Introduction

Most past and ongoing HIV eradication targeted clinical trials have narrowly focused on evaluating and targeting the long-lived memory CD4+ T cells, as these cellular reservoirs harbor the majority of the cell-associated HIV using current reservoir quantification assays. As treatment interruption of antiretroviral therapy (ART) remains the gold standard in determining the outcome of HIV curative designed clinical trials, the contribution of viral persistence in myeloid cells, such as blood monocytes and tissue macrophages, has not been at the forefront. This partly stems from the debate in the field on the scope of myeloid cells as viral reservoirs or contributors to viral persistence. However, recent studies now provide compelling evidence that blood monocytes and tissue macrophages in the lung [1, 2], adipose tissue [3], gut-associated lymphoid tissue (GALT) [4, 5], genital tract [6], semen [7], and bone marrow [8, 9] as well as central nervous system (CNS) cells including both myeloid originating microglia and astrocytes of the brain [10, 11]; all have been shown to harbor HIV in the setting of suppressive ART. Thus, in order to achieve full remission, we need to rethink the design and implementation of HIV cure-focused trials for elimination of all cellular reservoirs.

Two prominent cases that may shed light on the importance of myeloid cells as viral reservoirs in cure studies may be informative in this regard. In the case of the “Berlin Patient” [12], who received a bone marrow transplant and successfully achieved viral remission after stopping ART without subsequent viremic rebound for over a decade now, the exact mechanisms for this successful trial, which to date has not been replicated, remain undefined and speculative. With respect to monocytes and macrophages in this trial, prior administration of gemtuzumab (a myeloid cell depleting anti-CD33 monoclonal antibody) for his acute myeloid leukemia (AML) may have depleted HIV-infected myeloid reservoirs. It is also unclear whether the lack of a myeloid cell-targeted depletion contributed to the incomplete reservoir elimination and viral reemergence in other human [13] and non-human [14] trials that have attempted to recapitulate the case of the Berlin Patient.

In the case of the infamous “Mississippi baby,” and cases of acute and extremely early administration of ART [15,8,17], where a prolonged period of viral control without ART persisted, the viral reservoir monitoring studies were not designed for interrogation of viral persistence within myeloid cells. Interesting to note that in the case of the Mississippi baby following aggressive ART at birth, virus rebounded after 27 months after ART cessation. At 26 months, plasma virus was undetectable (< 2 copies/ml); however, the monocyte threshold for detection of cell-associated virus was higher than for the memory CD4 T cells, so it is unclear if the use of more sensitive assays would have detected any residual virus in the monocytes. Furthermore, lower monocyte counts were assessed as compared to CD4 T cells and viral outgrowth assays were not performed on monocytes. In our ongoing studies where ART was administered early during acute HIV infection [18], we are pursuing sensitive virological studies to better define the contribution of myeloid reservoirs to viral persistence in the earliest stage of infection.

This review will consider the contribution of HIV persistence in myeloid-derived cells in past, ongoing, and future cure clinical trials. Furthermore, we will review the sensitivity of myeloid reservoir measurement methods and argue for inclusion of monitoring myeloid cells harboring HIV as future primary and secondary end points in future HIV cure studies.

Evidence for Myeloid Cells as Contributors of HIV Persistence

Mounting evidence suggests that cells of the macrophage lineage, including blood monocytes subsets, play a significant role in HIV-1 persistence. Monocytes routinely survey tissues by transmigrating across the vascular endothelium from the bloodstream into tissues sites. Their capacity to harbor HIV DNA and RNA makes them suspect as critical contributors to HIV pathogenesis [19•]. Indeed, a variety of attributes make the monocyte and tissue macrophages, including microglia, ideal candidates for contributing to the HIV reservoir, both as carriers and replenishers of the viral reservoir. The macrophage reservoir half-life has historically been underdetermined, yet in the presence of ART, macrophages from SIV-infected rhesus macaques can sustain viremia for several months [20,13,22]. Furthermore, myeloid cells are relatively more resistant to apoptosis induced by HIV infection [23], and virus produced by macrophages may be more infectious than virus originating from CD4+ T cells [24].

The HIV envelope region undergoes more frequent sequence evolution in blood monocytes as compared to that of resting CD4+ T cells, suggesting a distinct contribution to plasma viremia. Indeed, phylogenetic analyses of HIV-1 sequences indicated that after prolonged ART, viral populations are related or identical to those found only in CD14+ monocytes [25]. The non-classical or patrolling monocyte subset (CD14lowCD16high) expresses higher levels of CCR5, a co-receptor for M-tropic HIV strains, and CD4, making them more susceptible to continual viral infection. Circulating blood monocytes traffic into tissues to later differentiate into tissue macrophages, but have the potential to undergo subsequent differentiation into migratory myeloid dendritic cells, which then traffic to other lymph tissues. Monocytes and macrophages disseminate into most tissues of the body and mediate HIV spread, particularly into the central nervous system and lymphoid tissues. Post-mortem brain tissue analysis has revealed that viral DNA is present in 3 to 19% of astrocytes [26] despite astrocyte infection being both relatively infrequent and unproductive [27]. Moreover, our group has used next-generation in situ hybridization RNAScope to identify HIV RNA in cerebellum macrophages of an infected individual who died with a undetectable plasma viral load [28•]. Perivascular macrophages have a half-life of ∼ 3 months [29] while microglia have a half-life of months-years to years-lifetime [30], and parenchymal microglia have been shown to represent two thirds of the infected cells in brain autopsies of HIV-infected persons with encephalitis [31]. Therefore, both CNS perivascular macrophages and microglia need to be considered as possible long-lived HIV reservoirs.

The extent of monocyte and macrophage tissue reservoir compartments contributing to viral recrudescence is poorly understood. Compared to the vast majority of studies tailored to evaluate the viral reservoir in the CD4 T cell compartment, few studies focus their effort to elucidate the relationship of monocyte/macrophages to HIV reservoir and its persistence. In humanized myeloid-only mice (MoM) infected with HIV, it has been shown that after ART-interruption, one of three mice experienced a delayed viral rebound [19•]. Due to the absence of human T cells in MoM mice, this study showed for the first time, in vivo, that persistent HIV infection exists in tissue macrophages during ART and the myeloid compartment can contribute to viral rebound after treatment interruption. Although investigators reported not being able to detect HIV DNA or viral outgrowth in peripheral monocytes isolated from viremic and ART-suppressed patients [19•, 32•], others have reported detection of HIV in circulating monocyte populations. CD16+ monocytes isolated from the blood of ART-treated individuals have been shown to harbor HIV DNA [33, 34]. In non-human primate studies, SIV was quantifiable in monocytes and macrophages from blood and tissues using a modified quantitative viral outgrowth assay (QVOA). The macrophage-QVOA (Mϕ-QVOA) is specifically tailored to quantify productively infected myeloid cells [35•]. The evidence in these reports stresses a need for HIV curative studies to include a more comprehensive evaluation of HIV persistence in the myeloid compartment in both blood and tissue. In doing this will we begin to fully understand the extent of the viral reservoir and the dynamics of its’ persistence, as well as to ascertain failures in past and current cure studies, which may better inform future curative endeavors (Fig. 1).

Fig. 1
figure 1

Current HIV eradication approaches. Overview of cure strategies that are being actively considered for CD4 T cell HIV clearance of relevance to non-CD4 T cell populations

Full size image

Biomarkers and Advances in Quantifying HIV Reservoirs in Myeloid Cells

CD4 T cells have been at the center of understanding HIV persistence dynamics and defining the latent reservoir. However, studies have utilized various HIV persistence and reservoir measurement techniques in characterizing these aspects in the myeloid compartment (Table 1). While there is evidence of HIV persistence in forms of proviral DNA in myeloid cells, measurement of replication-competent proviruses that produce infectious virions during ART-suppression is what will define the myeloid compartment as a reservoir. It is the existence of the latter that results in viral rebound after ART interruption and is a major barrier to curing HIV infection [39, 40]. The QVOA remains the gold standard for quantifying the latent reservoir, and its development was critical in defining resting CD4 T cells as latent reservoirs [39]. Overtime, modifications of the QVOA have been implemented to improve and streamline different steps of the assay. However, the majority of modifications done on the QVOA remained to be T cell centric [41,32,43, 44, 45•, 46, 47, 48•, 49, 50]. Clement et al. have modified the QVOA and tailored the assay to measure the viral reservoir size in monocytes and macrophages, addressing the need to evaluate the potential viral reservoir in myeloid cells [35•]. The Mϕ-QVOA assessed enriched myeloid cells that undergo cellular stimulation with TNF-α, a potent activator of myeloid cells and the U1 HIV latently infected monocytic cell line [51, 52]. Using the Mϕ-QVOA technique, macrophages were not only found to be productively infected in the SIV-infected non-human primate model, but also the number of productively infected macrophages varied throughout different tissue sites. Moreover, despite viral suppression by ART, tissue macrophages isolated from SIV-infected macaques continued to be productively infected [38•].

Table 1 Detection methods for HIV in myeloid cells. A reference for quantitative assays which can detect HIV provirus in myeloid cells
Full size table

With the ongoing use of these techniques in myeloid cells, it is of great value to evaluate potential biomarkers of viral persistence and reservoir in this cellular compartment. Much of the biomarkers currently proposed, such as CD2, CD30, CD32, and immune checkpoint receptors TIGIT, PD-1, and LAG-3, are still being investigated, but CD4 T cells were the focused cell type of these discoveries [53,42,43,56, 57•, 58]. The utility of these biomarkers as it directly relates to viral persistence and reservoir in myeloid cells has not been vigorously pursued, although much of the markers are T cell specific and highlight a need to further investigate myeloid-specific candidates. CD16, a Fcγ receptor, expression on myeloid cells has been found to be enriched for HIV DNA [33, 34]. In addition, higher expression of CCR2 on CD14+CD16+ monocytes isolated for ART-suppressed patients correlated with higher HIV DNA levels in peripheral blood mononuclear cells (PBMC) [59•].

Consideration of Myeloid Cells in HIV Cure Trials

Latency Reactivation Agents

The “shock and kill” strategy is currently one of the most widely discussed approaches to eliminate the viral reservoir [60, 61]. In this approach, drugs are administered to reverse HIV latency and induce viral production, ultimately resulting in the death of infected cells by direct viral cytopathic effects or immune-mediated clearance. Latency reversing agents (LRAs) are administered during suppressive ART, thereby preventing reactivated virus from replenishing the reservoir through infection of new cells. Clinical trials involving LRAs such as romidepsin, vorinostat, disulfiram, and panobinostat have failed to demonstrate significant reduction in reservoir size, although transient elevation in plasma viral RNA has been observed [62,51,64, 65•, 66, 67]. Numerous LRAs have been shown in vitro to have implications for driving HIV transcription in myeloid-derived cells, yet primary outcomes of completed clinical trials have thus far been entirely T cell focused, leaving our knowledge of myeloid-driven viral control to be underdeveloped. Below are some LRAs that may have relevance in driving viral transcription in myeloid cells.

Numerous clinical trials have used histone deacetylase inhibitors (HDACi) to reactivate HIV transcription, these include the following: panobinostat, vorinostat, valproic acid, and romidepsin. While HDACi may not alter the initial susceptibility of macrophages to HIV infection, they have been shown to decrease HIV release from macrophages ex vitro [68]. Thus far, there are limited studies that have assessed the impact of LRAs on myeloid contributions to pro-virus reactivation kinetics, although the importance for which has been published. In human primary astrocyte cell lines transfected with patient-derived HIV-1 LTR, treatment with HDACi did activate transcription of HIV LTRs [26, 69]. Furthermore, valproic acid (VPA) has been shown to alter activation states of the myelo-monocytic pathway [70] and therefore may have promise in latency reactivation in myeloid cells. In a separate study using CD34+ myeloid precursor cells, results from RNA microarray analysis revealed altered pathways for myeloid cell differentiation [71]. Pathways altered are implicated in changing homeostatic signals for both sustained cell persistence and pro-inflammatory activation of pre-cursor myeloid cells.

Disulfiram and a protein kinase C activator, bryostatin, are non-HDACi with LRA activity. Although well tolerated in clinical trials (NCT01286259, NCT02269605), HIV-1 reactivation was only considered in T cells. However, interestingly, DSF reactivates latent HIV-1 expression in U1 cells (a monocytic cell line) but not in ACH2 cells (a T cell line) [72]. Bryostatin has had moderate LRA activity in astrocyte cell lines both in vitro and in cultured primary astrocytes by inducing HIV-1 expression through NF-kB activation [73]. These LRAs may thus have value in targeting myeloid HIV reservoirs in the central nervous system. One concern however has been a risk for adverse CNS toxicity with LRA reactivating HIV in the CNS [74•, 75]; however, a recent clinical study has shown no long-term neuro-consequences though additional studies are needed.

Biologics

Immunotherapy with immune checkpoint blocking antibodies has been shown success in oncology through reversing T cell immune exhaustion and may have efficacy in the elimination of HIV. A recent study in patients with metastatic non-small-cell lung carcinoma (NSCLC) and HIV infection, who received multiple doses of PD-1 inhibitors while on suppressive ART, one participant showed a decline in viral reservoirs in CD4 T cells [76]. Moreover, of six HIV-positive individuals on therapy who received the anti-PD-L1 antibody BMS-936559, two patients demonstrated improvements in HIV-1-specific CD8+ T cell responses suggesting blockade of the PD-1–PD-L1 pathway has the potential to improve HIV-1-specific immunity and potentially eliminate HIV [77]. Given that PD-L1 is expressed on myeloid cells, the impact of PD-1:PDL-1 blockade on viral reservoirs in myeloid cells would be of interest to future cure initiatives.

3BNC117 and 10-1074 are two of most potent broadly neutralizing antibodies currently available. 3BNC117 targets the CD4-binding site and 10-1074 targets the base of the V3 loop of HIV-1 gp120. In phase 1 clinical studies (NCT03526848), the combination of 3BNC117 and 10-1074 has additive effects extending the median time-to-viral-rebound by 11–15 weeks than treating with 3BNC117 alone [78•]. However, the protective capacity of bnAbs not only is solely due to virus neutralization but can also be attributable to Fc-mediated function [79, 80]. Indeed, studies in HIV-1-infected humanized mice show that the therapeutic activity of anti-HIV-1 bNAbs requires Fc receptor (FcR) effector function to mediate anti-viral protection [81]. Therefore, investigating the role of FcR phagocytic expressors such as monocytes and macrophages for their capacity in direct sequestration and destruction of HIV but in stimulating the secretion of inflammatory mediators could be key in delineating time-to-rebound kinetics or an actual HIV elimination.

Multiple studies demonstrate that α4β7-expressing gut homing CD4 T cells represent early targets for HIV, and therapy targeting this integrin has been considered in the management of HIV-1 infection. Recent studies in the SIV-infected macaques show that treatment, during acute infection with ART and anti-α4β7 therapy, achieved long-term viremic control following treatment interruption [82•], suggesting this may be a novel therapeutic for HIV remission. While several clinical studies are ongoing targeting α4β7 (NCT03147859, NCT03577782, and NCT02788175), in our recent studies, we have observed significant attenuation of lymphoid aggregates in the terminal ileum in inflammatory bowel disease patients with concomitant HIV-1 infection receiving ART and anti-α4β7 therapy [83•]. While we also observed modest impact on persistent viral reservoirs in the gut of these donors, the study has not initiated ART treatment interruption to determine the extent of immunological control of residual HIV reservoirs. Furthermore, given that α4β7 is expressed on myeloid cells [84], monitoring the effects of α4β7 immunotherapy in myeloid cells should be pursued.

Immune-Based Modification Therapies

CD8 T cells targeting HIV-infected CD4 T cells are well described, but it is apparent that this is more challenging when virus resides in myeloid cells. SIV-specific CD8+ T cells can efficiently kill SIV-infected CD4+ T cells but not SIV-infected macrophages [85]. Clayton et al. recently show that HIV-infected macrophages are more resistant to cytotoxic T lymphocytes (CTLs) killing through a granzyme-B inhibitor–mediated mechanism [86•]. Rainho et al. show that CTL-mediated killing of CD4+ T cells and monocyte derived macrophages infected with SIV nef variants was more efficient when targeting CD4+ T cells than macrophages [87]. Finding strategies to enhance CTL activity against infected macrophages are therefore important. Pegu et al. have shown that a bi-specific immunomodulatory protein that stimulates CD8+ T cell effector function thereby initiating latent-infected cell lysis through recognition of Env [88•] may be retargeted towards HIV-infected macrophages. Studies evaluating immunotherapies targeting negative checkpoint receptors [89•, 90, 91] to improve CTL activity as well as harnessing NK cells [92] could also be needed in concert to overcome myeloid cell resistance to CTL killing. Advances have also been made using chimeric antigen receptors (CAR) to re-engineer CD8 T cells for specific lysis of HIV-infected CD4 T cells [93,82,95]. However, engagement of a CAR T cell to mediate monocyte and macrophage killing of infected cells has not been assessed to date.

Excising the CCR5 locus using zinc finger nucleases was the first-in-human application of genome editing accomplished in both T cells and hematopoietic stem/progenitor cells [96]. However, the CRISPR-Cas9 system has now markedly improved the precision by which direct edits to the host genome can be made. However, no clinical trial in the context of HIV cure has genetically engineered myeloid cells despite the feasibility having been published. Zhang et al. have reported using a dCas9-synergistic activation mediator (dCas9-SAM) system to reactivate HIV-1 in both CD4+ T cell and microglial cell lines [97•]. Other promising results in myeloid cells that are presented by Hu et al. were latently infected microglial, pro-monocytes, and T cell lines; they developed a Cas9/guide RNA system to eradicate the HIV-1 genome and immunize target cells against HIV-1 reactivation [98].

Conclusion: Potential Future Clinical Trials

Promising newer strategies are being pursued including the “block and lock” approach for HIV cure. Didehydro-Cortistatin A (dCA) is one compound that inhibits Tat which has been shown in vitro to decrease chromatin accessibility at the HIV LTR, reducing the transcriptional competence of latent HIV-1 provirus under ART [99, 100]. In mouse models, dCA treatment showed strong reductions in both systemic and brain tissue levels of viral RNA [101•, 102]. This may be particularly relevant for myeloid cells in the CNS. dCA therapy, by reducing low-level viremia and preventing viral reactivation from latent reservoirs, makes it a promising therapeutic for tissue sanctuaries like the CNS and where cell-based cytotoxicity strategies can have irreversible consequences. In the promonocyte cell line U1, dCA had no effect on viral mRNA production [102, 103]; however, because the U1 cell line has two integrated proviruses with Tat mutations, U1 cells are therefore already deficient for processive viral production. Further studies of dCA treatment activity in myeloid cells harboring virus should be pursued. Inclusion of a Tat inhibitor to current ART regimens may contribute to a functional cure in which no further immune cells are infected but would need to be paired with a strategy that eliminates already infected cells.

New therapeutic strategies that place emphasis on targeting HIV in non-CD4+ T cells will be imperative on the road to a successful cure. As shown in Table 2, ongoing HIV cure trials may need to be redesigned to better inform on their impact on myeloid HIV reservoirs.

Table 2 HIV curative clinical trials and potential impact on myeloid reservoirs. A list of ongoing HIV cure trials and their possible impact on myeloid cells
Full size table