HIV-Specific CD8⁺ T Cells from Elite Controllers Are Primed for Survival

Abstract

HIV-specific cytotoxic T lymphocytes (CTL) are preferentially primed for apoptosis, and this may represent a viral escape mechanism. We hypothesized that HIV-infected individuals that control virus to undetectable levels without antiretroviral therapy (ART) (elite controllers [EC]) have the capacity to upregulate survival factors that allow them to resist apoptosis. To address this, we performed cross-sectional and longitudinal analysis of proapoptotic (cleaved caspase-3) and antiapoptotic (Bcl-2) markers of cytomegalovirus (CMV) and HIV-specific CD8 T cells in a cohort of HIV-infected subjects with various degrees of viral control on and off ART. We demonstrated that HIV-specific CTL from EC are more resistant to apoptosis than those with pharmacologic control (successfully treated patients [ST]), despite similar in vivo conditions. Longitudinal analysis of chronically infected persons starting ART revealed that the frequency of HIV-specific T cells prone to death decreased, suggesting that this phenotype is partially reversible even though it never achieves the levels present in EC. Elucidating the apoptotic factors contributing to the survival of CTL in EC is paramount to our development of effective HIV-1 vaccines. Furthermore, a better understanding of cellular markers that can be utilized to predict response durability in disease- or vaccine-elicited responses will advance the field.

INTRODUCTION

Without antiretroviral therapy (ART), the majority of HIV-infected individuals progress to AIDS. However, a small portion of those infected demonstrate evidence of immune control of the virus, namely, long-term nonprogressors (LTNP) and elite controllers (EC). These individuals can maintain low or undetectable viral loads without ART. Elucidating the mechanism of viral control in these unique patients remains an area of intense investigation. Despite scientific efforts over the past 2 decades, the design of an effective preventive vaccine for HIV still eludes us. Given the important role of CD8 T cell responses in viral control seen in nonhuman primate and human data (1, 2) and the fact that HLA class I alleles are associated with differences in disease progression (3–7), many current efforts are focused on defining an optimal CD8 T cell immune response to guide effective vaccine design. Unfortunately, studies have failed to consistently demonstrate clear associations of the breadth or magnitude of the cytotoxic T lymphocyte (CTL) response with the plasma viral load (pVL) (8–11). However, polyfunctional T cell responses, including the capacity to secrete cytokines, degranulate, and proliferate in response to antigen, correlate with clinical markers of disease progression (12–16). Maintenance of these types of responses also appears to be important, as patients identified and treated early generate and maintain these responses, while chronically infected patients with uncontrolled viremia and progressive disease lose these responses over time (17–21).

Although HIV-specific CTL appear to control HIV replication in most patients in acute infection and in EC or LTNP, we still do not understand why these responses are lost in the majority of patients in chronic infection (17, 22). CD8 T cells in chronic HIV infection succumb to exhaustion and cell death in an environment of uncontrolled viremia and nonspecific immune activation (23–25). Surface markers, including PD-1, CD160, and 2B4, have provided insights into predicting exhaustion and correlate with clinical parameters of disease progression (26). Similarly, vaccine design must incorporate the capacity to generate effective responses and maintain cell-mediated immunity over time or with subsequent boosting. The results of the RV 144 Thai Trial demonstrated modest protection overall that tended to be greatest in the first year but waned over time (27). This waning of vaccine efficacy highlights the importance of gaining a better understanding of the mechanisms dictating immune memory and the persistence of both antibodies and T cells. Furthermore, a better understanding of cellular markers that can be utilized to predict response durability in disease- or vaccine-elicited responses would advance the field.

Apoptosis occurs through two main pathways. The extrinsic pathway is mediated by surface death receptors, such as Fas/FasL. The intrinsic pathway is an intracellular process that can be initiated by a variety of mechanisms, including lack of growth factors or cytokines, that result in mitochondrial damage (28). Pro- and antiapoptotic members of the Bcl-2 family of proteins regulate the subsequent mitochondrial release of cytochrome c to induce apoptosis (29–31). Antiapoptotic members of the Bcl-2 family, such as Bcl-2 and Bcl-X_L, negatively regulate the induction of the intrinsic apoptotic pathway (30). The intrinsic and extrinsic pathways converge on caspase-3, the effector caspase. Cleavage of caspase-3 results in a cascade of events that lead to programmed cell death (32). Thus, measurement of cleaved caspase-3 (CC-3) reflects cell death occurring by either the intrinsic or extrinsic pathway.

While most research on apoptosis and HIV has focused on the death of HIV-infected CD4 T cells, there is a growing body of literature investigating the death of CD8 T cells and how this may represent an escape mechanism for the virus. This statement is especially pertinent in light of the evidence that apoptosis of the total CD8 T cell population correlates with disease progression (23, 33, 34). Furthermore, analysis of total CD8 T cell apoptosis in LTNP has demonstrated that LTNP have fewer apoptotic CD8 T cells (35, 36). The importance of this phenomenon is more accurately reflected by studies of HIV-specific CTL. In a cross-sectional analysis of chronically infected patients, HIV-specific CD8 T cells were more susceptible to Fas-mediated cell death than cytomegalovirus (CMV)-specific CD8 T cells from the same patient (37) and had reduced levels of Bcl-2 and Bcl-X_L (38). Furthermore, despite similar levels of Fas on CMV- and HIV-specific cells from the same patient, HIV-specific CTL were 3-fold more prone to apoptosis (37). In vitro studies also demonstrate that HIV-infected, activated macrophages (MΦ) can induce apoptosis in HIV-specific CTL, suggesting that while engaging with HIV-infected targets, CTL are at risk of being killed by the targets they are attempting to destroy (37). While the study of apoptosis of cytotoxic T cells suggests a putative mechanism that explains the loss of viral control, studies to date have been limited, as cohorts have not been divided by immune status, pVL, or receipt of ART. In addition, previous studies have not analyzed EC in whom survival of optimal CTL may explain why these patients demonstrate durable viral control. We hypothesize that the capacity for prolonged survival rather than exhaustion and death may be a unique and desirable quality of HIV-specific T cells. Thus, we sought to define the role of apoptosis resistance via upregulation of survival molecules, such as Bcl-2, in the maintenance of effective CTL responses in EC and to elucidate the signaling mechanisms involved in generating those responses. This capacity to resist apoptosis may be an important defining advantage of EC in their control of viremia. Furthermore, CTL with increased survival capacity or apoptosis resistance may be protected from the deleterious effects of immune activation, exhaustion, and death.

MATERIALS AND METHODS

Ethics statement.

Certification by the UAB IRB indicating their review and approval of this activity in accordance with the Common Rule and any other governing regulations has been issued and is on file with the Department of Health and Human Services. Written, informed consent, approved by the international review board, was obtained from each of the study subjects.

Study subjects and HLA typing.

Peripheral blood samples were collected from healthy donors and HIV-infected volunteers at University of Alabama at Birmingham (UAB) 1917 HIV and Alabama Vaccine Research Clinics. HLA typing was performed using the Micro SSP HLA typing system. Table 1 lists the clinical parameters of the patient cohort, which was divided into the following groups: elite controllers (EC), i.e., subjects maintaining plasma HIV-1 viral RNA loads (pVL) below the limit of detection by commercial assays (<50 copies/ml) without antiretroviral therapy (ART) for at least 4 years; successfully treated patients (ST), i.e., subjects with undetectable pVL on ART for a mean of 22 months (range, 10 to 47 months); viremic controllers (VC), i.e., subjects with detectable pVL which always remains below 2,000 copies/ml without ART; and progressors (P), i.e., subjects with higher pVL (>2,000 copies/ml) off ART. Viral blips of <200 copies/ml were acceptable in subjects with an undetectable viral load if subsequent viral load measurements were <50 copies/ml. Healthy controls were HIV-seronegative subjects (n = 9). HLA-specific analysis of CD8 T cells was restricted to HLA-A*2 and HLA-B*57. Table 2 shows detailed information on each patient, including viral load, receipt of antiretrovirals, CD4 T cell count, age, gender, HLA type, gamma interferon (IFN-γ) enzyme-linked immunosorbent spot assay (ELISpot) data, and frequency of HIV- and CMV-specific T cells as measured by tetramer.

Table 1.

Clinical characteristics of HIV-1-infected patient cohort

Patient group	No. of patients	Plasma viral load (RNA copies/ml)	Absolute CD4 count [range (median no. of cells/ml)]
Aviremic
Elite controller	3	<50	850–1,750 (890)
Successfully treated on ARTa	5	<50	450–650 (573)
Viremic
Viremic controller	4	<50–955	400–650 (583)
Progressor	11	2,000–80,000	250–650 (424)

^aNote that these patients have undetectable VL on ART.

Table 2.

Clinical parameters of HIV-infected subject cohort^a

Patient group and identifier	Viral load (RNA copies/ml)	ART status	CD4 count (cells/ml)	Age	Gender	HLA type(s)	SL9 response		KF11 response		NV9 response (% tetramer+)
							SFU/million cells	% tetramer+	SFU/million cells	% tetramer+
EC 1655	<50	Naïve	1,741	45	F	A2, B57	90	1.07	70	1.76	1.4
EC 5823	<50	Naïve	890	51	M	A2, B57	2,285	3.79	2,815	2.59	2.84
EC 5915b	178	Naïve	850	44	M	A2	260	0.18	NA	NA	ND
ST 4081	<50	On	623	59	M	A2	105	0.52	NA	NA	3.43
ST 3814	<50	On	576	60	F	A2, B57	1,000	1.86	1,000	1.58	1.84
ST 4447	<50	On	458	52	M	A2, B57	1,000	1.27	1,000	1.99	ND
ST 4480	<50	On	550	31	M	B57	NA	NA	450	1.59	NA
ST 4496	<50	On	573	51	F	A2	310	0.51	NA	NA	ND
VC 2795c	<50	Off	404	48	M	B57	NA	NA	4,265	4.12	NA
VC 2976	453	Naïve	863	37	F	B57	NA	NA	125	0.37	NA
VC 4311	873	Naïve	524	47	M	A2	337	1.56	NA	NA	ND
VC 6351	955	Naïve	642	39	M	B57	NA	NA	680	2.22	NA
P 6410	1,728	Naïve	847	27	F	A2, B57	—	0.88	715	1.17	ND
P 4024	5,258	Naïve	289	34	M	A2, B57	300	0.41	310	0.83	ND
P 4447	12,768	Naïve	501	52	M	A2, B57	1,320	1.27	2,650	1.87	ND
P 4496	14,208	Off	260	51	F	A2	310	1.42	NA	NA	0.88
P 4081	41,681	Off	436	59	M	A2	105	0.6	NA	NA	3.21
P 6315	43,500	Naïve	571	39	M	B57	NA	NA	680	2.02	NA
P 2569	53,876	Off	406	42	M	A2, B57	—	0.36	1,058	3.15	0.62
P 4480	65,030	Off	424	31	M	B57	NA	NA	450	3.11	NA
P 4081	68,139	Naïve	408	59	M	A2	105	0.44	NA	NA	3.24
P 1581	79,286	Naïve	672	41	M	A2, B57	—	1.4	330	1.93	1.72
P 4480	264,120	Off	294	31	M	B57	NA	NA	450	1.48	NA

^aSFU, spot-forming units; F, female; M, male; NA, data not available (individual did not have the HLA allele restricting the specific peptide tested); —, response not detected (<50 SFU/million cells); ND, not done.

^bPatient had viral blip of 178 copies/ml.

^cAll prior viral load measurements ranged from 81 to 458 copies/ml.

Quantitation of HIV-1 RNA in plasma and absolute CD4⁺ T cell count.

Plasma HIV-1 RNA levels were measured using the Amplicor ultrasensitive HIV-1 monitor assay at UAB hospitals (version 1.5; Roche Diagnostics Systems). Peripheral blood CD4⁺ T cells were determined as described elsewhere (39). Patients are routinely seen at 3-month intervals, and CD4 counts and HIV viral load measurements are obtained at each clinic visit.

Antigens.

HLA-restricted HIV-specific responses were determined using the IFN-γ ELISpot and optimized 9- to 11-mer peptides (Los Alamos National Laboratory HIV Immunology Database). Peripheral blood mononuclear cells (PBMC) were stimulated using optimized HLA-restricted CD8 T cell epitopes, including HLA B*57-restricted Gag-p24 KF11 (KAFSPEVIPMF) and HLA-A*2-restricted Gag-p17 SL9 (SLYNTVATL) (10 μM), cytomegalovirus (CMV) pp65 NV9 (NLVPMVATV) (10 μmol/liter), or staphylococcal enterotoxin B (SEB) (1 μg/ml) (positive control). Peptides previously shown to elicit IFN-γ ELISPOT responses in the studied patients were chosen for further studies.

Ex vivo or in vitro analysis.

Cryopreserved PBMC were thawed, washed in complete RPMI medium (10% AB serum) (R-10 medium), and rested overnight at 37°C in a 5% CO₂ incubator. Cells were counted using Trypan blue staining, and 1 × 10⁶ PBMC were analyzed ex vivo or were cultured in 500 μl R-10 in short-term (6-h; ex vivo) and long-term (4-day; in vitro) cultures at 37°C and 5% CO₂. The peptides HIV Gag-p24 KF11 and -p17 SL9 or CMV pp65 NV9 were used at a concentration of 10 μM for 4 days of peptide stimulation where indicated below.

Surface staining and ICCS.

PBMC from ex vivo and in vitro studies were stained with tetramer (allophycocyanin [APC] HLA-B*57-KF11, APC HLA-A*2-SL9, or PE HLA-A*2-CMV-NV9) (Beckman Coulter) and surface markers (Pacific blue anti-CD3, APC-Alexa Fluor 750 anti-CD4, peridinin chlorophyll protein [PerCp]-Cy5.5 anti-CD8, and phycoerythrin [PE]-Cy7 [PECy7] anti-CD45RO antibodies) (BD Biosciences). Intracellular cytokine staining (ICCS) was performed as described elsewhere (40–42) and was utilized for the detection of cleaved caspase-3 (CC-3) and Bcl-2. Briefly, cells were washed with phosphate-buffered saline (PBS) and permeabilized using Cytofix/Cytoperm (BD Biosciences). Intracellular staining was performed using fluorescein isothiocyanate (FITC)–anti-cleaved caspase-3 (Cell Signaling) and PE–anti-Bcl-2 (BD Biosciences) antibodies according to the manufacturer's protocol.

Flow cytometric analysis.

CompBeads (BD Biosciences) were used to establish the fluorescence compensation setting for multicolor flow cytometric analysis. A minimum of 5 × 10⁵ CD3⁺ events were collected using an LSRII flow cytometer (Becton, Dickinson) and analyzed using FlowJo software version 8.5.2, as previously described (42). Since we were interested in studying apoptosis, we did not include a viability stain, as this could reduce the events of interest. Lymphocytes were gated on forward scatter area (FSC-A) versus side scatter area (SSC-A), and any cells that had low forward scatter with higher side scatter were excluded. Next, CD3⁺ CD8⁺ T cells were selected. From this population, total CD8 T cells or tetramer-specific T cells were analyzed for cleaved caspase-3 expression (see Fig. 1 for gating strategy). Bcl-2, which is expressed in naïve cells, decreases at the peak of the expansion phase and remains low in cells that undergo apoptosis (43). Memory cells typically upregulate Bcl-2 expression, a quality that aids in their survival (43). Thus, all Bcl-2 expression was analyzed from the CD45RO⁺ memory CD3⁺ CD8⁺ gate to exclude activation-induced Bcl-2 downregulation and allow comparison of the memory population (see Fig. 3A and Fig. 4A for gating strategies) (43).

Fig 1.

Frequencies of cleaved caspase-3⁺ CD8⁺ T cells in HIV-infected patients. (A) Representative results of indicated polychromatic flow cytometry gating scheme for measurement of cleaved caspase-3 levels on total and tetramer-specific CD8⁺ T cells are shown. (B) Percentages of CC-3^hi CD8⁺ T cells in healthy controls compared to those in HIV-infected subjects. In both groups, total CD8 T cells are compared to virus tetramer-specific (HIV⁺ or CMV⁺) CD8 T cells. (C) Frequencies of CC-3^hi total CD8 T cells in healthy, aviremic (elite controllers and successfully treated patients) and viremic (viremic controllers and progressors) subjects. (D) Frequencies of CC-3^hi (tetramer-specific [CMV⁺ or HIV⁺]) CD8 T cells in HIV-aviremic and HIV-viremic subjects. CMV-sp, CMV specific. Note that the symbols for panel D (diamonds, progressors; squares, viremic controllers; triangles, successfully treated patients; circles, elite controllers) apply to both HIV-specific and CMV-specific responses. *, P < 0.05.

Fig 3.

Ex vivo expression of Bcl-2 in total and HIV-specific CD8 T cells. (A) Representative results of indicated flow cytometric gating scheme for measurement of Bcl-2 expression on total CD8⁺ T cells are shown. (B) Total CD45R0⁺ CD8 T cells were examined for expression of the prosurvival molecule Bcl-2. The frequencies of Bcl-2^hi total CD8⁺ T cells in healthy controls versus HIV patients, including aviremic (elite controllers and successfully treated patients) and viremic (viremic controllers and progressors) subjects, are depicted. (C) HIV-specific (tetramer⁺) CD45R0⁺ CD8 T cells were examined for expression of the prosurvival molecule Bcl-2. Frequencies of Bcl-2^hi HIV-specific CD8 T cells are plotted across the cohort. *, P < 0.05.

Fig 4.

Coexpression of Bcl-2 and CC-3 on HIV-specific CD8 T cells. (A) Representative results of indicated polychromatic flow cytometry gating scheme for measurement of cleaved caspase-3 and Bcl-2 expression on CD45RO⁺ tetramer⁺ CD8⁺ T cells are shown. Gating from lymphocytes to CD3⁺ CD8⁺ T cells is the same as depicted in Figure 1A. The frequencies of CC-3^lo Bcl-2^hi HIV-specific CD8 T cells (primed for survival) or CC-3^hi Bcl-2^lo HIV-specific CD8 T cells (prone to death) were determined for each subject. (B) Coexpression of CC-3 and Bcl-2 is plotted using the frequencies of CC-3^lo Bcl-2^hi HIV-specific CD8 T cells (primed for survival) for each subject across the cohort. *, P < 0.05. (C) Four patients were examined longitudinally before the initiation of ART and 1 year after the initiation of ART. Frequencies of CC-3^lo Bcl-2^hi HIV-specific CD8 T cells (gated on tetramer⁺ cells), cells primed for survival, are plotted with connecting solid lines to demonstrate the effect of ART. Dotted line connecting open symbols demonstrates the frequencies of CC-3^lo Bcl-2^hi CMV-specific CD8 T cells from one of the ART-treated subjects before and after ART treatment. (D) Frequencies of CC-3^hi Bcl-2^lo HIV-specific CD8 T cells (gated on tetramer⁺ cells), cells prone to death, are plotted with connecting solid lines to demonstrate ART effect. Dotted line connecting open symbols demonstrates the frequencies of CC-3^hi Bcl-2^lo CMV-specific CD8 T cells from one of the ART-treated subjects before and after ART treatment.

Statistics.

Using Prism software (version 5 GraphPad), statistical analysis included the nonparametric Mann-Whitney U test to compare differences between categorical variables. Spearman rank was used as the nonparametric test to compare the relationship between continuous variables. Statistically significant differences were defined as those with a P value of <0.05.

RESULTS

Ex vivo expression of cleaved caspase-3 in HIV-specific CTL.

To address the potential for apoptosis resistance as a mechanism of maintaining effective CD8 T cells in the rare HIV-infected subjects capable of immune control of HIV, we analyzed subjects who demonstrated various levels of HIV control with and without antiretroviral treatment (Tables 1 and 2). We performed an ex vivo flow cytometric analysis on PBMC, analyzing cleaved caspase-3 (CC-3), a marker of apoptosis. Caspase-3 is the terminal effector caspase for both intrinsic (mitochondrial-dependent) and extrinsic (death receptor ligand-dependent) apoptosis pathways. Thus, measurement of CC-3 accurately reflects the induction of apoptosis resulting from cytokine deficiencies and/or surface signaling via Fas and TRAIL. We performed tetramer staining using HLA-B*57 Gag KF11, HLA-A*2 Gag SL9, and HLA-A*2 CMV NV9 tetramers to allow analysis of HIV- and CMV-specific responses. For HIV-specific T cells, we limited our analysis to HLA-B*57 and HLA-A*2 Gag-restricted responses. Doing so allowed us to compare responses restricted by a favorable HLA (i.e., B*57) and a neutral HLA (i.e., A*2). The HLA of each patient from each patient group can be found in Table 2. Representative results of the gating strategy to measure CC-3 on total and HIV-specific CD8 T cells are shown (Fig. 1A), as well as a summary of the results for the entire cohort (Fig. 1B). Similar to prior studies, we found that total CD8 T cells from HIV-infected persons are more prone to apoptosis than those from healthy controls (P = 0.0001) and that HIV-specific CD8 T cells (tetramer positive [tetramer⁺]) are 2 to 3 times more prone to cell death than CMV-specific cells from the same persons (P = 0.02) (Fig. 1B). When examining apoptosis of total CD8 T cells, cells from aviremic HIV-infected persons (EC and ST) were similar to those from healthy controls (Fig. 1C). In contrast, the total CD8 T cells from viremic patients (VC and P) were more prone to apoptosis than those of healthy controls (P = 0.005 and P = 0.0001, respectively) (Fig. 1C). When comparing HIV-specific responses in these cohorts, we found no differences in cell death between viremic individuals (VC and P), despite viremic controllers maintaining pVL of <2,000 copies/ml without ART (Fig. 1D). This suggests that the mere presence of circulating virus, rather than the level of viremia, resulted in an increased frequency of cells prone to death. In contrast, we found significant differences in the frequency of HIV-specific T cells prone to death as measured by CC-3 in the aviremic group, with EC having the lowest frequency of cells prone to apoptosis (P = 0.048) (Fig. 1D). We also examined HIV-specific CD8 T cells targeting either HLA A*2 Gag SL9 or HLA B*57 Gag KF11 to determine if responses restricted by favorable HLA were more likely to survive under unfavorable conditions. The levels of CC-3 on HIV-specific CD8 T cells were indistinguishable between these two HLA phenotypes within each of the groups (data not shown).

Cleaved caspase-3 levels on HIV-specific CTL after stimulation.

We found significant differences in the ex vivo levels of CC-3 in HIV-specific (tetramer⁺) CD8 T cells between groups. In vivo, CD8 T cells from EC and ST have minimal virus exposure. We next sought to extend these studies to determine the effect of HIV peptide stimulation over time on cell death and survival. For these studies, we cultured PBMC from HIV-positive subjects for 4 days in the presence (stimulated) or absence (medium alone) of HIV peptides (KF11 or SL9). In addition, PBMC from healthy HIV-seronegative controls were cultured under similar conditions in the presence or absence of CMV peptide NV9. In our flow cytometric analysis of CC-3 under ex vivo conditions, we consistently found that the subset of HIV-specific CD8 T cells that had downregulated CD8 (CD8^dim) were the most prone to cell death (as determined by increased levels of CC-3) (Fig. 2A and D). In fact, 80 to 90% of the CC-3^hi HIV-specific CD8 T cells originated from the CD8^dim subset of CTL (Fig. 2B and D). This was observed across all HIV subgroups. We therefore analyzed both populations (CD8^hi and CD8^dim T cells) for CC-3 levels within our viremic and aviremic cohorts in an ex vivo and a 4-day culture (medium alone or with simultaneous peptide stimulation) assay. Across all HIV subgroups, we observed lower levels of CC-3 in the CD8^hi T cell group (Fig. 2A and C). Interestingly, peptide stimulation of PBMC over 4 days resulted in statistically higher levels of CC-3 in the CD8^hi population in both viremic populations (Fig. 2B). Alternatively, we found that CD8^dim cells were much more prone to death across the HIV cohort and under all conditions, as evidenced by higher expression of CC-3 (Fig. 2A and D). These cells demonstrated sensitivity to cell death in long-term culture, even in the absence of peptide stimulation (Fig. 2D, hatched versus white bars), that was present regardless of patient group. However, the magnitude of the cell death was remarkably different across the cohort. Despite overall higher expression of CC-3 among the CD8^dim cells than in the CD8^hi cells, EC consistently maintained the lowest levels of CC-3 within the CD8^dim population across the cohort. This observation was noted under both of the 4-day culture conditions, where the expression of CC-3 by EC was found to be 4-fold lower than for viremic (VC and P) subjects and 2-fold lower than for ST subjects (Fig. 2D).

Fig 2.

Comparison of rates of cell death between ex vivo and long-term culture with and without peptide stimulation. (A) Representative results of indicated polychromatic flow cytometry gating scheme to measure the frequencies of CC-3^hi in HIV-specific CD8⁺ T cells (gated on tetramer⁺ cells) in ex vivo studies (left) and after 4 days of in vitro stimulation with HIV peptides (right) are shown. CD8 T cell population is divided to demonstrate CD8^hi versus CD8^dim, the subpopulation with CD8 downregulation. (B) Frequencies of CC-3^hi antigen-specific (Ag-sp) CD8^{hi + dim} cells (gated on tetramer⁺ cells) across the cohort under different culture conditions. Antigen-specific responses for healthy individuals are to CMV (pp65-NV9 tetramer) and for HIV-infected individuals to HIV-1 (Gag-p17 SL9 and Gag-p24 KF11 tetramers). (C) Frequencies of CC-3^hi antigen-specific CD8^hi cells (gated on tetramer⁺ cells) across the cohort under different culture conditions. (D) Frequencies of CC-3^hi antigen-specific CD8^dim cells (gated on tetramer⁺ cells) across the cohort under different culture conditions. *, P < 0.05.

Ex vivo expression of Bcl-2 in total and HIV-specific CD8 T cells.

Since the expression of antiapoptotic molecules can affect the fate of the cell, we sought to determine the ex vivo levels of antiapoptotic Bcl-2 in cells derived from the same patient cohort. When examining Bcl-2 levels in total CD8 T cell populations, VC were different from healthy individuals and exhibited significantly higher levels than P (Fig. 3B). The levels of Bcl-2 on total CD8 T cells observed in healthy, EC, and ST patients were comparable. However, upon analysis of HIV-specific CD8 T cells (Gag tetramer⁺), we found significant upregulation of Bcl-2 in EC and ST compared to the levels in viremic subjects (Fig. 3C). Among HIV-specific CD8 T cells, EC exhibited approximately 2-fold higher levels of Bcl-2 than ST and more than 3-fold-higher levels than both viremic groups.

HIV-specific CTL from elite controllers have a survival advantage.

Analysis of CD8 T cells in EC indicates increased levels of antiapoptotic Bcl-2 and reduced levels of proapoptotic CC-3 compared to the levels in cells obtained from other HIV-infected groups. We next utilized a CC-3 and Bcl-2 colabeling strategy to provide additional insight as to the percentage of cells in each population that were more effectively primed for survival (CC-3^lo Bcl-2^hi) (Fig. 4A). We found significant differences between each group, demonstrating that EC and ST have fewer cells at risk of apoptosis than viremic subjects. Interestingly, when comparing subjects with immune control of virus (EC) to those with pharmacologic control of virus (ST) in an ex vivo analysis, HIV-specific CTL from the EC exhibited a survival advantage, demonstrating preservation of cells primed for survival (CC-3^lo Bcl-2^hi) (for EC versus ST, P = 0.038) (Fig. 4B). This survival phenotype was not preserved in viremic controllers when compared to progressors, despite their having some level of immune viral control. In contrast, EC maintained the lowest frequency of cells prone to death (CC-3^hi Bcl-2^lo) (data not shown).

Treatment partially reverses the susceptibility of HIV-specific CTL to apoptosis.

Since we found that the pattern of pro- and antiapoptotic molecule expression in ST patients more closely mimicked that of the EC group but never achieved the same levels, we extended these studies by performing a longitudinal analysis of four chronically infected persons at least 1 year prior to initiating ART and at 1 year after ART initiation, once viremia was suppressed to undetectable levels (Fig. 4C and D). We demonstrated that control of viremia with ART resulted in upregulation of Bcl-2 and downregulation of CC-3 in HIV-specific CTL, a pattern consistent with improved cell survival (Fig. 4C). In addition, the frequency of HIV-specific CD8 T cells prone to death (CC-3^hi Bcl-2^lo) declined dramatically with suppression of HIV (Fig. 4D, closed symbols), approaching the frequency of CMV-specific CD8 T cells with the same phenotype (Fig. 4D, open symbols). Thus, ART, while unable to reduce apoptotic cells to the levels seen in EC, was able to partially reverse the survival defect demonstrated in HIV-specific CTL.

Longitudinal data before and after the initiation of ART suggest that circulating virus, in part, contributes to enhanced cell death of HIV-specific CD8 T cells. However, the finding that VC with pVL of <2,000 copies/ml have frequencies of cells prone to apoptosis similar to the frequencies in patients with poorly controlled virus (Fig. 1D) challenges the notion that circulating virus is the primary driver of cell death. To better understand the extent of viremia driving cell death, we compared the frequencies of total (Fig. 5A) and HIV-specific (Fig. 5B) CC-3^hi CD8 T cells to plasma viral loads and found that viral load had a positive correlation, albeit weak, with CD8 T cell CC-3 expression (R² = 0.20 and P = 0.03 for total CD8 T cells and R² = 0.29 and P = 0.001 for HIV-specific CD8 T cells). For a more stringent analysis, we excluded all subjects with an undetectable plasma viral load (pVL < 50), including elite controllers and successfully treated patients. This analysis suggests that viral load was a poor correlate of apoptosis (R² = 0.0007 and P = 0.738) (Fig. 5C).

Fig 5.

Correlation of frequency of CC-3^hi CD8 T cells with plasma viral load. (A) For each patient, the frequency of CC-3^hi total CD8 T cells in ex vivo analysis is plotted versus pVL. (B) CC-3^hi HIV-specific CD8 T cells are plotted versus pVL. (C) Aviremic subjects were excluded, and the analysis of CC-3^hi HIV-specific CD8 T cells is plotted versus pVL in viremic subjects only. Spearman rank correlation was performed.

CMV-specific apoptosis in HIV-seronegative and -seropositive subjects.

While the focus of this work was on HIV-specific CD8 T cells, we also examined the susceptibility of CMV-specific CD8 T cells from both HIV-seronegative subjects and HIV-infected subjects. CMV-specific CD8 T cells from HIV-positive subjects are more prone to apoptosis than CMV-specific CD8 T cells from HIV-seronegative subjects (Fig. 1B, closed squares versus open squares). Within HIV-infected subjects, we observed a modest range of susceptibility of CMV-specific T cells to apoptosis, with cells from EC and ST being the least susceptible and cells from progressors being the most susceptible (Fig. 1D, gray symbols). Despite the range, all CMV-specific CD8 T cells had relatively low CC-3 expression, comparable to that of HIV-specific CD8 T cells from EC and ST (Fig. 1D). This is perhaps a reflection that CMV viremia is controlled better than HIV viremia. In long-term culture, CMV-specific T cells from healthy HIV-seronegative controls had low levels of apoptosis under ex vivo and 4-day culture conditions (Fig. 2B). Cell death increased with CMV peptide stimulation, but the frequencies of CC-3^hi CD8⁺ cells were comparable between the CD8^hi and CD8^dim populations (Fig. 2C and D). Finally, we examined the coexpression of CC-3 and Bcl-2 by CMV-specific CD8 T cells from a progressor who was eventually treated with ART and was reanalyzed at a later time point when virus was controlled on ART (Fig. 4C and D, open symbols with dotted line). In this longitudinal analysis, the CMV-specific CD8 T cells had higher frequencies of cells primed for survival (CC-3^lo Bcl-2^hi) than the HIV-specific CD8 T cells when the subject was off ART (Fig. 4C). Once HIV viremia was controlled, the frequencies of CMV-specific cells primed for survival or prone to death did not change significantly.

DISCUSSION

The HIV-specific CD8 T cell response is capable of controlling viral replication in acute HIV infection and, rarely, in chronic infection (i.e., long-term nonprogressors [LTNP] and elite controllers) for years. However, this initial response fails to control virus long-term in the majority of chronically infected patients. In the search for correlates of immunity to HIV, several qualities of HIV-specific CD8 T cells have been identified that likely contribute to an effective antiviral immune response. We reason that survival of CD8 T cells may also represent a key factor of an effective immune response. Antigen-presenting cells provide costimulatory signals which are necessary for the generation of polyfunctional CD8 T cell responses and for the maintenance or long-term survival of these responses (44, 45). In the absence of these critical signals, cells are not only less functional with regard to cytokine secretion, proliferation, and lytic capacity but are also more susceptible to apoptosis. Prior studies have demonstrated that HIV-specific CD8 T cells are more prone to apoptosis than CMV-specific CD8 T cells within the same patient (37). The results of the present study demonstrate an important link in HIV immunopathogenesis that may explain in part the inability of HIV-specific CD8 T cells to maintain long-term viral control.

The present study represents the first analysis of cell death and prosurvival factors in a cohort of viremic patients with various levels of viral control, but more importantly, in a cohort of aviremic patients with either immune control (EC) or pharmacologic control (ST) of HIV-1. Similar to previous groups, we demonstrate that HIV-specific CD8 T cells are prone to apoptosis, with viremic patients exhibiting the greatest susceptibility to apoptosis (37). Interestingly, in our cohort, we determined that this was the case regardless of the level of viral control, as we did not observe significant differences in CC-3 levels between viremic progressors with higher viral loads and viremic controllers, those patients capable of maintaining a pVL of <2,000 copies/ml. Despite this high degree of susceptibility to apoptosis, it is notable that chronically infected viremic patients have a high frequency of viral-specific CD8 T cells (46). Some have significant proliferation of viral-specific effector populations, as well as rapid turnover due to apoptosis. Once treated, patients actually have a decline in the frequency of viral-specific CD8 T cells, due to the decrease in antigenic stimulation and proliferation of effector T cells (47).

To our knowledge, this is the first study to also examine aviremic patients, those with VL of <50 copies/ml, who exhibited the lowest levels of CC-3 on HIV-specific CD8 T cells. Interestingly, we found that elite controllers, those patients with immune control of HIV, exhibit lower levels of CC-3 than successfully treated patients with virologic control on ART. These findings suggest that, in chronic infection, the death of HIV-specific CTL may contribute to the loss of viral control and that resistance to apoptosis in CTL from EC may contribute to their ability to intrinsically control viremia. In support of this, longitudinal studies of persons evolving from acute to chronic HIV infection reveal that HIV-specific CD8 T cells with functional attributes associated with viral control, such as high-avidity CTL, are often deleted in chronic infection (48). Similarly, the preservation of immunodominant responses into chronic infection is associated with a lower viral load set point and slower CD4 decline (49).

The susceptibility of CTL to apoptosis is likely related to environmental and intrinsic cellular factors. HIV infection leads to a state of high immune activation, which is created by circulating uncontrolled viremia and nonspecific activation. The pathogenesis and demise of the immune system is attributed in large part to a breakdown in the mucosal barrier and subsequent gut translocation (50). This leads to further cellular activation, replication of virus, and loss of CD4 T cells (51). To better understand how viral and nonviral mechanisms of cell death may contribute to the loss of the CD8 T cell response, we examined the correlation of viral load and expression of CC-3 on CD8 T cells. We attribute one-third of the HIV-specific CD8 T cell death observed to circulating virus (Fig. 5B). The relatively poor correlation with viral load is supported by the observation that progressors with higher viral loads do not exhibit significant differences in expression of CC-3 compared to its expression in viremic controllers, patients capable of viral control to less than 2,000 copies/ml. Furthermore, viral load alone does not explain the observed differences between EC and ST patients, two groups with undetectable circulating plasma virus.

Nonspecific immune activation resulting from gut translocation likely contributes to immune cell death. Indeed, studies of large cohorts of HIV-infected patients have demonstrated a strong correlation of plasma lipopolysaccharide (LPS) with viral load and the activation state of CD8 T cells as measured by HLA-DR and CD38 (50). While gut translocation can account for some of the non-viral-specific causes of cell death in viremic patients, it does not account for the observed differences between EC and ST. If differences in gut translocation and immune activation explain why ST patients have CTL that are more prone to apoptosis, we would expect ST to have higher levels of LPS and immune activation than EC. In fact, this is not the case. In large cohort studies, EC have higher levels of plasma LPS and higher levels of activated CD8 T cells than ST patients (52). Thus, if nonspecific immune activation explained the observed differences in susceptibility, we would expect the EC to have higher frequencies of HIV- and CMV-specific T cell apoptosis than ST. Instead, within any given EC or ST subject, we observed comparable rates of HIV-specific and CMV-specific cell death (Fig. 1D). While beyond the scope of this work, the unique survival advantage of the HIV-specific CD8 T cells in EC may reflect differences in low-level immune activation from viral turnover in cellular reservoirs, as it has been demonstrated that integrated viral DNA is significantly lower in EC than ST (53).

We observed that decreased expression of CD8 on HIV-specific T cells resulted in a greater frequency of cells prone to death, as evidenced by increased expression of CC-3 (Fig. 2). The down-modulation of CD8 expression has been reported in other infection models, including influenza, and can result in T cell nonresponsiveness (54, 55). CD8 down-modulation has been postulated as a mechanism of immune regulation whereby activated T cells become less or nonresponsive. Such a response in a state of persistent viremia like HIV infection likely contributes to disease pathogenesis.

The events preceding cell deletion are an area of intense investigation in the field of chronic viral infections, including HIV infection. While the generation of an effective immune response requires a combination of antigen presentation with costimulatory signals, the persistence of the response depends in part on negative or inhibitory signals. A persistent high antigen load results in progressive CD8 T cell exhaustion, which is characterized by loss of proliferation, ex vivo cytotoxicity, effector cytokine secretion, and eventually, deletion (56–59). Through genome-wide array analysis comparing exhausted CD8 T cells from chronic lymphocytic choriomeningitis virus (LCMV) infection with those generated in acute LCMV infection, molecular signatures have provided insights into the signaling pathways that result in exhaustion and deletion (60). Several inhibitory molecules are overexpressed in exhausted cells, including PD-1, 2B4, TIM-3, Lag-3, and CD160, some of which correlate with clinical parameters of HIV disease progression (60–63). Selective blockade of PD-1 and TIM-3 has been shown to restore effector functions (64, 65). More recent data demonstrate that the accumulation of multiple inhibitory factors, such as double positive CD160 PD-1, is associated with progressive exhaustion and cellular dysfunction (66–68). In contrast, CD8 T cells that express PD-1 without other inhibitory molecules can still recognize cognate antigen and generate a polyfunctional T cell response (68). Similarly, CD8 T cells from EC express mostly single positive CD160 CD8 T cells, suggesting that this phenotype is more functional (68). Future studies incorporating pro- and antiapoptotic parameters with a comprehensive analysis of inhibitory factors would provide insight into the regulation of exhaustion and identify phenotypes which result in irreversible exhaustion and are destined for deletion.

Given the lack of evidence that external factors explain the observed differences in apoptosis between EC and ST, we looked to intrinsic factors to explain how cells might be primed for survival. Costimulatory and cytokine signals (including interleukin-7 [IL-7] and IL-15) regulate T cell homeostasis and the expression of prosurvival molecules in the Bcl-2 family (44, 69–72). In animal models, the timing of costimulatory signals via CD137 monoclonal antibodies (during the first few days of infection with either LCMV or influenzavirus) significantly altered the persistence or deletion of viral-specific CD8 T cells and the subsequent survival of mice (73). This compromised CD8 T cell response was accompanied by higher levels of IL-10 and tumor necrosis factor alpha (TNF-α), which were associated with the loss of CD4 and CD8 T cells, respectively (73). We therefore reason that rapid viral control in acute disease requires an optimal balance of costimulatory and cytokine signals. The observed differences in the survival advantage seen in EC over ST may be a result of the generation of cells primed for survival early in infection. In this scenario, ART-treated patients, despite having undetectable pVL, may have cells prone to apoptosis, in part due to compromised APC and lack of optimal costimulation. To begin to investigate this, we examined the expression of Bcl-2 in memory cells across the cohort. We focused our investigation on memory cells, as effector populations downregulate Bcl-2 expression (43). We found no differences between EC and ST when examining total CD8 T cells, but they had marked differences in HIV-specific CD8 T cells (Fig. 3). In fact, the highest expression was seen in elite controllers, with observed levels twice those seen in the ST cohort. Finally, colabeling analysis of CC-3 and Bcl-2 demonstrated a phenotype of cells primed for survival (CC-3^lo Bcl-2^hi) (Fig. 4B). In these analyses, EC possessed a larger population of HIV-specific CD8 T cells that are resistant to apoptosis by upregulation of antiapoptotic molecules. In the longitudinal analysis of the viremic patients before and after ART initiation, we demonstrated that although ART can result in upregulation of Bcl-2 on HIV-specific CD8 T cells, the levels in successfully treated patients never achieved the levels seen in EC. This finding suggests that the HIV-specific T cells were generated earlier, prior to viral control, rather than de novo after successful treatment with ART. Finally, to elucidate whether priming for survival was restricted to “good” alleles, such as B57 and B27, we analyzed responses restricted by a favorable HLA, B*57, and by the product of a neutral allele, A*2. We did not observe any selective advantage based on HLA alone. This observation is promising as we translate these data into vaccine development, where an optimal response would be desired in patients of any HLA type.

Our analysis of CMV-specific CD8 T cells demonstrates that (i) CMV-specific T cells from HIV-seropositive subjects are more prone to apoptosis than CMV-specific CD8 T cells from HIV-seronegative subjects (Fig. 1B), (ii) there is a range, albeit modest, of susceptibility to apoptosis of CMV-specific CD8 T cells across the HIV cohort (Fig. 1D), and (iii) ART does not significantly change the frequency of either cells prone to death (CC-3^hi Bcl-2^lo) or cells primed for survival (CC-3^lo Bcl-2^hi) (Fig. 4C and D, open symbols). Collectively, these data demonstrate that the survival of CMV-specific CD8 T cells is not compromised to the same degree as that of HIV-specific T cells within the same host. Interestingly, increasing data suggest that CMV-specific T cells in aging and HIV-infected populations are relatively resistant to apoptosis, leading to higher frequencies of CMV-specific T cells in these populations, a phenomenon which may contribute to immunosenescence and inflammation (74).

In summary, the results of the present study demonstrate differences in the survival of CTL in EC, a phenotype that is important for viral control. This work adds to a growing body of knowledge of the generation and maintenance of effective immune responses to HIV. Further work elucidating the pathways for achieving these responses will allow us to gain insight into mechanisms for manipulating the immune response to induce long-lived responses in vaccine recipients.