Targeting HIV myeloid and central nervous system reservoirs for HIV cure

Abstract

Purpose of review

Myeloid vs. CD4⁺ T-cell reservoirs have received less attention for HIV cure strategies, mainly due to more limited access to tissues and challenging in vitro and in vivo models, including modeling how myeloid cells affect HIV-associated neurocognitive disorder (HAND). This review highlights recent studies providing insights into myeloid viral reservoirs, new methods to study them, and the strategies to target them.

Recent findings

In addition to studies describing replication competent virus derived from blood monocytes, which correlates with HAND, myeloid-derived virus can be characterized in clinical samples, such as the blood, using virion immunocapture. Characterization of monocyte subsets and pro-inflammatory markers in the blood can also help detect HAND. New humanized mouse models and in vitro organoid models have improved our ability to study central nervous system (CNS) reservoirs and inflammation. Strategies targeting the CNS vs. peripheral reservoirs may need to be fundamentally different to limit inflammation and which may contribute to HAND.

Summary

Insights provided by these recent studies should challenge the field to employ these methods for myeloid reservoir and HAND detection in preclinical and clinical trial studies. Future HIV cure proposals can aim to include a myeloid reservoir component to help guide the design of strategies for inclusive cure strategies.

INTRODUCTION

While CD4⁺ T-cells are the main targets for HIV infection, myeloid cells can also be infected by HIV. However, knowledge of infected myeloid cells is limited compared to their T-cell counterparts, in part due to the challenge of accessing tissue samples from people with HIV (PWH). Macrophages are tissue-resident immune cells that are derived from either the yolk sac or hematopoietic stem cell (HSC) monocytes. In mice, the source of macrophages varies depending on the tissue and the age of the animal [1]. Importantly, while monocyte-derived macrophages that infiltrate into tissues during injury and infection are relatively short-lived, yolk sac-derived tissue macrophages can self-renew. Unfortunately, this myeloid composition within human tissues is still unclear. In the context of HIV, understanding what subsets of tissue-resident macrophages are infected, the mechanisms that contribute to their persistence during antiretroviral therapy (ART), and the consequences of infection in promoting inflammation are essential for developing inclusive cure strategies that target the reservoir and resolve inflammation-driven comorbidities. This is especially important for microglia in the brain that can drive inflammation and contribute to the development of HIV-associated neurocognitive disorder (HAND), which affects ~50% of PWH [2]. This review will focus on these points.

UNDERSTANDING MYELOID AND CENTRAL NERVOUS SYSTEM RESERVOIRS FOR HIV CURE

In addition to viral latency contributing to reservoir persistence, in vitro studies suggest that myeloid cells do not die from viral cytopathic effects [3–7] and are resistant to immune-mediated elimination by CD8⁺ T-cells (CTLs) and natural killer cells (NK cells) [8–12]. As many macrophage subsets express high levels of MHC-II, direct interactions with CD4⁺ T-cells are frequent, and can result in the formation of virological synapses and efficient cell-to-cell spread of the virus [13,14]. Despite this work, skepticism about the in vivo relevance of myeloid reservoirs can be attributed to several past studies. First, TCR/CD3 DNA has been found in infected tissue macrophages, suggesting the appearance of infection may be due to phagocytosis of infected CD4⁺ T-cells [15,16]. Second, most HIV envelope (Env) sequences derived from non-central nervous system (CNS) human samples require high density surface CD4 expression to establish cell-free infection (termed T-cell or T-tropic virus). In contrast, macrophage (M)-tropic viruses represent distinct viral lineages that have evolved to infect cells that express low levels of surface CD4 and are usually derived from CNS samples [17–19]. Below we will focus on recent studies that have characterized myeloid reservoirs and address the points described above (more comprehensive descriptions of myeloid reservoirs can be found in [20]). We will also discuss novel in vitro and in vivo models to characterize infection in myeloid cells and put these into context for testing immune- and drug-based strategies to target myeloid reservoirs. We finish with highlighting the challenges faced with studying and targeting the CNS, which go beyond the blood–brain-barrier (BBB).

New insights into myeloid reservoirs

Myeloid phagocytosis of infected CD4⁺ T-cells has presented a challenge for confirming the relevance of the myeloid reservoir, specifically whether these cells harbor inducible, infectious proviruses that can spread infection. Elegant work by Veenuis et al., building on 10+ years of assay development [21–24], employed a human monocyte-derived macrophage quantitative viral outgrowth assay (MDM-QVOA), which had little to no T cell contamination, to show that replication-competent virus, distinct from inducible CD4⁺ T-cell virus from the same donors, can be obtained from blood monocytes from ART-treated PWH [25▪▪]. Adapted intact proviral DNA assays (IPDA) confirmed that monocytes from up to 40% of ART-treated PWH contained intact proviral DNA, but not RNA, suggesting a latently infected circulating myeloid population in virally suppressed PWH. A recent follow up study-showed that 92% of ART-treated PWH had detectable HIV DNA in monocytes, with 38% having intact proviral DNA [26▪]. Higher frequencies of intact DNA in monocytes, but not CD4⁺ T-cells, were associated with frequencies of intermediate monocytes (CD14⁺CD16⁺) and poor cognitive function. Interestingly, intermediate monocytes exhibit significantly higher CCL2-mediated BBB transmigration in in vitro models, which is greater for samples obtained from PWH exhibiting HIV-associated neurocognitive impairment (HIV-NCI) [27▪]. In vivo, transmigration of infected cells may also be enhanced by serum ATP, which is elevated in PWH exhibiting neurological symptoms [28▪]. Whether HIV infection is enriched in intermediate monocytes that could contribute to reservoir seeding of the CNS warrants further investigation. As blood monocytes only persist for up to ~72 h in the blood, it is unclear how a reservoir could be maintained in this population. Reservoirs in HSC progenitors are controversial [29–32], thus additional studies are needed to address whether long-lived progenitors are the soucre of infected monocytes. The life of the monocyte does not end in the blood; latently infected monocytes that differentiate into macrophages may seed tissue reservoirs. While the lifespan of a monocyte-derived macrophage (MDM) may be short [33▪], if latency is reversed, the infection might prolong their lifespan in vivo once they differentiate in the tissues [3–7].

Many M-tropic viruses are derived from CNS samples and it is well reported that infected microglia/macrophages are found in the CNS of PWH [20]. However, this tissue site has been notoriously difficult to study. Recently, in brain samples from ART-treated PWH brain, Eddine et al. characterized some viral transcripts exhibiting blocked transcription elongation [34▪▪]. However, HIV Gag protein was detected in the CD68⁺ myeloid cells in CNS tissues from ART-treated individuals that expressed elongated transcripts. This aligns with a previous study describing outgrowth of virus from microglia isolated from rapid autopsy brain tissues of ART-treated PWH [35]. Understanding how CNS infection drives the development of HAND is key to combatting neuropathogenesis. A recent study characterizing inflammatory pathways in non-human primate (NHP) brains during acute infection revealed signatures of myeloid cell clusters/subsets (microglia and CNS macrophages) associated with prion disease, Parkinson’s disease, Alzheimer’s and Huntington’s disease, and Amyotrophic Lateral Sclerosis [36]. How this translates to ART-treated scenarios is unclear but provides a framework for understanding how infection contributes to HAND. Inflammation may arise from neurosymptomatic HIV-1 cerebrospinal fluid (CSF) escape, which Kincer et al. show is associated with partial virus resistance to antiretrovirals (ARVs), increased viral diversity in the CSF, and T-tropism, which together suggest active viral replication in CNS CD4⁺ T-cells [37▪]. Optimization of ART regimens with better CNS penetration may help impede this viral replication and ameliorate neurologic symptoms. Finally, the role of other infected cell subsets, including CNS pericytes, also need to be considered for a full understanding of reservoir persistence in this compartment [38].

Beyond the CNS, work by Johnson and colleagues recently described myeloid-derived virus in the semen of ~54% of ART-treated men [39▪▪]. Cell-source virus immunocapture, a technique also used by other groups [40,41], magnetically captured viral particles from samples using antibodies against lineage-specific surface proteins (e.g., CD3 and CD14) [42], which identified the cell subset from which the virus bud. Under suppressive ART, semen-derived virus was genetically distant from blood-derived virus, suggesting ongoing virion production from tissue reservoirs, supporting past studies describing infected macrophages in the urethral tissue of ART-treated PWH [43,44]. Importantly, given the very low copy number, virus in the semen of ART-treated PWH is unlikely to permit transmission, aligning with the treatment as prevention studies [45].

Additional studies of note include a comprehensive characterization of lymph node dendritic cells (DCs), which harbor intact provirus that was inducible and infectious following TLR7/8 stimulation [46▪]. While in vitro studies confirmed the ability of DCs to support replication of M-tropic virus, most clinical strains are T-tropic, raising the question of how DCs are infected in vivo. Indeed, myeloid cells harboring intact proviruses in the duodenum and colon of cART-treated PWH were found to express T-tropic Envs [31]. Separate groups have shown that phagocytosis of infected CD4⁺ T-cells can lead to productive infection of macrophages [47], even when the T cells are infected by T-tropic viruses [48–50]. An intriguing in vitro study by Woottum et al. showed that when infected CD4⁺ T-cells fuse with macrophages, the T-cell nuclei can remain transcriptionally active and produce progeny virions [51▪▪]. Thus, even myeloid cells positive for T-cell DNA/RNA could generate virus, but additional investigation is required to assess whether this translates in vivo.

In vivo and in vitro models to study myeloid and central nervous system infections

One advantage of the NHP model to study viral reservoirs and disease pathogenesis is the access to tissues populated by both yolk-sac and HSC-derived myeloid cells. However, the former is challenging to distinguish from the latter without prior genetic manipulation of gametes to introduce reporters. Rahmburg et al. have recently developed a model of NHPs transplanted with barcoded HSCs to track myeloid population of tissues [33▪]. Their experiments suggested that myeloid cell turnover in tissues is significant over the course of two weeks, however, only the LN and spleen were sampled and it is unclear whether this turnover would apply for other tissues, where the frequencies of yolk sac-derived macrophages are higher (as described in mice) [1]. The NHP/SIV infection model has also been useful for the characterization of viral neuro-pathogenesis (reviewed in [52]), but the high cost and limited access has necessitated development of additional models.

While humanized mice lack yolk sac-derived macrophages, HSC-derived myeloid cells engraft into the brains of humanized mice in certain models that recapitulate aspects of neuro-HIV pathogenesis (reviewed in [53]). Most recently, Ghosh Roy and colleagues describe HIS-DRAGA mice that exhibit reconstituted CNS microglia-like cells and brain-resident T-cells, a step towards modeling HIV pathogenesis in the CNS [^54▪]. To characterize the functional role of myeloid cells during HIV infection, Baroncini et al. developed the inducible human myeloid depletion (iHMD) model, which uses CRISPR/Cas9-mediated depletion of human myeloid cells [55]. Here, HSCs were transduced with a lentiviral construct containing an apoptosis inducible Cas9 under the synthetic promoter p47-SP107 (made of cis elements from native myeloid promoters [56]), which triggers cell death upon administration of the molecule AP1903 [57]. This model not only showed that myeloid cells are the source of many pro-inflammatory cytokines, but they also exhibit antiviral effects during viremic infection. Further mechanistic work is needed to characterize the roles of specific myeloid subsets, and which myeloid effector functions contribute to control of infection. This may likely include production of IFN and phagocytosis of infected cells.

In vitro models of HIV infection traditionally use human MDMs, which have provided valuable insights into the effects of infection on macrophage phenotype and function. Furthermore, MDMs permit co-culture with autologous CTL/NK cells, which we and others have used to show that macrophages vs. CD4⁺ T-cells are relatively resistant to cytolytic elimination [8–12]. Using these MDM co-cultures, Mensching et al. recently showed that HIV-infected macrophages primed NK cell cytokine production, but not cytotoxicity [58▪], which may be applicable for scenarios of latency reversal in macrophages. Latency studies in macrophages can be challenging given the difficulty of manipulating primary MDMs, but the THP-1 cell line, which is amenable to genetic manipulation, permits mechanistic latency studies for myeloid cells [59]. iPSC-derived macrophages were recently shown to support HIV infection to a similar extent as MDMs, and could be used to complement the MDM studies [60▪]. Microglia incorporation into brain organoids have yielded a more relevant model for HIV CNS research [61,62▪,63]. In one model, iPSC-derived microglia are infected with HIV and added to cerebral organoid slices. These studies showed that infection of microglia drives organoid type-I interferon (IFN) signaling and neuroinflammation, which is not reduced by ARVs [62▪,63]. Furthermore, iPSCs made with hematopoietic progenitor cells during formation of embryoid bodies supports the development of microglia, which are susceptible to HIV infection and exhibit pro-inflammatory responses [64▪]. This aligns with a recent study by Ramaswamy and colleagues, where MDA5 sensing of infection and activation of IRF5 drives IFN expression in MDMs [65▪]. Interestingly, IRF5 expression and pro-inflammatory responses to infection in myeloid cells were greater in older vs younger individuals. If applicable to microglia, this mechanism could further exacerbate HAND with age. Finally, ɑ-synuclein fibrils, associated with Parkinson’s Disease pathogenesis, enhance the susceptibility of macrophages, microglia, and CD4⁺ T-cells to HIV infection [66▪]. This highlights potential synergistic effects between HIV infection and the development of neurological disorders, which need to be accounted for in treatment strategies as PWH age. Together, insights from these models could guide future investigations using CNS samples acquired from recently deceased PWH who have contributed their tissues through the last gift program [67].

Drug and immune targeting of myeloid and central nervous system reservoirs

The development of HAND is largely attributed to persistent inflammation in the CNS, potentially due to poor ARV penetrance and subsequent detectable CSF viral loads, even when the virus is undetectable in the plasma [37▪,68,69]. Neuropsychological testing is used to measure cognitive impairment, and work is ongoing to identify biomarkers available in CSF or blood that could facilitate early detection of HAND [70]. A recent study describes extracellular ATP in the blood as a correlate of neurological symptoms [28▪], which could be a marker incorporated into HIV cure studies. While improving ARV CNS penetrance is an active area of research [71,72▪], small molecules that target immune signaling pathways that contribute to viremia, especially in myeloid cells, are also under development. A CSF1R antagonist, which reduced perivascular macrophages but not microglia, can reduce viral load in brain tissues during acute SIV infection in NHPs, however there was no reduction in CSF viral load [73▪]. As CSF1R antagonists have been used to broadly deplete microglia in mice, toxicities and potential worsening of CNS health with this compound should be explored [74]. The tyrosine kinase inhibitor dasatinib has also been shown to target HIV infection in both CD4+ T-cells and MDMs, with infection-induced proinflammatory cytokines reduced in the latter, but it remains to be seen whether this drug is relevant in the CNS [75]. Neuronal loss associated with severe HAND has been partially attributed to neurotoxic effects of the HIV protein Tat, which is secreted from infected cells and taken up by neurons [76]. A series of endocytosis inhibitors was recently developed to block uptake of Tat in the SH-SY5Y neuronal cell line, and may prove to be an effective treatment for HAND [77▪].

Many HIV cure approaches currently under investigation, including broadly neutralizing antibodies (bnAbs), latency reversing agents (LRAs), and biologics/vaccines to boost cellular immunity, encounter additional difficulties in targeting the CNS. Transport of antibodies and other macromolecules across the blood brain barrier is limited (~0.1% of circulating antibodies reach the CNS [78]). Optimization of antibody design to improve CNS penetrance is ongoing; many of these strategies fuse the antibody with transcytosis-enabling modules, which facilitate protein cargo uptake by receptors naturally positioned at the BBB, such as the transferrin receptor (reviewed in [79]). Even so, in the SIV/NHP model, development of neutralizing IgG was observed in animals that did not progress to SIV encephalitis, suggesting that antibodies may help limit CNS pathogenesis [80▪▪]. For “shock-and-kill” strategies, several LRAs, including epigenetic modulators (SAHA and CM272) and TLR stimuli (specifically TLR7/8), have been used to reverse latency in cultured microglia and DCs, respectively, however, results with other LRAs in myeloid cells have been mixed [35,46▪]. Furthermore, understanding how the LRAs affect both transcription initiation and elongation will be important for stimulating the cells to produce viral protein [34▪▪], which could then be targeted by the immune system. In addition to the expression of viral antigens, Tat expression in microglia upregulates TREM1, which enhances the survival of microglia. Strategies to target this protein might be viable strategies to deplete HIV-infected microglia [81].

A potential drawback of latency reversal in the CNS is the accompanying induction of inflammation due to viral RNA/proteins or the LRAs themselves, which could worsen HAND symptoms [82–84]. Therefore, “block-and-lock” strategies may be more appropriate for HIV reservoirs in the CNS for limiting neuroinflammation (recently reviewed in [85]). While not exclusively focused on myeloid cells, certain compounds affecting splicing of HIV transcripts show efficacy for both CD4⁺ T-cells and MDMs [86,87]. In addition, the BRD4 inhibitor, ZL0580, which can block HIV transcription in myeloid cells [88], could potentially be used in combination with LEDGINs, which were recently shown to enhance blockade of HIV transcription and reactivation in primary cells [89]. These strategies warrant further investigation in in vivo models.

CONCLUSION

Recent studies characterizing myeloid reservoirs support the need to consider the effects of HIV cure strategies beyond the T-cell compartment. Access to tissues, including the CNS, will limit the study of infected cells in these compartments. However, characterization of monocyte reservoirs in the blood, immunocapture of virions to distinguish myeloid vs. T-cell-derived virus, and assessment of markers associated with neurological symptoms, should be considered for analysis of therapeutic strategies. Improved humanized mouse models that recapitulate human myeloid cell reconstitution are an alternative to the NHP model, but brain organoids also provide the opportunity to study the effects of candidate therapies to target myeloid cells. While “shock-and-kill” strategies remain of interest for reservoir eradication, excessive immune activation may worsen neurological symptoms. Therefore, silencing the HIV reservoir through “block-and-lock” strategies may prove more desirable for targeting CNS reservoirs.

KEY POINTS.

• Latently infected monocyte reservoirs may contribute to viral seeding in the tissues and neuroinflammation.

• Genetic manipulation of hematopoietic stem cells allows tracking of myeloid cells in non-human primate tissues, which could be used to track viral reservoir longevity in these cells.

• New humanized mouse models provide additional tools to study human myeloid cells in the central nervous system (CNS), but still lack yolk sac-derived microglia.

• Latency reversing agents (LRAs) specific for myeloid cells will need to be included with T-cell targeting LRAs for “shock-and-kill” strategies.

• “Block-and-lock” strategies may help mitigate CNS inflammation associated with expression of viral products.