro.
Having demonstrated the effect of combined PD-L1/IL-10R blockade in individuals prior to initiation of therapy, we next sought to define whether NK cell activity could also be restored by this in vitro intervention in subjects with suppressed viremia on ART. Compared with isotype control Ab conditions, IFN-γ (Fig. 3A), IFN-γ/CD107 (Fig. 3B), and TNF-α (Fig. 3C) production was increased in NK cells upon blockade of the PD-1 and IL-10 pathways. These results paralleled the increased function observed in autologous CD4 T cells upon dual blockade (Fig. 3D). The fold increase in the expression of these cytokines upon anti–PD-L1 and anti–IL-10Rα blockade was similar between viremic and ART-suppressed individuals (Fig. 3E). Therefore, blockade of inhibitory immune checkpoints can also reinvigorate NK function via improved CD4 T cell help in aviremic, ART-treated individuals.
NK cell activation after HIV Ag stimulation is dependent on IL-2 and IL-12 secretion
Having shown that NK cell activation can be improved by CD4 T cell help by interruption of inhibitory pathways, we next sought to determine immune mediators that were involved in this cross-talk. We hypothesized that cytokines secreted earlier upon addition of Gag Ag to the CD8-depleted PBMCs led to the secondary activation of NK cells. Although the data from Figs. 1 and 2 show correlations between IFN-γ HIV-specific CD4 T cell responses and NK cell activity, we reasoned that this cytokine was likely a correlate of robust Th cell responses and that the cytokines IL-2 and IL-12 were strong candidates for the improvement seen in NK stimulation. We previously demonstrated the expression of IL-2 and IL-12 in our system by CD4 T cells (16) and monocytes (20), respectively. We thus determined the impact of IL-2 or IL-12 blockade on the augmented NK responses elicited by combined anti–PD-L1 and anti–IL-10Rα Abs, again using a delayed ICS assay (Fig. 4). Compared with the isotype control condition, addition of either anti–IL-12 or anti–IL-2Rα lead to significant decrease in the percentage of NK cells producing IFN-γ (Fig. 4A, 4B; p < 0.001 and p < 0.05, respectively) and coexpressing IFN-γ and CD107a (Fig. 4A, 4C, p < 0.05 and p < 0.01 respectively), whereas the inhibitory effect was also observed but somewhat less pronounced for TNF-α (Fig. 4D, p > 0.05 and p < 0.05, respectively). As in Fig. 2, we could not reliably quantitate the induction of single CD107a+ cells because of the high level of pre-expression of this molecule by NK cells. These results show that blockade of either the IL-2 or IL-12 pathways, which are important positive regulators of NK activity, can partly inhibit the robust NK responses induced by the HIV Gag–specific CD4 T cells reinvigorated by combined PD-1/IL-10 blockade.
To determine whether soluble mediators are the principal contributors of this NK-CD4 T cell cross-talk, we conducted a transwell experiment with samples from two untreated, chronically infected individuals and two ART-suppressed donors who were good responders to PD-L1 and IL-10 blockade as defined in Figs. 2 and 3. CD8-depleted PBMCs were placed in the bottom chamber and purified NK cells in the top chamber of the transwell (Fig. 4E). Upon Gag stimulation and PD-L1/IL-10 blockade, we observed a clear increase in IFN-γ expression both in the bottom (CD4 T and NK cells) and top (NK cells) compartments (Fig. 4F). These results show that secreted factors able to diffuse through the small-pore membrane efficiently stimulated the purified NK cells. Importantly, the expression of IFN-γ by the NK cells of the top chamber was close to that observed in the NK cells from the CD4 T cell–containing bottom chamber (Fig. 4G). We therefore conclude that NK-CD4 T cells cross-talk upon PD-L1 and IL-10 blockade depends mainly on soluble mediators.
Combined blockade of the PD-1 and IL-10 pathways restores NK cell killing
A critical goal of immune checkpoint blockade in infectious diseases and cancer is to improve target cell killing. Therefore, we investigated whether the NK cells stimulated by HIV-specific CD4 T cell help restored by dual PD-L1/IL-10R blockade had superior killing capacity compared with NK cells stimulated by the unmanipulated exhausted CD4 T cells (Fig. 5). Killing assays with primary clinical samples are technically challenging and tend to show greater variability than the ICS used in the experiments described above. To overcome this issue, we adapted to our purpose a recently described method of NK killing measurement, which uses internal controls to minimize interassay variability (32) (see experimental design, Fig. 5A). CD8-depleted PBMCs, which contain the NK effector cells and the CD4 Th cells, are first stimulated with the HIV Gag peptide pool or controls, in the presence or absence or PD-1 and IL-10 blockade. After incubation, the primary cells are mixed with fluorescently labeled NK killing–sensitive (K562, devoid of MHC expression) and NK killing–resistant (RAJI) cell lines. Differential lysis rate of these two lines, as measured by a LIVE/DEAD cell dye assay (Fig. 5B), allows calculation of specific NK lysis under the different conditions of stimulation by the following formula: (percent sample lysis [K562 cells] − percent background lysis [RAJI cells]) / (100 − percent background lysis [RAJI cells]). We observed that compared with the unstimulated conditions, addition of HIV Gag peptides elicited specific killing, whereas the positive control SEB confirmed the robustness of the assay (Fig. 5C). Importantly, combined PD-L1/IL-10Rα augmented specific killing by NK cells in the HIV Gag–stimulated conditions (Fig. 5D). Therefore, consistent with the impact observed on IFN-γ and degranulation (CD107a), inhibition of the immune checkpoints PD-1 and IL-10 improves the cytotoxic activity of NK cells stimulated by HIV-specific CD4 T cells.
Immune checkpoint blockade can restore T cell immunity in chronic infections and cancer, is currently a major therapeutic strategy in clinical oncology, and is considered a potentially attractive adjuvant therapy for HIV cure strategies. Although the benefit of such an approach remains to be evaluated for individuals on ART, reinvigorating the host’s immune system could maximize the effect of therapeutic vaccination and reservoir clearance (e.g., through “shock and kill” strategies) because both approaches rely on an efficient immune response to eventually succeed. In this study, we show that blockade of two major inhibitory pathways contributing to HIV-specific T cell exhaustion, PD-1 and IL-10, increases NK cell function through reinvigorated HIV-specific CD4 T cell help. These results suggest that reversion of CD4 T cell exhaustion in HIV infection can improve cooperation of adaptive immunity and a major effector arm of the innate response and may thereby facilitate durable viral control.
CD4 T cell responses play a critical role in the development of effective cellular and humoral antiviral immunity (41). Although the role of CD4 T cell help for CD8 T cell and B cell maturation has been extensively studied (42, 43), less is known about the role of CD4 T cell help in modulating NK cell function. Cytokines are important mediators of CD4 T cell help to NK cells: IL-2 secreted by virus-specific CD4 T cells can stimulate NK cells directly and elicit secretion of cytokines such as IL-12 by APCs, which in turn is a potent activator of NK cell function. Indeed, IL-2 is a potent stimulator of IFN-γ production by NK cells (44) and synergizes with IL-12 (45). Our results in HIV infection are consistent with such a model of cooperative interaction between CD4 T cells and NK cells and with previous results obtained with SIV-infected macaques. The ability of NK cells to respond to PBMCs by HIV Ag inversely correlated with viral load, directly correlated with the magnitude of the HIV Gag–specific CD4 T cells and, importantly, was strongly dependent on the presence of CD4 T cells. Although it is likely that some of the NK cells were initially stimulated by direct recognition by Killer Ig-like receptors (KIRs) of HIV peptides bound to HLA molecules (24, 46), a phenomenon that may have been below the sensitivity of our T cell depletion experiments, CD4 T cell help appears clearly to be a limiting factor in the potentiation of NK cell activation.
Restoration of HIV Gag–specific CD4 T cell responses by dual PD-1 and IL-10 blockade resulted in stronger enhancement of IFN-γ NK cell responses compared with interruption of either pathway, consistent with our previous results showing an additive effect of anti–PD-L1 and ani–IL-10Rα on Th cell function (20). Given the multiplicity of conditions, we thus chose the combined intervention in subsequent experiments and demonstrated that immune checkpoint blockade restored both HIV Gag–specific CD4 T cell responses and NK functions, with a direct correlation in the extent of exhaustion reversal by blockade for these two immune subsets. Combined PD-1/IL-10 blockade did not only result in increased IFN-γ expression by NK cells but also increased NK cell degranulation, suggesting improved killing capacity and TNF-α secretion. We observed the enhancement of NK responses during a time frame (48 h) during which no significant cell proliferation occurred (16). Although in vivo data suggest that some immune interventions, such as in PD-1 blockade in SIV infection (47), affect the magnitude of virus-specific CD4 T cell responses less than that of CD8 T cells, our results suggest that improved CD4 T cell help (e.g., improved IL-2 production) may contribute to enhanced NK cell antiviral activity. Of note, dysfunctional NK cells can also upregulate coinhibitory receptors (27). However, only a small fraction (usually <10%) of peripheral blood NK cells express PD-1 in HIV infection (48). Thus, direct impact of PD-L1 blockade on NK cells may only make a minor contribution, if any, to our observations. Our study was mostly focused on viremic subjects before initiation of ART, a group in which we expect to see the highest levels of CD4 T cell and NK cell exhaustion. However, a clinically important group of ART-treated individuals with suppressed viremia responds well to PD-1 and/or IL-10 blockade in vitro. We demonstrated that restored CD4 T cell help through immune checkpoint blockade can also improve NK function in these subjects, which suggests that immune checkpoint modulation that improves CD4/NK cell cooperation could be used as adjuvant therapy in persons receiving potent ART, the current standard of care.
One of the advantages of the delayed ICS technique used in this study is that it allowed us to examine CD4 and NK cell responses in the same sample. This avoided interassay variation that might unduly influence potential correlations. One limitation is that we could not reliably quantify IL-2 production by HIV-specific CD4 T cells at 48 h, given the faster expression kinetics of this cytokine. To address the role of IL-2 in the restoration of NK function by combined PD-1/IL-10 blockade, we thus used an anti–IL-2Rα Ab in similar functional assays and compared its impact to the effect of anti–IL-12 blockade. Both interventions decreased, without fully abrogating, the boosting of NK cell responses induced by combined PD-1/IL-10 blockade. The partial inhibition observed suggests some redundancy in the CD4/NK cross-talk. These major roles played by the cytokines IL-2 and IL-12 are consistent with our transwell experiments, which show that most of the observed effect of CD4 help on NK function is mediated by soluble factors in this experimental system. Although cell-to-cell contact–dependent factors may contribute to this cross-talk, our data suggest that their contribution is limited. Previous studies showed that other NK stimulatory cytokines besides IL-2 and IL-12, such as IL-15 and IL-18, can reverse NK cell dysfunction (49, 50), and they may also play a role in the improved NK cell activation obtained in this study.
Our report also shows that the enhanced NK cell activation elicited by combined PD-1/IL-10 blockade improved the killing capacity of these cells against MHC class I–negative targets. Although this experiment demonstrated improvement of NK-mediated cytotoxicity, further investigation will be needed to determine the relevance of our findings in vivo. A key question that will determine the impact of CD4/NK cross-talk in HIV infection is the level of anatomical colocalization of HIV-specific CD4 Th cells, NK cells, and HIV-infected cells. Although previous studies have shown that NK cells are detected in lymphoid tissues at rather low frequencies in healthy uninfected individuals (44), studies in SIV-infected macaques have shown an accumulation of exhausted NK cells in lymph nodes compared with uninfected animals (51), suggesting that reversing dysfunction of these effector cells by immunotherapy could contribute to viral control in situ. Tissue studies and in vivo investigations are warranted to address these important issues.
This study demonstrates a direct relationship between CD4 T cell impairment and NK cell exhaustion in HIV infection and provides an in vitro proof of principle that reversal of CD4 Th cell exhaustion by PD-1 and IL-10 blockade can boost an important arm of the innate immune response. Our results suggest that improved CD4/NK cell cooperation may contribute to potential antiviral effects of immune checkpoint blockade as adjuvant therapy in HIV infection.
We thank the staff at Massachusetts General Hospital and Centre Hospitalier de l’Université de Montréal and all study participants, Dr. Dominique Gauchat, the Centre de Recherche du Centre Hospitalier de l’Université de Montréal Flow Cytometry Platform, Dr. Olfa Debbeche, and the Centre de Recherche du Centre Hospitalier de l’Université de Montréal BSL3 Platform.
This work was supported by National Institutes of Health Grant R01 HL-092565 (to D.E.K.), National Institute of Allergy and Infectious Diseases Grants UM1AI100663 (Center for HIV/AIDS Vaccine Immunology and Immunogen Discovery) (to D.E.K.; Dennis Burton, principal investigator) and P01AI056299 (to G.J.F.), the Canadian Institutes for Health Research (CIHR) Project Grant 137694 (to D.E.K.), a Canada Foundation for Innovation grant (to D.E.K.), and the Fonds de Recherche du Québec Santé (FRQS) AIDS and Infectious Diseases Network. D.E.K. is supported by an FRQS Senior Research Scholar Award. A.F. is supported by a Canada Research Chair Award. J.R. and M.V. were supported by CIHR fellowships. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
The online version of this article contains supplemental material.
G.J.F. has patents and receives patent royalties on the PD-1 pathway. The other authors have no financial conflicts of interest.