SUMMARY
Deciphering the mechanisms underlying viral persistence is critical to achieving a cure for HIV infection. We uncovered molecular signatures of HIV latently-infected CD4+ T cells, identifying the adenosine-producing ectonucleotidase CD73 as a key marker of latent infection. Hypoxic conditioning, reflective of tissue microenvironments, increased the frequency of CD73+ CD4+ T cells and promoted HIV latency. Transcriptomic profiles of CD73+ CD4+ T cells indicated expression phenotypes favoring viral persistence, immune evasion, and cell survival. Further, we demonstrate that CD73+ CD4+ T cells harbor a functional HIV reservoir and are capable of reinitiating productive infection in vitro. Moreover, blocking of the A2A receptor facilitates HIV reactivation in vitro, linking adenosine signaling to viral quiescence. Finally, tissue imaging of lymph nodes from HIV-infected individuals on antiretroviral therapy reveal spatial association between CD73 expression and HIV persistence in vivo. Our findings warrant exploration of the hypoxia-CD73-adenosine axis in curative strategies to promote viral eradication.
INTRODUCTION
In recent decades, remarkable scientific and biomedical advances have turned the tides in the ongoing fight against the global human immunodeficiency virus (HIV) epidemic. However, current therapeutic regimens fail to completely eradicate HIV due to the persistence of latently-infected cells1. These viral reservoirs are established very early during infection, remain invisible to the host’s immune system, and persevere for decades despite effective antiretroviral therapy (ART)2–4. Although latently-infected cells are extremely rare5–7, they are able to reinvigorate spreading infection rapidly, and people living with HIV (PLWH) almost inevitably experience viral rebound within weeks of a treatment interruption2, 8.
The identification of reliable biomarkers or unique gene expression patterns in HIV latently-infected cells are key goals of current HIV research efforts. Such factors could contribute to the development of a cure by: 1) refining and broadening our understanding of HIV latency mechanisms and the biology of viral persistence, 2) enabling accurate quantification of viral reservoirs to assess viral burden and efficacy of therapeutic interventions, and 3) providing potential therapeutic targets to specifically eliminate viral sanctuaries. However, the heterogenous and dynamic nature of the viral reservoir greatly complicates this endeavor9.
Viral reservoirs are found in a variety of anatomical sites and cell types. Infected CD4+ T cells arguably constitute the most important HIV reservoir10 and within this highly diverse cell lineage, the expression of several cellular factors is associated with increased levels of integrated proviral DNA11. This includes immune checkpoint molecules such as programmed cell death protein 1 (PD1), cytotoxic T-lymphocyte-associated protein 4 (CTLA-4), lymphocyte-activation gene 3 (LAG-3), and TIGIT9, 12, 13. Moreover, expression levels of CD2 in CD4+ T cells were reported to identify HIV latently-infected cells14, while CD2015 and CD3016 expressing CD4+ T cells were found to be specifically enriched for HIV RNA. In 2017, Descours et al.17 proposed CD32a as viral reservoir marker and described an unprecedented 1000-fold enrichment in HIV DNA in CD32a+ cells as compared to CD32a-CD4+ T cells. However, this finding has been repeatedly challenged in subsequent studies and incited a controversial discussion18–23.
In the present study, we sought to provide a better understanding of the phenotypic nature of latently infected CD4+ T cells and the mechanisms involved in the establishment and maintenance of HIV reservoirs. To this aim we utilized a dual-reporter HIV construct that enables isolation and purification of uninfected, productively-infected, and latently-infected cells24, 25 by flow cytometry. We then characterized each of these purified cell populations using systems approaches to obtain comprehensive gene and protein expression profiles of HIV latently-infected, primary CD4+ T cells. Our data reveal a novel mechanism of HIV latency establishment and maintenance, identifying the hypoxia-CD73-adenosine signaling axis as a key mechanism and potential target for therapeutic intervention and diagnostic evaluation.
RESULTS
Sorting of HIVDFII-infected primary CD4+T cells enables isolation of latently-infected cells
In order to be able to characterize latent HIV reservoir cells we utilized a modified version of a single round, recombinant HIV dual-reporter virus called HIV Duo-Fluo II (HIVDFII) (Figure S1A). To achieve sufficient infection frequencies, we first activated blood-derived primary CD4+T cells obtained from six healthy donors in vitro via αCD3/αCD28 bead stimulation and then spinoculated them with HIVDFII (Figure S1B-D). Four days post infection (p.i.), productively- and latently-infected, as well as uninfected cells were purified by fluorescence activated cell sorting (FACS, Figure 1A). Expectedly25, low frequencies of latently-infected cells were found for all donors, whereas uninfected and productively-infected cells could be rapidly collected by FACS in large numbers (Figure 1B and C). After testing the sorted samples for sufficient enrichment of the desired cell populations (Figure S1E), we subjected them, together with a panel of control specimens (untreated samples; unstimulated, infected cells; and stimulated cells without HIVDFII infection) to downstream analysis using systems approaches (Figure 1A). All samples were characterized using mass cytometry by time of flight (CyTOF), measuring 40 surface proteins to provide immunophenotypic profiles at the single cell level. In parallel, we applied NanoString hybridization and fluorescence-based digital counting technology allowing for simultaneous detection of 770 mRNA and 30 protein targets.
Latently-infected cells are phenotypically diverse and include T-regs and Tfh cells
First, to establish an in-depth analysis of the phenotypic features of latently-infected cells, we implemented CyTOF, simultaneously quantifying expression levels of 40 different proteins with single cell resolution. Our labeling panel comprised T cell lineage and differentiation markers, activation markers, homing receptors, and several proteins that were described previously in the context of HIV latency9, 26, 27 (Table S1). We performed extensive high dimensional analysis and generated t-distributed stochastic neighbor embedding (tSNE) plots to visualize the data and assess specific T cell subsets and population phenotypes. Across all donors, we identified latent cells in various T cell compartments, including central memory (CD45RO+CD45RA-CCR7+CD27+), follicular helper (CD45RO+CD45RA-PD1+CXCR5+), regulatory (CD45RO+CD45RA-CD127-CD25+), and to a lesser extent naïve (CDRO-CDRA+CCR7+) T cells (Figure 2).
NanoString analysis reveals unique features of latently-infected cells that promote viral quiescence and cell survivorship
We next obtained high-dimensional NanoString data to comprehensively characterize gene and protein expression patterns of latent cells. Our initial analysis revealed that gene expression levels needed to be adjusted for donor effects that were revealed as confounding variables in an unsupervised clustering analysis (Figure 3A, see dendrogram). Importantly, we also found that expression patterns in uninfected and productively-infected cells within each donor clustered together while latently-infected cells clustered separately, suggesting that latently-infected cells exhibited distinct expression signatures (Figure 3A, see dendrogram).
Subsequently, we examined changes in regulatory and signaling pathways in latent cells based on undirected and directed global significance scores (Figure 3B). The pathways ‘Antigen Processing’, ‘Adhesion’, ‘Pathogen Defense’, and ‘Interleukins’ exhibited the most significant changes. The upregulation of ‘Interleukins’ and ‘Pathogen Defense’ in latently-infected cells pointed towards intracellular signaling cascades that antagonize productive infection. In contrast, the suppression of ‘Antigen Processing’ in latent cells indicated the ability to evade host immunity, a pro-survival effect that could enforce viral persistence, and an observation in line with previous studies reporting that HIV actively modulates antigen presentation pathways28, 29.
Next, overall target expression in the sorted samples was assessed by differential gene expression analysis (DGE, Figure 3C). Comparison of uninfected and latently-infected cells revealed 16 differentially expressed targets (adjusted p-value < 0.1), two of which were identified at the protein level: HLA-DRA protein was downregulated, and NT5E (CD73) protein was upregulated in latent cells. DGE analysis between productively- and latently-infected cells resulted in 36 significant hits (adjusted p-value < 0.1). These included CD4 protein and CD73 protein. CD4 protein was significantly downregulated in productively-infected cells, indicating active viral replication, and CD73 protein was again upregulated in latent cells.
Overlapping both individual, differential expression analyses revealed 27 targets that were uniquely modulated in latent infection (p-value < 0.05, Figure 3D). Among these, CD73 was the only hit detected on the protein level (Table S2). The list further included three mRNAs encoding for transcription factors, 6 encoding for cytokines and 11 encoding for surface proteins. Among these, NF-κBIA (NF-κB Inhibitor Alpha) mRNA, which encodes a master inhibitor of the transcription factor NF-κB30, which in turn is a crucial transcription factor of HIV31, 32, was significantly upregulated in latent cells. In addition, CASP1 mRNA was significantly downregulated in latently-infected cells. CASP1 (Caspase-1) is a key regulator of pyroptotic cell death, a highly inflammatory process that has been reported to be a major determinant of HIV pathogenesis and a potent driver of virus-dependent CD4+ T cell depletion33.
The relationship between targets specific for latent infection was then investigated by utilizing the open-source software Cytoscape to generate a detailed protein-protein interaction network (Figure 3E). Interestingly, all genes were part of one protein interaction complex. The chemotactic factor IL8 (or CXCL8) was found to have the highest number of immediate connections (17 first neighbors), followed by the transcription factor STAT1 (15 first neighbors) and the cytokine CSF2 (15 first neighbors).
Three key members of the adenosinergic pathway, IL8, CD39, and CD73, are uniquely expressed in latent cells
In a more stringent attempt to pinpoint unique features of latent cells, we incorporated non-activated and non-virus-exposed samples in our NanoString analysis, thereby emulating the in vivo landscape in ART-suppressed individuals, where the majority of CD4+ T cells are likely in a resting state and have not encountered HIV. Here, expression signatures were clearly dominated by in vitro activation (Figure S2A and B). We therefore examined first if T cell activation is a determinant of HIV latency and investigated expression levels of established T cell surface markers CD45, CD45RO, and CD4 (Figure S2C) as well as panel-specific activation markers (Figure S2D and E). No significant associations were discovered between the expression levels of these markers and latent infection.
Next, latent cells were compared by DGE in a pair-wise fashion to all other samples. Of note, only a few differentially expressed genes were found in relation to HIVDFII exposed samples (Figure 3F) whereas differences were markedly broader towards virus naïve and non-activated samples. This demonstrated that distinct, albeit subtle expression patterns of latent cells can be found following our experimental approach. The pair-wise DGE analyses were then again overlaid to carve out genes uniquely changed in latently-infected cells (Figure 3G). We found three hits that were significantly changed with respect to all other comparators: IL8 mRNA, CD39 mRNA and CD73 protein (Figure 3H). Strikingly, this result comprised CD73 protein, a key hit from our previous analyses. In addition, the three genes are known to be mechanistically connected through the adenosine signaling pathway34–38.
Hypoxia promotes CD73 expression and HIV latency in primary CD4+ T cells
To better understand the relevance of CD73 in the context of HIV infection and particularly in the establishment of the latent reservoir, we next focused on the regulation of CD73 expression39, 40. Importantly, at least one hypoxia-response element (HRE) has been previously identified in the CD73 promoter region41 allowing for direct binding of hypoxia-inducible factors (HIFs). In this context, HIF-1α was demonstrated to control CD73 expression, with CD73 typically being upregulated under hypoxic conditions42. We therefore hypothesized that our discovery of CD73 upregulation in latent cells suggested that hypoxia may be the underlying cause of both increased CD73 expression and HIV latency.
We first confirmed our NanoString data using in vitro infection with HIVDFII followed by immunofluorescence staining and flow cytometry, again observing a significant enrichment of CD73+ cells among latently-infected cells (Figure 4A and B). We then mimicked hypoxic conditions in our culture settings via administration of dimethyloxalylglycine (DMOG), followed by infection with HIVDFII and flow cytometry (Figure 4C).
DMOG treatment drastically altered CD73 expression in CD4+ T cells and led to a 2.5-fold increase of CD73+ cells compared to mock treatment (Figure 4D and E). Notably, while CD73 expression was generally increased upon induction of hypoxic responses (compare Figure 4B and Figure S3A), we observed a pronounced enrichment of CD73+ cells specifically in the latent compartment with up to 60% of latent cells being CD73 positive (Figure S3A). This was also reflected in CD73 expression per cell measured by mean fluorescence intensities (MFIs), which showed an overall increased CD73 expression in latent cells in both, mock and DMOG treated samples (Figure S3B). Importantly, latent infection became significantly more abundant under hypoxic conditions with a 33.6% increase upon DMOG treatment, while the frequency of uninfected or productive cells did not change significantly between culture conditions (Figure 4F and G).
Lastly, we examined whether CD73+ cells were enriched for latent virus (Figure 4H). We measured a considerably higher rate of latently-infected cells in CD73+ compared to CD73-cells with an average 3-fold enrichment of latent infection within the CD73+ T cell compartment (Figure 4I). Considering the average frequency of CD73+ cells among CD4+ T cells (∼10%, Figure 4E), these data suggest that approximately 25% of the overall, peripheral CD4+ T cell HIV reservoir may reside in CD73+ cells. Interestingly, the enrichment of latent cells in CD4+ CD73+ T cells did not significantly differ between mock- and DMOG-treated samples (Figure 4I), indicating that CD73+ cells may possess specific features that favor the establishment and/or maintenance of latent infection that are not further modulated by the induction of hypoxic responses.
CD73+ cells exhibit specific immunoregulation and tyrosine kinase signaling cascade patterns
The role of CD73 in the context of oncogenesis, tumor progression and survival is well described36. Hitherto however, relatively little is known about the relevance and function of CD73 in human CD4+ T cells. We thus characterized the transcriptome of blood-derived, primary CD73+ and CD73-CD4+ T cells by RNA sequencing. To this aim, CD73-/+ cells were isolated from primary CD4+ T cells by FACS, revealing highly variable CD73 surface expression on CD4+ T cells across donors, ranging from ∼3% to ∼23% (Figure S4A and B). Sorted cells were then subjected to RNA sequencing and analyzed for DGE, taking into account and adjusting for donor effects (Figure S4C and D). CD73- and CD73+ CD4+ T cells exhibited distinct transcriptional profiles and a clear clustering of the sorted populations (Figure 5A). Expectedly, CD73 showed the most significant DGE, and we found 145 additional genes being differentially expressed, with 111 upregulated and 34 downregulated genes in CD73+ cells (Figure 5B). CR1, ADAM23, ABCB1 and AUTS2 were among the top genes upregulated in CD73+ cells (Figure 5B, yellow dots). CLEC17A exhibited the highest fold change (> 5-fold), followed by CD73, LINC02397 and MACROD2 (> 4-fold) (Figure 5B, red dots). Multiple genes were markedly downregulated in CD73+ cells; however, only two of them reached the highest level of statistical significance: FCER1A and NPR3. A gene set enrichment analysis yielded a multitude of significantly changed pathways between CD73+ and CD73-cells including ‘immune responses’, ‘complement activation’ and ‘receptor-mediated signaling cascades’ (Figure 5C) indicating diverging immunoregulatory programs being active in CD73+ and CD73-cells. In addition, ‘positive regulation of angiogenesis’, ‘regulation of apoptotic processes’ and ‘tyrosine kinase signaling cascades’ were among the 40 most significantly changed pathways.
Next, all differentially expressed hits were uploaded into Cytoscape and the resulting protein-protein interaction network was then overlaid with log2 fold changes and adj. p-values of each gene, visualizing directionality and significance of interacting hits (Figure 5D). The analysis revealed 5 distinct interaction networks, with one network containing the majority of hits. This intrinsically cross-connected network pointed towards overlapping signaling cascades and biological processes shared by the entire CD73+ CD4+ T cell compartment. Within this network, a tight cluster (I) of cell surface receptors was apparent (CD22, CD79A, CR1, CR2, KIT, FCER1A, CD34), accompanied by several signaling factors (LYN, BLNK, BLK, BTK, PIK3AP1, TCL1A) and transcription factors (IRF8, EBF1, PAX5). Another cluster (II) comprised downregulation of cell surface proteins CXCR3, CCR4 and PTGDR2. Importantly, the network also contained two significantly downregulated genes, transcription factor GATA2, and DNA topoisomerase TOP2A, which are known to support active HIV transcription and replication43, 44.
CD73+ CD4+ T cells harbor an inducible HIV reservoir
The intactness of integrated proviruses and the inducibility of viral gene expression are key determinants of the latent HIV reservoir. We therefore sought to investigate the capability of the HIV reservoir in CD73+ T cells to support reactivation of viral transcription. To this end, we adapted a primary in vitro HIV infection model45 and infected FACS-sorted CD73- and CD73+ CD4+ T cells with the reporter virus HIVLuc (Figure 5E). After a resting period of 5 days, we stimulated cells with αCD3/αCD28 beads and observed a pronounced increase of viral transcription in both, CD73- and CD73+ CD4+ T cells as compared to unstimulated cells, reflected by a 10-fold and 14-fold increase in LTR-driven luciferase activity, respectively (Figure 5F). Interestingly, induction of viral transcriptional activity was significantly higher in stimulated CD73+ T cells as compared to CD73-T cells. These results demonstrate that the CD73+ CD4+ T cell compartment can harbor an inducible latent reservoir in vitro, indicating the potential of this compartment to contribute to a spreading infection in PLWH upon treatment cessation.
Blocking the adenosine receptor A2AR promotes HIV latency reversal
Bearing in mind the enzymatic function of CD73, we hypothesized that extracellular adenosine produced by CD73 could be mechanistically involved in the establishment and/or maintenance of HIV latency. To test this, we investigated the adenosine signaling cascades downstream of CD73 in well-established cell line models46, 47 of latency, the so-called J-Lat cells46, 47, exploring how modulation of adenosine receptors by small molecule drugs affects HIV transcriptional activity.
First, CD73 surface expression was measured in three J-Lat clones, 5A8, 6.3 and 11.1, as well as in the parental JurkatE6 cell line (Figure 6A). Between 10-20% of JurkatE6, J-Lat 6.3 and 11.1 cells expressed CD73 on the surface. In strong contrast, about 80% of J-Lat 5A8 cells were CD73 positive. Due to this remarkably high frequency of CD73-expressing cells, subsequent experiments were performed with this clone. J-Lat 5A8 cells were pretreated with the antagonist SCH-58261 (SCH), which blocks adenosine signaling, followed by viral reactivation with PMA/I, a strong mitogen and latency reversing agent. SCH pretreatment did not compromise cell viability at any dose compared to cells treated with PMA/I only (Figure 6B). Notably, pretreatment with SCH significantly promoted latency reversal and resulted in an evident dose response with a 2-fold increase in the frequency of GFP+ cells at the highest dose (Figure 6C).
A comprehensive model of CD73-dependent latency
Based on our data, obtained in in vitro infection models and cell lines, we devised a model that incorporates hypoxia, CD73 and adenosine signaling, linking it with HIV transcriptional regulation and persistence. Our model predicts that the hypoxia-CD73-adenosine (HCA) axis plays a vital role with CD73 as a central player in the maintenance of latent HIV infection (Figure 6D): As is known, HIV preferentially persists in lymphatic tissues48, which exhibit low oxygen levels and thus high levels of active HIFs. We surmise that the resulting upregulation of CD73 expression leads to adenosine-rich, immunosuppressive microenvironments, which promote the establishment and maintenance of latent reservoirs. Accumulation of extracellular adenosine thereby creates optimal conditions for HIV persistence by suppressing HIV host dependency factors through autocrine signaling cascades, while impairing effective host immune responses through paracrine signaling mechanisms. Therefore, viral transcription is repressed, and immune escape of HIV-infected cells is facilitated, altogether promoting viral quiescence and survival of the latent reservoir.
Detection of HIV and CD73 colocalization in patient-derived tissue
To evaluate the clinical relevance of our findings and to test the HCA model in vivo, we applied a comprehensive tissue imaging pipeline that enables simultaneous detection of integrated HIV-DNA, viral mRNA, and HIV protein, as well as lineage markers in situ, followed by quantitative image analyses48. We concomitantly detected CD73 expression and HIV molecules in lymph node tissues (peripheral and inguinal lymph nodes) obtained from HIV-infected individuals on suppressive ART, as well as in viremic untreated individuals, and uninfected controls (Figure 7A). CD3 staining was used as a lineage marker for T cells and DAPI was included as a counter staining of nuclear DNA. CD73 signals were found across all samples, while HIV detection expectedly differed greatly between ART-suppressed and viremic individuals and was absent in uninfected control samples (Figure 7A and B). The CD73 expression and distribution in uninfected lymph nodes was uniform (Figure 7A). In contrast, in the setting of HIV infection, CD73 was mostly concentrated in HIV-infected or surrounding cells (Figure 7A). Quantification of the viral reservoir in viremic individuals indicated that at least 40% of CD3+ cells harbored integrated DNA and most of these cells expressed viral RNA and HIV-p24 (Figure 7B, viremic). In contrast, in ART-suppressed individuals, 0.017% of CD3+ cells contained viral DNA, half of which expressed viral RNA and 10% of which expressed viral proteins (Figure 7B, ART-suppr.). No unspecific detection of viral DNA, RNA or protein were detected in uninfected tissues.
CD73 expression correlates with HIV persistence in lymph nodes from ART-suppressed individuals
Next, we assessed correlations between HIV infection at the cellular level and CD73 expression specifically in the CD3+/CD3-cell compartment (Figure 7C). In general, CD73 expression levels in CD3+ cells were significantly higher in ART-suppressed samples as compared to uninfected samples. In addition, HIV DNA+ CD3+ cells in ART-suppressed individuals possessed a significant, 1.6-fold higher CD73 expression compared to the overall CD3+ population. This difference in CD73 expression levels was not observed in viremic individuals and suggested that CD73 expression might offer a survival advantage for HIV-infected cells in the setting of ART. We surmise that in the absence of treatment, when HIV is not subject to pharmacologic selection pressure, active viral replication dominates and potentially overwrites CD73 effects. It is likely that the long-term persistence of HIV during ART unearths the association between CD73 and viral latency. We further hypothesized that the effects of CD73-mediated viral persistence would not be limited to the CD73-expressing cell itself, and the immunosuppressive effects due to adenosine production would cover a ‘neighborhood’ within the lymph node. To explore this hypothesis, we performed analyses of the imaging data measuring distance-dependent CD73 positivity in the vicinity of CD3+ cells. Our data indicated that HIV DNA+ CD3+ cells in lymph nodes from ART-suppressed individuals reside in areas with generally high CD73 expression, reinforcing the notion that CD73 via adenosine production is contributing to HIV persistence (Figure 7D).
DISCUSSION
The latent HIV reservoir is considered the main obstacle to achieving full remission from HIV infections. It is generally believed that integrated provirus persists in specific cellular compartments and tissue sanctuaries, which enable both, long-term survival and spontaneous reactivation of viral replication even after decades of successful ART. The exact identity of the HIV reservoir however remains elusive. Goal of this study was to thoroughly characterize primary HIV latently-infected cells using two comprehensive and complementary systems approaches. Our work focused on CD4+ T cells, the major target of HIV infection and an important source of viral rebound upon ART interruption.
The core finding of this work was the identification of an elevated cell-surface expression of CD73 as signature characteristic of HIV latently-infected CD4+ T cells – an observation that, to the best of our knowledge, has not been described before. Using different in vitro models of HIV latency, we found indications that CD73 is not just a passive marker of reservoir cells, but that the regulation of CD73 expression as well as its enzymatic function may be linked to HIV persistence.
Noteworthily, our NanoString analysis also indicated significant changes in the expression of CD39 and IL8, which are mechanistically connected. In particular the two ectonucleotidases CD39 and CD73, and their orchestrated functionality in purinergic signaling are well described in the literature36–38. In addition, IL8 expression has been shown to be stimulated by adenosine signaling, and is thus directly linked to the enzymatic function of CD7335. The distinctive expression of these three genes in latent cells suggest that CD73 and the adenosinergic pathway play a pivotal role in HIV latency and the establishment and/or maintenance of viral reservoirs.
Our initial in vitro infection model using HIVDFII focused on a rather narrow time window of HIV latency, thus recapitulating the establishment and early persistence of viral reservoirs within a few days after infection. In clinical settings and most infected individuals on the other hand, HIV persists even after decades of successful ART. In line with that, our in situ imaging of lymph node tissues demonstrated, for the first time, that a significant association between CD73 expression in T cells and HIV-DNA specifically in ART-suppressed individuals is being maintained over extended periods of time.
A key finding of our study is also that CD73 is significantly upregulated under hypoxic conditions in the CD4+ T cell compartment. This result is important for two reasons: 1) Oxygen tension levels are highly variable throughout the body, ranging from 19% in well oxygenated tissues49, down to 0-7%50 in the gastrointestinal tract, and 0.5-4.5% in lymphoid organs51, 52. As a consequence, immune cells including CD4+ T cells encounter and operate at varying oxygen concentrations as they traffic through the body52. Lymphoid organs in particular represent crucial viral sanctuaries and as these are characterized by low oxygen levels, our data suggest a causal link between HIV persistence, CD73 and hypoxia. 2) Hitherto, the role of oxygen levels and hypoxia in HIV infection remains intangible and somewhat obscure. In 2009 Charles et al.53 reported decreased HIV-1 RNA levels at 3% oxygen, while S. Deshmane and colleagues showed increased HIV transcription mediated by the interaction between HIV accessory protein Vpr and HIF-1α54, 55. Moreover, HIV-1 replication was shown to be promoted by HIF-1α which in turn was stabilized by reactive oxygen species56. Very recently it was demonstrated that hypoxia can promote HIV latency with HIF-2α as a direct inhibitor of viral transcription57. Our data now provided new insights into the connection between HIV latency and hypoxia, with CD73 emerging as key mediator between hypoxia and viral transcription.
As outlined above, the clear association between CD73 expression and HIV latency suggested a mechanistic involvement of CD73 via its enzymatic function in the adenosine signaling cascade. Adenosine signaling leads to a suppression of cellular transcription factors that are critical for active viral transcription58, 59 and may create an ideal immunological niche for infected cells to evade host immune clearance due to an adenosine-mediated immunosuppressive microenvironment. In fact, pharmacological blockade of A2AR in an HIV latency cell culture system clearly facilitated HIV latency reversal without impairing cell viability. Thus, our data demonstrate the potential of targeting the adenosinergic system as a therapeutic approach in the context of HIV infection and provides first evidence that the enzymatic activity of CD73 is directly involved in HIV persistence.
It is important to note that in recent years increasing circumstantial evidence has been gathered, independently suggesting a critical role for either hypoxia53–57, CD73 expression60–62 or adenosine signaling63–67 in HIV infections. Our data now implies a direct, causal connection that links the hypoxic regulation and adenosine-producing enzymatic activity of CD73 to the establishment and persistence of viral reservoirs.
Finally, the CD73-adenosine axis has received a great deal of attention in the context of cancer biology68–70 and the adenosinergic system has emerged as a promising new drug target in oncology35, 36, 71–76. Our data suggest that, like solid tumor cells, latent HIV reservoirs hijack the CD73-adenosine axis to subvert innate and adaptive immune responses, enhancing the survivorship of the infected cell as a persistence mechanism during ART. Therefore, CD73- and adenosine-focused anti-cancer therapeutics should be actively explored for the development of novel HIV cure approaches and host-directed therapies.
MATERIALS AND METHODS
Molecular Cloning
The 1st generation dual reporter virus construct R7GEmC24 (kindly provided by Dr. Eric Verdin) was adapted by ligation-based molecular cloning. Briefly, R7GEmC was linearized by enzymatic digest using FseI and AscI (New England Biolabs, Cat. #R0588S and Cat. #R0558S) in order to excise the mCherry open reading frame. Then, mKO2 was PCR amplified from the template mKO2-N1 (Addgene, Cat. #4625) using primers containing matching restriction sites. Subsequently, the lentiviral vector was dephosphorylated using Quick CIP (New England Biolabs, Cat. #M0525S), purified and subjected to ligation with the digested and gel-purified PCR product utilizing T4 DNA ligase (New England Biolabs, Cat. # M0202S). Finally, circularized plasmid DNA was transformed by heat-shock in chemically competent Stbl-3 E. coli (ThermoFisher Scientific, Cat. #C737303) for subsequent antibiotics selection and Sanger sequencing (Elim Biopharm) of positive clones.
Plasmid amplification and preparation
Transformed Stbl-3 cells were grown in 2-5 ml of lysogeny broth with 0.1 mg/ml Ampicillin (LB Amp) for 4-8 hours prior to inoculation of large-volume flask with 100-300 ml LB Amp for overnight growth. 16 hours later, cells were pelleted by centrifugation for 15 min at >3000 g and subjected to plasmid preparation using Plasmid Plus Maxi Kits (QIAGEN, Cat. #12963) following the manufacturer’s protocol. The DNA concentration of isolated plasmid DNA was spectrophotometrically determined using a NanoDrop 1000 (ThermoFisher, Scientific) and DNA aliquots were stored at 4°C.
Cell lines and cell culture
HEK293T were obtained from ATCC and were cultured in DMEM (ThermoFisher, Scientific, Cat. # 11965-118) supplemented with 10% FBS (Corning, Inc., Cat. #35-010-CV) and 10% Penicillin/1% Streptomycin (Fisher Scientific, Cat. #11548876) (DMEM complete = DMEM+/+) at 37°C, 5% CO2 unless stated otherwise. J-Lat 5A8 cells were kindly provided by Warner C. Greene (Gladstone Institutes). All other J-Lat clones were a gift from Eric Verdin (Buck Institute). Jurkat E6-1 cells were obtained from ATCC. All suspension cells were cultured in RPMI 1640 (ThermoFisher, Scientific, Cat. #11875-119) supplemented with 10% FBS and 10% Penicillin/1% Streptomycin (RPMI complete = RPMI+/+) at 37°C, 5% CO2 unless stated otherwise.
Virus production
HIV-1 viruses were generated by transfection of proviral DNA into HEK293T cells via polyethylenimine (PEI, Polysciences, Cat. #23966) transfection protocol. Env-pseudotyped HIVDFII stocks were produced by co-transfecting plasmids encoding HIVDFII and a plasmid encoding HIV-1 dual-tropic envelope (pSVIII-92HT593.1, NIH HIV Reagent Program, Cat. #3077) at a ratio of 3:1 into HEK293T cells at 50-60% confluency grown in 175 cm2 culture flasks. Each flask was transfected with a total amount of 30 μg DNA. The transfection mix was prepared in 2 ml Opti-MEM (ThermoFisher Scientific, Cat. #31985062) as follows: DNA plasmids were diluted in Opti-MEM first, then PEI was added at a ratio of 3:1 PEI:DNA (90 μg PEI). The transfection mix was vortexed for 15 sec and incubated for 15 min at RT. Culture medium was replaced with 20 ml fresh DMEM + 10% FBS without P/S, and 2 ml transfection mix was added to each flask. 16h post transfection, P/S-free medium was replaced with standard culture medium (DMEM+/+), and cells were incubated for another 24h at 37°C, 5% CO2. For replication competent HIV NL4-3 Luciferase (a kind gift from Dr. Warner Greene), lentiviral vectors were introduced by Fugene HD transfection (Promega, Cat. #E2311) according to the manufacturer protocols. Cell supernatants were collected 48h post transfection, centrifuged at 4°C for 10 min at 4000 rpm (∼ 3390 x g) and subsequently filtered using 0.22 µm membrane vacuum filter units (MilliporeSigma, Cat. #SCGP00525) to remove cell debris. Virus preparations were concentrated by ultracentrifugation at 20,000 rpm (∼ 50,000 x g) for 2h at 4°C and resuspended in complete media for subsequent storage at −80°C. Virus concentration was estimated by p24 titration (HIV-1 alliance p24 ELISA kit, PerkinElmer, Cat. #NEK050001KT).
J-Lat cell latency reversal
J-Lat 5A8 cells (seeded at 1×106 cells/ml) were incubated with CGS21680 (Sigma-Aldrich, Cat. #C141-5MG) or SCH-58261 (Sigma-Aldrich, Cat. #S4568-5MG) at 37°C for 1h at increasing doses in RPMI+/+, followed by stimulation with 20 nM PMA / 1 μM Ionomycin (PMA/I, Sigma-Aldrich, Cat. #10634-1MG and Cat. #10634-1MG). Untreated cells or cells treated with 0.5% DMSO (Sigma-Aldrich, Cat. #D2650-100ML) served as negative controls. 7h after PMA/I reactivation, cells were washed 2x with PBS (ThermoFisher, Scientific, Cat. #14190-250) and viral transcriptional activity, reflected by GFP expression was measured using LSR II flow cytometer (BD Biosciences).
Leucocyte isolation and primary cell culture
Peripheral blood mononuclear cells (PBMCs) from HIV-seronegative donors (Vitalant) were isolated by Ficoll-Hypaque density gradient centrifugation (Corning, Inc., Cat. #25-072-Cl) at 2000 rpm (∼ 850 x g) at RT for 30 min, without brake. PBMCs were immediately processed to isolate CD4+ T cells by negative selection using the EasySep Human CD4+ T Cell Isolation Cocktail (StemCell Technologies, Cat. #17952) according to manufacturer’s protocol. Purified CD4+ T cells were cultured in RPMI+/+.
CD4+ T cell in vitro activation and infection
CD4+ T cells from peripheral blood were stimulated with αCD3/αCD28 activating beads (ThermoFisher Scientific, Cat. #11132D) at a concentration of 1 bead/cell in the presence of 100 U/ml IL-2 (PeproTech, Inc., Cat. #200-02) in RPMI +/+ for 3 days (initial seeding concentration 1×106 cells/ml). At the day of infection, cells were spinoculated in 96-well V-bottom plates (MilliporeSigma, Cat. #M9686-100EA) in 50 μl RPMI+/+ with 100 ng (HIVDFII) of p24 per 1×106 cells with 5×106 cells total per well for 2 h at 2350 rpm (1173 × g) at 37°C. After spinoculation, all cells were returned to culture in the presence of 30 U/ml IL-2. Pre-stimulated CD4+ T cells stayed in αCD3/αCD28 activating beads during spininfection and subsequent cell culture. For hypoxia experiments, cells were treated with 500 μM DMOG (Sigma-Aldrich, Cat. #D3695) or mock-treated with 0.5% DMSO two days after αCD3/αCD28 bead stimulation and 24h before HIVDFII spininfection. Cells were kept in DMOG containing RPMI+/+, in presence of activation beads and IL-2 until sample collection 4 days post infection.
CD4+ T cell in vitro infection and latency reversal
Initially, CD4+ T cells were isolated from peripheral blood as described above and subjected to FACS of CD73+ and CD73-cells. To that aim, cells were stained with APC anti-human CD73 (BioLegend, Cat. #344006) diluted in PBS (1:20) in 100 μl final volume for 20 min at RT. Cells were then washed twice with PBS and resuspended in 500 μl – 1000 μl PBS to achieve high cell concentrations (20 −40×106 cells/ml) for FACS. Cells were sorted into 15 ml conical tubes containing 1.5 ml RPMI+/+. Cells were cultured for 24h, then infected and rested in the presence of ART to establish in vitro latency45. Briefly, 100 ng of purified NL4-3-Luciferase was added per 1×105 sorted cells, which were then infected by spinoculation as described above. 24h post virus exposure, 5 µM saquinavir (protease inhibitor, Sigma-Aldrich, Cat. #S8451-50MG) was added to the cell cultures to suppress spreading infection. 5 days later, cells were stimulated with αCD3/αCD28 beads (or left untreated) in the presence of 30 µM raltegravir (integrase inhibitor, Sigma-Aldrich, Cat. #CDS023737-25MG) to prevent new infections. 24h after stimulation, luciferase activity was quantified using the bright glo luciferase assay system (Promega, Cat. #E2610).
Cell staining and processing
Freshly isolated CD4+ T cells were stained for viability, using the fixable Zombie viability dye (1:100; BioLegend, Cat. #423113) according to the manufacturer’s protocol. Subsequently, antibodies APC anti-Human CD73 (1:50), APC-Cy7 anti-Human CD25 (1:100, BD Pharmingen Cat. #557753), and V450 anti-Human CD69 (1:100, BD Horizon Cat. #560740) for cell surface staining diluted in PBS were added and incubated for 20 min at RT. For flow cytometry, cells were washed and fixed in 1% paraformaldehyde (PFA) in PBS after the staining. FACS experiments were performed with live, unfixed cell samples. Flow cytometry analyses were performed on the LSR II flow cytometer (BD Biosciences) or MA900 Multi-Application Cell Sorter (Sony Biotechnologies). All fluorescent-based sorts were conducted on the latter instrument.
Flow cytometry gating and data analysis
Data were analyzed and visualized using the FlowJo software (v.10.7.1). Crosstalk compensations was performed using single-stained samples for each of the fluorochromes and isotype controls were employed to assess antigen positivity and enable specific gating. FACS and flow cytometry gating was performed as follows: first, single live cells were selected from FSC/SSC scatter plots, sub-gated on Zombie low/negative cells. Then, antibody gates (CD73, CD69 or CD25) were defined based on suited isotype controls. Gating of HIVDFII reporter expression was based on non-infected, mock-treated (in vitro activated) negative control samples to account for activation-dependent increase of cellular background fluorescence.
Expression Profiling via NanoString
Quantitative RNA and protein expression data were generated using the nCounter Vantage 3D RNA:Protein Immune Cell Profiling Assay (NanoString Technologies, Inc., Cat. #121100019) and the nCounter SPRINT profiler (NanoString Technologies), comprising 770 RNA and 30 protein targets as well as positive and negative controls. 100,000 viable, sorted cells were used per sample, which were processed according to the manufacturer’s instructions. RNA and protein expression values were normalized and analyzed using the nSolver Analysis Software 4.0 and the add-on Advanced Analysis Software 2.0.115 (NanoString Technologies). Samples that did not pass the default control performance and quality parameters were excluded from subsequent analysis. Normalization genes for each sample were automatically selected by the software based on the geNorm algorithm. Biological replicates were grouped according to sample type and the differential expression of each analyte-type (RNA or protein target) was determined in cross-comparisons among all sample types by considering inter-donor differences as confounding variable unless otherwise stated. Intersections of significant targets of individual differential expression analyses were visualized in a Venn diagram using an open source platform from Bioinformatics & Evolutionary Genomics77.
Based on the differential expression of each gene, gene sets pre-defined by nanoString, representing different pathways included in this assay, were analyzed by calculating global significance scores for each gene set within each sample as follows: undirected global significance statistic = , where ti is the t-statistic from the ith pathway gene. The directed global significance statistic is similar to the undirected global significance statistic, but rather than measuring the tendency of a pathway to have differentially expressed genes, it measures the tendency to have over- or under-expressed genes. It is calculated similarly to the undirected global significance score, but it takes the sign of the t-statistics into account: Directed global significance statistic = sign(U)|U|1/2 where U = and where sign(U) equals −1 if U is negative and 1 if U is positive (MAN-10030-0278).
CyTOF samples preparation and analysis
For live/dead discrimination, 0.1-1 million cells per sample were treated with cisplatin (Sigma-Aldrich) and fixed with paraformaldehyde (PFA) as previously described9, 26, 27. Briefly, Cells were washed once with contaminant-free PBS (Rockland) with 2 mM EDTA (Corning), centrifuged and resuspend with 25 μM cisplatin in 4 ml PBS/EDTA and incubated for 60 seconds at room temperature, and then quenched with CyFACS (metal contaminant-free PBS supplemented with 0.1% bovine serum albumin and 0.1% sodium azide). Cells were then centrifuged, fixed with 2% PFA in metal contaminant-free PBS and washed 3x with CyFACS. These fixed cells were stored at −80°C until CyTOF staining.
Prior to CyTOF staining, multiple samples were barcoded using the Cell-ID 20-Plex Pd Barcoding Kit according to manufacturer’s instructions (Fluidigm). Briefly, each sample was washed twice with Barcode Perm buffer (Fluidigm), and incubated for 30 min with the appropriate barcode at a 1:90 ratio. Cells were then washed with 0.8 ml Maxpar Cell Staining buffer (Fluidigm) in NuncTM 96 Deep-Well polystyrene plates (Thermo Fisher), followed by CyFACS. Barcoded samples were combined and blocked with sera from mouse (Thermo Fisher), rat (Thermo Fisher), and human (AB serum, Sigma-Aldrich) for 15 minutes at 4°C. Cells were washed twice with CyFACS buffer and stained with the cocktail of CyTOF surface staining antibodies (Table S1) for 45 min on ice. Subsequently, cells were washed 3X with CyFACS buffer and fixed overnight at 4°C with 2% PFA (Electron Microscopy Sciences) in metal contaminant-free PBS. For intracellular staining, cells were permeabilized by incubation with fix/perm buffer (eBioscience) for 30 min at 4°C and washed twice with Permeabilization Buffer (eBioscience). Cells were blocked again with sera from mouse and rat for 15 min on ice, washed twice with Permeabilization Buffer (eBioscience), and stained with the cocktail of CyTOF intracellular staining antibodies (Table S1) for 45 min on ice. Cells were then washed once with CyFACS and incubated for 20 min at room temperature with 250 nM Cell-IDTM DNA Intercalator-Ir (Fluidigm) in 2% PFA diluted in PBS. Cells were washed twice with CyFACS, once with Maxpar Cell Staining Buffer (Fluidigm), once with Maxpar PBS (Fluidigm), and once with Maxpar Cell Acquisition Solution (CAS, Fluidigm). Immediately prior to acquisition, cells were resuspended in EQTM calibration beads (Fluidigm) diluted in CAS. Cells were acquired at a rate of 250-350 events/sec on a CyTOF2 instrument (Fluidigm) at the UCSF Parnassus single cell analysis facility.
Data were normalized to EQTM calibration beads and de-barcoded with CyTOF software (Fluidigm). Normalized data were imported into FlowJo (BD) for gating (cell, intact, live, single events) and heatmap was generated in Cytobank.
RNA Sequencing
Freshly isolated CD4+ T cells from healthy blood donors were sorted based on their CD73 expression as described above. 3×106 cells were collected per sample and stored as dry cell pellets at −80 °C. RNA preparation, library preparation and mRNA sequencing were conducted at Genewiz (USA). Paired-end sequencing was performed using the Illumina NovaSeq 6000 instrument to obtain a minimum of 20 million read pairs per sample with a read length of 2×150 bp. Sequence reads were trimmed to remove possible adapter sequences and nucleotides with poor quality using Trimmomatic v.0.36. The trimmed reads were mapped to the Homo sapiens GRCh38 reference genome available on ENSEMBL using the STAR aligner v.2.5.2b. Unique gene hit counts were calculated by using featureCounts from the Subread package v.1.5.2. The hit counts were summarized and reported using the gene_id feature in the annotation file. Only unique reads that fell within exon regions were counted. After extraction of gene hit counts, the gene hit counts table was used for downstream differential expression analysis. Using DESeq2, a comparison of gene expression between samples was performed adjusting for the donor effect as confounding variable. The Wald test was used to generate p-values and log2 fold changes. Genes with an adjusted p-value < 0.05 (Benjamini-Hochberg method) and absolute log2 fold change > 1 were called as differentially expressed genes for each comparison. A gene ontology analysis was performed on the statistically significant set of genes by implementing the software GeneSCF v.1.1-p2. The goa_human GO list was used to cluster the set of genes based on their biological processes and determine their statistical significance. A list of genes clustered based on their gene ontologies was generated.
In Situ Detection of HIV and Cellular Markers
The experimental procedure for the immunofluorescence staining and parallel detection of viral nucleic acids has been described and gradually optimized in a series of previous publications48, 79, 80. In its ultimate version, the protocol enables the detection of HIV-integrated DNA, viral mRNA, viral proteins, and several cellular markers in the same assay. Sample preparation, data acquisition and subsequent analyses were conducted in the laboratory of Dr. Eliseo Eugenin.
Tissue samples
Tissues from ART-suppressed individuals who have been on treatment for at least 6 months and had viral loads below clinical detection limits (<50 RNA copies/ml), as well as tissues from HIV-negative and ART-naïve viremic individuals with high plasma loads (>50 RNA copies/ml) were part of an ongoing research protocol approved by University of Texas Medical Branch (UTMB). Further clinical data and additional information are available and will be provided upon request by the lead contact Eliseo Eugenin (eleugeni{at}utmb.edu). All tissues were obtained with full, written consent from the study participants and freshly collected specimens were immediately fixed with 4% PFA, then mounted into paraffin blocks and subjected to tissue sectioning and ultimately to analysis by immunostaining.
Staining procedures
Paraffin-embedded slides containing the tissue samples were consecutively immersed in the following solutions: xylene for 5 min (2 times), 100% EtOH for 3 min, 100% EtOH for 3 min, 95% EtOH for 3 min, 90% EtOH for 3 min, 70% EtOH for 3 min, 60% EtOH for 3 min, 50% EtOH for 3 min, miliQ H2O for 3 min. Then, tissue was encircled with ImmEdge Pen to reduce the reagent volume needed to cover the specimens. Finally, slides were immersed in miliQ H2O for 3 min. For Protein K treatment, tissues were incubated with proteinase K diluted 1:10 in 1X TBS (Fisher Scientific, Cat. #BP24711; and PNA ISH kit, Agilent Dako, Cat. #K5201) for 10 min at RT in a humidity chamber. Next, slides were immersed in miliQ H2O for 3 min, then immersed in 95% EtOH for 20 sec and finally, the slides were let air-dry for 5 min. For HIV DNA probe hybridization tissues were incubated with 10 µM PNA DNA probe for Nef-PNA Alexa Fluor 488 and Alu-PNA Cy5 (PNA Bio). Next, slides were placed in a pre-warmed humidity chamber and incubated at 42°C for 30 min, then the temperature was raised to 55°C for an additional 1 h incubation. Subsequently, tissues were incubated using Preheat Stringent Wash working solution (PNA ISH kit) diluted 1:60 in 1X TBS for 25 min in an orbital shaker at 55°C. Slides were equilibrated to RT by brief immersion in TBS for 20 sec. HIV mRNA detection followed the manufacturer’s protocol for RNAscope 2.5 HD Detection Reagent-RED (Advanced Cell Diagnostics, Inc., Cat. #322360). Probe for HIV Gag-pol was added to the tissue samples and incubated for 30 min at 42°C and then 50 min at 55°C. Next, samples were incubated in Preheat Stringent Wash working solution diluted 1:60 in 1X TBS (PNA ISH kit) for 15 min in an orbital shaker at 55°C. Finally, slides were immersed in 1X TBS for 20 sec. For HIV or cellular protein detection antigen retrieval was performed by incubating slide sections in commercial antigen retrieval solution (Agilent Dako, Cat. #S1700) for 30 min in a water-bath at 80°C. Next, slides were removed from the bath and allowed to cool down in 1X TBS. Samples were permeabilized with 0.1% Triton X-100 (Sigma-Aldrich, Cat. #X100) for 2 min and then washed in 1X TBS for 5 min three times. Unspecific antibody binding sites were blocked by incubating samples with freshly prepared blocking solution. Afterwards, sections were incubated overnight at 4°C using a humidity chamber (10 ml of blocking solution: 1 mL 0.5 M EDTA, 100 ul Fish Gelatin from cold water 45%, 0.1 g Albumin from Bovine serum Fraction V, 100 ul horse serum, 5% human serum, 9 mL miliQ H2O). A primary antibody was added to the samples diluted in blocking solution and incubated at 4°C overnight. Then, slides were washed in 1X TBS 5 min for three times to eliminate unbound antibodies. Secondary antibodies were added at the appropriate dilutions and incubated for 2h at RT. Slides were washed three times in 1X TBS for 5 min to eliminate unbound antibodies. Next, slides were mounted using Prolong Diamond Antifade Mount medium containing DAPI (ThermoFisher Scientific, Cat. #P36931). Slides were kept in the dark at 4°C.
Image acquisition and analysis
Cells were examined by confocal microscopy using an A1 Nikon confocal microscope with spectral detection and unmixing systems. Image analysis was performed using the Nikon NIS Elements Advanced Research imaging software (Nikon Instruments Europe B.V.). The automated image segmentation and analysis is based on the following premises: For detection of HIV-integrated DNA, first, automatic or manual detection of cells that are positive for HIV-DNA and second, the HIV-DNA probe has to colocalize with DAPI and Alu repeats staining with a Pearson’s correlation coefficient of at least 0.8 as described previously81. For HIV-integrated DNA, these two conditions are essential, or the signal is considered negative or unspecific. For detection of HIV-mRNA, first, low colocalization with DAPI or Alu-repeats (0.2 Pearson’s correlation coefficient or below) and second, presence in cells with HIV-DNA signal. The sensitivity, accuracy and specificity of the system was previously validated in the laboratory of our collaborator in two well characterized T cell lines A3.01 (uninfected) and ACH-2 (HIV-infected) and two monocytic cell lines, HL-60 (uninfected) and OM-10 (HIV-infected).
Quantification and Statistical Analysis
Statistical details are given in the figure legends. All statistical analyses were performed using GraphPad Prism software versions 9. P-values ≤ 0.05 were considered statistically significant. A Student’s two-tailed t-test was used for two-way column analyses. ANOVA tests were used for multiple comparisons. P-values are denoted in figure panels. Data are presented as means with error bars indicating standard error of the mean (SEM) unless otherwise stated.
DATA AVAILABILITY
The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.
COMPETING INTERESTS
The authors declare no competing financial or non-financial interests.
ACKNOWLEDGEMENTS
The authors would like to acknowledge Guorui Xie, PhD and Ashley George, PhD for helpful input regarding the CyTOF analysis. The authors thank Leonard Chavez, PhD and Shivani Desai for providing reagents, as well as Konstantinos Georgiou, PhD and Zain Y. Dossani, PhD for guidance and support in data analysis.
We would like to thank The National NeuroAIDS Tissue Consortium (NNTC) for providing all human samples and associated information. The NNTC is made possible through funding from the NIMH and NINDS by the following grants: Manhattan HIV Brain Bank (MHBB): U24MH100931; Texas NeuroAIDS Research Center (TNRC): U24MH100930; National Neurological AIDS Bank (NNAB): U24MH100929; California NeuroAIDS Tissue Network (CNTN): U24MH100928; and Data Coordinating Center (DCC); R01AI147777; R01AI127219; UM1AI64567; UM1AI64559. The staining and analysis were funded by the National Institute of Mental Health grant, MH096625 and MH128032 and the National Institute of Neurological Disorders and Stroke, NS105584 (to E. A. E.).