Memory T cells of EC subjects are resistant to apoptosis

We studied the survival of memory CD4⁺ T cell subsets in HIV-infected subjects who naturally control viral replication (EC subjects) upon treatment with multiple apoptotic stimuli and compared them to those from ST subjects and HIV^- donors. Supplementary Figure 1 online summarizes the clinical features of the two HIV⁺ subject groups. Sorted T_CM and T_EM cells from EC subjects were significantly more resistant to Fas-mediated apoptosis (4.9% 2.5% and 10.3% 2.1% respectively) when compared to those from ST subjects (14.5% 3.2% and 30% 7% respectively; P < 0.01, n = 5) and to those from HIV^- subjects (8.5% 2.6%, P = 0.016 and 21% 4.4%, P < 0.01, respectively; n = 5) (Fig. 1a). T_CM cells and T_EM cells from both groups of HIV-infected subjects expressed similar amounts of CD95, whereas HIV-uninfected subjects expressed significantly lower amounts of CD95 on both memory T cell subsets (P < 0.01 for both mean fluorescence intensity (MFI) and percentage of positivity for T_CM cells and P < 0.01 for MFI for T_EM cells; Fig. 1b).

Figure 1: Memory T_CM and T_EM cells from EC subjects are less sensitive to Fas-mediated apoptosis and show heightened in vitro persistence after multiple rounds of TCR stimulation compared to those from ST subjects or HIV^- donors.

(a) Sorted T_CM (CD4⁺CD45RA^-CD27⁺CCR7⁺) and T_EM (CD4⁺CD45RA^-CD27^-CCR7^-) cells from HIV⁺ and HIV^- subjects (each symbol indicates a different individual) were treated in the presence or absence of antibody to Fas. Percentage apoptosis was assessed by annexin V labeling (n = 5). (b) Percentages of ex vivo CD95-positive cells. (c–f) Proliferation and persistence of purified T_CM cells from ST, EC and HIV^- subjects. Sorted T_CM cells from the indicated HIV⁺ and HIV^- subjects were labeled with CFSE, cocultured with autologous mDCs in the presence of super antigens (SEA and SEB) and then restimulated at days 12 and 19 with fresh SEA and SEB–pulsed mDCs. (c) Percentages of total CFSE-low viable (7AAD^-) memory cells (CD3⁺CD4⁺CD45RA^-) and T_CM cells (CD3⁺CD4⁺CD45RA^-CD27⁺CCR7⁺) are shown at day 12. (d) Absolute numbers of total viable memory CD4⁺ T cells (left) and gated T_CM (right) from ST, EC and HIV^- subjects were determined by trypan blue exclusion. Results are expressed in log₂ scale (n = 5 independent experiments for HIV⁺ subjects and n = 3 for HIV^- donors). The underlined numbers represent the half-lives of memory T cells. (e,f) Distribution of memory T cell subsets on cultured 7AAD^-, CFSE-low cells from EC subjects, ST subjects and HIV^- donors at day 12 (e) and in viable cells from four of five EC subjects at day 26 (f).

Full size image (97 KB)

We further compared the ability of T_CM cells from EC subjects, ST subjects and HIV^- donors to proliferate and survive in response to multiple rounds of TCR triggering. Sorted T_CM cells were stimulated with autologous staphylococcal enterotoxin A (SEA) and staphylococcal enterotoxin B (SEB)–pulsed mature dendritic cells (mDCs) and then restimulated at days 12 and 19 in the presence of zidovudine (AZT) and ritonavir to prevent production of HIV particles in the culture. At days 12, 19 and 26 of culture, total numbers of viable memory CD4⁺ T cells were counted, the half-lives of these cells were calculated and the frequencies of T cell memory progeny (carboxyfluoroscein succinimidyl ester (CFSE)-low) subsets were assessed by flow cytometry.

By day 12 after the beginning of the culture, total memory T cells and gated T_CM cells from both HIV⁺ and HIV^- groups showed similar levels of proliferation, as the percentages of CFSE-low viable cells did not differ between these groups (n = 5; Fig. 1c). Notably, despite being able to proliferate efficiently, T_CM cells from ST subjects were not able to persist in culture, unlike T_CM cells from EC subjects (Fig. 1d). In fact, by day 4, significant apoptosis was observed in T_CM cells from ST subjects, whereas the frequencies of apoptotic cells were significantly lower in cultures from EC subjects (5.3% 3.4% for EC subjects and 15.3% 4.9% for ST subjects, P = 0.044, n = 3; Supplementary Fig. 2 online). Total counts of viable memory CD4⁺ T cells from EC subjects were approximately 2.5-fold higher at day 12 than those obtained from ST subjects (n = 5; Fig. 1d). Strikingly, almost all cells from ST subjects died by day 19, whereas cells from EC subjects continued to proliferate and survived for at least 26 d after three rounds of activation. The half-lives of gated T_CM cells obtained from EC and ST donors were 13.1 and 4.3 d, respectively (Fig. 1d).

The absence of T_CM cells and other memory T cell subsets in cultures of cells from ST subjects at day 19 was not due to their incapacity to differentiate, because at day 12 we could observe comparable percentages of proliferating CFSE-low T_CM cells in culture from ST and EC subjects (Fig. 1e). However, by day 26, a significant proportion of T_CM cells persisted only in cultures from EC subjects (4 of 5 EC subject cultures) (Fig. 1f). The percentages of cells with the T_CM phenotype at day 26 in cultures of cells from EC subjects ranged from 8.9% to 21.7%. Progressive decreases in cell numbers from EC subject cell cultures were most probably due to the fact that TCR-triggered T_CM cells differentiate into T_TM (CD45RA^-CD27⁺CCR7^- transitional memory CD4⁺ T cells) and T_EM cells, two subsets that are not endowed with long-term survival potential¹³. Purified T_CM cells from HIV^- donors (n = 3) showed an intermediate phenotype when compared to T_CM cells obtained from the two sets of HIV⁺ subjects. Indeed, although we could not find significant differences in proliferation at day 12 between HIV^- and HIV⁺ subjects (Fig. 1c; P > 0.05 and n = 3), and although all groups of subjects showed similar capacities to generate T_TM and T_EM cells (Fig. 1e), the kinetics of survival of gated T_CM cells from HIV^- donors were always significantly lower (almost all T_CM cells died after day 19) when compared to those from EC subjects (P = 0.03 at days 19 and 26, n = 3; Fig. 1d).

The inability of T_CM cells from ST and HIV^- donors to persist was not due to defects in DC maturation, as mDCs from EC, ST and HIV^- subjects expressed comparable amounts of maturation markers (including CD80, CD86, CD83 and HLA-DR molecules), were able to secrete similar amounts of several proinflammatory cytokines and chemokines, and further showed equal capacity to induce proliferation in a mixed lymphocyte reaction (Fig. 2).

Figure 2: mDCs from EC subjects, ST subjects and HIV^- subjects show similar phenotypic and functional patterns.

(a) mDCs derived from EC, ST and HIV^- subjects show a similar capacity to induce allogeneic responses in CD4⁺ T lymphocytes. CD4⁺ T cells were purified by autoMACs enrichment (98.1% purity) and then labeled with CFSE and cocultured for 6 d with allogeneic mDCs. These mDCs were generated in vitro from different HIV⁺ or HIV^- subjects, the same subjects used for previous persistence assays. Results are expressed as the percentage of CFSE-low CD4⁺ T cells from five independent ST, EC and HIV^- subjects. (b) In vitro mDCs generated from ST, EC and HIV^- subjects were labeled with different surface markers (as CD1a, CD14, CD80, CD83, CD86, HLA-ABC and HLA-DR) and analyzed by flow cytometry. Results are expressed as percentages of positively-stained cells from five distinct HIV⁺ and three HIV^- donors. Similar conclusions were obtained using MFI results (data not shown). (c) Supernatants of the mDCs from b were assessed for the presence of several secreted proinflammatory cytokines and chemokines with cytometric bead array. Data are depicted as levels of cytokine/chemokine production in pg/ml (n = 5 for HIV⁺ subjects and n = 3 for HIV^- donors).

Full size image (94 KB)

Moreover, T_CM cells from ST and EC subjects expressed similar amounts of T cell activation markers, including CD127 (the IL-7 receptor), CD71, CD38 and Ki67, suggesting a comparable ex vivo activation state of the T_CM cells for both groups of HIV⁺ subjects at the onset of long-term culture (Fig. 3a–e). In contrast, the expression of CD127 (MFI P < 0.05, n = 3), as well as the percentages of CD38⁺ and Ki67⁺ T cells, were significantly lower in T_CM and T_EM cells from HIV^- as compared to HIV⁺ subjects (Fig. 3a–e). Finally, ex vivo T_CM cells from all HIV⁺ and HIV^- subjects had similar TCR repertoires in response to SEA and SEB, as monitored by cell surface expression of several V molecules previously described to be involved in the response to these superantigens (Fig. 3f)¹⁵. Together, these results show that T_CM cells from EC subjects are distinct from those from ST subjects and HIV^- subjects because they have a superior capacity to persist in culture and resist Fas-mediated apoptosis and that T_CM cells from ST subjects, although they can proliferate and differentiate, do not persist in culture and quickly die by apoptosis. Of note, at day 12, T_EM cells from EC subjects were more viable when compared to T_EM cells from ST subjects (3.87-fold increase, P = 0.026, n = 4), although T_EM cells from both groups of subjects did not persist beyond day 12 of culture (data not shown). These differences in the survival of T_CM and T_EM cells between EC and ST subjects prompted us to define the involvement of the FOXO3a pathway in T_CM cell persistence during HIV infection, as we had already shown that this pathway is implicated in the survival and maintenance of CD4⁺ T_CM cells¹³.

Figure 3: Ex vivo T_CM and T_EM cells from EC and ST subjects show similar profiles of activation and proliferation markers.

PBMCs from ST subjects, EC subjects and HIV^- subjects were labeled with antibodies specific for CD3, CD4, CD45RA, CD27, CCR7 and then for (a) CD127, the IL-7 receptor, (b) CD25, (c) CD71 (d) or CD38 or were labeled with antibodies to Ki67 for intracellular staining with saponin permeabilization (e). Results shown represent the percentage of positive cells on gated T_CM and T_EM cells from EC, ST and HIV^- donors (n = 5). Similar results were also obtained using MFI data (data not shown). Viremic HIV⁺ subjects (vir. subjects, c–e) in primary infection (<3 months of infection) were used in parallel as positive controls, owing to their heightened immune activation state (n = 2). (f) Gated T_CM cells were also labeled with antibodies to different TCR Vs known to respond to SEA or SEB. Results shown are depicted as the percentage of V-positive T_CM cells s.d. of five independent EC subjects, five independent ST subjects and five independent HIV^- donors (n = 5).

Full size image (70 KB)

Cells from EC subjects have distinct FOXO3a phenotypes

We examined the phosphorylation levels of FOXO3a on ex vivo purified T_CM and T_EM cells obtained from EC, ST and HIV^- donors (n = 3). There was a significant increase in FOXO3a phosphorylation in both memory subsets obtained from EC subjects when compared to those from ST subjects (P = 0.04 for Ser253 and P < 0.01 Thr32 for both memory subsets; Fig. 4a,b). Similarly, we found higher levels of FOXO3a phosphorylation in EC subjects when compared to HIV^- donors (P = 0.045 and P = 0.016 for Ser253 residue; P = 0.032 and P < 0.01 for Thr32 residue in T_CM and T_EM cells, respectively). We found no significant differences in the amount of total FOXO3a in T_CM and T_EM cells (Fig. 4a,b) from all groups of subjects (P > 0.05, n = 3). As FOXO3a phosphorylation negatively regulates its own transcriptional activity^{10, 13, 16}, we measured the expression of FOXO3a transcriptional targets in HIV^- and HIV⁺ subjects. We found that T_CM and T_EM cells from ST subjects expressed significantly higher amounts of Bim (P = 0.02 and P =0.049, respectively) and p130 proteins (P < 0.01 and P < 0.046, respectively) compared to the same subsets obtained from EC subjects (Fig. 4a,b). In addition, Bim and p130 expression was significantly increased in T_EM from HIV^- donors when compared to those from EC subjects (P = 0.036 and P = 0.022, respectively). These data indicate that the lower amounts of phosphorylated FOXO3a expressed in T_CM and T_EM cells from ST subjects and in T_EM cells from HIV^- donors are responsible for the increased FOXO3a transcriptional activity, as observed by the upregulation of its target molecules. To identify upstream signals responsible for FOXO3a phosphorylation in T_CM and T_EM cells from HIV-infected and HIV^- donors, we quantified the expression of activated AKT and IKK and subunits, all kinases involved in FOXO3a phosphorylation^{13, 16, 17}. Western blot densitometric analysis showed significantly higher amounts of phospho–IKK-/ in memory T cells from EC subjects (P < 0.01 for both subsets) compared to ST subjects; notably, similar levels of phospho–IKK-/ were obtained when comparing T_CM cells from EC subjects and HIV^- donors (P > 0.05, n = 3), whereas phospho–IKK-/ abundance was significantly lower in T_EM cells from HIV^- donors when compared to those from EC subjects (P < 0.01, n = 3). Phosflow analysis of T_CM and T_EM cells from EC subjects showed higher amounts of phospho-AKT compared to cells from ST subjects and HIV^- donors (MFI P < 0.01 for both subsets; Fig. 4c). The differences in phospho-AKT suggest that it is quite likely that the activity of this kinase could be responsible for the different levels of both phosphorylated forms of FOXO3a observed when comparing cells from HIV^- and EC subjects. These results provide evidence for the contribution of the FOXO3a pathway to memory T cell survival during HIV infection.

Figure 4: T_CM and T_EM from EC subjects display higher phospho-FOXO3a phenotypes than those from ST subjects and HIV^- subjects.

(a) Total lysates of equal numbers of T_CM and T_EM cells obtained from HIV⁺ and HIV^- subjects were subjected to immunoblotting as indicated. Actin and total FOXO3a blots were performed in parallel as loading controls. The antibody recognizes phospho-IKK phosphorylated at Ser176 and Ser180. (b) Densitometric quantification of the spots was performed using ImageQuant software; levels of expression of each protein were first normalized to actin and were later expressed as the ratio of densitometric values of protein of interest divided by densitometric values of actin within the same blot. Similar results were obtained in three independent experiments. n.s., not significant; ^*P < 0.05; ^**P < 0.01. (c) Phosflow analysis of phospho-AKT (Ser473) levels on ex vivo gated T_CM and T_EM cells from HIV⁺ and HIV^- donors (n = 5). An average of 20,000 gated events was collected on a LSRII flow cytometer. Data were analyzed on DIVA software.

Full size image (85 KB)

Absence of FOXO3a protects cells of ST subjects from death

We investigated whether interfering with FOXO3a expression could protect memory T cells isolated from ST subjects from apoptosis. Isolated total memory CD4⁺ T cells from ST subjects were activated with antibodies to CD3 and CD28 for 24 h and were then transfected with small interfering RNAs (siRNAs) specific for FOXO3a (Fig. 5a). Quantification of FOXO3a protein by densitometry in three separate subjects showed more than 70% reduction in its expression upon transfection of FOXO3a siRNA (P < 0.03, n = 3; Fig. 5b). Quantitative analysis of Bim and p130 showed significant downregulation of their protein expression (70–80%) upon silencing of FOXO3a expression (P < 0.05, n = 3; Fig. 5a,b). Electroporation alone or transfection with degenerate siRNA did not affect FOXO3a expression or the detection of its targets (Fig. 5a,b). Of note, we did not observe any changes in levels of ZAP70 and phospho–S6 ribosomal kinase upon transfection of siRNAs, further confirming the specificity of FOXO3a silencing (data not shown). siRNA-transfected memory T cells from ST subjects were then treated with antibody to Fas for 48 h. Total memory cells or gated T_CM and T_EM cells were identified by surface staining with antibodies to CD3, CD4, CD45RA, CD27 and CCR7 and then assessed for apoptosis by annexin V staining. The results show that percentages of apoptotic cells in the presence of the FOXO3a siRNA were never greater than 20% of those obtained when FOXO3a's expression was not inhibited by specific siRNA treatment (P = 0.03, n = 3; Fig. 5c). In fact, less than 10% of cells were annexin V positive (Fig. 5c), a percentage very similar to that observed in cells from EC subjects. To our knowledge, this is the first demonstration that FOXO3a is implicated in the increased susceptibility of memory T cells to apoptosis during HIV infection and that silencing of FOXO3a increases the short-term survival of T_CM and T_EM cells, at least in part by increasing their resistance to Fas-induced apoptosis.

Figure 5: FOXO3a shutdown by siRNA rescues T_CM and T_EM cells from ST donors from Fas-mediated apoptosis.

Purified memory CD4⁺ T cells from ST subjects (>96% purity) were activated for 24 h with 1 g/ml of antibodies to CD3 and CD28, electroporated and transfected with degenerated negative siRNA or siRNA specific for FOXO3a for 24 h. Cells were then treated in the presence or absence of 1.25 g/ml CH11 for 2 d. (a,b) FOXO3a, Bim and p130 protein abundance was assessed by western blotting. (a) Blots shown are representative of three separate experiments, and densitometric analysis (b) was performed in parallel as above with ImageQuant software. (c) Percentage of Fas-mediated apoptosis was determined by annexin V–allophycocyanin staining in T_CM, T_EM and total memory CD4⁺ T-cell subsets. The flow cytometric results are depicted as percentage of apoptotic cells treated with antibody to Fas (anti-Fas) – percentage of apoptotic untreated cells for the same subject s.d. (n = 3).

Full size image (78 KB)

FOXO3a Nt improves survival of T_CM cells of ST subjects

The decreased ability of T_CM cells from ST subjects to persist in vitro may translate into a decrease in global antigen-specific memory T cell pools in vivo^{5, 18}. Of note, our own preliminary data show a progressive decrease in total T_CM cells in a longitudinal study with HIV⁺ subjects (N.C., unpublished data), in agreement with recently published studies in the SIV model^{4, 7, 19}. Here we further investigated whether interfering with FOXO3a would restore the capacity of T_CM cells from ST subjects to persist in vitro. Isolated memory CD4⁺ T cells were transduced with lentiviral vectors encompassing a dominant-negative form of FOXO3a (FOXO3a Nt virus) or an empty virus²⁰ and then cultured with autologous SEA and SEB–pulsed mDCs. The FOXO3a Nt virus includes the N-terminal fragment of FOXO3a (amino acids1–304), previously described to inhibit the pro-apoptotic function of FOXO3a²¹. FOXO3a Nt is able to translocate to the nucleus and compete with transcriptionally active FOXO3a for binding to DNA. The antiapoptotic function of this vector was first validated in Jurkat cells (n = 3; Fig. 6a,b). This FOXO3a Nt virus was able to generate FOXO3a Nt and prevent both spontaneous and FOXO3a-induced apoptosis (Fig. 6a,b). Indeed, Jurkat cells stably expressing FOXO3a Nt were protected from FOXO3a-mediated apoptosis when Jurkat cells were transduced with a FOXO3a triple mutant (FOXO3a TM), which constitutively translocates to the nucleus and induces apoptosis (Fig. 6b)^{16, 22}. Expression of FOXO3a Nt in Jurkat cells led to a significant reduction in spontaneous apoptosis (25.9% 2.4% apoptotic cells in uninfected Jurkat cells compared to 5.8% 1.8% apoptotic cells in FOXO3a Nt–transduced Jurkat cells; Fig. 6b, left) and a potent decrease in FOXO3a TM–mediated apoptosis (76.8% 11.6% apoptotic cells in FOXO3a TM–transduced Jurkat cells compared to 15.5% 4.8% apoptotic cells in FOXO3a TM– and FOXO3a Nt–cotransduced Jurkat cells; Fig. 6b, right).

Figure 6: FOXO3a Nt increases T_CM cell persistence of ST subjects.

Cytometry analysis was performed on Jurkat cells to determine GFP and c-myc expression (a) and apoptosis, as assessed by 7AAD and annexin V labeling (b) (n = 3). (c–f) Purified memory CD4⁺ T cells obtained from ST subjects were cocultured with autologous SEA and SEB–pulsed mDCs and then transduced with empty or FOXO3a Nt–encoding lentiviruses. (c) Representative contour plots of GFP⁺7AAD^- memory CD4⁺ T cells at day 12 or 26, as detected by flow cytometry (n = 3). (d) In vitro absolute numbers of total viable memory CD4⁺ T cells and of viable gated T_CM cells from ST subjects. These results are expressed as the average of three independent experiments s.d. in log₂ scale. The underlined numbers represent the half-lives of memory T cells. (e) Distribution of memory T cell subsets on cultured 7AAD^-, FOXO3a Nt–transduced cells from ST subjects at day 26 (n = 3). Others, CD27^-CCR7⁺ cells. (f) c-myc tag, FOXO3a, p130 and Bim protein abundance was also determined at day 12 in memory CD4⁺ T cells from ST subjects by western blotting (n = 3).

Full size image (98 KB)

We then assessed whether the presence of FOXO3a Nt could increase the persistence of T_CM cells from ST subjects in long-term in vitro cultures. The efficacy of infection with our lentiviruses on primary activated memory CD4⁺ T cells from ST subjects ranged between 20% and 25% at day 12 after infection as monitored by GFP expression (Fig. 6c). Almost all surviving T cells at day 26 expressed GFP (82.9% 4.2%, n = 3), indicating a selective survival advantage for FOXO3a Nt virus–transduced cells. At days 12, 19 and 26 after infection, memory CD4⁺ T cell viability and phenotype were assessed (n = 3). Both total memory CD4⁺ T cells and gated T_CM cells from ST subjects infected with FOXO3a Nt virus survived significantly longer (26 d) and were able to sustain multiple rounds of activation (Fig. 6d) when compared to cells infected with the control lentivirus. Expression of FOXO3a Nt increased the half-life of total memory CD4⁺ T cells and T_CM cells by 8.2 and 6.8 d, respectively; the half-life of ST subject T_CM cells transduced with FOXO3a Nt was comparable to that of T_CM cells from EC subjects. In addition, we also found that transduction of memory T cells from ST subjects increased the survival of T_CM cells after 26 days of culture, as their percentages of CD45RA^-CD27⁺CCR7⁺ T_CM cells were now comparable to those shown for EC subjects (Fig. 6e). The presence of FOXO3a Nt led to a significant reduction in the expression of FOXO3a transcriptional proapoptotic targets, as indicated by the decrease in Bim and p130, without interfering with FOXO3a protein expression levels (Fig. 6f).

We confirmed that FOXO3a Nt was expressed only in FOXO3a Nt virus–infected cells by c-Myc tag detection (Fig. 6a,f). These results demonstrate that interfering with FOXO3a expression and downregulation of FOXO3a transcriptional target expression increases long-term persistence of T_CM cells in ST subjects by protecting them from premature death.

HIV⁺ subjects

We selected ten subjects chronically infected with HIV and undergoing HAART treatment rendering them aviremic, called here ST subjects. We also included five EC donors who had been evaluated with a very stringent set of inclusion criteria in this study. This research was approved by the Office of Research Ethics, Royal Victoria Hospital, McGill University Health Center, and by the Comité d'Éthique de la Recherche, Centre Hospitalier de l'Universite de Montréal. The virological and immunological profiles for all HIV⁺ subjects are summarized in Supplementary Figure 1.

Reagents and antibodies

CH11 antibody to Fas is from Immunotech. We purchased all antibodies used for flow cytometry from BD Biosciences, except for the antibody to CD45RA-ECD, which was from Beckman Coulter. We purchased all primary antibodies used in western blots from Cell Signaling Technology, whereas we purchased antibodies to p130 and actin from Sigma Aldrich and antibody to FOXO3a from Abcam.

Memory CD4⁺ T cell purification

We purified peripheral blood mononuclear cells (PBMCs) into CD4⁺ T_CM and T_EM cell subsets as previously described¹³.

In vitro generation of mature dendritic cells

We cultured PBMCs for 2 h to remove nonadherent cells. We further cultured adherent monocytes in X-vivo medium (Cambrex) for 6 d in the presence of 500 U/ml of IL-4 and 800 U/ml of granulocyte macrophage colony–stimulating factor (GM-CSF) to generate immature DCs. We achieved DC maturation by adding IL-4, GM-CSF, 10 ng/ml tumor necrosis factor-, 10 ng/ml IL-1, 150 ng/ml IL-6 and 1 g/ml prostaglandin E2 for 24 h.

Assays to measure apoptosis

We first cultured purified T_CM and T_EM cells in complete RPMI 1640 and treated them in the presence or absence of 1.25 g/ml of the Fas-specific antibody CH11 for 24 h. We detected apoptotic cells with annexin V–FITC labeling according to the manufacturer's protocol (Bender MedSystems). We collected approximately 20,000 gated events on the BD LSRII flow cytometer (BD Biosciences) and analyzed the data using the DIVA software.

Ex vivo Phosflow analysis of AKT

The experimental protocol for phospho-AKT staining has been previously described¹³.

Western blotting

A detailed experimental protocol for T_CM and T_EM cell western blotting has been previously described¹³.

Proliferation and persistence assays

We sorted T_CM cells from HIV⁺ and HIV^- subjects and then labeled them with CFSE. 500,000 T_CM cells were cocultured with autologous mDCs (40:1 T cell to DC ratio) in complete RPMI 1640 with 2 ng/ml SEA and SEB. On days 12 and 19, we restimulated the cells by adding fresh mDCs and superantigens. We counted viable cultured CD4⁺ memory T cells by trypan blue exclusion and stained them with Alexa700-conjugated antibody to CD3, Pacific Blue–conjugated antibody to CD4, allophycocyanin-Cy7–conjugated antibody to CD27, Alexa647-conjugated antibody to CCR7 and 7-amino-actinomycin D (7AAD) for flow cytometry analysis. To avoid production of HIV viral particles (data not shown), we added ritonavir (5 ng/ml) and AZT (10 M) to the cultures. We obtained ritonavir (catalog number: 401986) through National Institutes of Health AIDS Research, whereas we purchased AZT from Sigma Aldrich.

Transfection and small interfering RNA assays

We purified CD4⁺ memory T cells from PBMCs from ST subjects by magnetic bead separation (Miltenyi Biotec; >96% purity). We preactivated the cells with 1 g/ml of antibodies specific for CD3 and CD28 for 24 h. Then, we electroporated 6 10⁶ cells per condition using Nucleofector II technology according the manufacturer's protocol (Amaxa Biosystems). We obtained FOXO3a- specific siRNA and negative control siRNA from Invitrogen (FOXO3a Validated Stealth DuoPak). We transfected 3 g of siRNA for each condition. After 24 h, we treated or did not treat cells with 1.25 g/ml CH11 antibody for 2 d. We kept some cells to measure protein levels by western blotting and labeled the rest of the cells with Alexa700-conjugated antibody to CD3, Pacific Blue–conjugated antibody to CD4, antibody to CD45RA-ECD (energic couple dye, a tandem composed of phycoerythrin and Texas Red), phycoerythrin-conjugated antibody to CD27, phycoerythrin-Cy7–conjugated antibody to CCR7 and allophycocyanin-conjugated antibody to annexin V to quantify apoptosis in total memory CD4⁺ T cells and in gated T_CM and T_EM cells.

Plasmids and production of recombinant lentivirus

The lentiviral vector pWPI (empty vector), packaging plasmid psPAX2 and envelope plasmid pMD2G were generously provided by D. Trono at the University of Geneva. We cloned the FOXO3a N-terminal fragment into pWPI using a PmeI restriction site. We produced the recombinant virion particles by transient polyethylenimine cotransfection of 5 10⁷ 293T cells in 175-cm³ flasks. We determined viral titers (which ranged from 0.8 to 6.6 10⁹ transduction units per ml) were determined by transduction and FACS analysis of 293T-cells.

Restoration of memory T cell persistence by FOXO3a N-terminal fragment

We cocultured 1 10⁶ memory CD4⁺ T cells (>96% purity) from ST subjects with their autologous SEA and SEB–pulsed mDCs and ritonavir. We then infected the cells with empty lentivirus (carrying just GFP) or FOXO3a Nt virus (carrying both GFP and c-myc tagged FOXO3a Nt chimera protein) at a multiplicity of infection of 20. We performed restimulation steps on days 12 and 19 of in vitro culture by adding fresh mDCs pulsed with SEA and SEB. We then counted cells by microscopy at different times and labeled them for flow cytometry analysis with the same memory CD4⁺ T cell cocktail antibodies used for the siRNA assays.

Statistical analysis

We analyzed all numeric data with Prism3 statistical software with the nonparametric Mann-Whitney test. We considered P values of less than 0.05 significant.

Note: Supplementary information is available on the Nature Medicine website.

Author contributions

J.v.G. performed most of the experiments and wrote the article. R.-P.S. and E.K.H. generated the concept, supervised the experiments and wrote the article. F.A.P., Z.H., Y.Z. and Y.S. worked on the siRNA and lentiviral construct; N.C. did the p24 ELISA; B.Y.-D., E.A.S., L.T., M.E.F. and C.R. worked on the in vitro system; S.G. performed sorting; G.B. and P.W. worked on the statistics; and J.-P.R. and R.S.B. recruited subjects and helped with the discussion.