Robust and persistent reactivation of SIV and HIV by N-803 and depletion of CD8+ cells

Abstract

Human Immunodeficiency Virus (HIV) persists indefinitely in antiretroviral therapy (ART)-treated individuals due to a reservoir of latently-infected cells harboring replication-competent virus. To better understand the mechanisms responsible for latency persistence and reversal, we used the interleukin-15 superagonist N-803 in conjunction with CD8+ lymphocyte depletion in ART-treated Simian Immunodeficiency Virus (SIV)-infected macaques. While N-803 alone did not reactivate virus production, its administration after CD8+ lymphocyte depletion induced the most robust and persistent virus reactivation ever observed in vivo under ART, with viremia >60 copies/mL in 14/14 animals (100%) and 41/56 samples (73.2%) collected each week after N-803 administration. Strikingly, concordant results were obtained in ART-treated HIV-infected humanized mice. In addition, we found that co-culture with CD8+ T-cells blocked the in vitro LRA effect of N-803 on primary human CD4+ T-cells latently-infected with HIV. These results advance our understanding of the mechanisms responsible for latency reversal and lentivirus reactivation during ART-suppressed infection.

Summary paragraph:

Human Immunodeficiency Virus (HIV) remains a major global health problem with ~1.1 million deaths worldwide annually¹. Despite the major declines in morbidity and mortality associated with the use of antiretroviral therapy (ART), there is still neither a vaccine nor a cure for HIV infection. The inability to eradicate HIV infection with current therapies is due to the presence of latently-infected cells harboring integrated replication-competent virus which persist indefinitely in HIV-infected individuals undergoing ART and contribute to rebound viremia when therapy is discontinued (i.e., the “viral reservoir”)^2–5. A key paradigm in the field of HIV cure, referred to as “shock and kill”^6,7, supposes that induction of virus expression (i.e., “virus reactivation”) in these latently-infected cells (i.e., “shock”) followed by immune-mediated clearing (i.e., “kill”) may substantially reduce the reservoir size and possibly lead to a functional cure for HIV infection. Unfortunately, no latency-reversing agent (LRA) tested to date has successfully perturbed the viral reservoir in human clinical trials. In particular, histone deacetylase (HDAC) inhibitors failed to induce either robust virus reactivation or reduction of the viral reservoir in ART-treated HIV-infected individuals^8–13. More encouragingly, in Simian Immunodeficiency Virus (SIV)-infected ART-treated rhesus macaques (Macaca mulatta) treatment with Toll-like receptor 7 (TLR7) agonists was linked to transient blips of plasma viremia¹⁴. However, this result was not reproduced in further studies^15,16. More recently, persistent remission was observed in a subset of ART-treated simian/human chimeric immunodeficiency virus (SHIV)-infected macaques receiving GS-9620 in combination with the broadly neutralizing PGT121 antibody¹⁷. In all, these published data indicate that novel, more potent approaches for latency reversal are needed to achieve a functional cure for HIV infection.

SIV and SHIV infection of rhesus macaques

Infection of rhesus macaques with SIV or SHIV is the most widely used animal model to study the mechanisms by which the viral reservoir is established and maintained under ART, and to pre-clinically test interventions aimed at reducing the viral reservoir in vivo¹⁸. In a previous study, we demonstrated that depletion of CD8+ lymphocytes in SIV-infected ART-treated macaques was consistently followed by increased plasma viremia, thus indicating that these cells contribute to viral suppression under ART¹⁹. While the precise mechanisms responsible for this observation remain unclear, phylogenetic analysis of the rebounding virus suggested that silencing of virus transcription contributes to this antiviral effect. Based on these observations, we hypothesized that CD8+ lymphocyte depletion may combine with LRAs to enhance virus production under ART. As shown in Fig.1A, the IL-15 superagonist N-803 is a complex of a mutant IL-15 and a dimeric IL-15 receptor αSu/Fc fusion protein²⁰. The engineered structure is at least 25-times more biologically potent than IL-15 as it mimics trans-presentation and the IgG-Fc component confers improved in vivo safety and bioavailability^21,22. In the setting of ART-suppressed lentiviral infection, N-803 may target the residual virus pool due to its ability to act in vitro as a potent LRA and to strengthen the antiviral immune responses mediated by T and natural killer (NK) cells²³.

Figure 1 |. Study design and phenotypic/transcriptomic effects of N-803 with or without CD8 depletion in rhesus macaques.

a, IL-15 Superagonist N-803 structure. b, Study design. At intervention phase, green arrows designate 100 μg/kg N-803 administration and blue arrows designate 50 mg/kg MT807R1 administration. c, Plasma viral load pre-intervention (n=35 macaques), including infection and initiation of antiretroviral therapy (gray bar). Limit of detection is 60 copies of SIV RNA/mL of plasma (black bar). d, Mean peripheral CD4+ T-cell (maroon), CD8+ T-cell (purple), and NK cell (gray) count and e, percentage of CD4+ and CD8+ T-cells in the lymph node, and f, Ki67 expression in cellular subsets post-intervention with N-803 (n=7 biologically independent samples). g, Ki67 expression in bulk CD4+ T-cells following N-803 alone (green, n=7 biologically independent samples), CD8 depletion alone (blue, n=14 biologically independent samples), and CD8 depletion with N-803 administration (red, n=14 biologically independent samples). Day 3 was included in peripheral blood analyses. h, Gene set enrichment analysis (GSEA) of RNA sequencing data from bulk CD4+ T-cells comparing gene sets enriched on day 3 post-intervention with N-803 alone (green, n=7 biologically independent samples), CD8 depletion alone (blue, n=7 biologically independent samples), or CD8 depletion with N-803 (red, n=7 biologically independent samples). Normalized enrichment scores for select upregulated gene sets are depicted, where normalization is group specific. A normalized enrichment score cutoff of greater than 1.35 for upregulated gene sets with a false discovery rate of less than 0.2 was used, in accordance with GSEA guidelines. i, Heat map detailing enriched genes in bulk CD4+ T-cells in the IL-2/STAT-5 signaling gene set after administration of N-803 alone (n= 7 biologically independent samples). Heat map colors represent log2 transformed library size normalized read counts scaled to unit variance across transcript vectors and normalized to the baseline median sample value of each transcript. Sample means are indicated (±SEM), and two-sided Kruskal-Wallis tests (d, f) and Friedman tests (e,g) were used to compare post-intervention values to pre-intervention baseline and approximate P value summaries are provided.

The current study included a total of 35 SIV-infected macaques that started ART at day 56 post-infection and were treated for at least one year prior to any further intervention. The animals were divided in three groups as follows (Fig.1B): seven macaques were treated with four weekly doses of 100 μg/kg of N-803 (group 1, N-803 alone), fourteen macaques received one dose of the CD8 depleting antibody, MT807R1 (anti-CD8α) at 50 mg/kg iv (group 2, CD8 depletion alone), and fourteen macaques received four weekly doses of N-803 starting at the time of CD8 depletion (group 3, CD8 depletion with N-803). After reconstitution of CD8+ T-cells (defined as >100 CD8+ T-cells/μL of blood), seven macaques in groups 2 and 3 received four additional weekly administrations of N-803. Peripheral blood samples, lymph node and rectal biopsies were collected at various time points before, during, and after these interventions. All macaques underwent analytical treatment interruption at week 3 after either CD8+ T-cell reconstitution or the last N-803 treatment. As shown in Fig.1C, all animals showed suppression of viremia after one year of ART, with plasma viral loads below the detectable limit of our standard assay (60 copies/mL of plasma)²⁴ at the time of the additional interventions in 33/35 animals (94.3%). We also measured residual plasma viremia using a more sensitive assay (limit of detection of 3 copies/mL²⁵) at three monthly sampling points prior to the interventions. Viremia was <3 copies/mL in 19/35 ART-treated macaques (52.3%), with 26/35 animals (74.3%) showing levels of residual viremia ≤10 copies/mL at the time of intervention (Extended Data Table 1). These results indicate that the level of virus suppression observed in our cohort of macaques was in most cases comparable to that of long-term ART-treated HIV-infected individuals^26,27,28.

As shown in Extended Data Fig.1A–D and consistent with previous studies^19,29, treatment with anti-CD8α MT807R1, with or without N-803, depleted on average 99.1% of CD3+CD8+ T-cells in peripheral blood, 97.9% in lymph nodes, and 99.5% in rectal biopsies. In addition, treatment with MT807R1 alone depleted 93.2% of NK cells in peripheral blood (Extended Data Fig.1E). As expected based on previous studies, N-803 administration alone resulted in an expansion of CD8+ T-cells in the blood and lymph nodes (Fig.1D–E), as well as increased proliferation of peripheral CD8+ T-cells, CD4+ T-cells, and NK cells (Fig.1F). Of note, while CD8 depletion alone did not result in a rapid increase in CD4+ T-cell proliferation (as measured by Ki67 expression), the combination of CD8 depletion and N-803 led to a significant increase in CD4+ T-cell proliferation (Fig.1G; Extended Data Fig.2R–V). The frequency of CD4+ T-cells co-expressing Ki67 and the HIV/SIV coreceptor CCR5, i.e., potential target cells for infection, was significantly increased across all groups by the third week after intervention (Extended Data Fig.2L). Additionally, CD8 depletion with or without N-803 administration resulted in the expansion of effector memory CD4+ T-cells and increased PD-1 expression on CD4+ T-cells (Extended Data Fig.2F–G).

To better characterize the biological effects of N-803, we conducted a transcriptional analysis using RNA-Seq of sorted CD4+ T-cells collected prior to intervention and day 3, week 2, and week 4 after the start of intervention. Regardless of concurrent CD8 depletion, N-803 induced significant upregulation of gene sets associated with cell cycling/proliferation, activation, antiviral responses, and cell signaling (Fig.1H). Specifically, we observed a significant enrichment for genes in the IL-2/STAT-5 signaling gene-set, which is also indicative of IL-15 signaling as the receptor for this cytokine shares two out of three subunits with IL-2 and uses STAT-5 as the key adaptor molecule (Fig.1I). In addition, we examined the expression of 25 genes specifically involved in the host-virus interaction during SIV infection and found that N-803 administration induced a consistent and transient upregulation of APOBEC3 in CD4+ T-cells, CD8+ T-cells, and NK cells (Extended Data Fig.3A–C).

As shown in Fig.2A and Extended Data Table 1, administration of N-803 was not associated with an increase of plasma viremia >60 copies/mL in any of the treated animals, indicating that the IL-15 superagonist is not sufficient to exert an in vivo LRA effect in ART-treated SIV-infected macaques when used alone. As expected based on previous studies¹⁹, macaques undergoing CD8 depletion showed a moderate but significant increase in virus production, with plasma viremia >60 copies/mL detected in 11/14 animals (78.6%) and 18/56 samples (32.1%) collected weekly after CD8 depletion (Fig.2B). Viremia >1,000 copies/mL was observed in 2/14 animals (14.2%) and 2/56 (3.6%) of the same samples (Fig.2B). In all cases, the level of virus production returned to below 60 copies/mL of plasma at the time of CD8+ lymphocyte reconstitution (Fig.2E). Overall, the level of increased viremia observed in this study was consistent with previous studies¹⁹, even though the magnitude of virus production post-CD8 depletion was somewhat less dramatic, possibly related to the longer period of ART treatment (12 months vs. 2–8 months)¹⁹. Most remarkably, macaques treated with N-803 during CD8 depletion showed robust and persistent levels of virus production, with viremia >60 copies/mL detected in 14/14 animals (100%) and 41/56 samples (73.2%) and viremia >1,000 copies/mL observed in 6/14 animals (42.9%) and 13/56 samples (23.2%) (Fig.2C). We emphasize that all seven macaques with full suppression of virus production at the time of intervention with CD8 depletion and N-803 administration (i.e., repeated viral load measurements <3 copies/mL of plasma) demonstrated clear virus reactivation with detectable levels in 26/28 time points one week after each N-803 administration, and >60 copies/mL in 16/28 of the same samples (Fig.2D). Similar results were obtained in a smaller pilot study in which N-803 administration during CD8 depletion was performed in five ART-suppressed SHIV_SF162P3-infected macaques (Fig.2G). To the best of our knowledge, the level of viremia observed in SIV-infected macaques treated with combined CD8 depletion and N-803 administration during long-term ART is the highest and most persistent ever described, as compared to the results of previous “shock and kill” cure strategies tested in humans and nonhuman primates^8–14.

Figure 2 |. SIV and HIV reactivation following CD8 depletion with N-803.

Plasma viral loads following intervention with a, N-803 alone (n=7 macaques), b, CD8 depletion alone (n=14 macaques), and c, CD8 depletion with N-803 (n=14 macaques). d, Longitudinal plasma viral loads macaques with fully suppressed viral load (<3 copies/mL of plasma) prior to CD8 depletion with N-803 administration. Comparison of viral load pre-intervention (PRE), post-intervention when CD8+ T-cells are <100 cells/μL blood (POST), and during CD8+ T-cell reconstitution >100 cells/μL blood (CD8 return) in macaques treated with e, CD8 depletion alone (n=14 macaques), and f, CD8 depletion with N-803 (n=14 macaques). Sample means are indicated (±SEM). g, Plasma viral loads following CD8 depletion with N-803 administration in SHIV_162P3-infected macaques after six months of ART (n=5 macaques). Viral suppression <60 copies/mL represented via black bar (a-d,g) or dashed line (e-f) and the limit of detection for these assays was 3 copies/mL plasma. h, RNAscope determination of the percentage of SIV RNA+ lymph node cells expressing high levels (>4 copies) of viral RNA/cell one week after CD8 depletion with N-803. Plasma viral loads of HIV-infected, ART-treated humanized mice treated with i, N-803 alone (green, n=7 mice), j, CD8 depletion alone (blue, n=8 mice), and k, CD8 depletion with N-803 administration (red, n=8 mice) (limit of detection 346 copies/mL). Statistical significance was calculated using a two-sided Kruskal-Wallis test (e-f) or Wilcoxon test (h). Macaque animal code key provided in Extended Data Fig.5.

After the last treatment with N-803, the level of viremia rapidly declined coincident with the reconstitution of CD8+ T-cells, and all macaques returned to <60 copies/mL by week 6 after CD8 depletion with N-803 administration (Fig.2C+F). As expected, CD8+ T-cell reconstitution was faster in CD8-depleted macaques co-treated with N-803 (Extended Data Fig.5E–G), due to the effect of the IL-15 superagonist in enhancing CD8+ T-cell proliferation (Fig.1D–F).

We next investigated the correlates of virus reactivation in ART-treated SIV-infected macaques that received CD8 depletion with N-803 administration, and observed that the post-depletion viral load (day 3 through week 6) was negatively correlated to the frequency of peripheral CD8+ T-cells (Extended Data Fig.5B). Additionally, the area-under-the-curve (AUC) of virus production during CD8 depletion and N-803 administration was directly correlated with pre-intervention viremia (Extended Data Fig.5D). Of note, no correlation was found between the level of virus production after CD8 depletion and/or N-803 treatment and either the size of the peripheral blood DNA reservoir (measured as fraction of SIV DNA+ CD4+ T-cells) or the level of CD4+ T-cell activation (measured as Ki67 or PD-1 expression on CD4+ T-cells) (data not shown). To assess whether combined CD8 depletion with N-803 administration induced SIV production in lymphoid tissues, we next analyzed the levels of SIV-RNA in lymph nodes using the RNAscope technology at pre-intervention and day 7 post-intervention in five representative macaques receiving the combination of CD8 depletion with N-803 administration. As shown in Fig.2H, and consistent with the measurements of plasma viremia, we found a statistically significant increase in the percentage of SIV-RNA+ cells with high levels of SIV-RNA following intervention. No changes were observed in the level of SIV-RNA in peripheral CD4+ T-cells (Extended Data Fig.4A–C), suggesting that lymphoid tissues are the main source of the observed increased viremia after combined CD8 depletion with N-803 administration.

HIV infection of humanized mice

To confirm the virus reactivation induced by CD8+ lymphocyte depletion with N-803 administration in SIV-infected ART-treated macaques in an in vivo model utilizing HIV, we next conducted a similarly designed experiment using bone marrow-liver-thymus (BLT) humanized mice (hu-mice) infected with HIV-1_JR-CSF and treated with ART. As shown in Fig.2I–K, HIV-infected hu-mice showed strikingly similar results to those obtained in SIV-infected macaques, with no plasma virus reactivation following administration of N-803 alone, a moderate level of virus reactivation following CD8 depletion alone, and a robust level of virus reactivation involving 7/8 (87.5%) hu-mice undergoing CD8 depletion with N-803 administration. Furthermore, we also noted a statistically significant increase in the levels of cell-associated HIV-RNA in the spleen and thymic organoid of hu-mice receiving CD8 depletion with N-803 administration (Extended Data Fig.6B).

CD8+ T-cells inhibit latency reversal in vitro

The combined data obtained in SIV-infected macaques and HIV-infected hu-mice indicate that the strong virus reactivation activity attributed to N-803 is revealed only in the absence of CD8+ T-cells, thus delineating a novel mechanism of latency maintenance and/or inhibition of latency reversal mediated by CD8+ lymphocytes. To recapitulate this observation in a reductionist in vitro model of HIV latency in human cells, we applied the recently developed Latency and Reversal Assay (LARA)³⁰ to evaluate how CD8+ T-cells affect the virus reactivation activity of N-803 in autologous memory CD4+ T-cells latently-infected with HIV89.6 (Fig.3A). Of note, this assay was conducted using cells derived from HIV-negative donors, thus HIV-specific cytotoxic CD8+ T-cells were absent. As shown in Fig.3B, while N-803 (and its biological counterpart, IL-15) is able to reactivate HIV expression in latently-infected CD4+ T-cell monocultures, co-culture with activated CD8+ T-cells significantly suppresses this ability. These data indicate that CD8+ T-cells effectively suppress the latency reversing activity of N-803, and therefore confirm the discovery of a previously unrecognized CD8+ T-cell-mediated activity that contributes to the maintenance of in vivo primate lentivirus latency.

Figure 3 |. In vitro co-culture of latently-infected human CD4+ T-cells with autologous CD8+ T-cells results in decreased expression of HIV-GAG during LRA administration.

a, Schematic of HIV latency model used in these experiments. Memory CD4+ T-cells (mCD4+) are enriched on day 0, and infected in vitro on day 3 with HIV_89.6. After infection, mCD4+ are maintained in the antiretroviral saquinavir to prevent viral spreading. On day 6, HIV-infected mCD4+ are cultured in the presence of TGF-beta, IL-7, conditioned medium from H-80 feeder cell line, and saquinavir, efavirenz and raltegravir (additional antiretrovirals). Cryopreserved autologous PBMC were thawed on day 8 and rested overnight (O/N) before enriching for total CD8+ (tCD8+) cells and then TCR activated for three days. On day 12, HIV-infected mCD4+ and TCR-activated tCD8+ are co-cultured in a 1:1 or 1:5 ratio in the presence of an LRA for three days (day 15). b, The frequency of HIV-GAG+ CD4+ T-cells was quantified by flow cytometry and the fold change compared to frequency in CD4 monocultures was calculated after exposure to anti-CD3/CD28, IL-15 superagonist N-803, or recombinant IL-15. Each color represents a unique donor (n=8 biologically independent samples) and sample averages (±SEM) are indicated by the grey bars. Statistical significance was measured using a matched one-way ANOVA.

Virus sequence analysis

To next investigate the viral dynamics associated with reactivation following CD8 depletion with N-803 administration in SIV-infected, ART-treated macaques, we performed a longitudinal sequence analysis of plasma virus using single genome PCR amplification (SGA) of the SIV_mac239-derived env genes. The viral env was sequenced at three pivotal time-points: (i) day 7 post-infection, i.e., peak viremia, (ii) day 56 post-infection, i.e., immediately prior to ART initiation (pre-ART), and (iii) during peak virus reactivation. We conducted this analysis in six macaques that exhibited robust virus reactivation (i.e., plasma viremia >800 copies/mL, Fig.2C). Extended Data Fig.7 displays phylogenetic analysis of the translated Env amino acid (a.a.) sequences of the circulating viruses. The peak viral load Env sequences were homogeneous and near identical to the input SIV_mac239 sequence; however, the diversity and number of informative sites at subsequent time-points were limited, such that sequences could not be clustered based on time-point with significant bootstrap support. The diversity at each time-point, while limited, was quantitated by determining the number of a.a. differences from the input SIV_mac239. Extended Data Fig.8A shows that, for all animals, the peak viremia Env sequences are least different from the input virus, as expected, while increased divergence was observed at the pre-ART time point and following reactivation. We next calculated the average number of a.a. differences from SIV_mac239 at each time-point, and compared with contemporaneous plasma viremia for each animal in a correlation matrix. The only significant association was a direct relationship between Env divergence and plasma viremia during reactivation (Extended Data Fig.8B). Finally, Extended Data Fig.9 shows the location of sequence changes at each time-point using highlighter plots of the Env a.a. sequences. Overall, this analysis supports the hypothesis that CD8 depletion with N-803 administration induces robust reactivation of a diverse population of viral variants. As no signs of virus evolution emerged from the longitudinal sequence analysis during high levels of latency reversal, it is unlikely that the rebounding virus is a product of de novo viral replication. In support, combined CD8 depletion with N-803 administration did not increase the levels of 2-LTR circles, considered a marker of recent lentivirus infection, in peripheral blood mononuclear cells (data not shown).

Analytic treatment interruption

To determine whether the interventions used induced a decline of the virus reservoir, we first longitudinally measured the level of cell-associated SIV-DNA in blood and lymph nodes. As shown in Fig.4A–F, none of the experimental groups showed significant changes in either the total fraction of circulating CD4+ T-cells or the calculated fraction of lymph node-derived cells harboring SIV-DNA. To functionally assess the impact of the treatment regimens on the reservoir size, we performed an analytical treatment interruption (ATI) in all macaques three weeks after either CD8 reconstitution and/or the last N-803 treatment. As shown in Fig.4G–I, all macaques rebounded within three weeks of ART interruption and most animals sustained high viral loads until the time of necropsy. It should be noted that in the current study ART was initiated at day 56 post-infection, thus substantially later than in other published macaque studies involving ATI, and therefore in the setting of a larger and more disseminated reservoir^17,31. The rapid rebound after ART interruption was therefore not unexpected as the experimental design was focused on assessing the “shock” effect of CD8 depletion with N-803 administration, with no anticipated impact on the reservoir size in absence of an intervention aimed at clearing the cells that have reactivated SIV production (“kill” phase of the “shock and kill” approach). The absence of a decrease in the level of SIV-DNA+ cells after combined CD8 depletion with N-803 administration may be due to the lack of CD8+ T-cell-mediated clearance of cells that have reactivated virus expression and/or the N-803-mediated proliferative expansion of infected CD4+ T-cells that have survived the events of virus reactivation.

Figure 4 |. CD8 depletion with N-803 administration does not decrease the size of the latent SIV viral reservoir.

Copies of total cell-associated SIV-DNA per 10⁶ cells was determined in a-c, sorted bulk peripheral CD4+ T-cells, and d-f, frozen bulk lymph node cells before and after intervention with N-803 alone (a, n=7 and d, n=3 biologically independent samples), CD8 depletion alone (b, n=14 and e, n=5 biologically independent samples), and CD8 depletion with N-803 administration (c, n=14 and f, n=3 biologically independent samples). Two-sided Friedman tests (a-c) and Wilcoxon tests (d-f) were used to determine statistical significance between pre-treatment and post-treatment time points. For all comparisons p-values were >0.05. The sample means are indicated by the gray bars on each graph (±SEM). Viral rebound following ART interruption in animals that received intervention with g, N-803 alone (n=7 macaques), h, CD8 depletion alone (n=14 macaques), and i, CD8 depletion with N-803 administration (n=14 macaques). Assay limit of detection of 3 copies of SIV RNA/mL plasma, black bar represents viral loads <60 copies/mL and gray box indicates time points prior to ART interruption.

Discussion

The current paradigm for “shock and kill” interventions for HIV cure predicates that reactivation of virus transcription in latently-infected cells is the first essential step to eliminate the persistent reservoir of replication-competent virus in ART-treated HIV-infected individuals. In this study, we have shown that the administration of the IL-15 superagonist N-803 in both SIV-infected macaques and HIV-infected humanized mice induces a robust and persistent reversal of latency only in the setting of CD8+ lymphocyte depletion, thus suggesting a substantial role for CD8+ lymphocytes in suppressing the LRA effect of N-803. Importantly, this novel role of CD8+ lymphocytes in maintaining virus latency was fully recapitulated in an in vitro experimental approach involving co-culture of activated, unprimed CD8+ T-cells with autologous, latently HIV-infected human primary CD4+ T-cells. To the best of our knowledge, this is the first instance in which a novel approach to manipulate latently infected cells has been independently confirmed in the two best validated and most widely used in vivo models for HIV cure interventions and then recapitulated in an in vitro experimental system of virus latency. In addition to this conceptual advance, this study defines a robust “shock” approach which could provide a key experimental system to compare and contrast directly in vivo the efficacy of different “kill” interventions, including those that may act in a CD8+ T-cell-independent manner (i.e., neutralizing antibodies, CD4 mimetics or immunotoxins). Further studies aimed at identifying the specific molecular pathways used by CD8+ T-cells to promote latency may allow the suppression of this activity, and therefore permit a full utilization of the virus reactivating potential of N-803 or other LRAs in the clinical setting without depleting CD8+ lymphocytes.

Methods

Rhesus macaque model

Animals, SIV-infection, antiretroviral therapy, CD8 depletion, and N-803 administration

This study was conducted using a cohort of 35 Indian-origin rhesus macaques housed at Yerkes National Primate Research Center (male and female, 2–3 years of age at the start of the study). All macaques were Mamu*B07- and Mamu*B17- with the following being Mamu*A01+: 77_13, RFr15, 208_13, RPb16, RJt15, RHv15, RAu15, RNa16, RNz15, ROr15, RRb16, RSt15, RAk16, RUs15, Rye16, RVz15, RBn16, and RRn16. All procedures were approved by the Emory University Institutional Animal Care and Use Committee (IACUC) and animal care facilities are accredited by the U.S. Department of Agriculture (USDA) and the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC) International.

Rhesus macaques were infected intravenously with 10³ TCID₅₀ of SIV_mac239 (nef open). SIV_mac239 stock was titrated in vitro for viral infectivity by standard endpoint titration on CEMx174 cells. The 50% tissue culture infectious dose (TCID₅₀) was calculated by the method of Reed and Meunch³². All animals were put on a three drug ART regimen at eight weeks post-SIV infection. Tenofovir (TDF at 5.1mg/kg/day or PMPA at 20mg/kg/day) and emtricitabine (FTC at 40mg/kg/day) were both kindly provided by Gilead Pharmaceuticals. Dolutegravir (DTG at 2.5mg/kg/day) was kindly provided by ViiV Pharmaceuticals. Drugs were administered daily by subcutaneous injection.

After over 12 months of ART, macaques were assigned intervention groups. Age, weight, sex, A01 status, peak post-infection viral load, and time to suppression after ART were all controlled for when allocated animals into intervention groups. Group sizes were determined via a power analysis based on previous studies. There was no blinding in this study. 28 animals were administered one dose of the anti-CD8α-depleting antibody, MT807R1 at 50 mg/kg. The initial 15 animals receiving the depletion antibody received the administration intravenously. Due to safety concerns, the administration was changed to subcutaneous for the remaining 13 animals. There was no observable effect of the different administration routes on the efficacy of depletion.

At the start of intervention, 21 animals were administered a dose of N-803 either in addition to CD8 depletion (n=14), or without (n=7). N-803 is administered subcutaneously in a cycle of 100 μg/kg once a week for four consecutive weeks.

The study design included a later four dose administration of N-803 in seven animals of groups 1 and 2 that was conducted at the time of CD8+ T-cell reconstitution to potentially accelerate the recovery of these cells and improve their antiviral cytotoxic potential. As expected, this second cycle of N-803 induced a faster recovery of CD8+ T-cells (data not shown) and was associated with an increase in T-cell activation and proliferation that was similar to that observed after the first N-803 cycle (data not shown). The late administration did not result in an increase in plasma viremia.

Animals were interrupted of ART three weeks after either the last dose of N-803 or three weeks after the reconstitution of CD8+ T-cells (which ever occurred first). Plasma viral loads were monitored for about 6 months until elective necropsy was performed.

CD8 depletion with N-803 in SHIV-infected, ART-treated macaques (pilot study)

An additional 5 Indian-origin rhesus macaques were included in this study as part of a follow up pilot studying using a SHIV-model of infection. These animals were also housed at the Yerkes National Primate Research Center and all procedures were approved by the Emory University IACUC. Animals were infected intrarectally with high-dose SHIV_162P3, administered as a 1:50 dilution of a 2032 TCID₅₀/mL, 109 RNA copies/mL, 182 P27 ng/mL stock. All animals were put on the same FTC/DTG/FTC ART regimen at twelve weeks post-SHIV infection. After 6 months of ART, animals were administered one dose of MT-807R1 at 50 mg/kg subcutaneously. N-803 was administered subcutaneously in a cycle of 100 μg/kg once a week for four consecutive weeks starting at the time of CD8 depletion.

Sample collection and tissue processing

Blood, lymph nodes (LN), and rectal biopsies (RB) were collected longitudinally including at the time of necropsy and processed for further analyses as previously described¹⁹.

Immunophenotype by flow cytometry

Multiparametric flow cytometry was performed according to a standard protocol on PBMC and LNMC using fluorescently labeled monoclonal antibodies cross-reactive in RM. The following antibodies were used at 37°C for 30 minutes: CCR5 APC (3A9) and CCR7 FITC (150503). Then the following antibodies were at room temperature for 30 minutes: CD3 APC-Cy7 (SP34–2), CD4 BV650 (OKT4), CD8α BV711 (RPA-T8), CD8β PE-Cy5 (SIDI8BEE), CCR5 APC (3A9), CCR7 FITC (150503), CD45RA Pe-Cy7 (5H9), CD62L PE (SK11), CD95 BV605 (DX2), PD-1 BV421 (EH12.2H7), CD16 BV421 (3G8), CD20 PE-Cy5 (2H7), CD14 PE-Cy7 (M5E2), NKG2A (CD159) PE (Z199), CD28 PE-Cy5.5 or ECD (CD28.2), CD56 FITC (NCAM16.2). Additional panels included CD69 Pe-Cy5 (FN50), and CD25 APC (M-A251). Cells stained for Ki67 were fixed and permeabilized with Perm II kit (BD) before being stained at room temperature for 30 minutes with Ki67 AF700 (B56).

All flow cytometry specimens were acquired on an LSR II (BD Biosciences) equipped with fluorescence-activated cell sorter software (FACS Diva), and analysis of the acquired data was performed using FlowJo software (Tree Star).

Determination of plasma SIV RNA, and cell-associated RNA and DNA

For pre-intervention time points, quantitative real-time RT-PCR was performed to determine SIV plasma viral load as previously described with a sensitivity of 60 copies/mL²⁴. For the three time points prior to intervention and all post-intervention time points plasma SIV gag RNA levels were measured using quantitative PCR, essentially as described using high sensitivity assay formats^33,34. Quantification of total cell-associated SIVmac₂₃₉ gag DNA was performed as previously described³⁵. The number of gag DNA copies per 10⁶ CD4+ T-cells was calculated by dividing the number of gag DNA copies/10⁶ PBMC by the percentage of CD3+CD8-CD4+ cells in the PBMC population. CD4+ T-cells were isolated from PBMC via a CD4+ T-cell isolation kit (Miltenyi) and cell-associated RNA was measured as previously described³⁶.

In Situ RNA analysis and quantification

Viral RNA (vRNA) detection via RNAscope and quantitative image analysis was performed on formaldehyde fixed, paraffin-embedded (FFPE) tissue sections (5μm) according to our previously published protocol³⁷, with the following minor modifications: heat-induced epitope retrieval was performed by boiling slides in 1x target retrieval (322000; ACD) for 30 min., followed by incubation at 40°C with a 1:10 dilution of protease III (322337; ACD) in 1x PBS for 20 min. Slides were incubated with the target probe SIVmac239 (312811; ACD) for 2 hours at 40°C and amplification was performed with RNAscope 2.5 HD Detection kits (322360; ACD) according to manufacturer’s instructions, with 0.5X wash buffer (310091; ACD) used between steps. The resulting signal was detected with Warp Red chromogen (WR806M; Biocare Medical). Slides were counterstained with CAT hematoxylin (CATHE-GL; Biocare Medical), mounted with Clearmount (17885–15; EMS) until dry, coverslipped using Permount (SP15–100; Fisher Scientific), and scanned at 40x magnification on an Aperio AT2 (Leica Biosystems). RNAscope images were analyzed for the total number of vRNA⁺ cells/10⁵ total cells (quantitative) and the relative amount of vRNA present (semi-quantitative) using the ISH module (v2.2) within Halo software (v2.3.2089.27; Indica Labs). The relative amount of vRNA within a single cell was first estimated by quantifying the total area of the vRNA signal spot size (μm²). Since the signal spot size is a function of several steps in the experimental procedures, module settings were established on concomitantly assayed, acutely infected SIV⁺ control slides. To estimate the signal spot size of a single vRNA molecule, we measured the signal area (min/mean/max) of >10 identifiable individual virions within B cell follicles, which corresponds to two copies of vRNA, and multiplied by 0.5. We set the vRNA minimum signal spot size within the analysis module to exclude detection of a single vRNA molecule and/or integrated viral DNA. Relative vRNA copy numbers present within vRNA⁺ cells was calculated as [signal spot size within vRNA⁺ cell (μm²)/(0.5 × mean signal size for a virion)].

Fluorescence activated sorting (FACS) of live cells

Mononuclear cells isolated from blood were stained with Live/dead, CD3 AF700 (SP34–2), CD4 BV650 (OKT4), CD8 APC-Cy7 (SK1), CD14 PB (M5E2), CD20 PB (2H7), and CD16 PB (3G8), for 30 minutes at room temperature. Aliquots of 50,000 CD4+ T-cells (live CD3+CD20-CD14-CD16-CD8-CD4+) and CD8+ T-cells (live CD3+CD20-CD14-CD16-CD4-CD8+) were then sorted using a FACS Aria II (BD Biosciences). Mononuclear cells were also separately stained with Live/dead, CD3 AF700 (SP34–2), CD4 BV650 (OKT4), CD8 APC-Cy7 (SK1), CD14 PB (M5E2), CD20 PB (2H7), and NKG2A PE (Z199) to sort aliquots of 50,000 NK cells (Live CD3-,CD20-,CD14-CD4-,CD8+,NKG2A+).

RNA-Seq and data analysis

Bulk CD4+ T-cells were sorted from fresh PBMCs prior to intervention, day 3, week 2, and week 4. Briefly, RNA from sorted cells was collected and extracted and DNA was digested. Libraries were prepared and normalized, pooled, and clustered on a flow cells for sequencing. RNA-Seq data were aligned to the MacaM v7.8 assembly of the Indian rhesus macaque genome. To identify pathways differentially modulated, Gene Set Enrichment Analysis (GSEA)³⁸ was performed on the ranked transcript lists using 1000 phenotype permutations and random seeding. Gene sets used included the MSigDB H (hallmark) gene sets³⁹.

Single genome PCR amplification of SIVmac239 env sequences

cDNA synthesis and 384-well single genome PCR amplification (SGA) were performed using an approach similar to those previously described^40–42. Briefly, RNA was extracted from cryopreserved plasma samples using the QIAmp viral RNA kit (Qiagen, # 52906), and reverse transcription was performed using the SuperScript III kit (Invitrogen, #18080–044) with reverse primer SM-ER1 (5’-CTA TCA CTG TAA TAA ATC CCT TCC AGT CCC-3’). cDNA was diluted to result in <30% positive wells for SGA. First round PCR was performed in a 15 μL volume using the Phusion Hotstart II High Fidelity DNA Polymerase (Thermo Scientific, #F537S) with forward primer H2SM-EF1 (5’-CCC TTG AAG GMG CMR GAG AGC TCA TTA-3’) and SM-ER1. Cycling conditions were 98°C for 2 min; 10 cycles of 95°C for 15 s, 54°C for 60 s, and 68°C for 4 min; 25 cycles of 95°C for 15 s, 54°C for 60 s, and 68°C for 4 min, adding 5 s to the extension per cycle; 72°C for 30 min; and 4°C hold. Second round PCR was performed with the same enzyme in a 10 μL volume with 1 μL of the first round PCR reaction as template and primers H2SM-EF2 (5’-CAC CTA AAA ART GYT GCT AYC ATT GCC AG-3’) and SM-ER2 (5’-ATA AAATGA GAC ATG TCT ATT GCC AAT TTG-3’). Cycling conditions were 95°C for 2 min; 30 cycles of 95°C for 15 s, 54°C for 60 s, and 72°C for 2.5 min; 72°C for 10 min; and 4°C hold. PCR amplicons were purified using Qiaquick PCR Purification Kit (Qiagen #28106).

Sequencing of env amplicons

On average, 26 SGA PCR amplicons per time-point (range 20 to 30) were sequenced by Eurofins Genomic DNA Sanger sequencing using the following primers: SIVmac251seqF1 5’-GGGATATGTTATGAGCAGTCACG-3’; SIVmac251seqF2 5’-ATCCAAGAGTCTTGTGACAAGC-3’; SIVmac251seqF3 5’-AAGAGAGGGAGACCTCACG-3’; SIVmac251seqF4 5’-AGGCCAGTGTTCTCTTCC-3’; SIVmac251seqR1 5’-CTTGTTCCAAGCCTGTGC-3’; SIVmac251seqR2 5’-CCTCTGCAATTTGTCCACATG-3’; SIVmac251seqR3 5’-TCCAAGAAGTCAACCTTTCGC-3’; SIVmac251seqR4 5’-AGCTGGGTTTCTCCATGG-3’¹⁹. Sequencher v5.1 was used to generate nucleotide sequence contigs, and sequences with mixed peaks in the chromatogram were excluded from further analysis.

Sequence analysis

Geneious v9.1.7 was used to translate nucleotide sequences into amino acids and generate alignments. Amino acid alignments were exported from Geneious in FASTA format and used to generate Highlighter plots to visualize amino acid mismatches (http://www.hiv.lanl.gov/content/sequence/HIGHLIGHT/highlighter_top.html). Phylogenetic Neighbor-Joining consensus trees (Jukes-Cantor, resampling with 100 bootstrap replicates) were created in Geneious using amino acid alignments, and were exported in NEXUS format into Figtree v1.4.4 for further modification (Andrew Rambaut,Institute of Evolutionary Biology, University of Edinburgh.http://tree.bio.ed.ac.uk/). Phylogenetic trees were presented as unrooted or were rooted on the midpoint. Bootstrap values of greater than 80% are considered significant. Pairwise differences between the infecting SIVmac239 clone and each SGA-derived Env amino acid sequence were determined in Geneious.

Bone marrow-liver-thymus humanized mouse model:

Experimental design

The bone marrow-liver-thymus humanized (BLT hu-mice) mouse model of HIV infection was used to determine the efficacy of CD8 depletion alone, CD8 depletion in combination with N-803, or N-803 alone as a LRA. BLT mice (15–19 weeks post humanization surgery) were exposed to HIV-1_JR-CSF intravenously. ART was initiated 4–5 weeks later. Viremia was durably suppressed by ART for 4 weeks. A single dose of N-803, CD8 depleting antibody or the combination of N-803 and CD8 depleting antibody was administered to HIV-infected and suppressed animals as indicated below. HIV RNA induction was measured on days 4 and 7.

Construction of BLT humanized mice

BLT humanized mice were prepared as previously reported^39–42. Briefly, a 1–2 mm piece of human liver tissue was sandwiched between two pieces of autologous thymus tissue (Advanced Bioscience Resources) under the kidney capsule of sub-lethally irradiated (200 cGy) 12–15 week-old NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ (NSG; The Jackson Laboratory) mice. Following implantation, mice were transplanted intravenously with hematopoietic CD34⁺ stem cells isolated from autologous human liver tissue. Human immune cell reconstitution was monitored in the peripheral blood of BLT mice by flow cytometry every 3–4 week as previously described^39–42. Mice were maintained under specific pathogen-free conditions by the Division of Comparative Medicine at the University of North Carolina, Chapel Hill. Animal experiments were conducted in accordance with NIH guidelines for the housing and care of laboratory animals and in accordance with protocols reviewed and approved by the IACUC at the University of North Carolina, Chapel Hill.

Production of HIV and infection of BLT mice

Stocks of HIV-1_JR-CSF were prepared as previously reported^39–41. The proviral clone was transfected into human embryonic kidney (HEK)_293T-cells using Lipofectamine™ 2000 (Invitrogen #11668030) following manufacturer’s protocol. Viral supernatant was collected 48 hours after transfection and tittered on TZM-bl indicator cells in triplicate to determine the tissue culture infectious units (TCIU) per ml. At least two different titer determinations were performed for each virus stock. BLT mice were exposed to 3×10⁴ TCIU HIV-1_JR-CSF via tail vein injection.

Analysis of HIV infection in BLT mice

The peripheral blood plasma viral load was monitored longitudinally by quantitative real-time PCR using a TaqMan® RNA to-C_T™ 1-step kit (Applied Biosystems #4392656). The sequences of the forward and reverse primers and the TaqMan™ probe for PCR amplification and detection of HIV gag RNA were: 5′-CATGTTTTCAGCATTATCAGAAGGA-3′, 5′-TGCTTGATGTCCCCCCACT-3′, and 5′-FAM CCACCCCACAAGATTTAAACACCATGCTAA Q −3′, respectively. Known quantities of HIV gag RNA were run in parallel, creating a standard curve for HIV gag and sample RNA was quantified by extrapolation from the standard curve. All samples were run and analyzed on an ABI 7500 Fast Real-time PCR System (Applied Biosystems).

HIV DNA levels were measured in tissue cells collected at harvest and cryopreserved in cryopreservation media (10% DMSO: 90% fetal bovine serum). Cells were thawed slowly, counted by trypan exclusion, aliquoted, and pelleted. DNA was extracted from cell pellets using the QIAamp DNA Blood Mini Kit (Qiagen) according to the manufacturer’s instructions. RT-PCR was performed with a TaqMan Fast Universal PCR Master Mix (Applied Biosystems). The sequences of the forward primer, reverse primer, and the TaqMan™ probe for amplification and detection of HIV gag DNA were: 5′-CATGTTTTCAGCATTATCAGAAGGA-3′, 5′-TGCTTGATGTCCCCCCACT-3′, and 5′-FAM CCACCCCACAAGATTTAAACACCATGCTAA Q −3′ respectively. As a control, Homo sapiens hemoglobin subunit gamma-2 was ran to quantify the presence of human DNA in each sample. The sequences of the forward primer, reverse primer, and the TaqMan™ probe for amplification and detection of hemoglobin subunit gamma-2 were 5’-CGCTTCTGGAACGTCTGAGATT-3’, 5’-CCTTGTCCTCCTCTGTGAAATGA-3’, and 5’-FAM TCAATAAGCTCCTAGTCCAGAC-3’ respectively. All samples were run and analyzed on an ABI 7500 Fast Real-time PCR System (Applied Biosystems).

ART administration

ART was administered to BLT hu-mice as previously described^43–45 via 1/2″ pellets of irradiated Teklad chow consisting of emtricitabine (1500 mg/kg), tenofovir disoproxil fumarate (1560 mg/kg), and raltegravir (600 mg/kg) (Research Diets).

N-803 and MT807R1 administration

N-803 (0.2 mg/kg in PBS) and control vehicle (PBS) were administered to mice intravenously in a total volume of 200 μL. MT807R1 (3 mg/kg in PBS) and the control vehicle (PBS) were also administered intravenously in a total volume of 200 μL.

Immunophenotypic analysis of BLT mice

Immunophenotyping was performed on peripheral blood samples longitudinally and at harvest on blood and mononuclear cells isolated from the tissues of BLT mice. All flow cytometry data were collected on a BD FACSCanto instrument using BD FACSDiva software (version 6.1.3) and data were analyzed with FlowJo Software (version 10.5.0). Antibodies for the analysis of human immune cell levels include: CD45 APC (clone HIT3a; BD Biosciences #555485), CD3 FITC (clone HIT3a; BD Biosciences #555339), CD4 APC-Cy7 (clone RPA-T4; BD Biosciences #560158), CD33 PE (clone P67.6; BD Biosciences #340679); CD19 PE-Cy7 (clone SJ25C1; BD Biosciences #557835) and CD8 PerCP (clone SK1; BD Biosciences #347314). Flow cytometric gating for expression of lineage specific antigens on human leukocytes was performed as follows: (step 1) forward and side scatter were utilized to set a live cell gate; (step 2) live cells were then analyzed for expression of the human pan-leukocyte marker CD45⁺; (step 3) human leukocytes were then analyzed for human CD3⁺ T-cells and CD19⁺ B-cells and (step 4) T-cells were analyzed for human CD4⁺ and CD8⁺ expression. The following flow cytometry antibody panel was also used to analyze HLA-DR, CD38 and CD25 expression: CD3 BV421 (clone UCTH1; BD Biosciences #562426), CD4 BV605 (clone RPA-T4; BD Biosciences #562658), CD45 FITC (clone 2D1; BD Biosciences #347463), HLA-DR PerCP (clone L243; BD Biosciences #347364), CD69 PE (clone FN50; BD Biosciences #555531), anti-CD38 PE-Cy7 (clone HB7; BD Biosciences #335790), CD25 APC (clone 2A3; BD Biosciences #340938), CD8 APC-Cy7 (clone SK1; BD Biosciences #557834), and AQUA (ThermoFisher #L35957). Flow cytometric gating was performed as follows: (step 1) forward scatter height and forward scatter area were used to eliminate doublets; (step 2) side scatter area and forward scatter area were used to distinguish leukocytes based on morphology; (step 3) the viability dye AQUA was use to discriminate live cells from dead cells; (step 4) live cells were analyzed for the expression of the human pan-leukocyte marker CD45; (step 5) human leukocytes were then assessed for human CD3 expression to identify T-cells; (step 6) T-cells were evaluated for expression of human CD4 and CD8; (step 7) human CD4⁺ or CD8⁺ T-cells were examined for expression of HLA-DR and/or CD38, or CD25. Gates were set with fluorescence minus one controls. Non-specific binding was assessed with isotype controls.

CD8 in vitro suppression assay

Experimental Design

In vitro latently-infected memory CD4+ T-cells were generated using the LARA method as previously described³⁰ with the following modifications. On day 0, after peripheral blood mononuclear cells (PBMC) were isolated from HIV-naïve buffy coats (New York Blood Center) using SepMate density gradient centrifugation (StemCell, 85460), a portion of PBMCs from each HIV naïve donor were cryopreserved in fetal bovine serum (VWR Life Science Seradigm, 97068–085) + 10% DMSO and stored in liquid nitrogen. On day 8, PBMCs were thawed and rested overnight in RPMI 1640 medium (Fisher Scientific, SH3002701.01) supplemented with 10% fetal bovine serum, 1% penicillin-streptomycin (Corning, 45000–650), and 1% HEPES (Gibco, 15630080); cRPMI) before total CD8+ T-cell positive enrichment on day 9 (Miltenyi, 130-045-201). CD8+ T-cells were stimulated with anti-CD3/CD28 beads (Dynabeads, 11141D) at a 1:1 ratio plus 30 U/mL IL-2 (R&D system, 202IL050CF) for three days. On day 12 of LARA, CD4+ and CD8+ were prepared for co-culture. HIV latently infected memory CD4+ T-cells were washed, counted, and plated for latency reversal in cRPMI in the presence of 100 nM efavirenz, 200 nM raltegravir and 5 μM saquinavir (antiretroviral therapeutics, ART). Activated total CD8+ T-cells were removed from the anti-CD3/CD28 beads, washed and resuspended in cRPMI plus ART. CD4+ and CD8+ T-cells were co-cultured at a 1:1 or 1:5 ratio at a final density of 1 × 10⁶ cells/mL. CD4+ monocultures were also maintained in parallel. Mono- and co-cultures were then left unstimulated, TCR-activated with 1 μg/mL plate-bound OKT3 and 1 μg/mL soluble CD28 (Biolegend, 302933), or treated with 14 nM N-803 (provided by NantKwest) or 500 ng/mL IL-15 (R&D Systems, 247-ILB). Cells were harvested after 72 hours and analyzed by flow cytometry and qPCR.

Flow cytometry

Multicolor flow analysis of cell surface and intracellular marker expression was performed with a BD FACSymphony flow cytometer. Between 200,000 and 600,000 events were acquired for each sample using the live cell gate. The data was analyzed with FlowJo (v.10).

Antibodies used in this study: CD3 Alexa Fluor® 700 (UCHT1, BD Biosciences, #557943), CD8 BUV737 (SK1, BD Horizon™, #564629), HIV-1 core antigen-FITC (KC57, Coulter Clone, #6604665), CD4 BV421 (SK3, BD Horizon™, #565997), CD45RA APC-eFluor™780 (HI100, Invitrogen, #47045842), CD27 BV650 (O323, Biolegend, #302828), CCR7 Pe-Cy7 (3D12, BD Pharmingen™, #557648), Fixable Viability Dye eFluor™ 506 (Invitrogen eBioscience #65-0866-18).

Statistics and Reproducibility

Statistical analyses, including two-way Kruskal-wallis tests, two-way Friedman tests, one-way ANOVA, and Spearman r, were performed using Prism v7.0 or v8.0. Data are represented as mean ±SEM as indicated.

Data and code availability

Illumina sequencing reads for RNA-Seq experiments were submitted to the NCBI SRA repository and are available at Accession #SRP188630. RNA-Seq datasets were submitted to the NCBI GEO repository and are available at accession number GSE128415. Env nucleotide sequences have been deposited into Genbank under the accession number MK922999-MK923550.

Extended Data

Extended Data Figure 1 |. MT807R1 effectively depletes CD8+ T-cells in peripheral blood, lymph node, and rectum in addition to NK cells from the blood at day 7.

The percentage of CD8+ cells in the CD3+ population seven days post-depletion was compared to pre-depletion levels. Sample flow cytometry shows the absence of a CD8β+ cells as part of the CD3+ T-cell population after depletion in a, the peripheral blood, b, rectum, and c, lymph node and similar results were found across all CD8-depleted macaques (n=28 biologically independent samples). d, The percentage of CD8β+ cells as compared to pre-depletion baseline was calculated in all CD8-depleted animals (+/− N-803, n=28 macaques) across blood and tissue (no differences in CD8+ T-cell depletion were observed between groups 2 and 3 at day 7). A two-sided Friedman test was used to calculate statistically significance changes from the baseline across tissues. e, Depletion of NK cells in the peripheral blood was assessed one week following CD8 depletion alone (n=14 macaques) as compared to baseline. Statistical significance was calculated using Wilcoxon test. Mean ± SEM are shown.

Extended Data Figure 2 |. Phenotypic changes to CD4+ T-cells following intervention.

Longitudinal flow cytometry analysis following N-803 alone (green, n=7 macaques), CD8 depletion alone (blue, n=14 macaques), and CD8 depletion with N-803 (red, n=14 macaques). a, CD4+ T-cell frequency. Percentage of b, naïve, c, stem cell memory (SCM), d, central memory (CM), e, transitional memory (TM), and f, effector memory (EM) CD4+ T-cells. Percentage of bulk CD4+ T-cells expressing g, PD-1, h, CD25, i, CD69, j, HLA-DR, k, CCR5, and l, CCR5 and Ki67 co-expression. m-q, CCR5 and r-v, Ki67 expression on CD4+ T-cell subsets. Sample means are indicated (±SEM), and two-sided Kruskal-Wallis tests were used to compare post-intervention values to pre-intervention baseline and (approximate P value summaries are provided).

Extended Data Figure 3 |. SIV-associated genes and IL-15 subunit genes show a transient change in expression following treatment with N-803 alone.

RNA was extracted from sorted peripheral a, bulk CD4+ T-cells (CD3+, CD4+ CD8-,CD20-,CD14-), b, bulk CD8+ T-cells (CD3+,CD4-CD8+,CD20-,CD14), and c, NK cells (CD3-CD20-CD14-NKG2A+) and libraries were prepared, normalized, pooled, and clustered on flow cells for sequencing. RNAseq data was aligned to the MacaM v7.8 assembly of the Indian rhesus macaque genome. Transcripts were analyzed for alignment against a custom gene set with SIV host restriction factors, PPIA (capsid folding protein), SIV receptors, SIV receptor agonists, NFkB subunits (involved in mediating LTR transcription), IL-15 receptor subunits, and NFAT subunits.

Extended Data Figure 4 |. Quantification of levels of cell-associated SIV RNA in peripheral CD4+ T cells prior to and following interventions.

Changes in expression of SIV-RNA in relation to the number of copies of CD4 following intervention with a, N-803 alone (n=7 macaques), b, CD8 depletion alone (n=7 macaques), and c, CD8 depletion with N-803 (n=7 macaques). Sample means are indicated (±SEM), and two-sided Wilcoxon tests were used to compare post-intervention values to pre-intervention baseline.

Extended Data Figure 5 |. Level of virus reactivation correlated with the absence of CD8+ T-cells.

Correlation between CD8+ T-cell counts and viral load (SIV RNA copies/mL plasma) on day 0, day 3, and weekly through week 6 following intervention with a, CD8 depletion alone (n=103 samples from 14 macaques), and b, CD8 depletion with N-803 (n=112 samples from 14 macaques). The area-under-the-curve (AUC) and the average pre-intervention viral load following c, CD8 depletion alone (n=14 macaques), and d, CD8 depletion with N-803 (n=14 macaques). Correlation coefficients are calculated using the Spearman’s rank-order correlation (two-tailed, no adjustments). e-g, Longitudinal viral loads (top) and CD8+ T-cell counts (middle) following e, N-803 alone (n=7 macaques), f, CD8 depletion (n=14 macaques), and g, CD8 depletion with N-803 (n=14 macaques). The bottom panels provide animal color keys.

Extended Data Figure 6 |. HIV-DNA, HIV-RNA, and human T-cell activation levels in HIV-infected, ART-suppressed BLT humanized mice intervened with N-803, CD8 depletion, or N-803 with CD8 depletion.

Seven days post-intervention with N-803 (green, n=4 BLT humanized mice), CD8 depletion (blue, n=4), or N-803 with CD8 depletion (red, n=4) to HIV-infected, ART-suppressed BLT humanized mice, a, total HIV-DNA, and b, cell-associated HIV-RNA were extracted from mononuclear cells isolated from the spleen, human thymus (huThy), and lymph node (LN, HIV-RNA only). Percentage of HLA-DR+, CD38+, CD25+, or HLA-DR+/CD38+ cells was measured in human CD4+ (left) or CD8+ (right) T-cells isolated from the c, spleen, or d, human thymus of HIV-infected, ART-suppressed BLT mice seven days post-intervention with N-803 (green, n=4), CD8 depletion (blue, n=4), or N-803 with CD8 depletion (red, n=4). Treatment groups were compared using a two-tailed Student’s t-test (a), or a Kruskal-Wallis test with a false discovery rate correction (b-d). Sample means are indicated by a horizontal bar (+/−SEM).

Extended Data Figure 7 |. Phylogenetic trees of longitudinal SGA-derived Env amino acid sequences.

Phylogenetic trees were generated for six macaques receiving CD8 depletion with N-803 administration using Env sequences from the peak VL (red), pre-ART (blue), and reactivation time points (green). The Env sequence of the SIVmac239 clone used for infection is included in each tree (black). The horizontal bar below each tree indicates the genetic distance. Sequence clusters that are supported with bootstraps greater than 80% are indicated by an asterisk. Env sequences that contain a stop codon are indicated by an arrow.

Extended Data Figure 8 |. Longitudinal Env amino acid divergence from the input virus and relationship with plasma viral load.

The number of amino acid differences between the infecting viral clone SIV_mac239 and each SGA amplicon was determined using Geneious. a, A violin plot was created to show the frequency distribution of the number of amino acid differences between sequences at each time point in each macaque. The solid line indicates the median number of amino acid differences for each individual Env sequence, while the dotted lines indicate the quartiles. Peak VL (red), pre-ART (blue), and reactivation (green) time points are shown. The animal ID and the 3 time points are indicated below the graph. Statistical differences between time points for each macaque were determined by performing multiple comparisons using a Kruskal Wallis test with Dunn’s correction. b, The average number of sequence differences for each animal at the reactivation time point is plotted on the y-axis, and the corresponding plasma viral loads are plotted on the x-axis on a log10 scale. Correlation coefficients are calculated using the Spearman’s rank-order correlation (two-tailed, no adjustments).

Extended Data Figure 9 |. Highlighter plots of longitudinal SGA-derived Env amino acid sequences.

Highlighter plots were generated for six representative macaques receiving CD8 depletion with N-803 administration using Env sequences from peak VL (red box), pre-ART (blue box), and reactivation (green box) time points. The Env sequence of the SIVmac239 clone used for infection is included as the master (reference) sequence in each plot. The position of N-linked glycosylation sites on the master sequence are indicated by pink circles. Each tick represents an amino acid difference from the master sequence, as is indicated by the legend. Blue diamonds indicate the loss of an N-linked glycosylation site.

Extended Date Table 1 |.

Viral loads from macaque and humanized mice studies

Model: ART-treated, SIV-infected rhesus macaques (limit of detection = 3 copies of SIV RNA/mL of plasma)
		Pre-intervention				Post-intervention
Intervention	Macaque	Month −3	Month −2	Month −1	Day 0	Day 3	Week 1		Week 2	Week 3	Week 4	Week 5	Week 6
N-803 alone	REf16 R	7	< 3	< 3	7	< 3	< 3		< 3	< 3	7		< 3
	RVz15 R	7	25	19	7	25	19		< 3	< 3	1600		40
	RAf16 R	< 3	4	< 3	< 3	< 3	32		7	< 3	< 3	10	10
	REi16 R	7	5	7	< 3	7	10		< 3	< 3	< 3	< 3	< 3
	RIt16 R	18	15	8	< 3	15	< 3		7	10	7	< 3	< 3
	RBn16 R	7	4	< 3	< 3	7	< 3		< 3	< 3	< 3	< 3	< 3
	RRn16 R	25	40	25	25	50	10		15	15	25	15	15
CD8 depletion alone	RNz15 R	< 3	< 3	< 3	7	7	50		490	< 3	30	20	< 3
	ROr15 R	7	65	< 3	< 3	7	25		25	7	25		1200
	RKs15 R	< 3	< 3	< 3	< 3	10	10		65	< 3	< 3		< 3
	RRb16 R				< 3	65	40		90	65	50	Nx
	REs16 R	65	18	25	15	40	65		65	65	450	230	40
	RSt15 R	7	7	10	< 3	40	40		50	170	7	7	32
	RVe16 R	3	15	50	20	50	230		330	430	40	7	32
	REg16 R	10	7	10	7	30	50		90	10	50	25	25
	RAk16R	< 3	6	< 3	< 3	7	25		19	40	7	< 3	< 3
	RJz15 R	15	< 3	15	< 3	15	270		400	< 3	19	19	< 3
	RRa16 R	< 3	< 3	< 3	< 3	10	< 3		7	< 3	19	< 3	< 3
	ROs15 R	7	< 3	10	< 3	65	50		30	7	310	32	< 3
	RUs15 R	15	90	29	7	40	10		50	19	3200	12000	1800
	RYe16 R	24	10	40	40	260	30		90	7	1200	1200	470
CD8 depletion with N-803	RJt15 R	< 3	< 3	600	90	12000	24000		13000	2700	32	10	15
	RHv15 R	< 3	< 3	< 3	< 3	< 3	10		7	20	660	< 3	< 3
	106_13 R	7	7	< 3	< 3	30	90		420	190	40	< 3	< 3
	RUa16 R	65	7	220	25	1100	90		1200	1700	1900	10	40
	RAu15 R	7	< 3	7	10	610	810		90	510	520	7	10
	RNa16 R	< 3	15	< 3	< 3	25	50		210	150	300	< 3	< 3
	REu16 R	40	130	40	25	557	1610		1000	1200	1000	19	32
	77_13 R	15	7	< 3	10	230	280		290	30	4600	540	< 3
	RFr15 R	< 3	10	< 3	< 3	15	50		90	50	7	7	7
	208_13 R	< 3	< 3	< 3	< 3	19	< 3		15	65	50	< 3	< 3
	RBc16 R	50	10	90	65	510	260		65	170	90	10	25
	RCa16 R	7	7	< 3	< 3	240	320		580	270	3000	< 3	10
	RCr15 R	< 3	< 3	< 3	< 3	< 3	32		200	5100	< 3	< 3	10
	RPb16 R	25	50	7	25	330	400		200	90	90	7	< 3
Model: ART-treated, SHIV-infected rhesus macaques (limit of detection = 3 copies of SIV RNA/mL of plasma)
		Pre-intervention				Post-intervention
Intervention	Macaque	Month −3	Month −2	Month −1	Day 0	Day 3	Week 1	Day 10	Week 2	Week 3	Week 4	Week 5	Week 6
CD8 depletion with N-803	RKm16	10	9	7	3	430	30	470	480	3600	90	3	90
	CB91	65	7	9	3	460	60	660	90	810	50	30	3
	RPp16	10	7	7	3	20	7	10	7	7	25	15	15
	RRl16	3	7	7	3	3	3	19	3	7	65	15	3
	RYr16	15	7	7	3	20	50		50	25	40	3	7
Model: ART-treated, HIV-infected bone marrow-liver-thymus humanized mice (limit of detection = 346 copies of SIV RNA/mL of plasma)
Intervention	Mouse	Pre-intervention	Day 4	Day 7
N-803 alone	1	<346	<346	<346
	2	<346	<346	<346
	3	<346	<346	<346
	4	<346	<346	<346
	5	<346	<346	<346
	6	<346	<346	<346
	7	<346	<346	<346
CD8 depletion alone	1	<346	560	<346
	2	<346	<346	378
	3	<346	<346	378
	4	<346	<346	<346
	5	<346	<346	<346
	6	<346	<346	<346
	7	<346	<346	<346
	8	<346	<346	<346
CD8 depletion with N-803	1	<346	1300	1488
	2	<346	1079	<346
	3	<346	779	<346
	4	<346	546	574
	5	<346	354	<346
	6	<346	<346	1504
	7	<346	<346	1981
	8	<346	<346	<346