Latency reversal plus natural killer cells diminish HIV reservoir in vivo

Abstract

HIV is difficult to eradicate due to the persistence of a long-lived reservoir of latently infected cells. Previous studies have shown that natural killer cells are important to inhibiting HIV infection, but it is unclear whether the administration of natural killer cells can reduce rebound viremia when anti-retroviral therapy is discontinued. Here we show the administration of allogeneic human peripheral blood natural killer cells delays viral rebound following interruption of anti-retroviral therapy in humanized mice infected with HIV-1. Utilizing genetically barcoded virus technology, we show these natural killer cells efficiently reduced viral clones rebounding from latency. Moreover, a kick and kill strategy comprised of the protein kinase C modulator and latency reversing agent SUW133 and allogeneic human peripheral blood natural killer cells during anti-retroviral therapy eliminated the viral reservoir in a subset of mice. Therefore, combinations utilizing latency reversal agents with targeted cellular killing agents may be an effective approach to eradicating the viral reservoir.

Introduction

Even though antiretroviral therapy (ART) can effectively halt HIV replication, ART must be maintained for life because latent HIV-1 infected cells persist and initiate active virus replication if ART is discontinued^1,2,3. Experimental approaches for eliminating the latent reservoir have included myeloablation followed by transplantation of cells lacking co-receptors for virus infection, genome editing of the latent provirus, or use of viral inducers in the presence of ART (reviewed in ref. ⁴). Various latency reversing agents (LRAs) have been used to kick (or shock) infected cells out of latency and induce their death by virus-induced cytopathic effects, but a kick alone may not be sufficient in eliminating the HIV reservoir⁵. Indeed, we have previously described a synthetic bryostatin analog SUW133, which by itself reversed latency and induced the death of a subset of previously latent cells^6,7,8, suggesting that the addition of a dedicated kill agent could further diminish the viral reservoir. Kill approaches have included immunological therapies that target HIV-infected cells by enhancing endogenous antiviral immune responses or harnessing of antibody-based or effector cell therapies (reviewed in⁹). One combinatorial kick and kill approach comprised of multiple LRAs combined with broadly neutralizing antibodies (bNAbs) 3BNC117, 10-1074, and PG16 administered during ART decreased viral rebound after ART interruption in humanized mice infected with HIV-1¹⁰. Other kick and kill approaches utilizing a TLR7 agonist and Ad26/MVA vaccine¹¹ or bNAb PGT121¹², decreased viral rebound after ART interruption in rhesus monkeys infected with SIV or SHIV, respectively. These preclinical studies demonstrate the enhancement of endogenous antiviral immunity or the addition of bNAbs as a dedicated kill agent to LRAs diminishes viral rebound from the reservoir. Other potential strategies towards an HIV cure, including the combination of latency reversal with anti-HIV cellular therapies, represent promising approaches that are currently understudied. Here we show injections of allogeneic human peripheral blood natural killer (NK) cells alone slow and sometimes prevent a viral rebound in HIV-infected humanized mice. We also demonstrate a single administration of the protein kinase C modulator and latency reversing agent SUW133 followed by injections of these NK cells during ART further decreases rebound frequency and delays rebound when it occurs, which provides proof-of-concept that a kick and kill strategy can effectively target the HIV reservoir. Importantly, we utilize genetic barcoded virus technology to show these treatments also reduce the diversity of the HIV reservoir, thus demonstrating adjunct interventions may eliminate the reservoir towards a complete functional cure.

Results

NK cells are activated by allogeneic HIV-infected CD4 + T cells

NK cells rapidly target and kill HIV-infected cells, which is important to early control of HIV infection and AIDS^{13,14,15,16,17,18,19}. However, endogenous cytotoxic effector cells including NK cells may be diminished or impaired during chronic HIV infection and not fully restored despite achieving viral suppression^20,21,22,23. Allogeneic peripheral blood NK cells have been shown to efficiently inhibit viral replication in vitro²⁴ and acute HIV infection in a humanized mouse model²⁵. Recently, activated NK cells primed with IL-15 inhibited latently infected cells from propagating infection in vitro²⁶, suggesting NK cells as a promising therapy to treat viral rebound. Thus, we hypothesized that exogenously administered NK cells might be a valuable component of HIV reservoir elimination strategies intended to delay or prevent HIV rebound after ART interruption.

We first sought to investigate whether peripheral blood NK cells could be utilized as an efficient kill agent against cells productively infected with HIV. We obtained human peripheral blood NK cells from the peripheral blood mononuclear cells (PBMCs) of four healthy donors using magnetic bead isolation. We used Uniform Manifold Approximation and Projection (UMAP) to visualize subpopulations among the CD56⁺CD3⁻ cells from concatenated and individual samples by flow cytometry (Supplementary Fig. 1a,b). We next manually assigned clusters based on the differential expression of eleven surface markers and identified five subpopulations, including the canonical NK cell subsets: CD56^dimCD16⁺ (populations 1-4) and CD56^brightCD16^– cells (population 5) (Supplementary Fig. 1c). As expected, CD56^dimCD16⁺ NK cells were the predominant cell type in the peripheral blood (Supplementary Fig. 1d)^27,28,29. Among the CD56^dimCD16⁺ NK cells, we identified four subpopulations based on the differential expression of CD57, KIR2DL1/S1/S3/S5, and natural cytotoxicity receptor (NCR) NKp44 (Supplementary Fig. 1c)^29,30. In comparison to CD56^dimCD16⁺ NK cells, the CD56^brightCD16^– cells demonstrated increased expression of NCR NKp46 and inhibitory receptor NKG2A and decreased expression of KIR2DL1/S1/S3/S5 and CD57, which was consistent with an immature phenotype as previously reported^29,30,31. All subpopulations expressed high levels of the activating co-receptors 2B4 and NKp80 and cell stress sensing activating receptor NKG2D and low levels of NCR NKp30²⁹. These results were consistent with our current understanding that the peripheral blood contains the classical subsets of CD56^dimCD16⁺ NK and CD56^brightCD16^– cells as well as additional heterogeneous subpopulations.

To assess the function of these NK cells, we isolated allogeneic CD4⁺ T cells from the PBMCs of healthy donors and performed CD3 and CD28 co-stimulation for three days prior to infection with 800 ng of p24 of X4-tropic NL4-3 or R5-tropic NFNSX for 24 h. NFNSX is derived from NL4-3 with the envelope cloned from the CCR-5 tropic JR-FL³². The infected or uninfected CD4⁺ T cells were washed and cocultured with allogeneic NK cells at an effector-to-target ratio of 1:1 in media containing 20 ng per ml of recombinant human IL-2 (Supplementary Fig. 2a). Cocultures were analyzed 24 and 48 h later to assess activation of NK cells and frequency of live infected T cells by flow cytometry. After 24 h of coculture the allogeneic NK cells exhibited higher levels of IFN-γ production and CD107a degranulation when cocultured with HIV-infected CD4⁺ T cells compared to allogeneic NK cells cocultured with uninfected CD4⁺ T cells or NK cells cultured alone (Supplementary Fig. 2b), suggesting that allogeneic NK cells were specifically activated by HIV-infected CD4⁺ T cells. These results were consistent with previous findings that allogeneic NK cells are efficiently activated by HIV-infected CD4⁺ T cells^33,34. In addition, NK cells treated with varying doses of cell-free HIV supernatant in the absence of CD4⁺ T cells demonstrated no notable increase in levels of intracellular IFN-γ or degranulation (CD107a) after 24 h or intracellular p24 staining after 48 h (Supplementary Fig. 2c, d), suggesting that cell-free virus did not efficiently activate or replicate in NK cells. As expected, the frequency of infected p24⁺ CD4⁺ T cells was significantly lower among cocultures of allogeneic NK cells with infected CD4⁺ T cells compared to cultures containing infected CD4⁺ T cells without NK cells (Supplementary Fig. 2e, f), indicating that allogeneic NK cells decreased HIV infection in CD4⁺ T cells.

NK cells delay viral rebound of R5-tropic HIV after ART interruption

Next, to study HIV latency in NSG-BLT mice, which achieve a high level of human immune cell reconstitution³⁵, we intravenously injected NFNSX and monitored acute viremia by qRT-PCR in plasma samples for four weeks. Infected mice were administered ART comprised of raltegravir (RAL), emtricitabine (FTC), and tenofovir disoproxil fumarate (TDF) in the animal feed for six weeks (Fig. 1a)⁶. Once viremia was suppressed, ART was interrupted in the presence or absence of 5 × 10⁶ allogeneic human peripheral blood NK cells, which were administered at one and six days post-ART interruption. The donors for NK cell injections were randomly selected based on donor supply and not chosen based on KIR/HLA genotyping. Injections of NK cells in a mouse cohort were from the same human donor. Because there are no robust virologic or immunologic correlates to definitively identify all HIV reservoir cells, monitoring viral rebound after ART interruption was used as a primary endpoint in vivo. One week after ART interruption, rebound viremia occurred in five out of six (83%) mice in the control group and one out of six (17%) mice in the group receiving NK cells (Fig. 1b). After two weeks of ART interruption, all mice in both groups rebounded. The group receiving NK cells exhibited a delay in viral rebound compared to the control group (P = 0.027, Fig. 1c). As expected control mice displayed low frequencies of human CD56⁺CD3⁻ cells in the blood due to poor human NK cell development in BLT mice³⁶ (Fig. 1b, top). Compared to control mice, the mice receiving NK cells had six-fold higher, but overall low levels (~0.8%) of human CD56⁺CD3⁻ cells in the blood five days after the administration of NK cells (P = 0.0022, Fig. 1d), which then declined thereafter (Fig. 1b, bottom). Indeed, the adoptive transfer of allogeneic NK cells has limited engraftment^37,38. There were no notable differences between the frequencies of CD56⁺CD3⁻ cells in the blood, spleen, and bone marrow at necropsy between the two groups (Supplementary Fig. 3a). Although NK cells can downregulate CD56 expression and subpopulations of long-lived memory-like NK cells can exist³⁹, our results likely indicate the overall transient survival of adoptively transferred non-engineered NK cells^40,41.

The levels of pre-ART viral infection and human immune cell engraftment likely did not account for differences in viral rebound because notable differences were not observed between the two groups (Fig. 1e and Supplementary Fig. 3b–e). In addition, mice receiving NK cells did not demonstrate diminished frequencies or absolute counts of human CD45⁺ and CD4⁺ T cells compared to control mice, suggesting allogeneic NK cells did not indiscriminately kill host immune cells, which is consistent with clinical studies demonstrating the overall safety of allogeneic NK cells in contrast to allogeneic T cells, which induce graft-versus-host disease^42,43,44.

In addition, we found the level of viremia, cell-associated HIV RNA, and total HIV DNA of the spleens and bone marrow at necropsy were not significantly different between the mice treated with or without NK cells, which was expected given the rebounding virus was allowed to replicate and reseed new target cells during ART interruption (Fig. 1f–h). These results indicate total HIV RNA and DNA measurements several weeks after ART interruption did not capture the dynamic effect that NK cells had on the early viral rebound.

NK cells also delay rebound of X4-tropic barcoded HIV

Next, we investigated whether HIV-1 barcoded technology could quantify the effect that allogeneic human peripheral blood NK cells had on rebounding viral clones after ART interruption^8,45. As such, we recently developed a genetically barcoded HIV-1 containing a 21 bp genetic barcode inserted upstream of a hemagglutinin tag in the nonfunctional vpr region of the HIV strain NL-HA^8,46, which is an X4-tropic near full-length, replication-competent, pathogenic strain of NL4-3 (Fig. 2a and Supplementary Fig. 4a–c)…