Using animal models to overcome temporal, spatial and combinatorial challenges in HIV persistence research

A limited number of animal research studies have employed the qVOA to quantitate the latent viral reservoir. These include a pair of NHP studies and a trio of humanized mouse studies that defined the viral reservoir during ART. In the NHP studies by Shen et al. and Dinoso et al., either 2- or 3-drug ART was administered for 5–6 months during which low level plasma viremia remained detectable in most animals [17, 23]. Replication competent persistent virus was detected in cells isolated from peripheral blood and lymph nodes at very low IUPM in both studies. In the three humanized mouse studies, the qVOA was performed on resting memory CD4⁺ T cells that were pooled from multiple organ systems within the same animal [65, 67, 69]. The humanized mice utilized by Choudhary et al. and Denton et al. harbored a full complement of human hematopoietic cells while the mice used by Honeycutt et al. were reconstituted solely with human T cells [T-cell only mice (TOM)] (Fig. 3). In all 5 studies, resting memory CD4⁺ T cells harboring replication competent virus were recovered at IUPM levels analogous to patients on ART. As no animal study to date has continued ART beyond 6 months, and some of the NHP data was generated in untreated animals, the published IUPM values in these models include both stably integrated persistent virus as well as unintegrated virus [9].

While several NHP studies have focused on neurological tissues and the potential role of myeloid-lineage cells as viral reservoirs, the presence of T cells in the animals increases the challenges associated with defining the cell lineage(s) harboring persistent virus [13, 22, 27, 28]. To address this, a novel use of humanized mice has been recently described by Honeycutt et al. [77]. The mice used in this study are NOD/SCID animals that received a bone marrow transplant of hHSCs (Fig. 2) such that the animals are systemically reconstituted with human monocytes/macrophages, B cells and dendritic cells [108]. Notably, these mice lack human T cells [109]. These mice, referred to as myeloid-only mice (MoM), exhibit a general lack of detectable viral DNA in their peripheral blood cells during HIV infection. This is reversible by the administration of granulocyte colony stimulating factor (G-CSF) which leads to blood cell viral DNA becoming detectable within 6 days of administration. This finding could mean that circulating monocytes are generally not infected in this model, but mobilization of the tissue macrophages with G-CSF leads to the presence of infected cells in the blood [77]. MoM have only been presented in a conference abstract to date. The publication of comprehensive data from this model is likely to increase our understanding of the role played by the myeloid lineages in HIV persistence.

Hematopoietic stem cells are long-lived, self-renewing cells that have the potential to perpetuate integrated HIV DNA for a very long time. Carter et al. showed that hHSC infected with CXCR4-tropic HIV and then transplanted into immunodeficient mice are capable of multi-lineage engraftment of these animals [79]. Subsequently, Nixon et al. showed that CXCR4-tropic HIV can infect hHSC in vivo in humanized mice [78]. However, the bulk of HIV variants do not use CXCR4 as a co-receptor for viral entry which should limit the contribution of HSC in virus persistence. This conclusion is consistent with the observation that HIV DNA was not detected in highly purified CD34⁺ HSC from patients on long-term ART [110].

In addition to hHSC, the Zack Laboratory has also examined the role of splenocytes [71] and thymocytes [61‐64, 70, 72] in HIV persistence in humanized mice. Viral persistence in splenocytes infected with a murine heat-stable antigen (HSA)-reporter virus was studied by Marsden et al. in the absence of ART because this reporter system allowed them to evaluate viral persistence in the context of ongoing virus replication in the untreated animals [71]. To identify persistently infected cells ex vivo, HSA⁺ splenocytes were immuno-depleted. The HSA^neg splenocytes were then stimulated in culture in the presence of raltegravir to inhibit virus spread and to limit unintegrated viral DNA from confounding the data interpretation. Culture supernatants were examined for Gag^p24 by ELISA and it was determined that splenocytes did harbor latent HIV in vivo but the specific cell lineage(s) that harbored persistent virus were not elucidated in this study. Years earlier, this same reporter virus was used by Brooks et al. to demonstrate that thymocytes also could harbor persistent HIV [61]. The mice used in the earlier study were engrafted with a human thymus, but did not exhibit systemic human engraftment, such that the only anatomical compartment examined in this report was the thymus. A subsequent study on HIV persistence in thymocytes showed that harboring a transcriptionally silent HIV provirus did not alter cell surface phenotypes during differentiation and maturation of T cells [62]. In contrast to the humanized mouse studies, however, Shen et al. reported that thymocytes were not the source of persistent virus in a NHP model when the animals were undergoing ART [23].

Together, these studies indicate that hematopoietic stem cells, splenocytes and thymocytes all have the potential to serve as cellular reservoirs of persistent virus, although these cell types likely do not harbor the bulk of virus that persists during long-term ART in patients. In addition, the replication competence of persistent virus in these cell lineages remains an open question.

Multiple research groups have sought to characterize the anatomical distribution of persistent virus during ART in NHP models. Three groups performed extensive survey studies and reported on the levels of persistent virus present in tissues throughout the body [20, 21, 29, 35]. In these papers, viral DNA and RNA quantitations from lymphoid, gastrointestinal, neurological and reproductive tissues as well as other tissues from the urinary, respiratory and circulatory tracts showed that the bulk of persistent viral nucleic acids were present in lymphoid and gastrointestinal tissues during ART. In a more tissue specific study, Matusali et al. found that virus persisted at higher levels throughout the male genital tract and in the semen of adult cynomolgus macaques on ART than was observed by North et al. in juvenile rhesus macaques [21, 33]. Matusali et al. also found that virus shedding in semen during ART was reduced in some animals following ART intensification with raltegravir. Viral persistence in the central nervous system of NHP was the focus of a series of manuscripts from the Clements Laboratory [13, 22, 27, 28]. Beyond the importance of identifying persistent virus in this anatomical location, a key finding in these works was the observation that virus coding for cytotoxic T lymphocytes (CTL) escape mutations were archived in the central nervous system during ART [22]. Surprisingly, given the intense focus on the intestines as a major anatomical site harboring persistent virus [20, 21, 29, 35, 111‐124], there is no published evidence from humans or animal models that replication competent, latent HIV proviruses persist in the intestines during ART.

Common to these anatomical location analyses were their reliance upon nucleic acid measures of viral persistence. The HIV DNA measures do not distinguish between defective archived virus and replication competent virus. Thus, these nucleic acid measures overestimate the amount of replication competent virus that must be targeted by any successful HIV eradication strategy [9]. The presence of HIV RNA indicates that expression has occurred recently, but neither nucleic acid determination can measure the level of replication competent, transcriptionally silent proviruses [88].

Lymphoid tissues, particularly lymph nodes, were the focus of several other virus persistence studies [17, 18, 23]. These studies reiterated the importance of lymphoid tissues as anatomical locations that harbor persistent virus during ART. Fukazawa et al. examined the architecture of lymph nodes as a means to gain mechanistic insights into the reason that viruses persist in these organs [18]. They noted that virus producing CD4⁺ T follicular helper cells in the B cell follicles were not susceptible to immune clearance. They identified B cell follicles as regions of the lymph nodes that experience very limited infiltration by CTL. CTL exclusion from B cell follicles was described in elite controller NHP although this phenotype would also be expected during ART. In addition to nucleic acid measures of viral persistence, this study included the use the qVOA to demonstrate that the persistent viral reservoirs in the lymph nodes include replication competent virus.

Santangelo et al. recently described a novel immunoPET scan technique that they were able to use for non-invasive identification of the anatomical locations with ongoing virus production during ART in NHP [41]. This study showed that cells in lymphoid tissues and the intestines continue to exhibit virus protein expression during ART which is in keeping with other studies in animal models and patient biopsy data. The key contribution of this study is its description of a technique for the detection of virus production in multiple anatomical sites simultaneously that has the potential to be refined and applied broadly in HIV cure research.

The concept of viral reactivation in the context of a “kick and kill” eradication strategy is being pursued aggressively in clinical trials [129] and most agents in testing for HIV therapy are being developed for or are already in clinical use for oncology interventions [104, 130‐133]. Animal models are ideal in vivo experimental platforms for the efficacy evaluation of not-yet-approved pharmaceuticals for their capacity to reactivate latent viruses. In a series of papers from the Zack laboratory, humanized mice were used to generate latently infected cells which were subjected to a range of stimulatory agents ex vivo (e.g., prostratin, PMA, PHA, and cytokines) to identify T cell signaling pathways involved in latency reversal [63, 64, 70, 72]. In a separate ex vivo study, Shen et al. determined that reactivation conditions for SIV in NHP cells (from animals described in Ref. [23]) did not overlap fully with conditions used for reactivating HIV from human cells in the context of the qVOA [24]. This outcome indicates that caution should be exercised when translating latency reactivation in vivo in NHP to clinical expectations; however, similarities between NHP and human reactivation data have been recently reported. Specifically, vorinostat given to patients [104, 130, 133] and NHP [37, 38] during ART caused modest increases in cell-associated viral RNA. The similar results in both in vivo systems are encouraging, but the characterization of latency reversal efficacy in NHP and patients during more potent latency reversal therapy will provide stronger confirmation of the translational nature of the NHP to clinical outcomes in such studies.

A robust example of the utility of animal models in overcoming combinatorial research challenges is the work of Halper-Stromberg et al. in humanized mice. This group combined multiple strategies to reactivate latent virus with strategies to improve antiviral immunity and killing of infected cells [74]. A key experiment in their study was to serially dose (five times) HIV-infected humanized mice with a mix of three broadly neutralizing antibodies (i.e., 3BNC117, 10-1074 and PG16) which suppressed plasma viremia. Then the animals received three separate doses of a combination of three virus reactivating agents. The three virus inducers used in the combination were vorinostat, an HDAC inhibitor; I-BET151, a BET protein inhibitor; and anti-CTLA-4 antibody, a T cell inhibitory pathway blocker. The remarkable observation made during this experiment was: even in the absence of ART, the combinatorial eradication strategy employed prevented viral rebound in 57 % of animals. Even though the dosing and timing of dosing were not yet optimized in this study, these observations strongly suggest that combinations of viral inducers appear to be synergistic in vivo.

Active immunization, as well as passive immunization, are strategies being investigated for improving antiviral immunity during virus eradication therapy. Hansen et al. performed an active immunization of NHP using a cytomegalovirus (CMV) vector expressing SIV proteins [19]. This study generated excitement because long-term control of viremia was observed in the vaccine arm of the study. Amazingly, a cohort of eight vaccinated animals ceased to exhibit any plasma viremia by ~70 days post immunization and remained viremia free until necropsy (>160 days post immunization). This effect has been attributed to atypical CD8⁺ T cell targeting by the CMV vector [42]. At harvest the qVOA was utilized to demonstrate that no replication competent virus could be recovered from multiple tissues in the immunized animals. This outcome is very encouraging and the hurdles to implementing this strategy in clinical trials are being addressed.

Passive immunization with broadly neutralizing antibodies has been evaluated for its therapeutic potential in humanized mouse and NHP HIV models. The antibodies were administered either in combination with [75] or in the absence of [16, 73, 76] ART. In all cases, significant reductions in viremia were associated with antibody administration. Furthermore, broadly neutralizing antibodies combined with ART also led to a reduction in cell-associated viral DNA. Immunotoxins are a form of passive immunization where Fab fragments of antibodies are fused with a toxin to deliver the deadly payload to virus expressing cells [134]. Brooks et al. showed that an immunotoxin that specifically targets HIV envelope-expressing cells kills virus expressing cells following latency reversal ex vivo [63]. Subsequently, Denton et al. used this immunotoxin to reduce viral persistence in vivo. The immunotoxin was administered to humanized mice over a 2 week period in conjunction with continued ART [68]. The outcome of this relatively short immunotoxin therapy in vivo was an additional 1 log₁₀ reduction in tissue viral RNA levels beyond the impact of ART alone.

In addition to immunization, immunomodulatory strategies also have the potential to boost antiviral immunity against cells where latent virus has recently been reactivated. When toll-like receptors (TLR) recognize distinct foreign molecular patterns, they activate innate and adaptive immune cells. Whitney et al. recently published a conference report describing the effect of a TLR7 agonist given to ART-suppressed NHP [31]. In their study, they observed both reduced viral DNA in blood and tissues and a marked induction of plasma viremia following multiple oral administrations of the inhibitor. Furthermore, the viral set point in these animals was lower during ATI versus animals that did not receive the TLR7 agonist. The surprising outcome of this study suggests that it may be possible for a single molecule to function as both the “kick” and the “kill” component of a virus eradication strategy. At the least, such a TLR agonist could augment the reactivation of latent virus while simultaneously functioning as an immune enhancer.

Recently, Deng et al. showed that it is not sufficient to simply boost antiviral immunity, especially in patients who initiated ART during chronic infection [66]. The experimental conditions were highly controlled in this study as three key experimental components were independently derived from a single human donor who initiated ART during chronic infection. The three components were: (1) virus grown out from the latent reservoir using the qVOA; (2) ex vivo stimulated CD8⁺ T cells; and (3) humanized mice generated from bone marrow-derived HSC. The humanized mice were infected with virus from the patient’s latent viral reservoir and then received an adoptive transfer of the ex vivo stimulated CD8⁺ T cells. The key variable was the reagent used for the ex vivo CD8⁺ T cell stimulation. When the stimulation included immune-dominant gag-epitopes based upon the patient’s HLA genotype, the adoptively transferred cells were unable to control viremia in the recipient humanized mice. However, when the stimulation did not include epitopes for which escape mutations were pre-existing, the adoptively transferred cells were capable of controlling viremia in the recipient humanized mice. This study yielded two very valuable contributions. First, when ART is initiated during chronic infection, the latent virus reservoir harbors CTL escape variants. Second, vaccination strategies in patients may need to be tailored for individual HLA genotypes to avoid stimulating ineffective antiviral immunity.

Different strategies to suppress residual virus replication during ART have been tested in animal models. North et al. evaluated ART intensification regimens in NHP [32]. Animals received either 3-, 4-, or 5-drug ART regimens, but no major differences in viral decay kinetics were observed and all animals experienced viral rebound during ATI. Using a complicated delivery strategy, Shytaj et al. combined ART intensification in NHP with auranofin and buthionine sulfoximine [14, 15, 25]. The net effect of this regimen was a substantial alteration in the natural course of infection that appeared to limit disease progression to AIDS, but did not limit viral rebound during ATI. Dunham et al. coupled an indoleamine 2,3-dioxygenase inhibitor with ART in NHP, but did not observe any alteration of the inflammatory environment in vivo, as hypothesized, nor did they observe an impact on persistent infection [36]. Moniuszko et al. performed a similar study with glucocorticoid to reduce the number of proinflammatory CD16⁺ monocytes present in NHP on ART [40]. This regimen did reduce the target cell population, but did not alter the persistent viral DNA levels relative to ART alone. Mavigner et al. performed HSC transplants in NHP during ART using autologous cells to determine the impact of myeloablative total body irradiation plus stem cell transplant on viral persistence [34]. One to two months following the transplant, ART was discontinued in the experimental and control arms of the study. Rapid viral rebound and similar tissue viral nucleic levels characterized in the animals from both groups at necropsy indicated that neither the act of conditioning for a stem cell transplant nor the transplant itself were sufficient to significantly impact viral persistence. Holt et al., Neff et al., Kitchen et al. and Shimizu et al. employed various gene editing strategies to successfully reduce viral burden by reducing the number of human T cells susceptible to viral entry in HIV infected humanized mice in the absence of ART [81‐84]. While these pioneering studies did not achieve the ultimate therapeutic goal of virus eradication, they confirm the feasibility of performing preclinical combinatorial trials in animal models. Regarding future animal model studies in HIV persistence, it is noteworthy that high costs and animal availability are major concerns for researchers conducting HIV persistence studies in animal models—particularly for investigators utilizing NHP. The creation of a core facility that supplies ART-suppressed animals to HIV persistence researchers could address these concerns, accelerate the work and lower the costs associated with these lines of research.