Changes to the Simian Immunodeficiency Virus (SIV) Reservoir and Enhanced SIV-Specific Responses in a Rhesus Macaque Model of Functional Cure after Serial Rounds of Romidepsin Administrations

ABSTRACT

HIV persistence requires lifelong antiretroviral therapy (ART), calling for a cure. The histone deacetylase inhibitor, romidepsin, is used in the “shock and kill” approach with the goal of reactivating virus and subsequently clearing infected cells through cell-mediated immune responses. We tested serial and double infusions of romidepsin in a rhesus macaque (RM) model of SIV functional cure, which controls virus without ART. Off ART, romidepsin reactivated SIV in all RMs. Subsequent infusions resulted in diminished reactivation, and two RMs did not reactivate the virus after the second or third infusions. Therefore, those two RMs received CD8-depleting antibody to assess the replication competence of the residual reservoir. The remaining RMs received double infusions, i.e., two doses separated by 48-h. Double infusions were well tolerated, induced immune activation, and effectively reactivated SIV. Although reactivation was gradually diminished, cell-associated viral DNA was minimally changed, and viral outgrowth occurred in 4/5 RMs. In the RM which did not reactivate after CD8 depletion, viral outgrowth was not detected in peripheral blood mononuclear cells (PBMC)-derived CD4⁺ cells. The frequency of SIV-specific CD8⁺ T cells increased after romidepsin administration, and the increased SIV-specific immune responses were associated, although not statistically, with the diminished reactivation. Thus, our data showing sequential decreases in viral reactivation with repeated romidepsin administrations with all RMs and absence of viral reactivation after CD8⁺ T-cell depletion in one animal suggest that, in the context of healthy immune responses, romidepsin affected the inducible viral reservoir and gradually increased immune-mediated viral control. Given the disparities between the results of romidepsin administration to ART-suppressed SIVmac239-infected RMs and HIV-infected normal progressors compared to our immune-healthy model, our data suggest that improving immune function for greater SIV-specific responses should be the starting point of HIV cure strategies.

IMPORTANCE HIV cure is sought after due to the prevalence of comorbidities that occur in persons with HIV. One of the most investigated HIV cure strategies is the “shock and kill” approach. Our study investigated the use of romidepsin, a histone deacetylase (HDAC) inhibitor, in our rhesus macaque model of functional cure, which allows for better resolution of viral reactivation due to the lack of antiretroviral therapy. We found that repeated rounds of romidepsin resulted in gradually diminished viral reactivation. One animal inevitably lacked replication-competent virus in the blood. With the accompanying enhancement of the SIV-specific immune response, our data suggest that there is a reduction of the viral reservoir in one animal by the cell-mediated immune response. With the differences observed between our model and persons living with HIV (PWH) treated with romidepsin, specifically in the context of a healthy immune system in our model, our data thereby indicate the importance of restoring the immune system for cure strategies.

INTRODUCTION

Antiretroviral therapy (ART) is one of the most effective therapeutic strategies of the 20th century, with a remarkable ability to suppress HIV replication and an exquisite impact on extending the life expectancy of persons living with HIV (1). However, ART is only virostatic, i.e., it does not actively eradicate proviruses from infected individuals, which persist in the latent reservoirs, calling for new therapeutic approaches aiming for an HIV cure (2). Over 10 years of intensive research, only two cases of HIV cure, defined as prolonged, complete viral remission and absence of detectable HIV proviruses in tissues after cessation of ART, were documented: the Berlin patient and the London patient (3, 4). Both occurred after allogeneic stem cell transplantations using donors homozygous for the CCR5 Δ32 allele (3, 4). Conversely, when allogeneic stem cell transplantation with intact CCR5 was performed, remission was only transient, as reported for the Boston patients (5), pointing to the importance of the use of a nonfunctional CCR5, although this is still not sufficient to guarantee viral remission (6). Furthermore, a case of very early and aggressive ART in a newborn (the “Mississippi baby”) also had some impact on the viral reservoir (illustrated by a significant delay in virus rebound after cessation of ART) but in the meantime demonstrated that ART alone cannot cure HIV (7, 8). Importantly, circulating CD4⁺ T cells from both the Boston patients and the Mississippi baby were negative for viral DNA at the time of treatment interruptions, yet recrudescence of infection occurred in all these subjects. The reason for these failures is that HIV has the ability to persist in rapidly established latent reservoirs which, upon ART cessation, reactivate and produce recrudescent infection (9–12). Studies of early reservoir dynamics in rhesus macaques (RMs) reported that even initiation of ART as early as 3 days postinfection (dpi), i.e., prior to detectable viremia, does not prevent reservoir seeding (13–15). Further, proviruses are found in resting CD4⁺ T cells (2, 9, 10, 16–18): central memory (19–21), transitional memory (20, 21), stem cell memory T cells (22), regulatory T cells (Tregs) (23), and follicular T helper CD4⁺ cells (24). Corroborated by the early reservoir establishment and the lack of a specific surface marker for latently infected cells (25), this explains why interventions to eliminate the latent reservoir are astonishingly onerous.

Multiple strategies to reduce/eradicate the latent reservoir have been proposed, yet with relatively modest success: (i) ART intensification (26–29), (ii) permanent transcriptional silencing of HIV (30), (iii) use of checkpoint inhibitors to enhance HIV-specific immune responses (31–34), (iv) gene editing of CCR5 (35), (v) genetic engineering of anti-HIV chimeric antigen receptor (CAR) T cells (36–38), (vi) bone marrow transplantation (5), (vii) broadly neutralizing antibodies (39, 40), (viii) Treg manipulation strategies (23), and (ix) “shock and kill.” However, these strategies have all experienced issues with efficacy or additional problems. For example, ART intensification does not reduce the reservoir size (41), nor do checkpoint inhibitors (31, 32), and Treg manipulation results in insufficient viral reactivation (42). Broadly neutralizing antibodies are susceptible to resistance and immune escape (43), while chimeric antigen receptor (CAR) T cells and gene therapies do not have long-term stability and lack sufficient efficacy (38, 44). Bone marrow transplantation has unacceptable toxicity and morbidity combined with no success beyond the Berlin and London patients (5, 45). Even shock and kill has had issues with a lack of reservoir reactivation and clearance (46–48), as well as the potential for detrimental immune activation (49, 50).

Over the last 10 years, the shock and kill approach was one of the most pursued avenues for an HIV cure, with multiple agents being tested as potential latency reversing agents (LRAs): histone deacetylase inhibitors (HDACis) (51–53), protein kinase C (PKC) agonists (54, 55), bromodomain inhibitors (JQ1) (56), second mitochondrial activator of caspases (SMAC) mimetics (57), stimulator of interferon genes (STING) agonists (58), Toll-like receptor (TLR) agonists (59–64), and ingenol derivatives (65–68) alone or in combinations (69, 70). Shock and kill operates on the premise that LRAs reverse viral latency, which allows for immune surveillance to recognize antigens of the reactivated virus and subsequently clear the infected cells. This strategy is performed in virus-suppressed individuals on ART, which prevents de novo infections of susceptible cells (71–74). The premise is that repeated cycles of virus reactivation and clearance would significantly curb/clear the HIV reservoir.

Theoretically, HDACi are strong candidates for latency reversal. Among genome modifications, deacetylation of histones at the integrated HIV proviral structure around the long-terminal repeats has been shown to inhibit transcription of the provirus by tightening the DNA around the histone, thereby driving the provirus toward latency (75–79).

The depsipeptide romidepsin (RMD) is a potent HDACi (80, 81) that has been extensively studied in shock and kill approaches and has been shown to be the most potent, both in vitro and ex vivo (82). In rhesus macaques (RMs), RMD reactivates latent SIV, with subsequent boosts in T-cell activation (83, 84); in humans, RMD reactivates latent HIV (85, 86). However, reports of the RMD impact on the cytotoxic T lymphocyte response to viral antigens yielded conflicting results (83, 87, 88). As such, RMD-induced cytotoxic T lymphocyte (CTL) suppression may have detrimental effects on the important “kill” step of this cure strategy (83, 85).

Here, we investigated the use of RMD as a latency-reversing agent in RMs infected with SIVsab, a nonhuman primate (NHP) model of functional cure (89, 90). We utilized repeated rounds of RMD infusion with the hypothesis that repeated infusions will progressively pare down the reservoir and enhance the SIV-specific immune response with each subsequent reactivation round.

RESULTS

Study design.

To assess RMD efficacy in reactivating the latent SIV, we used RMs that were infected with SIVsab (the virus that naturally infects African green monkeys). These infected RMs present with a phenotype of functional cure: to date, all animals infected with SIVsab have controlled virus to below detectable limits, recovered immune populations, and restored control of immune activation, while maintaining reactivation capacity through CD8⁺ T-cell depletion (89, 90). The ages, weights, and sexes of the animals are detailed in Table 1. The study design was as follows: all RMs received three rounds of RMD (7 mg/m²) through slow intravenous perfusion over 4 h, separated by at least 1 month. As no viral reactivation was observed after the second and third rounds of RMD administration in two RMs, the group was split, and the RMs showing no virus reactivation received CD8⁺ cell-depleting antibody (M-T807R1; 50 mg/kg). For the remaining three RMs, the first three rounds of RMD administration were followed by two rounds of “double dosing,” consisting of two RMD administrations (7 mg/m²) at 48 h interval, and finally, a single infusion due to an animal requiring euthanasia due to clinical signs of severe allergic reaction 1 day postinfusion. Since SIVsab is spontaneously completely controlled (<1 copy/mL of plasma) in 100% of RMs (89, 90), the model can be employed to study virus reactivation without ART. In this study design, with the animals not receiving ART, the reactivated virus can complete cycles of replication, which results in robust virus rebounds detectable with conventional viral load (VL) assays.

TABLE 1.

Sex, age, and weight of RMs in the study

Animal no.	Sex	Age (yrs)	Wt (kg)
RM85	Male	8	10
RM89	Male	8	11
RM94	Male	6	11
RM95	Male	6	11.5
RM100	Male	12	10

RMD reactivates latent SIV in RMs.

At each round of RMD administration, relatively robust virus rebounds occurred in RMs in the absence of ART (Fig. 1A). After the first round, SIV rebounded in all of the five RMs receiving RMD. The average rebound of virus reached 1,130 copies/mL and occurred between 7 and 28 days posttreatment (dpt), with the highest peak of 3,342 copies/mL occurring in RM89 at 26 dpt.

FIG 1.

Romidepsin treatment induces reactivation of latent SIV. Five rhesus macaques were treated with RMD and experienced viral reactivation postinfusion. (A) Subsequent infusions decreased reactivation potency, whereas the addition of a second infusion 48 h postinfusion (double infusion) increased viral reactivation. After no reactivation was observed in two RMs (RM85 and RM89), upper panels, CD8-depleting antibody M-T807R1 was administered. (B to D) CD8⁺ T-cell depletion (B) and CD4⁺ T-cell activation (C) in the M-T807R1-treated animals were confirmed, and plasma viral loads were measured (D). RMD, romidepsin; DI, double infusion; SCA, single copy assay.

After the second round of RMD administration, SIV rebounded in three of five animals; no detectable virus was observed in RM85 and RM89 at any time point after the second RMD administration. The average zenith of viral rebound was 504 vRNA copies/mL of plasma and occurred between 5 and 29 dpt, with the highest peak occurring in RM94 at 29 dpt (700 viral RNA (vRNA) copies/mL).

After the third treatment, SIV rebounded in the same three of five animals; again, no detectable viremia was observed in RM85 and RM89. The average of the virus rebound peaks was 4,304 copies/mL and occurred between 5 and 21 dpt, with the highest peak occurring in RM100 at 29 dpt (12,194 vRNA copies/mL). However, the massive reactivation observed in RM100 drastically increased the average rebound, whereas RM94 and RM95 peaked at 507 and 211 vRNA copies/mL, maintaining a decrease in reactivation relative to the prior RMD infusion, with RM85 and RM89 undetectable.

Because no detectable virus rebound was observed in RM85 and RM89 after two consecutive rounds of RMD administration (infusions 2 and 3), these RMs were excluded from further romidepsin infusions. The remaining three RMs (RMs 94, 95, and 100) received double infusions, i.e., two administrations of 7 mg/m² RMD at a 48-h interval to test whether the additional dosage would increase viral reactivation and immune activation. The conversion to double infusions increased peak viral reactivation in all three RMs at the fourth round of treatment (10,834 vRNA copies/mL), but these increases did not reach statistical significance (P > 0.05). We observed a similar trend to the single infusions whereby subsequent treatments yielded lower levels of viral reactivation at the fifth round of treatment. The peak viral rebound was 6,356 copies/mL for RM100, while RM95 rebounded up to 139 copies/mL, both drastic decreases from the round 4 peaks of 30,730 and 1,260 copies/mL, respectively. Interestingly, RM94 did not reactivate after the fifth round, but did have a viral blip of 52 copies/mL after the sixth round of RMD administration, while RM95 did not reactivate after the sixth round. However, the sixth RMD treatment was discontinued after the first infusion due to a severe allergic reaction that resulted in the euthanasia of RM100 at 1 dpt. Altogether, these rounds showed a tending decrease in the viral reactivation after subsequent infusions, although the changes in peak viremia at each infusion did not reach statistical significance (P > 0.05).

For RM85 and RM89, instead of continuing with RMD treatments after two rounds without detectable viremia, we proceeded to administer a CD8⁺ cell-depleting antibody (M-T807R1) to test for the presence of replication-competent virus. The CD8 depletion is used as a surrogate method for analytic treatment interruption in the model of functionally cured RMs and consistently results in dramatic viral reactivation (91), although its mechanism of action is different from that of analytic treatment interruption. Specifically, the loss of CD8-mediated suppression and activation of CD4⁺ T cells promotes reactivation of virus, as opposed to the removal of viral RT/integration blocks. With our model that has enhanced SIV control and does not use ART, this is a suitable alternative to analytic treatment interruption. After infusion of M-T807R1, CD8⁺ T cells were depleted by >99% (Fig. 1B), and a massive increase in the fraction of CD4⁺ T cells expressing K_i-67, a marker of cell proliferation, occurred (Fig. 1C). Plasma virus in RM85 rebounded up to 459,918 copies/mL at 33 dpt (Fig. 1D), mirroring the expression of K_i-67 in the CD4⁺ T cells. Surprisingly, RM89 did not reactivate virus above our limit of quantification (1 copy/mL) (Fig. 1D), even with the drastic increase of K_i-67 expression in CD4⁺ T cells.

RMD administration induces acute, transient hepatotoxicity.

We have previously shown that RMD infusion is associated with moderate acute toxicity. Similar to our previous study (88), markers of liver toxicity (alanine aminotransferase [ALT] and aspartate aminotransferase [AST]), kidney toxicity (urea nitrogen and creatinine), and general toxicity (lactate dehydrogenase [LDH] and creatine kinase [CK]) transiently increased post-RMD infusion. ALT and AST increased rapidly after each infusion, peaking within 2 days of infusion (at approximately 3 and 5 times the baseline levels for ALT and AST, respectively) and returning to near pretreatment levels by 7 dpt at each infusion (Fig. 2A and B). Interestingly, when switching to double infusions, we did not observe an additive effect on RMD toxicity (Fig. 2A and B). It is possible that the lack of creatine may reflect increased protein catabolism.

FIG 2.

Romidepsin treatment is associated with acute toxicity. (A to F) Serum from animals was used to determine levels of liver markers alanine aminotransferase (ALT) (A) and aspartate aminotransferase (AST) (B), kidney markers urea nitrogen (C) and creatinine (D), and general toxicity markers lactate dehydrogenase (LDH) (E) and creatinine kinase (CK) (F). The gray shaded area represents a reference range from uninfected rhesus macaques. RMD, romidepsin; DI, double infusion. n = 3.

Nephrotoxicity was minimal, as demonstrated by minimal increases in urea nitrogen (Fig. 2C), while creatinine remained virtually unchanged (Fig. 2D). When regimens were switched to double infusions, we did not observe a larger increase in urea nitrogen compared to single infusions but did see a prolonged elevation.

General toxicity markers LDH and CK increased postinfusion. LDH levels increased from baseline up to 2-fold (Fig. 2E). Switching to double RMD infusions did not result in further increases in the LDH levels (Fig. 2E). CK increased more dramatically (up to 20-fold compared to baseline levels) (Fig. 2F). Switching to double infusions did not result in more prominent increases compared to a single infusion.

Although one animal required euthanasia during the sixth round of treatment due to a severe allergic reaction, the increases in toxicity markers are in line with what we have seen in uninfected animals previously (88).

RMD induces transient loss of circulating lymphocytes.

Following the RMD treatments, the circulating CD3⁺ (Fig. 3A), CD4⁺ (Fig. 3B), and CD8⁺ (Fig. 3C) T-cell populations experienced substantial, yet transient, decreases, in line with our previously reported studies demonstrating an increase in cell homing and migration to the intestines and lymph nodes after RMD infusions (83, 88). These decreases were rapid, occurring as early as 1 dpt (Fig. 3), and were significant (P < 0.05) after rounds 1 and 2. RMD administration in double infusions resulted in slightly more pronounced reductions of the lymphocyte counts, but these differences did not reach statistical significance (P > 0.05) relative to the other lymphocyte nadirs (Fig. 3). Circulating lymphocytes were then rapidly restored close to pretreatment levels. In some animals (i.e., RM100) the rebounds of the circulating lymphocytes even largely exceeded the pretreatment levels (Fig. 3).

FIG 3.

Romidepsin treatment induces transient but pronounced loss of peripheral T lymphocytes. (A to C) RMs consistently experienced a transient loss of peripheral CD3⁺ (A), CD4⁺ (B), and CD8⁺ T lymphocytes (C) after romidepsin infusions. RMD, romidepsin; DI, double infusion. n = 5.

RMD administration induces significant increases of the activated T-cell fraction.

Because T-cell immune activation is related to both SIV reactivation and generating the cell-mediated immune responses, we used multiple flow cytometry biomarkers to assess the levels of T-cell activation. Both CD4⁺ T-cell and CD8⁺ T-cell fractions expressing the general immune activation marker CD69 mildly increased after each RMD infusion (Fig. 4A and B). There was a general trend toward an increase in the CD4⁺ T-cell fraction expressing CD69 within 1 dpt, followed by a rapid return to baseline (Fig. 4A). Yet, starting with round 3, we observed statistically significant increases in the CD69 expression by CD4⁺ T cells between 2 dpt and 40 dpt (P < 0.05). The fraction of circulating CD8⁺ T cells expressing CD69 was greater than that of CD4⁺ T cells (Fig. 4B), and the increases in immune activation tended to last longer for CD8⁺ T cells than CD4⁺ T cells (Fig. 4A and B). After switching to the double infusions, we did not observe an additive increase in CD69 expression on CD4⁺ or CD8⁺ T cells compared to single infusions.

FIG 4.

Romidepsin treatment induces immune activation in peripheral T lymphocytes. (A to F) RMs consistently experienced increases in three separate markers of immune activation: CD69 expression on CD4⁺ (A) and CD8⁺ T cells (B), HLA-DR⁺CD38⁺ coexpression on CD4⁺ (C) and CD8⁺ T cells (D), and K_i-67 intracellular expression on CD4⁺ (E) and CD8⁺ T cells (F). RMD, romidepsin; DI, double infusion. n = 5.

Unlike the CD69 expression, consistent increases of HLA-DR and CD38 coexpression on both CD4⁺ and CD8⁺ T cells were observed, albeit less robust for the CD4⁺ T cells (Fig. 4C and D). The increases in the T-cell fractions expressing HLA-DR and CD38 occurred later than those for CD69 and lasted longer (between 5 dpt and 19 dpt), peaking at 19 dpt. Similar increases were seen for rounds 2 and 3 (Fig. 4C and D). Yet, when RMD administration was switched to the double infusions, marked increases (P < 0.05) in the T-cell fractions coexpressing HLA-DR and CD38 occurred, which were more dramatic for the CD8⁺ T cells.

Finally, we monitored the fractions of the circulating T cells expressing the proliferation marker K_i-67. K_i-67 expression dramatically increased on both CD4⁺ and CD8⁺ T cells, peaked between 9 and 14 dpt, and returned to pretreatment levels after 19 dpt (Fig. 4E and F). These increases were statistically significant (P < 0.05). The switch to double infusions resulted in a prolonged increase in K_i-67 expression, with similar trends being observed for both CD4⁺ T cells and CD8⁺ T cells (Fig. 4E and F).

RMD administration decreases the short-term functionality of SIV-specific immune response.

As we have shown previously that RMD results in a diminishment of CD8⁺ T-cell reactivity and even cytokine production without stimulation (88), we were interested in the changes in RM85 and RM89. Because these two RMs did not have viral reactivation after the second and third treatments, we investigated whether RM85 and RM89 had different modifications to the short-term CD8⁺ T-cell responses than historic animals, which may explain the differences in viral reactivation, e.g., no abrogation of the SIV-specific responses. However, after RMD administration, both animals experienced decreases in the total cytotoxic production of their CD8⁺ T cells, as illustrated by the total fraction of CD8⁺ T cells expressing effector molecules (e.g., gamma interferon [IFN-γ], macrophage inflammatory protein 1β [MIP-1β], interleukin-2 [IL-2], tumor necrosis factor alpha [TNF-α], and the surrogate killing marker CD107a [LAMP-1]) after Env and Gag peptide pool stimulation (Fig. 5). Between the two animals, there were differences in the initial cytokine production by the CD8⁺ T cells, such that the pretreatment expressing cytotoxic effector molecules decreased from 1.09% at 0 dpt to 0.68% at 7 dpt for RM85 and 1.90% at 0 dpt to 0.93% at 7 dpt for RM89 when stimulated with the Gag2 pool. Further decreases were seen at 9 dpt, when the frequency of SIV-specific CD8⁺ T cells decreased to roughly half of the baseline levels: to 0.47% in RM85 and to 0.81% in RM89. At 29 and 34 dpt, functionality varied between the two RMs. The frequency of SIV-specific CD8⁺ T cells further decreased in RM85 to 0.35% at 29 dpt, with subsequent recovery to 1.02% at 34 dpt. In RM89, the frequency of SIV-specific CD8⁺ T cells increased at 29 dpt (1.21%) but then decreased at 34 dpt (to 0.26%). Similar trends were seen with the Env and Gag1 peptide pools, with decreases at 7 and 9 dpt followed by recovery to near the pretreatment levels at later time points (see Fig. S1 in the supplemental material). Thus, the short-term changes in SIV-specific immune responses in these two animals still demonstrated decreased reactivity after romidepsin infusion, in line with what was previously seen (88).

FIG 5.

Romidepsin treatment is associated with acute loss of CD8⁺ T-cell functionality. (A and B) After the second round of treatment, PBMCs from RM85 (A) and RM89 (B) were stimulated with Gag peptide pools (peptides 69 to 136) and assayed for IFN-γ (I), MIP-1β (M), IL-2 (2), TNF-α (T), and CD107a (LAMP-1) (7). Total percentages are presented inside the circles, while color coding of the rings indicates whether 1, 2, 3, 4, or 5 effector molecules are being produced. n = 2.

Repeated RMD treatments boost SIV-specific immune responses.

Because RMD was administered serially, we believed that subsequent infusions with viral reactivation would result in enhanced SIV-specific immune responses, contrary to the short-term changes. We have previously shown that repeated RMD administration does not decrease histone acetylation (83). Therefore, we stimulated PBMCs taken prior the first infusion and 1 month after the first, second, and third infusions with Env, Gag1, and Gag2 peptide pools to assess the cell-mediated immune response against SIV for all 5 RMs with pooled percentages (Fig. 6). We found that after treatment with the Env peptide pool (Fig. 6), the percentage of CD8⁺ T cells expressing effector molecules related to the cell-mediated response (e.g., IFN-γ, MIP-1β, IL-2, TNF-α, and CD107a) increased from preinfusion at each round tested. At baseline, the average percentage of CD8⁺ expressing effector molecules after stimulation was 0.21 ± 0.08%. After round 1, it was 0.90 ± 0.28%, which then increased to 0.93 ± 0.26% and 1.78 ± 0.54% after rounds 2 and 3, respectively. Similar to Env stimulation, with Gag1 and Gag2 stimulation, we observed an increase from baseline after rounds 1 and 3 (Fig. 6). In fact, cytokine expression increased from 0.60 ± 0.24% at preinfusion to 0.89 ± 0.23%, 0.54 ± 0.21%, and 1.84 ± 0.44% after rounds 1, 2, and 3, respectively, with the Gag1 peptide pool. CD8⁺ T cell stimulation with the Gag2 peptide pool increased their expression of functional effector molecules from 0.87 ± 0.37% at preinfusion to 1.41 ± 0.35%, 0.56 ± 0.18%, and 1.79 ± 0.48% following infusions 1, 2, and 3, respectively (Fig. 6). However, these increases in SIV-specific immune responses did not reach significance (P > 0.05).

FIG 6.

Repeated romidepsin treatments are associated with bolstered CD8⁺ T-cell functionality. Prior to treatment and around 30 days after each round, PBMCs from the RMs were stimulated with peptide pools of Env, Gag1 (peptides 1 to 68), and Gag2 (peptides 69 to 136) and stained for IFN-γ (I), MIP-1β (M), IL-2 (2), TNF-α (T), and CD107a (LAMP-1) (7). Total percentages of CD8⁺ T cells producing effector molecules are presented inside the circles ± the standard error of the mean (SEM), while color coding of the rings indicates whether 1, 2, 3, 4, or 5 effector molecules are being produced. n = 5.

SIV-specific immune responses are associated with decreased viral rebound.

Interestingly, the two animals that did not reactivate after two consecutive rounds of single-infusion RMD (RM85 and RM89) had greatly increased CD8⁺ T-cell reactivity compared to RM94, RM95, and RM100, such that even the preinfusion time points were at 0.42 ± 0.23%, 1.25 ± 0.88%, and 2.05 ± 1.5% for the Env, Gag1, and Gag2 peptide pools, respectively (Fig. 7). Meanwhile, RM94, RM95, and RM100 were at 0.069 ± 0.03%, 0.17 ± 0.08%, and 0.086 ± 0.05% for the Env, Gag1, and Gag2 peptide pools, respectively (Fig. 8). Further, the percentages of reactive SIV-specific cells after infusions were also higher in RM85 and RM89, providing a potential explanation for the viral control in these two animals. Of RM94, RM95, and RM100, RM94 had the greatest increases: from 0.15% to 0.78% and 1.9% for pretreatment, round 3, and round 5, respectively, as an average of the three peptide pools. Interestingly, RM95 after round 6 also had a large increase to 1.3% of CD8⁺ T cells expressing effector molecules and did not reactivate virus. However, RM100 had its highest CD8 values after round 3, 0.95%, and reactivated virus nonetheless.

FIG 7.

Repeated romidepsin treatments increase CD8⁺ T-cell functionality. RM85 and RM89 were stimulated with Env, Gag1 (peptides 1 to 68), and Gag2 (peptides 69 to 136) peptide pools and assayed for IFN-γ (I), MIP-1β (M), IL-2 (2), TNF-α (T), and CD107a (LAMP-1) (7). Total percentages are presented inside the circles, while color coding of the rings indicates whether 1, 2, 3, 4, or 5 cytokines are being produced. n = 2.

FIG 8.

Repeated romidepsin treatments increase CD8⁺ T-cell functionality and maintains it after six rounds of treatment. RM94, RM95, and RM100 were stimulated with Env, Gag1 (peptides 1 to 68), and Gag2 (peptides 69 to 136) peptide pools and assayed for IFN-γ (I), MIP-1β (M), IL-2 (2), TNF-α (T), and CD107a (LAMP-1) (7). Total percentages are presented inside the circles, while color coding of the rings indicates whether 1, 2, 3, 4, or 5 cytokines are being produced. n = 3; round 6, n = 2.

RMD treatment diminished the replication-competent reservoir in one RM.

Using longitudinal sampling after each round of RMD administration, we tracked changes in cell-associated SIV DNA (CA-vDNA). Samples were taken from time points closest to 30 days postinfusion to avoid potential interference from active viremia. Although, on average, there was a small decrease in CA-vDNA over time in RMs receiving RMD, individual animals had variable responses to treatment (Fig. 9A). Thus, three RMs (RM85, RM89, and RM95) experienced a decrease in the CA-vDNA levels over the course of RMD treatments, CA-vDNA levels increased in RM94, and finally, in RM100 the CA-vDNA levels remained virtually unchanged during the follow-up. After the first RMD treatment, a decrease in the viral burden was observed in three out of five RMs, with the viral reservoir decreasing below the detection threshold for RM89. However, RM89 did contain detectable viral DNA after the other rounds of treatment, nonetheless remaining below its baseline level of 170 copies/10⁶ cells. Further, RM85 had a large increase in viral DNA after the CD8 depletion, most likely due to reseeding of the reservoir as a result of the massive viral reactivation. After the switch to double infusions, increased CA-vDNA copies were observed for all three RMs (RMs 94, 95, and 100), with a subsequent decrease after round 5. CA-vDNA decreased again after round 6 in RM95, but not in RM94 (Fig. 9A). These changes in CA-vDNA were not statistically significant.

FIG 9.

Romidepsin infusion in a rhesus macaque model of SIV functional cure does not yield consistent decreases in cell-associated viral DNA or replication-competent virus. (A) Cell-associated vDNA was quantified by qPCR at preinfusion and 1 month postinfusion for each round. (B and C) Viral outgrowth was measured by p27 intracellular staining after coculture of negatively selected CD4⁺ T cells with MOLT-4 cells at 1 month postinfusion for each round of treatment (B) and in tissues at necropsy (C). Dashed lines in panel A are baseline CA-vDNA levels for each animal. n = 5.

Due to the inherent limitations of using CA-vDNA to measure the reservoir (92), we further characterized the dynamics of the reservoir in RMs treated with RMD in a viral outgrowth assay. CD4⁺ T cells isolated from PBMCs through negative selection were cocultured with MOLT-4 clone 8 cells, which are susceptible to SIVsab infection, after stimulation with phorbol myristate acetate (PMA) and ionomycin. The PBMCs were taken at time points closest to the start of the following RMD administration when no viremia was detected to allow for normalization after potential reactivation and immune activation. After 17 to 19 days in coculture, the presence of intracellular p27 in MOLT-4 cells was used as the determining marker of viral outgrowth. While viral reactivation decreased during the latter rounds of RMD treatment, the presence of replication-competent virus did not complementarily decrease as expected. Instead, only two RMs (RM89 and RM94) demonstrated a marked decrease in the levels of replication-competent virus (Fig. 9B). RM89 did not reactivate virus after the second or third RMD rounds or following CD8⁺ cell depletion. However, only after CD8⁺ T-cell depletion did the amount of replication-competent virus significantly decrease, with 0/10 CD4⁺ T-cell cocultures staining positive for intracellular p27 (P < 0.01). Similarly, RM94 displayed viral outgrowth in only 2/7 CD4⁺ T-cell cocultures after the fifth round of treatment (P < 0.05), indicating a decrease in the levels of replication-competent proviruses. However, after round six, 5/5 CD4⁺ T cell cocultures presented with viral outgrowth. Given the changes in viral reactivation, we further investigated necropsy-collected tissues for RM89, RM94, and RM95 which presented with RMD infusions that did not result in plasma reactivation. RM89 and RM95 did not reactivate prior to necropsy, yet in the necropsy samples, virus outgrowth occurred in both animals, as well as in RM94, which had a viral spike of 52 copies/mL at 16 dpt of round 6, pointing to the presence of replication-competent proviruses within the superficial and mesenteric lymph nodes (LNs), jejunum, and spleen (Fig. 9C), with only the jejunum of RM95 showing a potentially diminished viral reservoir at 3/5 wells positive for p27.

DISCUSSION

In this study, by using an SIV/RM model of spontaneous functional cure and healthy immune responses, we found that multiple rounds of RMD administration bolstered the immune responses, gradually decreased viral reactivation, and in one animal, diminished the circulating reservoir of replication-competent virus. Our results also show that RMD administration has the capacity to alter the reservoir in our model, resulting in a discrepancy in the levels of replication-competent proviruses between blood and tissues.

In this model, in which functional cure is achieved spontaneously, without ART, the viral rebound observed after RMD administration was due not solely to viral reactivation, but also to de novo infections, which had an additional effect of boosting the healthy immune response against SIV. Though the data were variable, the SIV-specific immune responses were slightly hindered in the days immediately following treatment, yet not significantly, in agreement with previous data demonstrating a nonsignificant decrease in CD8⁺ T-cell functionality directly after RMD administration (88). Nonetheless, the longitudinally bolstered SIV-specific immune responses post-RMD infusion were likely due to the virus reactivation, allowing for an additional strong immune response to potentially activate new naive CD8⁺ T cells or to induce additional expansion of SIV-specific CD8⁺ T cells. The concept of improved viral control with repeated reactivation events is supported by increased viral control of HIV after structured analytic treatment interruptions which allowed for plasma viremia to become detectable prior to reinitiating ART (93). Further, the increase in T-cell activation after each treatment, despite undetectable viremia, points to another possible reason for the observed decreased viral reactivation after RMD, whereby CD8 cells were better primed to react to any virus-producing cell. In fact, the fraction of CD8⁺ T cells coexpressing HLA-DR and CD38 correlates with increased HIV-specific responses in PWH (94). It is thus not surprising that every subsequent round of RMD administration did not elicit the same reactivation as the preceding rounds. This points to the critical role of the cellular immune responses for virus control, off and on ART (95), or after viral reactivation. This conclusion is further supported by the case of one animal (RM85), in which repeated RMD administrations gradually abolished virus reactivation, yet CD8⁺ cell depletion led to a massive virus rebound, in the range of the viral levels observed during the acute infection.

Upon investigation into the cell-associated viral DNA, we found that the CA-vDNA copies per million PBMCs changed in three different manners. RM100 was virtually unchanged throughout the infusions, with only a small increase in CA-vDNA after a large reactivation event at the fourth infusion. RM94 initially increased after the first two infusions of RMD but then decreased at each round toward baseline, concomitant with decreases in viral reactivation events after infusions. Finally, three of the five RMs maintained decreased CA-vDNA through round 3 for RM85 and RM89 and round 6 for RM95. RM85 and RM89 showed discrepant responses after CD8⁺ T-cell depletion: RM85 greatly reactivated virus, and CA-vDNA increased past baseline after depletion, while RM89 had minimal changes in CA-vDNA after CD8⁺ T-cell depletion. The increase for RM85 is unsurprising due to the nature of the study in which the animals are ART-naive; thus, the viral reactivation after CD8⁺ T-cell depletion for RM85 likely initiated de novo infections, thereby expanding the reservoir and increasing the CA-vDNA. Although the CA-vDNA is not a perfect marker, there are trends between the changes in CA-vDNA and viral replication: massive or prolonged virus reactivation resulted in increased levels of CA-vDNA (e.g., RM100 after the fourth round of infusions and RM94 after the second round of infusions). Further, when subsequent reactivation events were diminished, particularly beginning after the second infusion, CA-vDNA was decreased in RM94 and RM95. Finally, RM85 and RM89 did not reactivate after the second and third infusions, and their CA-vDNA was diminished relative to baseline. Interestingly, RM89 had a large and prolonged reactivation after the first RMD infusion and resulted in undetectable CA-vDNA, which matched the massive increase in SIV-specific CD8⁺ T-cell functionality and polyfunctionality measured after the first infusion.

While the CA-vDNA indicated small changes in the size of the viral reservoirs in functionally cured SIVsab-infected RMs treated with RMD, the PCR-based assays do not take into consideration the replication competency of the viral sequences amplified. To address this issue and assess the replication competence of the virus in the reservoirs, we utilized a flow cytometry-based viral outgrowth assay (14, 96, 97). This assay showed a loss of replication-competent proviruses in the peripheral blood of both RM94 and RM89 after round 5 and CD8 depletion, respectively. By comparing viral loads and outgrowth with the SIV-specific CD8⁺ T cells, our data strongly implicate CD8⁺ T cells in the viral control. (i) The functionality of the SIV-specific CD8⁺ T cells greatly increased in both these RMs in later rounds of RMD administration; (ii) after round 6, RM95 produced more effector molecules and did not reactivate but still had viral outgrowth; (iii) after round 6, RM94 had a much lower percentage of SIV-specific CD8⁺ T cells and reactivated after RMD treatment; (iv) RM85 did not reactivate after the second or third infusion but reactivated after CD8⁺ T-cell depletion. Thus, these results support the conclusion that CD8⁺ T cells are playing a large role in the control of virus and may have cleared a portion of the circulating reservoir.

An alternative explanation is that NK cells play a substantial role in the functional cure. M-T807R1A does not discriminate between CD8⁺ T cells and NK cells, due to the presence of CD8α on NK cells (95), resulting in partial depletion.

Meanwhile, after experimental depletion, recovering CD8⁺ cells yielded a significantly greater fraction that were major histocompatibility complex (MHC)/Gag tetramer-positive (98). As a result, they may provide greater suppression, establishing a stranglehold on infected cells and preventing viral expansion or eliminating cells that may have escaped previous detection by the NK cells. Yet, this also puts a damper on the SIV-specific CD8⁺ T-cell response assays after CD8 depletion, as these cells may be artificially boosted through this mechanism, and the true reactivity to the antigens may be lower if given more time for the cells to stabilize MHC/Gag.

Viral outgrowth was observed from the gut and LN samples from RM89, RM94, and RM95 at necropsy. The observed discrepancies between the lack of inducible proviruses in CD4⁺ T cells and outgrowth from tissues in RM89 and RM94 may be due to the different immunophenotypes of the CD4⁺ T cells in blood (where the pool comprises more naive cells) and the intestine (where the pool comprises more effector and memory cells). Note also that the gut comprises 60% of the T lymphocytes in the body (99, 100). Additionally, it has a higher viral burden, both DNA and RNA, than the blood (92, 101–103), contributing up to 83 to 95% of the HIV-infected cells in humans, with different mechanisms driving latency in the gut versus the blood (104). We also reported that RMD drives CD3⁺ T-cell homing to the gut and LNs in RMs (88), which could further diminish the peripheral SIV reservoir if tissue-homed infected cells do not reenter circulation. Altogether, these rationales may explain the presence of replication-competent virus in the gut and LNs, while no viral outgrowth can be detected peripherally. It also can explain how RM94 decreased in viral reactivation and outgrowth after the fifth round of infusions yet still reactivated the virus after the sixth RMD round, as there was likely reentry of reservoir cells into the circulation from tissues. The mechanism by which tissue sites harbor replication-competent virus but do not reactivate after RMD or M-T807R1 is currently not known. We have previously demonstrated that the RMD half-life in the intestine and LNs is longer than in blood (88). However, in patients with cutaneous T-cell lymphoma the compartmental response of the LNs to RMD is not as strong as the blood or skin responses (105). Thus, the overall effects of RMD may be diminished in the gut and LN, supporting the presence of replication-competent virus at these sites.

Nonetheless, the discrepancy between blood and tissue that we document here is an essential piece of data, which is also supported by previous case studies. The two famous cases of the Mississippi baby (7) and the Boston patients (5) focused upon measurements of the levels of virus in PBMCs and plasma to determine whether the patients were functionally cured. However, as demonstrated in these cases (106–108), the lack of viremia and cell-associated DNA/RNA from plasma and PBMCs at the cessation of ART, and even months later, was not predictive of virus rebound. In our cases the obvious explanation is the presence of replication-competent virus in tissues. Note, however, that the posttreatment controllers in the Visconti study, which is one of the largest cohorts of functionally cured patients, had detectable HIV DNA in PBMCs yet achieved sustained viral remission (109). As such, our results suggest that the functional cure should be predicted by testing tissue samples, in addition to circulating cells and plasma.

A novel use of the double infusion regimen in RMs 94, 95, and 100 showed further benefits. We demonstrated that the double infusions did not extend the delay for immune recovery, and there was no significant increase in toxicity. Although one animal was euthanized, this was due to a severe allergic reaction and not specific toxicity from RMD. Importantly, we identified a prolonged, increased CD8⁺ T-cell activation after double infusions, which resulted in the greatest increase in SIV-specific CD8⁺ T-cell functionality for these three animals. This is not surprising, as the enhanced CD8⁺ T-cell activation likely bolstered responsiveness to antigen.

An important novelty of our study is the use of the SIVsab-infected RM model of functional cure. In this model, the virus is naturally controlled by healthy immune responses, the immune populations are restored after an acute infection that inflicts a significant damage on the immune system, and chronic immune activation and inflammation are at baseline levels (89, 90). Recently, two papers have reported cases of elite controllers with detection of only defective proviruses (110, 111), with one claiming a possible sterilizing cure (111). Interestingly, a prospective mechanism of action for the slow clearance of virus is through preferential elimination of proviruses that are more permissive to transcription, which the authors deemed “autologous shock and kill” (110). Thus, this model presents a unique opportunity to test experimental agents in a similar background. The fact that the virus is controlled off ART allows, in the case of the reactivation with LRAs, the reactivated virus to complete cycles of replication. The resulting de novo infections bolster viral production, gleaning better resolution of viral reactivation. Furthermore, the model also provides the unique perspective of how experimental agents may alter the reservoir in the context of a restored immune system relative to other SIV-infected models. While one may argue that such a model is not illustrative for the HIV-infected individuals on ART, the advantage of performing reactivation studies in a system with a healthy immune system is that it allows us to focus on the effects on virus reactivation. As such, it is possible that the diminishment of viral reactivation and changes in ca-vDNA observed in our model are due to an autoviral vaccination mechanism, by which the repeated reactivation events allow for expanded SIV-specific responses via antigen recognition. This further eliminates transcriptionally permissive proviruses and may work to expedite the previously mentioned “autologous shock and kill” (110). Because of these caveats, our data do not necessarily promote the use of RMD, but instead suggest that resolving immune dysfunction prior to latency reactivation in PWH is likely to result in better therapeutic outcomes. Given the acute decreases in T-cell reactivity immediately after RMD infusion in both our study and others (87, 88) and the lack of virological effects in PWH (112, 113), RMD is not likely to be the LRA of the future. Instead, the inclusion of immunological therapies in shock and kill, such as IL-15 superagonists, IL-21, or TLR agonists, may improve outcomes. In fact, recent studies on N-803 with broadly neutralizing antibodies (bNAbs) (114) and TLR agonists with bNAbs (115) in nonhuman primates have shown the potential for establishing posttreatment control, further supporting this concept.

In conclusion, we report that repeated RMD administrations resulted in a gradual diminishment of viral reactivation in a model of spontaneous functional cure of HIV infection. Further, the switch to RMD double infusions also increased viral reactivation with the same pattern of subsequent diminishment, and surprisingly, further boosted the SIV-specific immune responses. Remarkably, no viral reactivation was observed in an animal in which CD8⁺ cells were experimentally depleted in vivo. Compounded with the lack of viral outgrowth from PBMCs, this feature provides evidence for a successful reduction of replication-competent SIV in circulation. Further testing in tissues, however, presented viral outgrowth, suggesting that the correlates of the HIV remission must be investigated beyond blood to conclude a cure/functional cure. Nevertheless, these data provide a glimpse of potential viral control through the cell-mediated immune response when in an environment of a healthy immune response. The associations between greater CD8⁺ T-cell activity and decreased viral reactivation and the eventual loss of viral outgrowth from PBMCs in one RM thereby suggest that when deploying HIV cure strategies, the first aspect to tackle should be the restoration of the immune system and reversion of T-cell exhaustion in chronically infected PWH, thus setting the scene for better control and clearance.

MATERIALS AND METHODS

Ethics statement.

All animals were housed and manipulated at the Plum Borough Research Facility of the University of Pittsburgh, following the guidelines of the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC) and the Animal Welfare Act (116). The University of Pittsburgh approved these experiments with the Institutional Animal Care and Use Committee (IACUC) protocols 15045866 and 18042404. All RMs included in this study were socially housed (paired) indoors in stainless steel cages, had a 12-h/12-h light cycle, and were fed twice daily, and water was provided ad libitum. A variety of environmental enrichment strategies were employed, including housing of animals in pairs, providing toys to manipulate, and playing entertainment videos in the animal rooms. In addition, the animals were observed twice daily, and any signs of disease or discomfort were reported to the veterinary staff for evaluation. For sample collection, animals were anesthetized with 10 mg/kg ketamine HCl (Park-Davis, Morris Plains, NJ, USA) or 0.7 mg/kg tiletamine HCl and zolazepan (Telazol, Fort Dodge Animal Health, Fort Dodge, IA, USA) injected intramuscularly. The animals were sacrificed by intravenous administration of barbiturates prior to the onset of any clinical signs of disease.

Animals and treatments.

Five RMs (Macaca mulatta) were intravenously (i.v.)-infected with 100 tissue culture infectious infusion of SIVsab92018 and allowed time to control infection. The ages, weights, and sexes of the animals are included in Table 1. Two of these RMs received 3 rounds of RMD (7 mg/m²) through slow intravenous perfusion over 4 h. These animals then received a CD8⁺ cell-depleting antibody (M-T807R1; 50 μg/kg) due to lack of viral reactivation upon the RMD treatments. The remaining three RMs received three rounds of RMD under the same conditions as the first group, followed by two rounds of double infusions of RMD, i.e., two infusions (7 mg/m²) through slow intravenous perfusion over 4 h separated by 48 h, and a final single infusion due to an animal requiring euthanasia due to clinical signs of severe allergic reaction one dpt.

Cell separation from whole blood and tissues.

PBMCs were separated from whole blood as described previously (83, 117). Briefly, blood was centrifuged for 20 min at 2,200 rpm to separate the plasma, which was subsequently removed and frozen. Blood was layered over lymphocyte separation medium (LSM; MPBIO, Solon, OH, USA), and layer separation was achieved by centrifugation for 20 min at 2,200 rpm. The PBMC layer was pipetted off and washed with 1× phosphate-buffered saline (1× PBS; Lonza, Basel, Switzerland). Washed PBMCs were then counted and frozen at 5 million cells/mL using freezing medium (95% fetal bovine serum [FBS; VWR, Radnor, PA, USA], 5% dimethyl sulfoxide [DMSO; Thermo Fisher Scientific]), with 2 million cells dry-frozen.

Lymphocytes were separated from LNs and spleen by extensive mincing of the tissue, followed by pushing through a 70-μm cell strainer, as previously described (118). Cells were frozen the same way as the PBMCs.

Lymphocytes were separated from intestinal sections at necropsy as previously described (119). Briefly, tissues were trimmed of fat and cut open longitudinally. The blunt, plastic safety covering of the safety scalpel was used to gently remove excess mucus and waste from the tissues. The gut sections were then cut into 1-cm² pieces and incubated twice in 8 mM EDTA solution (Fisher Scientific, Pittsburgh, PA, USA) with Hanks’ balanced salt solution (HBSS; Lonza) for 30 min at 37°C on an orbital shaker at 300 rpm. Tissue pieces were then cut to one-half to one-quarter size and incubated with the same conditions in an RPMI 1640-collagenase solution (0.75%; Sigma-Aldrich, St. Louis, MO, USA). The resulting cell suspension was layered over Percoll (Sigma-Aldrich) gradients of 60% and 35% and spun at 2,200 rpm for 20 min. Cells were frozen the same way as the PBMCs.

Flow cytometry.

Whole blood or cells separated from LNs or intestinal tissue were stained with antibodies for flow cytometry, as described (120). BD Trucount (BD Biosciences) was utilized to quantify absolute cell counts for CD3⁺ T cells as described (120). CD4⁺ and CD8⁺ T-cell absolute counts were determined by acquired percentages of CD3⁺ T cells from other flow cytometry panels. Antibodies used for staining are listed in Table 2. Intracellular staining for K_i-67 was completed as previously described (121). Acquisition was completed on the LSR-II flow cytometer (BD Biosciences), and data were analyzed with FlowJo software (Treestar, Ashland, OR, USA).

TABLE 2.

Antibodies used in the study

Antibody	Clone	Fluorophore	Manufacturer
CD3	SP34-2	V450	BD Biosciences
CD4	L200	APC	BD Biosciences
CD8	RPA-T8	PE-CF594	BD Biosciences
CD14	M5E2	PE-Cy7	BD Biosciences
CD20	2H7	APC-H7	BD Biosciences
CD25	2A3	PE	BD Biosciences
CD28	CD28.2	PE-Cy7	BD Biosciences
CD38	AT-1	FITC	Stemcell
CD39	eBioA1	PE	Invitrogen
CD45	D058-1283	PerCP	BD Biosciences
CD69	FN50	APC-Cy7	BD Biosciences
CD73	AD2	PerCP-Cy5.5	BD Biosciences
CD95	DX2	PE-Cy5	BD Biosciences
CCR4	1G1	PE-Cy7	BD Biosciences
Ki-67	B56	PE	BD Biosciences
HLA-DR	L243	PE-Cy7	BD Biosciences
Annexin V	-	FITC	BD Biosciences
NKG2a	Z199	PE	Beckman Coulter
FoxP3	259D	AF488	BioLegend
LIVE/DEAD	-	UV	Life Technologies

SIV-specific immune response assay.

The SIV-specific immune responses were assayed as previously described (42, 83, 122). Briefly, frozen PBMCs were thawed, counted, and either unstimulated or stimulated with the Env, Gag1 (peptides 1 to 68), or Gag2 (69 to 136) peptide pools (NIH Reagent Program). After 2 h, cells were treated with monensin and brefeldin A to prevent acidification of endocytic vesicles and exocytosis of cytokine-containing vesicles, respectively. PBMCs were then stained and acquired on an LSR Fortessa flow cytometer (Becton, Dickinson, New Jersey, USA). Flow cytometry data were analyzed with FlowJo version 10.6.1 (Treestar, Ashland, OR, USA).

Viral quantification.

SIV plasma VLs were measured using quantitative reverse-transcription PCR (qRT-PCR), as described (120, 122, 123). Briefly, RNA was extracted from plasma with the RNeasy minikit as per the manufacturer’s instructions (Qiagen, Valencia, CA, USA). RNA was then reverse transcribed with a Superscript III first-strand synthesis supermix for qRT-PCR kit (Invitrogen, Carlsbad, CA, USA) and run on a SimpliAmp thermal cycler (Applied Biosystems, Foster City, CA, USA). TaqMan gene expression master mix (Applied Biosystems) was added to the resulting cDNA with primers and probes as previously described and run on either an ABI 7900HT (Applied Biosystems) or QuantStudio 7 (Applied Biosystems) system.

Viral outgrowth assay.

Inducible, replication-competent virus was detected by intracellular staining of SIV p27 in MOLT-4 clone 8 cells (obtained through the NIH HIV Reagent Program, Division of AIDS, NIAID, NIH; MOLT-4 cells, clone 8, ARP-175, contributed by Ronald Desrosiers [124]) cocultured with PBMCs, as described (14, 96, 97). Briefly, CD4⁺ T cells were negatively selected from PBMCs according to the manufacturer’s instructions (CD4⁺ T-cell isolation kit; Miltenyi Biotec, Auburn, CA, USA). CD4⁺ T cells were then stimulated with PMA (50 ng/mL) and ionomycin (1 μg/mL) for 6 h, washed, and plated at 100,000 to 250,000 cells per well, depending on cell availability, with 100,000 MOLT-4 clone 8 cells. Cocultures were followed for 17 to 19 days, at which time the cells were harvested, washed, and stained for CD3-V450, CD4-APC, and intracellular SIV p27-FITC (ARP-2321) conjugated to FITC using the FluoReporter fluorescein isothiocyanate (FITC) protein labeling kit according to the manufacturer’s instructions (Life Technologies) at 100 ng/replicate. ARP-2321 was obtained through the NIH HIV Reagent Program, Division of AIDS, NIAID, NIH and Centre for AIDS Reagents, NIBSC, UK, supported by EURIPRED (EC FP7 INFRASTRUCTURES-2012–INFRA-2012-1.1.5; grant number 31266; www.nibsc.org/). Anti-simian immunodeficiency virus (SIV) SIVmac gp41 monoclonal (KK41), and ARP-2321 (CFAR number 3005) were contributed by Karen Kent and Caroline Powell. Cultures positive for SIV p27 were designated as having ≥0.1% SIV p27 in the MOLT-4 populations.

Statistics.

Graphical and statistical analyses were completed with Prism version 9.1.2 (GraphPad Software, Inc., San Diego, CA, USA). Statistical significance was defined as P < 0.05. To analyze immune cell dynamics and phenotypes and toxicities, a mixed-effects model with Geisser-Greenhouse correction and Tukey’s multiple-comparison test was used due to the difference in time points between some of the animals. To compare differences between the percentage and total SIV-specific responses producing one or more effector molecules at a time as well as changes in viral reactivation, we utilized the Friedman test with Dunn’s multiple comparisons. To compare the cell-associated DNA and viral outgrowth between rounds of treatment, we utilized the Kruskal-Wallis analysis of variance (ANOVA) with Dunn’s multiple comparisons.