Therapeutic CD8⁺ T cell tissue retention and immunomodulation during ART interruption fail to prevent SIV rebound

Significance

HIV therapeutic and cure strategies have all operated under the assumption that CD8⁺ T cells can become cytotoxic in lymph nodes to eliminate infected cells from reservoir repositories. Using FTY720 to prevent lymphocyte egress from lymph nodes, we demonstrated that lymphoid tissue-restricted CD8⁺ T cells are insufficient to prevent Simian immunodeficiency virus (SIV) rebound during antiretroviral treatment interruption, even in the presence of immunomodulators such as IL-15 receptor superagonist and PD-1 blockade. These findings call into question the ability of lymph node CD8⁺ T cells to control or eliminate SIV/HIV infection from lymphoid tissues without substantial immunomodulation, a notion contrary to current efforts to engage CD8⁺ T cells for therapeutic elimination of HIV viral reservoirs during antiretroviral therapy.

Abstract

A primary obstacle for HIV elimination is the long-term viral reservoir in lymphoid tissues (LT) that can cause rebound viremia if therapy is stopped. Cytotoxic CD8⁺ T cells are critical for control of HIV and Simian immunodeficiency virus (SIV) viremia; however, CD8⁺ T cells that migrate to LT are primarily noncytotoxic, calling into question whether these cells could reduce the viral reservoir on antiretroviral therapy (ART) or control viral replication when therapy is halted. To determine whether CD8⁺ T cells can inhibit viral replication when retained in LT, we inhibited lymphocyte egress from LTs in ART-treated SIV-infected rhesus macaques (RMs) during analytic treatment interruption (ATI) using the S1PR modulator FTY720 alone or in combination with anti-PD1 antibody (αPD1) and the IL-15 receptor superagonist N-803 to increase cytolytic function. FTY720 retained migrating CD4⁺ and CD8⁺ T cells in LT, whereas cytotoxic CD8⁺ T cells remained in the vasculature. After ATI and viral rebound, activated SIV-specific CD8⁺ T cells increased in frequency in LT of FTY720-treated RMs but failed to become cytotoxic or control plasma viremia compared to controls, even when combined with αPD1 and N-803. These findings indicate that LT-localized CD8⁺ T cells alone may be insufficient to delay or prevent plasma viral rebound during ATI.

Elimination of the latent HIV reservoir remains the key challenge for immune-mediated clearance of HIV infection and the focus of HIV cure-based research. While antiretroviral treatment (ART) can suppress viral replication, the long-lived nature and homeostatic renewal capability of infected CD4⁺ T cells perpetuates the viral reservoir for the lifetime of the host (1). The HIV-infected CD4⁺ T cell reservoir resides predominantly in lymphoid tissues [lymph nodes (LNs), and gut-associated lymphoid tissues] irrespective of treatment status (2–4). In the absence of therapy, LN T follicular helper cells (TFH) serve as a predominant cell type for viral infection and propagation within lymphoid tissue (5, 6). Indeed, in the Simian immunodeficiency virus (SIV) rhesus macaque (RM) model of HIV infection, >98% of SIV-infected cells are found within the gut and lymphoid tissues (2, 7). However, many infected CD4⁺ T cells in tissues and peripheral blood have migratory and tissue-homing properties, and are therefore capable of continual redistribution across the body (1). Development of strategies to clear the HIV reservoir remains paramount to HIV eradication efforts.

Numerous studies have shown the importance of CD8⁺ T cells and perforin/granzyme B-mediated cytotoxicity to HIV/SIV immune control (8–13). Given that fully differentiated cytotoxic CD8⁺ T cell subsets are largely restricted to the vasculature and blood penetrating tissues such as the spleen and bone marrow, due in part to the absence of CCR7 and CD62L expression (14, 15), their role in controlling lymphoid tissue HIV reservoirs is unclear. We and others have shown that cytotoxic CD8 activity is lost in LNs rapidly after acute SIV/HIV infection (16, 17), and remains significantly lower than blood during chronic infection, irrespective of treatment or progression status (17–21). However, despite the lack of cytotoxicity, HIV- and SIV-specific CD8⁺ T cells can readily be observed in lymphoid tissues (either as migrating or resident cells) during all phases of infection (16–19, 22, 23). While it remains unclear what role tissue-homing or tissue-resident CD8⁺ T cells play in controlling plasma viremia, migration of infected CD4⁺ T cells into lymphoid tissues likely facilitates the circumvention of surveillance by blood cytotoxic CD8⁺ T cells.

To directly address this issue, we used the lymphocyte migration inhibitor fingolimod (FTY720) alone or in combination with immunomodulators anti-PD1 antibody (αPD1) and the IL-15 receptor superagonist (IL15R SA) N-803 to inhibit lymphocyte egress from lymphoid tissues during analytic therapy interruption (ATI) in SIV-infected RMs, with the goal of increasing the cytotoxic properties of LN SIV-specific CD8⁺ T cells. FTY720 is a sphingosine-1 phosphate receptor (S1PR) modulator that blocks the interaction of S1P with four of its receptors (S1PR1, S1PR3, S1PR4, and S1PR5), thereby rendering T cells unable to respond to the S1P gradient and impeding their egress from lymphoid tissues (24). FTY720 does not selectively induce recruitment of lymphocytes to lymphoid tissues but rather prevents S1P-dependent egress from any tissue in which they are present. FTY720 has previously been used in ART-suppressed, SIV-infected RMs, where it resulted in a decrease of SIV DNA content in blood as well as in LN follicular helper T cells in most treated animals (25). Based upon this, we hypothesized that FTY720-induced migration inhibition would trap migrating infected CD4⁺ T cells in secondary lymphoid tissues, allowing us to determine the differential role of vascular-restricted vs. tissue homing and resident CD8⁺ T cells in the control of SIV replication. We further hypothesized that SIV-specific CD8⁺ T cells trapped in LNs would acquire cytotoxic function in response to SIV reactivation and the performed immune-based interventions, potentially leading to enhanced viral control.

While FTY720 effectively redistributed migrating T cells into tissues, including both CD4⁺ and CD8⁺ T cells, the fully differentiated cytotoxic T cell subsets appeared to remain in the vasculature. Contrary to our expectations, we did not find increased accumulation of cytotoxic SIV-specific CD8⁺ T cells in LNs of FTY720-treated animals during ATI irrespective of the addition of αPD1 and N-803, despite activation and expansion within tissues. In addition, we observed a trend toward higher plasma viremia in FTY720-treated animals during ATI compared to controls, particularly for the group treated also with αPD-1 and N-803. Together, these results imply that tissue-localized CD8⁺ T cells cannot delay or prevent plasma viral rebound during ATI and suggest a direct role for vascular-restricted cytotoxic T cell subsets in the control of HIV/SIV.

Results

FTY720-Mediated Tissue Redistribution of Recirculating T Cells in SIV-Infected RMs.

To assess the ability of tissue homing and resident CD8⁺ T cells to directly control viral replication in tissues, we conducted an analytic treatment interruption (ATI) study in the context of FTY720 administration after barcoded SIVmac239M infection in n = 14 RMs (Fig. 1A). Peripheral blood, LN samples and rectal biopsies (RB) were collected at predetermined time points during the study, with additional LN biopsies at dynamic time points immediately following viral rebound during ATI. The window of LN sampling of the viral rebound timepoint (Vreb) in relation to the timing of rebound was between 1 and 6 d, depending on the animal, with an average of 3.9 d. All RMs had similar peak viral loads (10⁷ to 10⁸ viral copies/mL) and achieved rapid control of viremia after ART initiation at day 14 p.i. (Fig. 1B). As previously shown in ART-suppressed SIV-infected RMs (25, 26), FTY720 administration led to rapid and near complete redistribution of T cells from blood into tissues, with circulating CD4⁺ and CD8⁺ T cell numbers falling rapidly and remaining low throughout the FTY720 treatment period (CD4⁺ T cells: pre-FTY720 834 ± 304 vs. post-FTY720 13 ± 11; CD8⁺ T cells pre-FTY720 596 ± 284 vs. post-FTY720 77.2 ± 101; mean values ± SD, Fig. 1C, representative gating strategy shown in SI Appendix, Fig. S1). Using MHC class I tetramers, we further quantified the absolute numbers of Mamu-A*01 restricted SIV-Gag CM9 and Tat TL8 specific CD8⁺ T cells (27, 28). Previous studies indicated that these epitopes had not escaped during the first ATI (29). During ART, circulating SIV-specific cell frequencies remained low in all animals and increased in frequency in control animals after viral reactivation during ATI. No increase in circulating SIV-specific CD8⁺ T cells was observed in FTY720-treated RMs during ART or after ATI (Fig. 1C), suggesting SIV-specific CD8⁺ T cells that expand in response to viral rebound were maintained in lymphoid tissues in the treated RMs.

Fig. 1.

FTY720 reduces levels of circulating T cells in SIV-infected RMs during ART and at ATI. (A) Study design (i.v. = intravenous, p.i. = post infection, ART = antiretroviral treatment, pVL = plasma viral load – SIV RNA copies/mL, Vreb = viral rebound). (B) Left: Plasma SIVmac239M RNA levels during acute infection and first 6 mo of ART expressed as copies/mL (limit of detection = 60 copies/mL, dashed line) are shown for each individual animal from control (black lines) or RMs treated with FTY720 from 190 to 310 d p.i. (red lines). Right: Peak plasma viral load during acute infection for control and FTY720-treated RMs. Lines correspond to median values. (C) Absolute count of blood CD4⁺ T cells, CD8⁺ T cells, CM9- and TL8-specific CD8⁺ T cells shown as cells/μL of blood. (D) Representative flow plots of CD28 vs. CD95 staining to evaluate memory phenotype and absolute counts of blood naive, CM, and EM CD8⁺ T cells of control (Left) or FTY720-treated RMs (Right). (E) Representative flow plots of perforin versus granzyme B in FTY720-treated RM, pre- and posttreatment (Left). Expression of perforin, granzyme B, and Tbet (frequency and absolute count, middle and right, respectively) on circulating memory CD8⁺ T cells of control or FTY720-treated RMs overtime (n = 7 for each group, except at day 220 p.i. where FTY720-treated group n = 5; for absolute count n = 4 and n = 3 for Vreb). Fresh PBMC were used to assess CD4⁺, CD8⁺, and CM9-specific CD8⁺ T cell absolute counts and CD8⁺ T cell memory phenotype, and frozen PBMC for TL8-specific CD8⁺ T cell absolute count and perforin, granzyme B, and Tbet expression. Data are presented as the mean ± SD. Gray rectangles indicate time points under ART. Statistical differences in C and D were assessed by applying a two-way ANOVA with Dunnett’s multiple comparison tests (every time point vs. day 190 p.i.; black asterisks correspond to statistical analysis within the control arm, and red asterisks within FTY720 arm; n = 7 per group except Vreb where n = 3, which was excluded from the statistical analysis). *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

While redistribution of CD4⁺ T cells from blood into tissues was nearly absolute in FTY720-treated animals, a population of CD8⁺ T cells remained in blood during FTY720 treatment. To further characterize these nonmigrating CD8⁺ T cell populations, we examined T cell memory subset distribution using CD28 and CD95 expression patterns (Fig. 1D). FTY720 strongly decreased naïve and central memory (CM) CD8⁺ T cell counts but had limited effect on effector/effector memory (EM) CD8⁺ T cells, which constituted the majority of the remaining CD8⁺ T cells (Fig. 1D). In FTY720-treated animals, approximately 70% of the residual circulating CD8⁺ T cell population expressed perforin and granzyme B and 63% expressed the cytotoxicity-associated transcription factor Tbet after 4 mo of FTY720 administration, although the absolute count of these cell subsets in blood remained largely unchanged during treatment (mean values, Fig. 1E). We also observed an increase in the frequency of cytotoxicity-associated CX3CR1⁺ cells and a decrease in the frequency of tissue-trafficking CXCR3, CXCR5, and CCR7⁺ memory CD4⁺ and CD8⁺ T cells in the blood of FTY720-treated animals (SI Appendix, Fig. S2 A and B).

Expansion and Activation of SIV-Specific CD8⁺ T Cells in LNs of FTY720-Treated RMs After ATI.

To assess the impact of FTY720 on CD8⁺ T cells in tissues during FTY720 treatment, we measured the frequency of LN CD8⁺ T cells. By flow cytometry, we observed no change in the overall proportion of LN CD8⁺ T cells during FTY720 treatment compared to controls (SI Appendix, Fig. S3A). However, by immunohistochemistry (IHC) we observed an accumulation of CD8⁺ cells in the LN T cell zone of FTY720-treated RMs compared to controls after 120 d under FTY720 treatment (Fig. 2A). No change in CD8⁺ cell infiltration was observed in B cell follicles, nor did we detect differences between treatment groups in CXCR5 expression within memory CD8⁺ T cells (Fig. 2A and SI Appendix, Fig. S3B). While the LN CD8⁺ T cell memory subset distribution of control animals remained stable over time, EM CD8⁺ T cells increased in proportion during ATI, as did CM CD8⁺ T cells after ART reinitiation in FTY720-treated animals (SI Appendix, Fig. S3C). Increased proliferation, as measured by Ki67, was observed during ATI in LN CD8⁺ T cells for FTY720-treated RMs, but no differences over time or between groups were observed for LN bulk memory or germinal center (GC)-TFH CD4⁺ T cells (SI Appendix, Fig. S3 D–F).

Fig. 2.

FTY720 treatment accumulated CD8⁺ T cells and activated CM9-specific CD8⁺ T cells in lymph nodes and gut mucosa during ATI. (A) LN CD8⁺ cell IHC analysis during FTY720 treatment (n = 7 for each group, except day 190 p.i., Vreb and day 310 p.i. for FTY720-RMs n = 6). Inset showing the proportion of LN CD8⁺ cells per tissue area in the T cell zone at day 310 p.i. Representative IHC staining of T cell zone CD8⁺ cells at day 310 p.i is shown. C = control and F = FTY720. (B) Representative flow plots showing CM9-specific CD8⁺ T cell identification in LN samples by flow cytometry. Summary plot showing the proportion of CM9-specific cells in LNs of control or FTY720-treated RMs (n = 7 for each group, except day 280 and 310 p.i. where n = 6 for FTY720-treated RMs). Inset showing the proportion of CM9-specific cells in LNs of control or FTY720-treated RMs at day 310 p.i. Far Right: proportion of TL8-specific cells in LNs of FTY720-treated or control RMs (n = 7 for each group, except for day 190 and 310 p.i. where n = 6 for FTY720-treated RMs). (C) Frequency of Ki-67⁺ cells within CM9-specific CD8⁺ T cells from LNs of control or FTY720-treated RMs (n = 7 for each group, except day 280 and day 310 p.i. for FTY720-treated RMs where n = 6; and day 280 p.i. for control RMs where n = 6). (D) Left: Representative flow plots showing TNF-α vs. IFN-γ expression within memory CD8⁺ T cells after stimulation with GAG-SIVmac239 peptide pool. US = unstimulated, negative control. SEB = staphylococcus enterotoxin B, positive control. Right: Summary plots showing the proportion of CM9-specific cells either positive for IFN-γ, TNF-α and/or IL-2 within LN memory CD8⁺ T cells, of control or FTY720-treated RMs over time (n = 7 for each group, except day 190 p.i. and 310 p.i. where n = 6 for FTY720-treated RMs). Background-subtracted values are shown. (E) CD8⁺ T cell analysis from RB by IHC over time (Control RMs: d190, n = 5; d220: n = 3; d280, n = 5, d310, n = 3. FTY720-treated RMs: d190, n = 6; d220: n = 5; d280, n = 3, d310 n = 2). (F) Summary plot showing the proportion of CM9-specific cells in RB of control or FTY720-treated RMs over time (n = 7 for each group). (G) Proportion of Ki-67⁺ cells within CM9-specific CD8⁺ T cells of RB of control or FTY720-treated RMs (n = 7 for each group, except day 190 p.i. n = 6 and day 220 p.i. n = 5 for FTY720-treated RMs). Fresh cells from LNs and RB were used for flow cytometric analysis to analyze the frequency and phenotype of CM9-specific CD8⁺ T cells, and frozen samples were used to analyze the proportion of TL8-specific cells and cytokine production by LN SIV-specific CD8⁺ T cells. Gray rectangles indicate time points under ART. Data are presented as the mean ± SD, except when comparing between arms at a given time point, where lines correspond to median values. Statistical differences were assessed by applying a two-way ANOVA or mixed-effects model (if there were missing values) with Dunnett’s multiple comparison test for comparisons between time points within each arm (every time point vs. day 190 p.i.; when plotted on the same graph, black asterisks correspond to statistical analysis within the control arm, and red asterisks within FTY720 arm). The Mann–Whitney test was used to compare between arms at day 310 p.i (A and B Insets). *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

After viral reactivation during ATI, LN Gag CM9-specific CD8⁺ T cells increased in frequency for both control and FTY720-treated RMs and remained elevated after ART reinitiation, with FTY720-treated animals trending higher (Fig. 2B: day 310 p.i., control vs. FTY720-treated animals P = 0.051). In contrast, the proportion of LN Tat TL8-specific CD8⁺ T cells remained steady in both animal groups (Fig. 2B). Increased proliferation of LN Gag CM9-specific CD8⁺ T cells was observed in both animal groups during ATI, but at significantly higher levels in the FTY720 treated group (Fig. 2C). No differences were observed between groups regarding the expression of the coinhibitory receptors PD1, CTLA-4, and TIGIT on CM9-specific CD8⁺ T cells (SI Appendix, Fig. S3 G–I). LN CM9-specific CD8⁺ T cells also produced cytokines after ex vivo stimulation with GAG_SIVmac239 peptides at similar levels in both animal groups (Fig. 2D). Together, these data demonstrate that SIV-specific CD8⁺ T cells were present in the LNs of FTY720-treated animals, and showed increased frequency and activation compared to controls while retaining functional properties.

We further addressed the effects of FTY720 on SIV-specific CD8⁺ T cells in mucosal sites. We observed an increase in the proportion of total and EM CD8⁺ T cells in rectal tissue during ATI in both treatment groups (SI Appendix, Fig. S3 J and K). We also found a trend toward a higher proportion of CD8⁺ T cells during ATI and after ART reinitiation in rectal tissue follicular aggregates but not mucosa of FTY720-treated RMs (Fig. 2E); however, we could not conduct a statistical analysis due to several missing samples. The proportion of CM9-specific CD8⁺ T cells increased in rectal tissue during ATI for both treatment groups (Fig. 2F), with heightened proliferation selectively observed in the FTY720-treated group during the ATI (Fig. 2G).

Inhibition of Lymphocyte Migration Slightly Limits Control of Viremia During ATI.

FTY720 during ART was safe and well tolerated, and no viral reactivation was observed (Fig. 3A). However, following ATI, all animals rebounded at similar kinetics irrespective of FTY720 treatment (Fig. 3 A and B; Control: 12.9 ± 3.5 d; FTY720: 14.9 ± 5.3 d). Furthermore, while the peak plasma viral load (pVL) and area under the curve (AUC) during the entire initial ATI were similar between control and FTY720-treated RMs (SI Appendix, Fig. S4 A and B), control animals more effectively controlled peak viremia during the first ATI relative to acute phase peak pVL and had a lower viral set point (Fig. 3 C and D). Absolute quantification of viral DNA+ or RNA+ cells in LNs, measured by DNA and RNAscope, respectively, showed no significant changes or differences between groups at any time point analyzed. However, the SIV RNA signal was undetectable for several samples (SI Appendix, Fig. S4 C and D).

Fig. 3.

FTY720 administration did not show any therapeutic benefit on SIV infection. (A) Plasma SIVmac239M RNA levels during ATI #1. Thick lines correspond to geometric mean values for each arm. (B) Survival curve showing time to viral rebound during ATI#1 for control and FTY720-treated animals. (C) Log reduction of peak pVL during ATI#1 compared to acute infection. (D) Setpoint viral load during ATI#1. Lines correspond to median values. (E) Plasma SIVmac239M RNA levels during ATI#2 and after CD8⁺ T cell depletion with anti-CD8α antibody (two animals per group were euthanized at viral rebound (per protocol design, following Emory National Primate Research Center (ENPRC) guidelines), and anti-CD8α antibody was administered to the remaining five animals per group). Thick lines correspond to geometric mean values for each arm. (F) Survival curve showing time to viral rebound during ATI#2 for control and FTY720-treated animals. (G) Log reduction of peak plasma viral load during ATI#2 compared to acute infection. (H) Setpoint viral load during ATI#2. Lines correspond to median values. Statistical differences were calculated with the log-rank (Mantel-Cox) test (survival curves) and Mann–Whitney (peak pVL comparisons). *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

To assess the potential longer-term impact of FTY720 during treatment interruption on immune control and viral reservoir stability, we reinitiated ART in all 14 animals at 2 mo post ATI #1, reducing pVLs to undetectable levels within 1 wk of ART reinitiation (Fig. 3A). Daily dosing of FTY720 was continued for 1 mo following the reinitiation of ART in the 7 animals assigned to the treatment arm and ART was interrupted for a second time 3 mo after being reinitiated. Following this second ART interruption, two animals (one from each group) exhibited post-treatment control, but otherwise no delay in viral rebound was observed in animals previously treated with FTY720 compared to control (Fig. 3 E and F). Two animals from each group were euthanized as soon as they rebounded. The rest of the RMs underwent experimental depletion of CD8⁺ T cells with an anti-CD8 antibody and as expected, all animals demonstrated substantial viral load increases between 4 to 8 d after CD8⁺ T cell depletion, showing the relevance of CD8⁺ T cells in the control of SIV infection (Fig. 3E). Previous FTY720 treatment during the first ATI did not impact viral rebound levels or kinetics during the second ATI (Fig. 3F, SI Appendix, Fig. S4 E–G), and the reduction in the peak plasma viral load compared to acute infection as well as the viral set point was similar between arms (Fig. 3 G and H).

FTY720 Prevents Egress of CD8⁺ T Cells from Viremic Lymphoid Tissues but Does Not Enable the Acquisition of Cytotoxic Properties.

Given that FTY720 induced expansion and activation of SIV-specific cells but did not exert any beneficial control of the plasma viremia, we next examined whether forced tissue retention affected the cytotoxic properties of CD8⁺ T cells in lymphoid and mucosal sites. The frequency of perforin+ cells within LN total memory, CM9- and TL8-specific CD8⁺ T cells remained low compared to values observed in peripheral blood, and similar between control and FTY720-treated RMs, even after viral rebound during ATI (Fig. 4 A–C). We found no association between pVL and the phenotype of LN bulk memory and SIV-specific CD8⁺ T cells during ATI (SI Appendix, Table S1; and pVL values per animal at Vreb and day 280 p.i. are shown in SI Appendix, Fig. S4H). While at Vreb it is possible that there was insufficient time for the activation of CD8⁺ T cells to induce cytotoxicity, we were readily able to observe SIV-specific CD8⁺ T cell proliferation at day 280 p.i, in both control and FTY720-treated animals (Fig. 2C). Despite this, there was not specific accumulation of cytotoxic SIV-specific cells in LNs of FTY720-treated RMs compared to control (Fig. 4 B and C).

Fig. 4.

FTY720-enforced tissue retention does not enable acquisition of cytotoxic properties in CD8⁺ T cells. Longitudinal analysis of perforin expression of LN bulk memory (A), CM9-specific (B) and TL8-specific CD8+ T cells (C) compared to blood compartment during ATI (at viral rebound (Vreb) and day 280 p.i.). Fresh cells from LNs and PBMC were used for flow cytometric analysis of perforin expression except for TL8-specific CD8⁺ T cells analysis where frozen cells were used. Memory and CM9-specific CD8⁺ T cell analysis: n = 7 for each group, except for day 280 where n = 6 for FTY720-treated RMs and n = 4 for FTY-treated RMs at Vreb. TL8-specific cells analysis: n = 7 for each group. (D) Longitudinal analysis of perforin expression by IHC on LN B cell follicles and T cell zone. N = 7 for control and n = 6 for FTY720-treated RMs. (E) Data quantification of redirected killing assays at different target:effector ratios for LN CD8⁺ T cells from control and FTY720-treated RMs during ATI#1 (at Vreb), compared to spleen cells. Data were pregated on single live TFL4⁺ cells. N = 5 for each group. Frozen LNMC were used to assess T cell functionality. Representative dot plots of active Caspase-3 staining under different conditions are shown. The numbers represent the proportion of TFL4⁺ target active caspase-3⁺ apoptotic cells within single live TFL4⁺ cells. (F and G) Longitudinal analysis of perforin expression of bulk memory and CM9-specific CD8⁺ T cells of RB. Fresh samples were used for flow. N = 7 for each group. Gray rectangles indicate time points under ART. Lines in the plots correspond to median values. For the longitudinal analysis, statistical differences were assessed by applying a two-way ANOVA or mixed-effects model (if there were missing values) with Dunnett’s multiple comparison test (every time point vs. day 190 p.i.). For the pairwise comparison between ATI timepoints within each arm, Wilcoxon matched-pairs signed rank test was used; and the Mann–Whitney test was used to compare between LNs and blood during ATI. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

Due to LN sample availability, only the Vreb time point could be analyzed by IHC. There was no differential increase of perforin in the B cell follicles and the T cell zone, but we did observe a statistically significant small increase in the proportion of granzyme B+ cells in the B cell follicles of control RMs at viral rebound (Fig. 4D and SI Appendix, Fig. S5A). We further measured total cytotoxic potential during ATI using redirected caspase-3 based killing assays and observed comparably low cytotoxic activity for LN CD8⁺ T cells within control and FTY720 treated groups (Fig. 4E). RB analysis showed similar results with a slightly higher proportion of perforin+ memory but not Gag CM9-specific CD8⁺ T cells in FTY720-treated compared to control RMs during ATI (Fig. 4 F and G). IHC analysis showed no changes in the proportion of granzyme B+ cells in the mucosa of control and treated RMs (SI Appendix, Fig. S5B).

CD8⁺ T Cell Immunomodulation During FTY720 Treatment Is Insufficient to Enable Viral Control During ATI.

Previous studies have shown that reversal of immune exhaustion or enhancement of immune function via PD1 blockade (30, 31) or IL-15R SA N-803 (32), respectively, can improve T cell responses and viral control in HIV/SIV infection. To test whether these immunomodulators mediate their effects directly on T cells in LNs, we infected seven additional RMs with 10,000 IU of barcoded SIVmac239M i.v., treated with ART at d14 p.i. for 6 mo, and then administered FTY720 plus N-803 and a rhesusized αPD1 (combined treatment labeled as FNP) prior to and during ATI (Fig. 5A). The study design mirrored the control and FTY720-treated RMs arms to be able to compare between the three treatment groups. Peak plasma viral load during acute infection was similar among all animals within and between the different study arms, and all animals maintained undetectable plasma viral loads during ART (Fig. 5 B and C). One month prior to ATI, all animals began receiving daily doses of FTY720. One week later, and 3 wk pre-ATI, they received the first dose of N-803 and αPD1, with additional doses administered throughout the study as depicted in Fig. 5A. FTY720+N-803+αPD1 treatment was safe and well tolerated with the exception of a few instances of dermatitis and N-803 injection-site reactions. As expected, all seven RMs showed profound lymphopenia during FTY720 treatment (Fig. 5D), and the majority of the remaining CD8⁺ T cells in blood were EM with a cytotoxic profile, similar to the FTY720-only-treated RMs (Fig. 5E and SI Appendix, Fig. S6 A and B). Additionally, PD1 receptor occupancy remained high after αPD1 monoclonal antibody administration in most animals in both LN and blood CD8⁺ T cells, preventing the detection of the receptor by flow cytometry (Fig. 5F and SI Appendix, Fig. S7A). We also observed transient changes in circulating and LN T cells after N-803 and/or αPD1 administration, including increased Ki-67⁺ on memory and SIV- specific CD8⁺ T cells (Fig. 5G) and CD4⁺ T cells (SI Appendix, Fig. S7B) in the peripheral blood, and a minor increase in the proportion of proliferating memory CD4⁺ T cells in the LNs of FNP-treated RMs (SI Appendix, Fig. S7C). This trend was not observed in LN memory CD8⁺ T cells at the same time point, although some animals showed an increase in the proportion of proliferating LN CD8⁺ T cells (SI Appendix, Fig. S7D). GC-TFH CD4⁺ T cells also expanded during FTY720+αPD1 and N-803 administration, prior to ATI (SI Appendix, Fig. S7E). Furthermore, at viral rebound during ATI, the proportion of proliferating LN memory and GC-TFH CD4⁺ T cells was higher in FNP-treated RMs compared to control and FTY720-treated RMs (SI Appendix, Fig. S7 F and G). Together, these data indicate that the N-803 and αPD1 immunotherapies activated CD4⁺ and CD8⁺ T cells in the context of FTY720.

Fig. 5.

Redistribution of circulating T cells into tissues and proliferation of blood effector cells during combined FTY720, N-803, and αPD1 administration. (A) Study design (i.v. = intravenous, p.i. = post-infection, ART = antiretroviral treatment). Numbers on top of the arrows (days p.i.) correspond to time points where αPD1 and/or N-803 were administered. Numbers (days p.i.) below the timeline correspond to time points where blood and/or LN samples were taken for mononuclear cell analysis. (B) Plasma SIVmac239M RNA levels (pVL) during acute infection and first 6 mo of ART expressed as copies/mL (limit of detection = 60 copies/mL, indicated with a dashed line) are shown for each individual animal. (C) Summary plot showing the peak plasma viral load during acute infection for control, FTY720, and FTY720-αPD1-N-803-treated RMs. Lines correspond to median values. (D) Absolute count of blood CD4⁺ T cells and CD8⁺ T cells and (E) absolute count of naïve, CM, and EM CD8⁺ T cells in FTY720-N-803-αPD1-treated RMs at selected time points pre- and post-FTY720-administration. (F) Summary plots showing PD1 receptor occupancy in LNMC and PBMC CD8⁺ T cells over time. Arrows indicate αPD1 administration and numbers on top or below the arrows are days p.i. The 3/7 animals that showed intermittent control of the infection during ATI are depicted in brown. (G) Summary plot showing the changes in Ki67 (surrogate for proliferation) in peripheral memory CD8⁺ T cells and SIV-specific CD8⁺ T cells over time. Arrows indicate αPD1 and/or N-803 administration. Numbers below the arrows are days p.i.; Mean and SD values are shown. Cryopreserved mononuclear cells were used to assess T cell phenotype. Gray rectangles indicate ART. Statistical differences in (C) were calculated using the Kruskal–Wallis test with Dunn’s post comparisons versus control group; in (D–G) one-way ANOVA with Dunnett’s post comparisons every time point vs. day 185 p.i. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

Following ATI, all animals rebounded, with three animals (Fig. 6A, brown lines) achieving intermittent post-ATI control of viremia. Despite these animals demonstrating control, the FNP combination did not otherwise yield a beneficial effect as there was no delay in viral rebound, and we observed a trend toward higher peak viral loads compared to the control group (Fig. 6 B–D). Furthermore, the reduction in peak pVL during ATI compared to acute infection was larger for control compared to FNP-treated RMs, and the viral set point during ATI was similar to FTY720-treated RMs, although the difference was not statistically significant compared to control animals (Fig. 6 E and F).

Fig. 6.

FTY720, N-803, and αPD1 administration impacts plasma viral load during ATI. (A) Plasma SIVmac239M RNA (pVL) levels during ATI are expressed as copies/mL (limit of detection = 60 copies/mL, indicated with a dashed line). 3/7 animals that controlled the infection during ATI are shown in brown. (B) Survival curves showing time to viral rebound during ATI#1 for control, FTY720, and FTY720-N-803-αPD1-treated animals. (C) Plasma SIVmac239M RNA levels during ATI#1 for control and FTY720-treated RMs. For comparison purposes, pVL from equivalent timepoints of FTY720-N-803-αPD1-treated RMs were included in the summary plot. Thick lines correspond to the geometric mean pVL levels for each group. (D) Summary plot showing the peak pVL during ATI#1 for control, FTY720, and FTY720-αPD1-N-803-treated RMs. (E) Log reduction of peak pVL during ATI#1 compared to acute infection for each treatment group. (F) PVL setpoint. Lines correspond to median values. Statistical differences in (B) were calculated with the log-rank (Mantel-Cox) test; and in (D–F), using the Kruskal–Wallis test with Dunn’s post comparisons vs. control group. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

Finally, we evaluated whether αPD1 and N-803 treatment combined with FTY720-enforced retention increased the cytotoxic ability of LN CD8⁺ T cells using samples collected at various stages during the treatment period (Fig. 5A). While we observed an increase in the frequency of LN SIV-specific CD8⁺ T cells during ATI (SI Appendix, Fig. S8A), the proportion of LN cytotoxic SIV-specific CD8⁺ T cells remained low and constant over time by flow cytometry and imaging analysis (SI Appendix, Fig. S8 B and C). We also did not observe an increase in cytotoxic ability of total LN CD8⁺ T cells in redirected killing assays (SI Appendix, Fig. S8D). Together, these data suggest that αPD1 and N-803 are unable to modulate CD8⁺ T cell cytotoxicity directly within the LNs.

Discussion

While the role of HIV/SIV-specific CD8⁺ T cells in control of HIV/SIV is well established (8, 33–35), CD8⁺ T cells remain unable to fully eliminate the HIV/SIV reservoir, resulting in chronic viremia and disease progression in the absence of ART. This shortcoming may be due in part to impaired perforin/granzyme B-mediated cytotoxic CD8⁺ T cell activity in lymphoid tissues (17, 18, 36, 37), and the differential LN migration capabilities of early differentiated (naïve and CM) vs. late differentiated (cytotoxic effector) CD8⁺ T cells in the blood. While each of these mechanisms normally helps to protect the host from immune-mediated destruction of antigen-presenting cells, migration of infected CD4⁺ T cells into lymphoid tissues potentially allows evasion of peripheral cytotoxic CD8⁺ T cell surveillance. Even with lymphoid CD8⁺ T cell subsets that correlate with better viral control, such as those expressing both TCF-1 and CD39, their antiviral role in vivo seems independent from canonical cytotoxic functions (23).

Here, we tested this concept using FTY720 to inhibit T cell tissue egress in SIV-infected RMs, resulting in physical separation of the SIV-infected CD4⁺ T cells from cytotoxic SIV-specific CD8⁺ T cells in blood. In this scenario, the only CD8⁺ T cells that could encounter and potentially eliminate SIV-infected CD4⁺ T cells after ATI would be those trapped within the same tissues. We found that FTY720 was highly effective at preventing T cell migration in SIV-infected RMs, with nearly all T cells remaining in tissues except for late-differentiated cytotoxic CD4⁺ and CD8⁺ T cells. Importantly, LN and mucosal CD8⁺ T cells became activated after ATI, indicating target cell recognition, but did not acquire cytotoxic ability. It is also possible that the biopsy time points failed to capture cytotoxic T lymphocyte differentiation, if such events were transient or counterregulated. However, it is clear from our data that such responses cannot accumulate or be sustained if they do occur.

We observed no viral blips or T cell activation during FTY720 prior to ATI, indicating that general immune quiescence persists during this redistribution event. Nevertheless, following ATI, FTY720-treated animals demonstrated viral rebound with similar kinetics to controls and trended toward higher peripheral blood viremia, especially in animals treated with αPD-1+N-803 in combination with FTY720. Previous publications have shown a modest impact on viral control after FTY720 administration, at least in a subset of SIV-infected RMs (25, 26), however, FTY720 was administered during ART in these previous studies, not during ATI.

Based on the near absence of CD4⁺ T cells in the blood of the FTY720-treated RMs, plasma viremia in these animals must originate from infected cells in tissues and directly implicates either poor immunosurveillance or no immune-mediated clearance in these sites. Normally, infected CD4⁺ T cells migrate between the vasculature and tissues, subjecting a portion of the infected cells to vascular immune surveillance. However, in the FTY720-treated animals, nearly all infected cells became trapped in tissues and thus were not subject to vascular surveillance. Alternatively, sequestration into the tissues may have provided activation or proliferative cues to the CD4⁺ T cells, leading to increased viral replication. However, we did not observe increased activation or cell cycle entry (Ki-67) of bulk LN CD4⁺ T cells or GC-TFH CD4⁺ T cells after FTY720 administration prior to or after ATI, nor did we detect an increase in the frequency of infected cells in LNs of FTY720-treated RMs. It is also worth mentioning that we were unable to detect viral RNA in several LN samples at viral rebound after ATI. This could be due to limitations in the depth of RNAscope analysis for each tissue or that viral rebound/dissemination had not yet occurred in the sampled LN.

Taken together, these results could indicate that lymphoid tissue-CD8⁺ T cells alone are insufficient to delay or prevent plasma viral rebound during ATI, and further suggest that the little evidence of cytotoxic CD8⁺ T cell activity in tissues plus the physical separation of SIV-infected CD4⁺ T cells from the cytotoxic blood CD8⁺ T cells were the primary driver of the increased viral load in FTY720-only treated animals.

We did indeed observe an increase in the proportion of proliferating CD4⁺ T cells and GC-TFH in LNs of FNP-treated animals. Immunomodulation with αPD-1 or IL15R SA has individually been shown to induce CD4⁺ T cell activation (30, 31, 38, 39); thus, it is possible that the heightened viral load in the RMs treated with FTY720 and αPD-1+N-803 resulted in part from increased viral replication or target cell availability. However, limitations of this study include evaluating CD4⁺ and CD8⁺ T cell phenotype in gut mucosa and peripheral LNs, such as axillary and inguinal, but not in mesenteric LNs, which have been described as a major site of viral reservoir (40).

These results have significant implications for the development of viable strategies to engage CD8⁺ T cell–mediated immune clearance of tissue viral reservoirs. First, our findings suggest that vascular-restricted cytotoxic CD8⁺ T cells play a direct role in the control of viremia but require interaction with infected CD4⁺ T cells for their elimination within the vascular space. Infected CD4s in tissues must migrate into the blood for clearance through this mechanism, and strategies designed to mobilize infected cells into circulation may enhance reservoir clearance. Second, simply expanding the number of CD8⁺ T cells in lymphoid tissues does not enable control of viral replication after ATI, because tissue trafficking and resident T cells cannot efficiently eliminate infected cells due to limited cytotoxic differentiation. Furthermore, only a very small proportion of memory CD8⁺ T cells within lymphoid and mucosal tissues may acquire cytotoxic functions after exposure to renewed viral replication within the tissue. The reasons for this are currently unclear but may be due to both cell intrinsic and cell extrinsic regulatory effects. For example, TGF-β may negatively modulate cytotoxic properties in lymphoid tissues CD8⁺ T cells, as it was demonstrated that in vitro neutralization of TGF-β partially restored perforin expression in gut CD8⁺ T cells (36). Finally, despite evidence of PD1 blockade and IL-15R SA bioactivity within peripheral blood and lymphoid tissues, we did not observe increased CD8⁺ T cell cytotoxicity in tissues or evidence of viral control. These findings contrast with previous publications (30–32, 38), likely because FTY720 administration during the immunotherapy phase prevented the interaction between SIV-infected CD4⁺ T cells and cytotoxic SIV-specific CD8⁺ T cells in blood. We also did not find evidence of T cell exhaustion in tissues, as SIV-specific CD8⁺ T cells in LNs of FTY720-treated RMs became activated and expanded in vivo and produced cytokines upon stimulation in vitro. This further reinforces a previously unappreciated role for T cell trafficking and anatomical differences in T cell function in the control of HIV/SIV disease progression.

HIV cure-based strategies designed to engage CD8⁺ T cell–mediated elimination of HIV reservoirs in tissues have long been in development, including therapeutic vaccine, viral reactivation, exhaustion reversal, and chimeric antigen receptor (CAR) based modalities (41, 42). While these strategies have shown promise in vitro or in small animal models, immunological clearance of tissue reservoirs in HIV/SIV infection has thus far remained elusive. Here, we have identified a critical issue that undermines the potential of these strategies, specifically that most SIV-specific CD8⁺ T cells in LNs and gut tissues of SIV-infected RMs do not become cytotoxic after interaction with virally infected cells, even when provided state-of-the-art immunotherapy. As such, our results suggest that engagement of CD8⁺ T cell cytotoxic activity for eradication of tissue HIV reservoirs may require mobilization and reactivation of tissue-localized infected CD4⁺ T cells into the peripheral blood, or modification of the trafficking ability of cytotoxic CD8⁺ T cells in the peripheral blood to enable entry into tissue reservoir sites. These strategies, in combination with potent latency reversal agents, exhaustion reversal, and therapeutic vaccine-mediated expansion, may enable CD8⁺ T cell–mediated clearance of the HIV reservoir.

Materials And Methods

Animal Models and Study Design.

Indian-origin RMs (Macaca mulatta) were housed in an animal biosafety level 2 (BSL-2) facility at the Emory National Primate Research Center in Atlanta, Georgia as previously described (43). All animals were infected intravenously with 10,000 copies of SIVmac239M, a genetically tagged virus with a 34-base genetic barcode inserted between the vpx and vpr accessory genes of the infectious molecular clone SIVmac239 (44). Additional details of the study design are shown in Figs. 1A and 5A and further described in SI Appendix, Supplemental Materials and Methods. Some aspects of this study were previously reported elsewhere (29). All procedures were approved by Emory University Institutional Animal Care and Use Committee (IACUC) under permit PROTO201700655.

FTY720 Administration.

FTY720 (Sigma-Aldrich, CAS # 162359-56-0) was reconstituted in HPLC water to a concentration of 12.5 mg/mL and administered orally at a dose of 500µg/kg with food unless animals were anesthetized for another procedure, in which case FTY720 was delivered via an orogastric feeding tube.

Plasma Viral Load Quantification.

Plasma SIVmac239M loads were quantified at the Virology Core of the Emory Center for AIDS Research using a standard qPCR assay (with a limit of detection of 60 copies/mL) as described previously (45). Additional details can be found in SI Appendix, Supplemental Materials and Methods.

Fluorescence Cytometry and Intracellular Cytokine Staining.

Fresh or cryopreserved peripheral blood mononuclear cells (PBMC) and LN mononuclear cells (LNMC), and fresh gut-derived lymphocytes were used for flow cytometry analysis using previously described techniques (16). Nine-hour T cell stimulation assays examining the response to SIVmac239 Gag Peptide Pool (2 μg/mL; NIH AIDS Reagent Program, Division of AIDS, NIAID, NIH) were conducted from cryopreserved cells using previously published methodology (16). Additional details can be found in SI Appendix, Supplemental Materials and Methods. Data shown in the figures have been background-subtracted.

Immunohistochemistry and In Situ Hybridization.

Immunohistochemical staining and quantification were performed as described in SI Appendix, Supplemental Materials and Methods (46). To assess the location and number of cells harboring vRNA and vDNA across the study we performed highly sensitive and specific in situ hybridization RNAscope and DNAscope on two tissue sections per animal and per time point. RNAscope and DNAscope were performed on fully automated BOND-RX instrument following protocol previously described (47). Whole tissue slides were scanned using the Akoya Fusion microscope at 40X and were analyzed using the QuPath software. Cell segmentation was obtained using StarDist script on whole sections and region of interest were manually selected to define B cell follicles. The number of cells harboring vRNA or vDNA was reported per million nuclei for each region of interest.

Redirected Killing Assay.

Cytolytic activity of LN CD8⁺ T cells was evaluated using an adapted redirected killing assay for nonhuman primates as previously described (17) and in SI Appendix, Supplemental Materials and Methods. Killing capacity was calculated by subtracting the frequency of active caspase-3⁺ TFL4⁺ live/dead- P815 cells in target-only wells from the frequency of active caspase-3⁺ TFL4⁺ live/dead- P815 cells in wells containing effector cells.

Statistical Analysis.

Statistical significance of viral data between study groups was performed using Mann–Whitney or Kruskal–Wallis with Dunn’s multiple comparisons test vs. control group when comparing three groups in GraphPad Prism. Differences in time to viral rebound were assessed using the Log-rank (Mantel-Cox) test. Statistical significance of absolute cell count and immunophenotyping were assessed by applying a one- or two-way ANOVA or mixed-effects model (if there were missing values) with Dunnett’s multiple comparison test for comparisons between time points within each arm (every time point vs. day 190 p.i., unless stated otherwise). For immunophenotype pairwise comparisons at a given time point, Mann–Whitney test was used to compare values between two arms or between LNs vs. blood compartments, or Kruskal–Wallis with Dunn’s multiple comparisons test vs. control group was used when comparing three groups. A P value less than 0.05 was considered statistically significant. In the figures, asterisks were used to denote statistically significant differences (*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Study data is not publicly available. The raw flow cytometry files generated in this study can be provided upon reasonable request.