Irreversible Loss of HIV-1 Proviral Competence in Myeloid Cells upon Suppression of NF-κB Activity

ABSTRACT

Although antiretroviral therapy (ART) sustains potent suppression of plasma viremia in people with HIV-1 infection (PWH), reservoirs of viral persistence rekindle viral replication and viremia if ART is halted. Understanding the nature of viral reservoirs and their persistence mechanisms remains fundamental to further research aiming to eliminate them and achieve ART-free viral remission or virological cure. CD4⁺ T-cell models have helped to define the mechanisms that regulate HIV-1 latency as well as to identify potential latency manipulators, and we similarly hoped to extend this understanding to macrophages given the increasing evidence of a role for myeloid cells in HIV-1 persistence under ART (T. Igarashi, C. R. Brown, Y. Endo, A. Buckler-White, et al., Proc Natl Acad Sci U S A 98:658–663, 2001, https://doi.org/10.1073/pnas.98.2.658; J. M. Orenstein, C. Fox, and S. M. Wahl, Science 276:1857–1861, 1997, https://doi.org/10.1126/science.276.5320.1857). In the pursuit of a primary cell model of macrophage latency using monocyte-derived macrophages (MDMs), we observed that NF-κB inhibition, originally intended to promote synchronous entry into a latent state, led to an irreversible loss of proviral competence. Proviruses were refractory to latency reversal agents (LRAs), yet host cell functions such as phagocytic capacity and cytokine production remained intact. Even after NF-κB inhibition was relieved and NF-κB action was restored, proviruses remained refractory to reactivation. Agents that interfere with the NF-κB–HIV-1 axis in myeloid cells may provide an approach with which to render myeloid cell reservoirs inert.

IMPORTANCE Although HIV-1 infection can be suppressed using antiretroviral therapy, it cannot yet be cured. This is because HIV-1 integrates itself into host cells and may become dormant but also remains ready to emerge from such reservoirs when antiretroviral therapy stops. The CD4⁺ T cell has been the most actively investigated cell type in reservoir research due to its prominent role in hosting HIV-1; however, HIV-1 can infect and fall latent in myeloid cells, and therefore, their role must also be assessed in pursuit of a cure. Here, we show that caffeic acid and resveratrol, two nontoxic chemicals, both of which interfere with the same set of host mechanisms, can each prevent HIV-1 reactivation from latency in myeloid cells even after either chemical is removed and previous cell functionality is restored. Strategies to interfere with latency underlie the future of HIV-1 cure research, and our findings help to focus such strategies on an important but often neglected cell type.

INTRODUCTION

Human immunodeficiency virus type 1 (HIV-1) is able to persist in the face of potent antiretroviral suppression (1, 2). This viral persistence during long-term antiretroviral therapy (ART) is caused by a combination of at least three mechanisms: (i) maintenance of intact but transcriptionally silent proviral genomes within long-lived CD4⁺ T cells, which are established early and believed to confer lifelong persistence to the virus (3–6); (ii) homeostatic proliferation of latently infected cells leading to a stable and infectious reservoir through self-regeneration (7, 8); and (iii) tissue-based foci of viral replication and cell-to-cell spread that may be driven in part by innate immune (inflammatory) responses during microbial translocation, concurrent infections, or antigenic stimulation (9–11). While in its broadest sense, the term “viral reservoir” encompasses all virally infected cells, most viral genomes contain inactivating mutations, frameshifts, and deletions that render them biologically inactive and incapable of producing replication-competent viruses (12–14). As a result, the term “rebound-competent reservoir” has been coined for the HIV-1 reservoir in which the virus is biologically competent to replicate and, thus, capable of fueling viral replication and rebound after antiviral treatment is interrupted.

While most mechanistic research exploring viral persistence under ART has focused on the nature of persistent viral reservoirs in CD4⁺ T cells, there is increasing evidence that myeloid cells maintain viral infection in people with HIV-1 infection (PWH) on effective ART. For example, it has been found that urethral macrophages from HIV-1-infected individuals on ART harbored viral nucleic acids and released replication-competent virus upon exposure to the macrophage activator lipopolysaccharide (LPS) (15). Furthermore, we have previously demonstrated that rebound viremia from individuals undergoing analytic treatment interruption (ATI) comprised highly macrophage-tropic viruses that were phylogenetically distinct from T-cell-tropic viruses obtained from the same sample (16). Virion immunocapture with myeloid-specific ligands indicated that macrophage-tropic viruses in post-ATI plasma had a myeloid cell origin, while molecular clock analysis affirmed that these macrophage-tropic variants predated treatment interruption, as opposed to having been created during explosive viral outgrowth over the treatment interruption interval (16). These observations suggest that myeloid cell reservoirs may need to be considered in the design of approaches to promote ART-free viral remission or virological cure.

Studies with CD4⁺ T-cell models of viral latency have established foundational strategies with which to promote the clearance of viral reservoirs in infected individuals. Some of these strategies center on the use of latency reversal agents (LRAs), which are meant to reinitiate viral transcription and enable the clearance of the reservoir cells by CD8⁺ T cells and/or viral cytopathic effects (17, 18). Understandably absent from these efforts with T cells are considerations regarding the unique biology of macrophages, which remains comparatively unexplored but could present distinctive challenges to reservoir clearance strategies in this cell type.

The transcription factor NF-κB has known activity as a regulator in the long terminal repeat (LTR) region of HIV-1, where two tandem NF-κB binding sites are among the most conserved elements of the HIV-1 genome (19). In CD4⁺ T cells, the canonical NF-κB pathway has been shown to modulate HIV-1 transcription and portions of latency alongside NFAT and SP1; however, the dynamics of latency in general, let alone the role of NF-κB in that latency, remain poorly understood in macrophages (20, 21). Given the evidence implicating myeloid cells in viral persistence under ART, we set out to establish a primary macrophage model of viral latency that could guide future approaches for latency reactivation and/or host cell clearance. A surprising finding from our study is that NF-κB inhibition, originally intended to promote host cell quiescence and the synchronous establishment of macrophage latency, led to an irreversible loss of proviral competence; proviruses were refractory to LRAs, even though host cell function remained intact. This suggests that approaches perturbing the interaction between NF-κB and HIV-1 might serve as a strategy with which to render myeloid reservoirs in PWH inert.

RESULTS

In order to identify the mechanisms underlying HIV-1 persistence in myeloid cells, we first set out to refine a macrophage model of HIV-1 latency. A number of cell lines recapitulate the attributes of natural macrophage populations, for example, THP‐1, U937, HL‐60, and Mono Mac; however, none of these convincingly capture the full suite of characteristics found in primary macrophages or monocyte-derived macrophages (MDMs), which therefore comprise the bulk of our research (22, 23). As has been shown for primary CD4⁺ T cells (24, 25), we adopted low-serum conditions to promote a quiescent state that might favor the establishment of latent infection in primary macrophages (see Fig. 1a for the schematic). When MDMs were infected with the macrophage-tropic variant HIV-1_BaL, there was a prolonged interval of virus production (extracellular genomic viral RNA and Gag p24) extending up to 8 to 10 weeks postinfection (wpi) (Fig. 1b and c) as well as successful establishment of cell-associated HIV-1 DNA pools (Fig. 1d). By week 8 postinfection, these extracellular genomic viral RNA and Gag p24 levels declined to negligible levels, while the numbers of viral genomes (as determined by cell-associated viral DNA) declined approximately 2-fold. These viral genomes were responsive to reactivation by LPS and suberoylanilide hydroxamic acid (SAHA), as demonstrated by the renewed production of virus in the culture supernatants (Fig. 1e and f), while the number of viral genomes remained relatively unchanged after activation (Fig. 1g). Gag p24 and viral nucleic acids were not detectable in the tenofovir (TFV) control groups, indicating that the presence of Gag p24 and viral nucleic acids in non-TFV-treated cultures was the result of bona fide infection.

FIG 1.

HIV-1 latency is successfully established under low-serum conditions in MDMs. Extracellular HIV-1_BaL RNA copies (b and c), extracellular HIV-1_BaL surface Gag p24 protein (d and e), and cell-associated HIV-1_BaL DNA (f and g) were detected by RT-qPCR, qPCR, or an ELISA, compared to TFV-treated controls. Panels b, d, and f display the three respective virological metrics over the entire time course of the low-serum latency procedure (a), while panels c, e, and g present each metric during the shorter time scale of an LRA challenge by LPS (1 μg/mL) or SAHA (1.5 μM) compared to a mock-treated control (−). Extracellular HIV-1_BaL RNA copies and Gag p24 surface protein correlate with one another in the long term (b and d) and increase following LRA stimulation (c and e). Conversely, cell-associated DNA rose during initial infection (f) and remained constant during reactivation (e). Where appropriate, statistical analysis was done using Tukey’s multiple-comparison test, with at least 2 replicates per panel (*, 0.05 > P > 0.005; **, 0.005 > P > 0.0005; ***, 0.0005 > P > 0.00005; ns, not significant).

Because HIV-1 latency is often associated with “quiescent” cells of low transcriptional activity, and because such cells have sparser availabilities of the transcription factors necessary for basal proviral transcription, we next examined whether limiting key transcription factors would accelerate the establishment of a latent state (26, 27). The HIV-1 LTR contains NF-κB response elements important for basal proviral transcription until Tat-mediated transcriptional activation occurs (28). We selected NF-κB inhibitors for which there is available in vivo experience. Resveratrol and its metabolites such as piceatannol are naturally occurring stilbenes that inhibit NF-κB activity via IκBα kinase and p65 phosphorylation (29). Due to its effect on NF-κB action, resveratrol and its derivatives have been explored clinically for a variety of conditions, including inflammation and metastasis (30). The structurally unrelated agent caffeic acid is a phenolic acid that likewise has seen broad in vivo use in inflammation and cancer (31). When macrophages were infected and then incubated with clinically relevant concentrations of caffeic acid and resveratrol (Fig. 2a), active (phosphorylated) levels of NF-κB were greatly reduced (Fig. 2b), while total NF-κB (including phosphorylated and unphosphorylated forms) remained relatively constant (Fig. 2c), resulting in a suppressed activation ratio (Fig. 2d). The caffeic acid group displayed a modest increase in total NF-κB levels at the later time point, likely as a result of sample lysate variability; however, the NF-κB activation ratio remained constant throughout the duration of the experiment. When MDMs were maintained in the presence of either caffeic acid or resveratrol beginning at 2 weeks postinfection (Fig. 3a), the interval over which viral RNA and p24 in the culture supernatants declined to undetectable levels was slightly reduced, from 8 weeks to 6 weeks (Fig. 3b and c). We next examined whether proviruses in the caffeic acid- and resveratrol-treated macrophages could be reactivated with LRAs (as in Fig. 1). Unexpectedly, despite an abundance of viral genomes in these cultures (Fig. 3d), infected macrophages maintained with either inhibitor were completely refractory to reactivation by the well-characterized LRAs SAHA and LPS (Fig. 3e and f). This phenomenon held true upon stimulation with a range of other LRAs during preliminary testing, including tumor necrosis factor alpha (TNF-α), phorbol myristate acetate (PMA), and phytohemagglutinin (PHA) (data not shown).

FIG 2.

Treatment with caffeic acid or resveratrol successfully attenuates NF-κB activation within the low-serum MDM HIV-1 latency system. A dual phos(p65)/pan-NF-κB ELISA was used to gauge the NF-κB activity level at the baseline over the time course of the revised latency procedure. (a) Macrophage timeline, with arrows corresponding to time points taken for panels b to d beginning at 2 wpi and just prior to NF-κB antagonism (0 h). (b to d) Zero hours, 24 h, 3 days, 1 week, and 4 weeks after treatment with caffeic acid (20 μM) and resveratrol (50 μM), compared to mock-treated controls, relatively constant total NF-κB (pan) (b), fluctuations in phosphorylated NF-κB p65 (c), and the two combined into an activation ratio (d) were determined. Statistical analyses were done using Tukey’s multiple-comparison test, with at least 2 replicates per panel (*, 0.05 > P > 0.005; ***, 0.0005 > P > 0.00005; ****, P < 0.00005). OD, optical density.

FIG 3.

The NF-κB inhibitors caffeic acid and resveratrol inhibit the reactivation of latent HIV-1 in MDMs. At 6 weeks, LPS (1 μg/mL), SAHA (1.5 μM), or a mock challenge (−) was applied overnight (12 h). (b, d, and f) The levels of production of cell-associated HIV-1_BaL DNA (b), de novo HIV-1_BaL supernatant RNA (d), and extracellular HIV-1_BaL Gag p24 protein (f) were monitored using qPCR, RT-qPCR, and an ELISA, respectively, with results shown at day 4 postchallenge. TFV was used as an infection control for carryover plasmid. (a, c, and e) Extracellular HIV-1_BaL RNA (c) and Gag p24 protein (e) were also assessed throughout the time course of the MDM HIV-1 latency procedure (a) and remained consistent with one another. Statistical analyses were done using Tukey’s multiple-comparison test, with at least 3 replicates per panel (*, 0.05 > P > 0.005; ***, 0.0005 > P > 0.00005; ****, P < 0.00005). CA, caffeic acid; R, resveratrol.

Since NF-κB is involved in the transcriptional regulation of many cellular genes, we assumed that the lack of responsiveness of resveratrol- or caffeic acid-treated macrophages to latency-reactivating agents simply reflected a global impairment of host cell function. However, this did not appear to be the case. We first assessed the dynamics of interleukin-10 (IL-10) and CXCL2 cytokine production in infected macrophages. IL-10, which is, in part, regulated by NF-κB (32–35), was inhibited in the presence of both resveratrol and caffeic acid (Fig. 4b). In contrast, the levels of CXCL2 were indistinguishable between NF-κB-treated and untreated macrophage cultures (Fig. 4c). The production of monocyte colony-stimulating factor (M-CSF) was also assessed and was similar in the presence and absence of caffeic acid (data not shown). Macrophage viability, as assessed from the levels of lactate dehydrogenase (LDH) in the culture supernatants (36, 37), showed modest changes with either inhibitor (Fig. 4d). Finally, the phagocytic activity of infected macrophages, assessed by the uptake of pH-dependent fluorescent nanoparticles (pHrodo), was not impaired by either agent (Fig. 4e and f). Taken together, these results indicate that NF-κB inhibition renders latent HIV-1 genomes in macrophages refractory to reactivation and that this loss of proviral competence is not due to a generalized impairment of macrophage function by NF-κB inhibition.

FIG 4.

Prolonged treatment of HIV-1-infected MDMs with caffeic acid or resveratrol does not dramatically impact viability, nor does it disable major functionality over the time course of the model. (a to d) General viability over the model time course (a) was assessed via LDH release relative to a time point-matched 10% DMSO control curve (d), while ELISAs for both IL-10 and CXCL2 (b and c) demonstrate key cytokine responses to prolonged NF-κB antagonism. (e and f) The uptake of pH-dependent fluorescent nanoparticles (pHrodo) was employed to explore major MDM functionality over the same period relative to nuclear DAPI. “Week 0” refers to day 3 postinfection. NF-κB inhibitor treatment uniformly occurred at 2 weeks postinfection. Tukey’s multiple-comparison test was performed as appropriate, with at least 3 replicates per panel (*, 0.05 > P > 0.005; **, 0.005 > P > 0.0005; ****, P < 0.00005). CD, cytochalasin D, a phagocystosis inhibitor and negative control; CA, caffeic acid; R, resveratrol.

We next assessed whether the lack of responsiveness of viral genomes to LRAs in NF-κB-depleted macrophages could be restored after the cessation of NF-κB antagonism (see the experimental scheme in Fig. 5a). Genomic viral RNA and Gag p24 levels in the culture supernatants as well the levels of cell-associated viral DNA were quantitated before and after the removal of the NF-κB antagonists. While viral reactivation was evident in infected macrophages maintained in the absence of NF-κB inhibitors, there was no reactivation in NF-κB inhibitor-treated macrophages after inhibitor removal (Fig. 5b and c), even though the viral genome copy number remained constant prior to and following the cessation of NF-κB inhibition (Fig. 5d). The removal of NF-κB inhibitors completely restored the levels of active NF-κB to those seen in cultures maintained in the absence of NF-κB inhibitors (Fig. 6a). Furthermore, while CXCL2 levels remained constant as before, the cessation of NF-κB antagonism restored the production of IL-10 to the levels observed in cultures maintained in the absence of NF-κB inhibitors (Fig. 6b and c). Collectively, these results indicate that latently infected macrophages become refractory to LRAs after NF-κB inhibition and that this unresponsive state is not restored upon restoring active NF-κB levels.

FIG 5.

The silencing of HIV-1 reactivation by caffeic acid or resveratrol is irreversible and independent of the presence of an inhibitor. (a) NF-κB antagonism was halted at week 6 in half of the groups previously treated with caffeic acid or resveratrol, and a reactivation challenge was then performed 1 week later with LPS (1 μg/mL), SAHA (1.5 μM), or a mock challenge (−). (b to d) The levels of production of de novo HIV-1_BaL supernatant RNA (b), extracellular HIV-1_BaL Gag p24 protein (c), and cell-associated HIV-1_BaL DNA (d) were monitored using RT-qPCR, an ELISA, and qPCR, respectively, with results shown at day 4 postchallenge. Statistical analyses were done using Tukey’s multiple-comparison test, with at least 3 replicates per panel (**, 0.005 > P > 0.0005; ****, P < 0.00005).

FIG 6.

Interrupting the NF-κB antagonism caused by either caffeic acid or resveratrol restores the previous activity of NF-κB, IL-10, and CXCL2. Given the same treatment interruption parameters and subsequent challenge as the ones detailed in the Fig. 5 legend, NF-κB activation was assessed before and after challenge by a dual ELISA (a), while the levels of IL-10 and CXCL2 were also each detected by the respective ELISAs (b and c). Where appropriate, statistical analyses were done using Tukey’s multiple-comparison test, with at least 2 replicates per panel (*, 0.05 > P > 0.005; ****, P < 0.00005).

DISCUSSION

We recreated HIV-1 latency in primary macrophages in order to identify mechanisms that could be coopted to reactivate viral transcription and promote the clearance of myeloid cell reservoirs. Although viral latency could be recapitulated in primary macrophages, it required the maintenance of infected macrophages for intervals of 8 to 10 weeks before a latent state, one in which virus production was negligible yet LRA susceptible, was established. We assumed that entry into viral latency might be accelerated if cellular transcription factors shown to promote HIV-1 transcription were reduced. While the maintenance of HIV-1-infected macrophages in the presence of the NF-κB inhibitors caffeic acid and resveratrol slightly accelerated the time to the establishment of latency, it also led to a marked and irreversible loss of proviral competence. Although NF-κB activity could be fully restored after the removal of either NF-κB inhibitor, proviruses remained refractory to LRA treatment.

While it is tempting to speculate that NF-κB inhibition indirectly subverted proviral competence due to a generalized impairment of the host cell, several important macrophage functions, such as the phagocytic index and cytokine production (in the cases of CXCL2 and M-CSF), did not appreciably suffer. While CXCL2 is controlled by both STAT1 and NF-κB, and this redundancy could explain the unaltered levels regardless of inhibitor use (38), IL-10, in contrast, showed a marked but reversible drop in its level under the conditions of NF-κB inhibition despite also being regulated by multiple pathways. This could be explained by cell type given that NF-κB has been strongly implicated in the optimal production of IL-10 in mouse macrophages but not CD4⁺ T cells (35, 39). In all, while NF-κB antagonism might be expected to globally reduce cytokine release, the considerable redundancy in the pathways regulating cytokine production likely underscores the differential impact of NF-κB inhibition on regular cytokine levels. More importantly, the fact that IL-10 production but not proviral activity could be restored upon the removal of NF-κB inhibition strengthens our assertion that NF-κB suppression selectively impaired proviral competence in a manner independent of global effects on the host cell.

From a mechanistic standpoint, the results obtained here using LRAs with well-defined mechanisms of action on proviral activity suggest that NF-κB inhibition exerts its effect at the level of proviral competence, although an alternate possibility is that the effects are instead exerted at the level of endosome dynamics. Previous studies indicate that HIV-1 virions in macrophages can bud into, and be archived within, vesicles that exhibit characteristics of nonacidic endosomes (40–42). While it is possible that virions were being produced and retained within endosomes during NF-κB inhibition and that LPS treatment simply led to an evacuation of these endosomal components, we know of no direct evidence in the literature to suggest that LPS promotes fusion between the endosomal and plasma membranes that would be required for the liberation of virions. Furthermore, the same effect was observed with SAHA, which is a histone deacetylase inhibitor that is chemically distinct from LPS. It also bears noting once again that virion production stimulated by LPS or SAHA was not observed after long-term treatment with either caffeic acid or resveratrol. Taken together, these lines of evidence argue that the effects of NF-κB inhibition are exerted at the level of proviral activity rather than vesicle dynamics. Ultimately, our study alone has not elucidated the mechanism by which the observed silencing effect occurs but only linked and associated it with the NF-κB pathway using two structurally distinct inhibitors and related cytokines. Further research will be required to explore the precise role of NF-κB, investigate possible epigenetic changes rendered with these inhibitors, and assess the extent of endosomal virion release within the refined model.

In reference to epigenetics, loss of proviral competence following interference with viral transcription has been observed following treatment of infected CD4⁺ T cells with cortistatin A. This agent inhibits the Tat-TAR interaction to suppress transcriptional elongation (43, 44), which effectively reduces the frequency of transcriptional events at the HIV-1 promoter, leading to the deposition of chromatin marks at the viral promoter chromatin (45). While the mechanism underlying the loss of proviral competence imparted by resveratrol and caffeic acid is not yet understood, we suspect that under conditions where NF-κB is rate limiting, a reduction in transcriptional activation events from the HIV-1 promoter might similarly lead to the deposition of epigenetic marks and the eventual loss of proviral competence. We have, as yet, been unable to assess whether the effects of caffeic acid and resveratrol might also manifest in infected CD4⁺ T cells. Quiescent CD4⁺ T cells remain viable for short intervals in vitro, and as such, we were unable to sustain the viability of infected CD4⁺ T cells over the intervals needed to assess the impact of caffeic acid or resveratrol on latency and reactivation from latency.

Observations with elite controllers might further enforce a relationship between the transcriptional activity of the provirus and proviral competence. Elite controllers are individuals with very low levels of viral reservoirs and who exhibit indefinite intervals of viral remission without the use of ART. A distinguishing feature of proviruses from elite controllers relative to those from individuals on long-term ART was their increased distance from transcriptional start sites and accessible chromatin as well as an enrichment in repressive chromatin marks (4). It is possible that susceptibility to modification by epigenetic modification is increased for proviruses that, due to their location in inaccessible chromatin, undergo fewer transcriptional events. Therefore, limiting transcriptional events in the viral reservoir may warrant further study as an approach to render viral reservoirs inert.

Many current HIV-1 reservoir clearance strategies are tailored to the biology of T cells and center on viral reactivation to promote the neutralization of the reservoir cells by CD8⁺ T cells and/or viral cytopathic effects. However, the impacts of these approaches on myeloid cell reservoirs are likely to be limited due to basic biological differences between macrophages and T cells. Some of these differences may not yet be fully appreciated and will require further investigation, such as the role of macrophage virus-containing compartments (VCCs) in latency and overall reservoir persistence, but others are well recognized (46, 47). For example, we and others have demonstrated that productive infection of myeloid cells is noncytopathic (48) and that macrophages are resistant to cytotoxic CD8⁺ T-cell killing (49, 50) and to killing by natural killer (NK) cells (51; reviewed in references 46 and 52). If these in vitro observations extend to myeloid reservoirs in vivo, the strategies being explored for the attrition of CD4⁺ T cell reservoirs might have a limited impact on myeloid cell reservoirs. Furthermore, some viral cure strategies rely on ex vivo manipulation of circulating CD4⁺ T cells with gene-editing enzymes that render the cells resistant to infection or that excise proviral segments (53, 54). However, macrophages are tissue-resident cells that would not be readily amenable to approaches requiring their ex vivo manipulation (55, 56). As such, the use of small-molecule agents that limit the transcriptional activity of the provirus and increase its susceptibility to epigenetic modification might prove more effective in targeting viral reservoirs in tissue-resident macrophages.

Finally, the experiments presented in this study focused on the use of monocyte-derived macrophages. This was necessitated by the need to have a constant supply of primary macrophages in sufficient quantities for the experiments featured in our study. While we believe that monocyte-derived macrophages provide good physiological representations of tissue macrophages, there is extensive heterogeneity in the phenotypes of tissue macrophages in different anatomic compartments and the central nervous system (CNS), and we therefore cannot predict whether the response of infected, monocyte-derived macrophages to NF-κB antagonists would be recapitulated in primary macrophages. Additional studies employing macrophages representative of the major myeloid compartments that might harbor HIV-1 reservoirs (such as CNS microglia) or affirmation using animal models of HIV-1 would be needed to assess whether NF-κB antagonism might generally impair proviral competence in different myeloid compartments and whether this might form the basis of a strategy to silence myeloid reservoirs in vivo.

MATERIALS AND METHODS

HEK293T and TZM-bl cell culture.

The human embryonic kidney cell line (HEK293T) was originally purchased from the American Type Culture Collection (ATCC) (Manassas, VA). TZM-bl cells (also called JC53BL-13 cells) were originally obtained from the NIH AIDS Research and Reference Reagent Program (catalog no. 8129). TZM-bl and HEK293T cells were maintained in Dulbecco’s modified Eagle medium (DMEM) (Gibco) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (HyClone), 2 mM l-glutamine (Gibco), 100 U/mL penicillin, and 100 μg/mL streptomycin (Sigma) at 37°C with 5% CO₂ (57, 58). Trypsin-EDTA passaging was performed at ~80% confluence.

Monocyte-derived macrophages.

Monocyte-derived macrophages (MDMs) were isolated from the elutriated blood of HIV-1-seronegative donors and were negatively selected using magnetic particles from the EasySep human monocyte CD14⁺ without CD16 depletion enrichment kit (StemCell Technologies). After plating with a seed density of ~6 × 10⁶ monocytes/plate, monocytes were differentiated in MDM medium, containing DMEM supplemented with 10% heat-inactivated human serum (Sera Care Life Sciences), 2 mM l-glutamine (Gibco), and 10 μg/mL gentamicin (Sigma-Aldrich). Additional M-CSF (12.5 ng/mL; R&D Systems) was used for the differentiation process only. Differentiation in MDM medium continued for 7 days at 37°C with 5% CO₂ before subsequent infection and experimentation under the same incubation conditions, with medium alterations as specified for each assay (59, 60).

The LDH assay, the p24 enzyme-linked immunosorbent assay (ELISA), the CXCL2 ELISA, the IL-10 ELISA, and extracellular HIV-1 RNA reverse transcription-PCR (RT-PCR) (see the relevant sections below) were performed on the macrophage culture supernatants. For the phagocytic assay, the NF-κB dual ELISA, and quantitative PCR (qPCR) for total cell-associated HIV-1 DNA (see the relevant sections below), MDMs were processed directly on the plate according to purification or kit instructions. In unrelated troubleshooting instances where MDM detachment was necessary, 45-min trypsinization and power washes were used.

The induction of HIV-1 latency in MDMs and subsequent experimentation were performed according to a standard time course that was initially dictated by the interval required for the viral output to decline to the baseline (Fig. 1a). Fully differentiated monocytes (7 days postplating) were infected after 3 days of acclimatization in reduced-serum medium (base MDM medium with 0.5% rather than 10% human serum) and maintained with weekly medium replacement. Beginning at 2 weeks postinfection (wpi) to allow sufficient HIV-1 establishment, medium was recurrently supplemented with either resveratrol (50 μM; Sigma), caffeic acid (20 μM; LSBio), or a dimethyl sulfoxide (DMSO) mock treatment control (Fig. 2a) (61–63). Reactivation of HIV-1 transcriptional activity was attempted between 6 and 8 wpi, depending on the procedure conditions, using incubation overnight with lipopolysaccharide (LPS) (1 μg/mL) or suberoylanilide hydroxamic acid (SAHA) (15 μM) compared to a negative DMSO mock treatment control (Fig. 3a) (64–66). When NF-κB antagonism was halted (or not) at 6 wpi (Fig. 5a), MDMs were given 1 week of acclimatization in resveratrol- or caffeic acid-free medium before further LRA challenge.

Viral constructs.

The HIV-1 HXB3/BaL infectious molecular clone (pWT/BaL) was obtained through the NIH HIV Reagent Program, Division of AIDS, NIAID, NIH: human immunodeficiency virus type 1 (HIV-1) WITO4160.27 Env optimized expression vector (pWITO4160.27 gp160-opt), ARP-11408, contributed by Beatrice H. Hahn, Jesus Salazar-Gonzalez, and Denise L. Kothetr (67). Replication-competent HIV-1_BaL stocks were prepared by transfection of HEK293T cells with the Lipofectamine 2000 system (Invitrogen). Culture medium was changed at 24 h and 72 h, and the virus-containing supernatant was collected, filtered (0.45-μm syringe filter; Millipore), aliquoted, and stored at −80°C. Infectivity was ascertained by a TZM-bl-based luciferase assay (Promega) as previously described (59), and the viral copy number per microliter was quantified using RT-qPCR with Volcano 2G DNA polymerase (see the relevant section below).

According to laboratory protocols, all MDMs were infected with concentrations of the viral stock equaling 3 × 10⁸ viral genome copies per 1 × 10⁶ MDMs or 1.5 × 10⁶ viral genome copies per 1 × 10⁶ TZM-bl cells for the purposes of the luciferase assay. After incubation overnight (16 h), inoculation medium was removed, and cells were washed twice with phosphate-buffered saline (PBS) before the addition of fresh medium. The nucleotide reverse transcriptase inhibitor tenofovir was used as an infection control to gauge the presence of carryover plasmid DNA from the transfection process; in all experiments involving tenofovir control groups, a 2-h pretreatment with either tenofovir (10 μM) or an equivalent DMSO control was performed on all cells immediately prior to infection.

HIV-1 RNA RT-qPCR.

Procedures for RT-qPCR used to quantify extracellular HIV-1 RNA copies employed Volcano 2G DNA polymerase (myPOLS Biotec). PCR amplification was done in 20-μL reaction mixtures using 2 μL of the supernatant sample (first diluted in water 1:10 to reduce PCR inhibition). Master mix consisted of 5× Volcano buffer, 10 mM deoxynucleoside triphosphates (dNTPs), 5 U/μL anti-Taq antibody, 20 μM LA9115/f primer, 20 μM 8T22 primer, 100 μM 2nr4nrB probe, Rox, and RNase-free water, all on ice (data not shown). The mix and samples were distributed into 96-well plates (hard shell, catalog no. HSP9601; Bio-Rad) and amplified using a Bio-Rad CFX Connect instrument set for cycling conditions of 95°C for 1 min and 40 cycles of 95°C for 10 s and 63.5°C for 1 min. The output was analyzed using CFX software and compared to Rtotal-3T27 PCR product standards ranging from 10² to 10⁶.

DNA purification.

DNA harvest and purification were required prior to the dual NF-κB ELISA as well as cell-associated HIV-1 DNA qPCR and were performed with DNA purification kits (Promega) according to the manufacturer’s instructions. DNA was quantified for downstream applications using a Bio-Rad plate reader.

Total HIV DNA qPCR.

The qPCR used to quantify total cell-associated HIV-1 DNA copies (data not shown) compared to a concurrent CCR5 control (data not shown) was performed using a regular FastMix procedure (1 s and 40 cycles of 20 s and 1 s) and basic TaqMan 2× master mix (Applied Biosystems). Samples were distributed into 96-well plates (hard shell, catalog no. HSP9601; Bio-Rad) and amplified using a Bio-Rad CFX Connect instrument. The output was analyzed using CFX software and compared to in-house standards ranging from 10² to 10⁶.

p24, IL-10, CXCL2, and pan/phos-NF-κB ELISAs.

ELISAs for HIV-1 p24 (TaKaRa), the cytokine IL-10 (Abcam), and the cytokine CXCL2 (LSBio) were done according to the respective manufacturer’s instructions using aliquots of the harvested supernatant. The NF-κB dual ELISA (Ray Biotech, LSBio) required prior DNA harvest and purification (see the relevant section above). All ELISA results were quantified using a Bio-Rad plate reader with the output specifications delineated by each kit.

Viability.

LDH was quantitated using the LDH-Glo cytotoxicity assay (catalog no. J2380; Promega) on gathered supernatant samples according to the manufacturer’s instructions. A final optical density readout was obtained using a Bio-Rad plate reader at 450 nm and compared to standard LDH curves derived from positive controls as well as 10% DMSO positive controls derived from sacrificed cells at each time point (68, 69). Preliminary comparison to cell counts across time points was used to validate baselines, particularly to account for naturally diminishing cell numbers over time (data not shown).

Phagocytosis.

Phagocytic analysis was performed using pH-dependent fluorescent bioparticles (pHrodo red zymosan bioparticle conjugate for phagocytosis, catalog no. P35364; Thermo Fisher) according to the manufacturer’s instructions. MDMs were seeded as detailed above in plates set aside for single use at each time point, including control groups to be treated with the phagocytosis inhibitor cytochalasin D (5 μM; Thermo Fisher). Images were processed with ImageJ (NIH) and normalized to nuclear 4′,6-diamidino-2-phenylindole (DAPI) staining to create a phagocytic index and determine percent phagocytosis, as outlined by standard NIH image processing protocols (70, 71) (https://imagej.nih.gov/ij/docs/guide/user-guide.pdf).

Statistics.

All data were analyzed using GraphPad Prism 9 as indicated by the design of each experimental set. Tukey’s multiple-comparison test in particular was used for many data sets given the characteristics and arrangement of most experimental groups. P values of 0.05 or lower were considered statistically significant.