The chaperone protein p32 stabilizes HIV-1 Tat and strengthens the p-TEFb/RNAPII/TAR complex promoting HIV transcription elongation

Significance

Results presented here highlight: 1) a chromatin affinity purification strategy to identify host factors regulating HIV transcription; 2) identification of a few yet unknown host factors regulating HIV transcription, namely the scaffold protein p32; 3) p32 enhances the half-life of the HIV transactivator Tat protein and strengthens its association with trans-activation response element (TAR), p-TEFb, and RNAPII to promote polymerase transcriptional elongation and HIV amplification; 3) the Tat inhibitor dCA, besides blocking Tat–TAR interaction, was found to disrupt Tat binding to p32 partly explaining its potency; 4) disrupting p32 interaction with Tat may be explored for therapeutic interventions against HIV.

Abstract

HIV gene expression is modulated by the combinatorial activity of the HIV transcriptional activator, Tat, host transcription factors, and chromatin remodeling complexes. To identify host factors regulating HIV transcription, we used specific single-guide RNAs and endonuclease-deficient Cas9 to perform chromatin affinity purification of the integrated HIV promoter followed by mass spectrometry. The scaffold protein, p32, also called ASF/SF2 splicing factor-associated protein, was identified among the top enriched factors present in actively transcribing HIV promoters but absent in silenced ones. Chromatin immunoprecipitation analysis confirmed the presence of p32 on active HIV promoters and its enhanced recruitment by Tat. HIV uses Tat to efficiently recruit positive transcription elongation factor b (p-TEFb) (CDK9/CCNT1) to TAR, an RNA secondary structure that forms from the first 59 bp of HIV transcripts, to enhance RNAPII transcriptional elongation. The RNA interference of p32 significantly reduced HIV transcription in primary CD4⁺T cells and in HIV chronically infected cells, independently of either HIV splicing or p32 anti-splicing activity. Conversely, overexpression of p32 specifically increased Tat-dependent HIV transcription. p32 was found to directly interact with Tat’s basic domain enhancing Tat stability and half-life. Conversely, p32 associates with Tat via N- and C-terminal domains. Likely due its scaffold properties, p32 also promoted Tat association with TAR, p-TEFb, and RNAPII enhancing Tat-dependent HIV transcription. In sum, we identified p32 as a host factor that interacts with and stabilizes Tat protein, promotes Tat-dependent transcriptional regulation, and may be explored for HIV-targeted transcriptional inhibition.

HIV persists in latently infected CD4⁺T cells in individuals on suppressive antiretroviral therapy (ART) for prolonged periods of time. Stable cell reservoirs harbor proviral DNA, which are thought to be the source of viremia upon ART interruption (1–3). Viral expression and transcriptional reactivation from latently infected cells are not suppressed by current antiretrovirals. Our central premise is that HIV transcriptional inhibitors can be used as latency-promoting agents (LPAs) in block-and-lock functional cure approaches, aimed at reducing residual viremia during ART and limiting viral rebound upon treatment interruption (TI) (4). The nucleosome (nuc) positioning at the latent HIV promoter (long-terminal repeat, LTR) is typical of a promoter poised to reactivate (5). Blocking residual HIV transcription that occurs during suppressive ART may allow accumulation of repressive epigenetic marks, creating an unfavorable chromatin environment for RNA polymerase II (RNAPII) recruitment. A transcriptional inhibitor would not simply replace ART, but rather through epigenetic silencing a drug-free functional cure could ensue. This hypothesis was developed with the discovery and characterization of the Tat inhibitor, didehydro-Cortistatin A (dCA) (6, 7). In in vitro and in vivo models of HIV latency, combining ART with dCA accelerates HIV-1 suppression to below detection limits and inhibits viral rebound upon TI or stimulation with latency-reversing agents (LRAs) (4). Transcriptional repression by dCA resulted in the heterochromatinization and loss of RNAPII at the HIV promoter, characterized by tighter nuc/DNA association, increased deacetylated H3 occupancy at nuc-1, decreased Switch/Sucrose Non-Fermentable (SWI/SNF) chromatin-remodeling complex polybromo-associated BAF complex (PBAF) (necessary for Tat-mediated transactivation), and enhanced recruitment of the repressive BRG1/BRM associated factor (BAF) complex (8). Whether long-term transcriptional inhibition can prompt a permanent block in vivo remains to be determined; but as with ART, multiple LPAs that target different aspects of HIV transcription are likely required to lock the HIV promoter in an effectively irreversible state. Recently we described another LPA, the generic drug Spironolactone (SP), which degrades the host xeroderma pigmentosum type B (XPB) subunit of the general transcription factor TFIIH, blocking HIV transcription and viral reactivation without strongly affecting the host cellular transcriptome (9). This study highlights that a host factor can be targeted to selectively suppress HIV transcription over host cell gene expression. HIV-1 transcriptional inhibitors have immense potential in functional cure approaches; the more knowledge we gain on the interconnectivity between Tat, transcription factor regulatory mechanisms and how these affect nuc positioning, the more we can leverage transcriptional regulators as antiviral targets.

The HIV promoter contains binding sites for multiple transcription factors (TFs) that, upon binding, initiate the formation of the preinitiation complex (PIC) [reviewed in ref. 10]. Tat is produced from rare full-length HIV transcripts, and once it accumulates above a certain threshold promotes exponential HIV RNA production via a positive feedback loop (11). Tat efficiently recruits p-TEFb (composed of CDK9/CCNT1) to the HIV TAR RNA, a dynamic stem-loop-bulge secondary structure that forms from the first 59 bp of HIV transcripts (12). The initial recruitment of RNAPII to the transcriptional start site (TSS), PIC formation, and the first “burst” of initiation that triggers viral mRNA production is of key importance, since these preliminary rounds of transcription initiation allow the production of Tat and establishment of the Tat/TAR feedback loop (13, 14). Tat also promotes transcription in other ways, e.g., by recruiting the superelongation complex (SEC), histone acetylases, and chromatin remodelers that are thought to reposition nuc-1 into a transcriptionally favorable position (15–20). In essence, Tat exponentially increases HIV gene expression, thus therapeutically targeting any process that can hinder Tat activity or production will inevitably inhibit viral production. The reliance of HIV transcription on Tat as well as a range of host factors suggests that an optimal approach for viral silencing may entail targeting several factors simultaneously. There may be unique therapeutic opportunities to explore, but these require the identification of host factors or specific interactions with Tat that can safely be targeted for HIV inhibition, with minimal toxicity.

To expand knowledge on mechanisms regulating HIV transcription, we sought to identify proteins associating with the HIV promoter. For this, we used a modified version of the chromatin affinity purification with mass spectrometry (ChAP–MS) (21), to determine enrichment or loss of proteins in transcriptionally active or silenced HIV promoter loci. ChAP–MS has been successfully used to identify the proteomics of a specific yeast genomic locus (21–24). ChAP–MS was adopted to purify the HIV–LTR DNA from the genome using CRISPR/dCAS9 followed by MS analysis. We identified the protein p32 in actively transcribing HIV promoters and depleted in dCA-treated promoters where HIV was silenced. p32, also known as ASF/SF2 splicing factor-associated protein, p33, gC1q receptor (gC1qR), or hyaluronic acid-binding protein 1 (HABP1), is a multifunctional and multicompartmental protein, which plays important roles in inflammation, infection, and cancer (25). p32 is a ubiquitously expressed conserved cellular protein that is highly charged and highly acidic [pI = 4.15] (26). The cellular p32 is a propolypeptide, and cleavage of the N-terminal 73 amino acids forms the mature polypeptide. The pro-p32 protein is found in the cytosol, and the mature p32 in the cytosol and nucleus (27). p32 interacts with multiple factors and has been proposed as a chaperone protein (28, 29). For instance, it can bind the globular heads of complement component C1q (30), hyaluronic acid (31), and numerous other proteins (32–35), including proteins from a wide range of virus such as EBV, Herpes simplex, Adenovirus, HIV, and Rubella virus [reviewed in ref. 36]. p32 was shown to interact with some HIV-1 proteins (37–39): gp41 binding to p32 renders uninfected CD4⁺T cells susceptible to autologous NK killing (37); p32 interaction with HIV Rev modulates HIV splicing increasing viral production (39–41); and p32 was identified as preferentially binding acetylated Tat peptides in vitro and p32 partially colocalized with Tat in HLM cells (38). However, p32 has never been directly implicated in HIV-1 transcriptional regulation.

Using chromatin immunoprecipitation (ChIP) we confirmed that p32 associates with actively transcribing HIV promoters in chronically infected HeLa and Jurkat T cells, and we demonstrate that p32 recruitment is dependent on Tat. We show the modulatory role of p32 in HIV transcription, i.e., inhibition of p32 by RNA interference (RNAi) significantly reduces HIV transcription in both Jurkat and primary CD4⁺T cells, while p32 overexpression specifically increases Tat-mediated transcription. Run-on experiments confirmed this role of p32 in Tat-mediated transcription. Mechanistic studies revealed that p32 specifically interacts with Tat’s basic domain, significantly stabilizing Tat and increasing its half-life. This interaction between p32 and Tat results in an enhanced Tat association with the HIV-LTR, RNAPII, and CyclinT1/CDK9. In return, p32 suppression reduces the association of Tat with RNAPII. Interestingly, dCA blocks Tat–p32 interaction, supporting p32’s absence in dCA-silenced promoters. In summary, through an unbiased affinity purification approach, we identified p32 as a host factor that modulates HIV transcription by stabilizing and enhancing Tat transcriptional activity, revealing a potential target to explore in HIV-silencing approaches.

Results

ChAP–MS Analysis of Transcriptionally Active and dCA-Silenced HIV-LTR Promoters.

We previously reported that the Tat inhibitor dCA blocks HIV transcrition in chronically infected HeLa–CD4 cells (8). In these cells, long-term dCA treatment reduces p24 production to undetectable levels and interruption of dCA treatment does not result in viral rebound (8). To ensure uniformity in HIV gene expression prior ChAP–MS, we derived a single-cell clone of infected HeLa–CD4 cells by limiting dilution. All clones showed more than one integration, and the M1 clone selected for further studies, included HIV proviral insertions in chromosomes 1 (121227513) and 22 (240020092). As with the parental line, dCA reduced viral production to below the detection limit within 24 d of treatment (Fig. 1A), and precluded viral reactivation (performed on day 44) in response to multiple LRAs such as TNF-α, Trichostatin A (TSA), suberoylanilide hydroxamic acid (SAHA), and PMA (SI Appendix, Fig. S1A). Loss of RNAPII recruitment to the HIV promoter and open reading frame (ORF) was also observed in dCA-treated M1 cells with or without PMA (SI Appendix, Fig. S1B). To perform the ChAP–MS, we transfected M1 cells with a catalytically inactive protein A-tagged version of deficient Cas9 (dCAS9) (SI Appendix, Fig. S1C), combined with gRNAs targeting the HIV promoter (Fig. 1B). Cells expressing dCAS9 and gRNA were crosslinked, and chromatin was sheared and purified using rabbit IgG. Four guide RNAs were tested for their ability to individually enrich HIV–LTR sequences after purification (Fig. 1C). Of note, the dCAS9/gRNA complex association to the promoter did not significantly alter HIV transcription in cells treated or not with dCA (SI Appendix, Fig. S1D). A combination of the 4 gRNAs was chosen to maximally enrich HIV–LTR chromatin pull-down, while gRNA to GFP was used as a negative control. Importantly, similar LTR enrichment was observed by qPCR between ART and ART+dCA-treated cells (Fig. 1D). HIV gag at position 1,959 nt, 1 kb downstream from the LTR, was not detected, confirming selective enrichment of the LTR (Fig. 1D). Similarly, no significant enrichment was detected of most similar host genome sites to the first 12 base-pairs of the 20-bp gRNA target, known to strongly influence gRNA-directed binding specificity (42), and marked as off-target sites (OT) (Fig. 1D). Next, dCAS9 expression was confirmed before and after purification by western blot (WB), using histone H3 as control (SI Appendix, Fig. S1E). Isolated proteins were resolved by SDS-PAGE (SI Appendix, Fig. S1F ), gel excised, in-gel trypsin digested and subjected to mass spectrometry analysis (LC-MS/MS).

Fig. 1.

Identification of HIV promoter (5′LTR)-associated proteins by ChAP–MS. A. HeLa–M1 cells chronically infected with HIV NL4-3 were treated with ART or ART + dCA (100 nM) over time and viral capsid (p24) production in the supernatant monitored by p24 ELISA. Cells were split on average every 4 d. Data are n=3. B. Schematic of the HIV promoter (5′LTR) region 1 to 1,100 nt. and location of the four gRNAs. Nuc locations are represented by gray circles. TSS (green arrow): Transcription start site. C. HIV promoter enrichment after purification with the indicated gRNAs. Cells were transfected with dCAS9 and a single gRNA and collected 48 h later, crosslinked, and subjected to chromatin purification. HIV-specific primers used to quantify HIV promoter enrichment by qPCR. Fold enrichment: % of input DNA of a specific gRNA divided by % of input DNA of gRNA to GFP (negative control). Data represent mean ± STE (n=3). D. HIV promoter DNA enrichment in both ART + DMSO and ART + dCA-treated M1. Cells were transfected with dCAS9 and a mixture of gRNAs B, N, E, and I or control gRNA to GFP. HIV-specific primers used to analyze HIV promoter enrichment by qPCR. Enrichment of similar sites in the host genome to the first 12 base-pairs of the 20-bp gRNA target were measured to determine off-target enrichment. Data represent mean ± STE. E. Proteins bound to the HIV–LTR were resolved by SDS-PAGE, gel excised, in-gel trypsin digested, and analyzed by mass spectrometry analysis (LC-MS/MS). Specific protein enrichment in different promoter conditions was determined first by removing gRNA background noise based on the spectral counts. Normalized abundances of triplicates were averaged, and the Log2 fold changes were calculated for DMSO versus dCA and ranked. Top hits enriched in DMSO (active 5′LTR, blue spots) or dCA (latent 5′LTR, red spots) are highlighted. F. mRNA levels of CCNT1, FUBP3, p32, and PA2G4 in HIV chronically infected HeLa–CD4 M1 treated by gene-specific shRNAs or control shCD8. G. HIV gag mRNA level in cells from panel F. F and G. Data represent mean ± STE, independent Student’s t test, *P<0.05, **P<0.01.

A total of 161 proteins were identified and ranked based on enrichment from ChAP immunoprecipitates of cells treated with DMSO versus dCA (Fig. 1E). Among the top 30 candidates enriched in DMSO-treated cells (active LTR), we identified several known Tat-interacting proteins such as AIMP1 (43), PABPC1 (44), and XRCC6 (45). The top 30 candidates identified in dCA-treated cells (latent LTR) included several histone proteins (H1, H2, H4), an expected outcome since dCA promotes increased histone/nuc occupancy on the HIV promoter (8). The E3 SUMO–protein ligase TRIM28 (also named KAP1), which contributes to HIV-1 latency by manipulating suppressive epigenetic modifications (46), was also enriched in dCA promoters (Fig. 1E), suggesting ChAP–MS sucessfully discriminates known regulators of active and inactive HIV proviruses. We pursued a subset of candidate proteins reproducibily detected in three independent experiments and for which a priori knowledge of their biology was available, while discarding proteins commonly classified as contaminants in affinity enrichments (21, 22, 24). As such, FUBP3, p32, and PAG2G4 were selected for shRNA interference studies alongside CCNT1 as positive control to determine their impact on HIV transcription. Retroviral vectors (pMKO-puro) expressing short hairpin RNAs (shRNAs) against p32 (shp32), FUBP3 (shFUBP3), PA2G4 (shPA2G4), CCNT1 (shCCNT1), and CD8 (shCD8, as the negative control) were built, packaged into virus like particles (VLPs) and transduced into HeLa-M1 cells. Upon puromycin selection all gene-specific shRNAs successfully reduced the mRNA levels of target genes (Fig. 1F), and their depletion reduced HIV mRNA expression (Fig. 1G), suggesting a role for these factors in HIV transcriptional regulation. Taken together, we used ChAP–MS to analyze the proteomic profile of transcriptionally active and dCA-induced silent HIV–LTR promoter and identified several potential host factors regulating HIV mRNA production.

p32 Supports Efficient HIV Replication.

Here, we characterize the activity of the p32 protein on HIV transcription, while FUBP3 and PA2G4 will be reported elsewhere. First, we suppressed p32 protein expression by transducing primary CD4⁺T cells from 4 healthy donors by two means, with retroviral vectors pMKO-puro or pLMPd-Ametrine that express shRNAs/shRNAmirs specific for p32 (shp32) or CD8 (shCD8, as negative control). Activated primary CD4⁺T cells from donors 1 and 2 were infected with NL4-3 and transduced 8 h later with pMKO-puro vector expressing shCD8 or shp32. Transduced cells were selected with puromycin (1.5 µg/mL) and cultured for 9 d. Cells of donors 3 and 4 were transduced with pLMPd-Ametrine vector expressiing shCD8 or shp32 and transduced cells isolated by fluorescence-activated cell sorting (FACS) based on ametrine expression. Sorted cells were then infected with HIV NL4-3 and cultured for 9 d. Cell-associated RNA was isolated from all donors at 9 dpi. Using these two knockdown methods, p32 was efficiently depleted in the CD4⁺T cells from all 4 donors, ~80% in donors 1 and 2 and ~90% in donors 3 and 4 (Fig. 2 A and B). The HIV mRNA production from p32-depleted cells was efficiently decreased by 40 to 50% in donors 1 and 2 (Fig. 2C) and 70 to 90% in donors 3 and 4, suggesting p32 is required for HIV infection in primary CD4⁺T cells (Fig. 2D). Cytopathic effects caused by HIV replication are typically observed several days post infection. Cell viability was monitored by trypan blue staining at 9 dpi and was either unaffected by p32 KD (in donors 1 and 2) as compared with control, or improved (in donors 3 and 4), likely due to HIV suppression (SI Appendix, Fig. S2A). Next, we examined the effect of depleting p32 as compared with other known positive regulators of HIV replication such as Cyclin T1 (CCNT1) or cyclin dependent kinase 9 (CDK9) in Jurkat CD4⁺T cells using pLMPd-Ametrine vectors and isolated transduced cells by FACS. The mRNA levels of p32, CCNT1 and CDK9 were quantified by RT-qPCR (Fig. 2E), and protein depletion confirmed by WB (SI Appendix, Fig. S2B). Next, cells were infected with HIV NL4-3, and HIV mRNA expression (Fig. 2F), and cell viability (SI Appendix, Fig. S2C) assessed on day 12 post infection, as well as p24 production in a time course (Fig. 2G). p32 RNAi strongly impaired both HIV mRNA and p24 production to levels just above suppression of either CCNT1 or CDK9 (Fig. 2 F and G), without impacting cell viability (SI Appendix, Fig. S2C). Together, these results suggest p32 is an important factor required for efficient HIV transcription and replication both in Jurkat and primary CD4⁺T cells.

Fig. 2.

p32 is required for HIV replication. A–D. p32 RNAi and HIV infection in primary CD4⁺T cells of four independent donors. Purified CD4⁺T cells were activated by anti-CD3/CD28 antibodies for 2 d followed by transduction with two independent RNAi vectors to knockdown p32 or control CD8B. Donors 1 and 2 were infected with NL4-3, transduced 8 h later with shRNAs in the pMKO-puro vector and selected with puromycin (1.5 µg/mL). Donors 3 and 4 were transduced with shRNAs embedded in a microRNA backbone in the pLMPd-Ametrine vector. Transduced (Ametrine⁺) cells were isolated by FACS and infected with NL4-3. The p32/HIV mRNA expression was analyzed by RT-qPCR 9 d post-infection. E. Relative mRNA levels of CCNT1, CDK9 and p32 in stable Jurkat CD4⁺T cells transduced with gene-specific shRNA or control shCD8. F. Cells from E were infected with HIV NL4-3 (10 ng p24 per million cells) and passaged for 20 d. HIV mRNA production determined at 12 dpi. G. HIV p24 production in the supernatant of cells from E analyzed by p24 ELISA from 4 to 20 dpi. Data represent mean ± STE, independent Student’s t test, *P < 0.05, **P  < 0.01, ***P < 0.001.

p32 Associates with Transcriptionally Active HIV-LTR Promoters.

To investigate p32 association with the HIV promoter, we performed ChIP in M1 cells. Consistent with the ChAP–MS results, HIV inhibition by dCA abrogated p32 recruitment to the HIV promoter, with or without PMA activation (Fig. 3A). Neither dCA nor PMA treatments changed p32 enrichment at the promoter of the housekeeping gene GAPDH, suggesting p32 is preferentially recruited upon stimulation to the transcriptionally active HIV promoter (Fig. 3A). To exclude cell type-specific effects, we also performed ChIP in a clonal HIV latently infected Jurkat CD4⁺T cell line named Jurkat D6. D6 cells contain a single integrated copy of NL4-3 at chromosome 15 (72355395). As with M1 cells, dCA reduces viral p24 production in D6 cells (SI Appendix, Fig. S3A), and inhibits viral reactivation by PMA (SI Appendix, Fig. S3B). ChIP to RNAPII and p32 revealed that viral reactivation by PMA readily increases RNAPII recruitment to the HIV DNA (Fig. 3B), and concomintantly p32 (Fig. 3C). PMA treatments did not change RNAPII or p32 enrichment to GAPDH, again suggesting p32 is preferentially recruited upon stimulation to the transcriptionally active HIV promoter (SI Appendix, Fig. S3 C and D). Next, since p32 was reported as binding Tat (38), we investigated whether the enrichment of p32 is mediated by Tat. We assessed p32 recruitment to the HIV-5’LTR in the presence or absence of transfected Flag-tagged Tat (Flag-Tat) in HeLa cells stably expressing the HIV promoter driving Luciferase (HeLa–LTR–Luciferase). ChIP revealed that overexpression (OE) of p32 alone was not sufficient to enrich p32 at the LTR, however Tat expression increased p32 recruitment by approximately 20-fold (Fig. 3D). In agreement with this result, Tat expression also increased endogenous p32 recruitment to the HIV–LTR (~5-fold), alongside RNAPII (~10-fold) (Fig. 3E). Meanwhile, treatment with the HDAC inhibitor SAHA enriched RNAPII at the LTR by ~12-fold, but barely enriched p32 at the LTR (non-statistically significant 2-fold change) (Fig. 3E). p32 recruitment to the Luciferase (Luc) ORF (~1 kb downstream the 5′LTR) was unchanged, even when RNAPII recruitment to Luc was increased by about 10 to 15-fold with Tat transactivation or SAHA treatment (Fig. 3E). Alongside data from Fig. 3D, these results suggest that p32 robustly associates with the transcription complex at the promoter upon Tat transactivation.

Fig. 3.

p32 is enriched on transcriptionally active HIV 5′LTR promoters. A. p32 ChIP in HeLa–M1-treated with ART or ART+dCA with/without reactivation with PMA. HIV or GAPDH promoters’ specific primers used to quantify HIV-5′LTR enrichment by qPCR. B and C. ChIP to RNAPII (B) and p32 (C) performed with latently infected Jurkat D6 cells with/without reactivation with PMA. HIV genome-specific primers used to quantify p32 and RNAPII enrichment at the HIV genome by qPCR. Results represent percent immunoprecipitated DNA over input after IgG control background subtraction. D. p32 recruitment to the HIV LTR (analyzed by primer 550 in panel A) with/without Tat. HeLa-LTR-Luc reporter cells were transfected with p32-GFP with or without Flag-Tat. ChIP to p32 performed with anti-GFP antibody and bound DNA quantified by qPCR with HIV–LTR-specific primers. p32 and Tat expression detected by WB. E. Endogenous p32 recruitment to the HIV–LTR and Luciferase ORF in HeLa–M1 cells. Top: Diagram of the LTR–Luciferase reporter in stable HeLa cells (HeLa–LTR–Luc) and primers positions for qPCR. Bottom: Cells were stimulated with SAHA or transfected with Tat to activate transcription from the LTR, followed by ChIP to endogenous p32 and RNAPII. F. p32–RNAPII interaction in presence/absence of Tat. The lysate from HEK293T cells expressing Tat-Flag (or not) were subjected to p32 IP. Mouse IgG used as the negative control. Data are n=3. G. Densitometry of samples from F. RNAPII band normalized to p32. Data represent mean ± STE, independent Student’s t test, *P<0.05, **P<0.01.

We next investigated p32 association with RNAPII by co-immunoprecipitation (co-IP) in the presence or absence of Tat in HEK293T cells (Fig. 3F). In the absence of Tat, we detected an interaction of p32 with RNAPII (Fig. 3F, Lane 5), which may be direct or indirect. Importantly, in the presence of Tat, the amount of RNAPII associating with p32 increases by 50% (Fig. 3F, Lane 6 and Fig. 3G), again consistent with increased p32 recruitment to the HIV promoter by Tat (Fig. 3E). These results suggest that p32 may minimally associate with the transcription complex but is significantly recruited by Tat onto the HIV promoter to strengthen the transcription complex.

p32 Regulates HIV Transcription.

Next, we investigated the role of p32 in HIV transcription in M1 cells containing full-length HIV. The shRNA depletion of p32 (Fig. 4A) resulted in approximately 50% inhibition of HIV total mRNA production and concomitantly of unspliced and multiple spliced HIV transcripts (Fig. 4B). Depletion of p32 also resulted in an increase in the ratio of multiple spliced mRNAs over unspliced mRNAs (Fig. 4C), an expected result given previous reports that human p32 inhibits excessive HIV transcript splicing to support HIV production (40). In addition, ChIP revealed a 50% loss of RNAPII recruitment to the HIV promoter and ORF (Fig. 4D), without affecting RNAPII recruitment to the housekeeping gene GAPDH promoter and ORF (Fig. 4E). These results suggest that p32 is necessary for RNAPII occupancy on the HIV genome and may positively regulate HIV transcription.

Fig. 4.

p32 depletion by RNAi inhibits HIV transcription. A. Protein expression of p32 upon shRNA knockdown in HIV-infected HeLa–CD4 M1 by WB. B. Effects of p32 KD on all HIV RNA species on cells from A. C. Ratio of multiply spliced mRNA to un-spliced HIV mRNA from panel B. D and E. RNAPII recruitment onto the HIV genome (D) or GAPDH (E) in HIV-infected HeLa-CD4 M1 as determined by ChIP. Results presented as percent immunoprecipitated DNA over input, after IgG control background subtraction. Data represent mean ± STE, independent Student’s t test, *P<0.05, **P<0.01.

p32 was reported to inhibit mRNA production from the adenovirus major late transcription unit by disrupting the balance of Ser-5-phosporylation (Ser5p) and Ser-2-phosporylation (Ser2p) of the RNAPII carboxy-terminal domain (CTD) (47). As such, we investigated p32’s role in RNAPII phosphorylation in M1 cells. Depletion of p32 reduced overall amounts of both Ser5p and Ser2p RNAPII CTD across the HIV genome, consistent with the general loss of RNAPII observed (SI Appendix, Fig. S4 A, C, and D); however, it did not alter the normal pattern of RNAPII-Ser5 enrichment near the TSS, nor the RNAPII-Ser2 enrichment near the 3′end the HIV genome (SI Appendix, Fig. S4 B and D) (10). These results suggest that in the context of HIV, p32 does not seem to disrupt RNAPII phosphorylation levels.

p32 Interacts with Tat’s Basic Domain.

To define the p32/Tat interaction, we cotransfected p32 and full-length Tat into HEK293T cells. Consistent with other studies (38, 48), endogenous p32 and overexpressed p32-GFP readily co-IPed with Tat (Fig. 5A and SI Appendix, Fig. S5A), confirming their interaction (Fig. 3 F lane 6). The p32/Tat interaction does not seem to be bridged by RNA, such as TAR, since RNase treatment did not affect the co-IP between Tat and transfected or endogenous p32 (Fig. 5A lanes 3–4 and SI Appendix, Fig. S5A lanes 3–4). Tat has been reported to interact with p32 via Tat’s core domain (32 to 48 a.a.) (49), but a subsequent study, using biotin labeled peptides, identified region 36–53 a.a. as necessary for p32 binding (38). To refine our understanding of this interaction, we built a series of Tat deletions (Fig. 5B) and performed co-IP studies with either overexpressed and endogenous p32 (Fig. 5C and SI Appendix, Fig. S5B). Lower protein expression was observed for Tat deletions 22–37 a.a. and 48–57 a.a., possibly due to stability issues (Fig. 5C lanes 3 and 4). However, the interaction of p32 with Tat was clearly lost when Tat’s basic domain was deleted (∆48–57, Fig. 5C lane 4), supporting Tat’s basic region as the interacting domain. Additional point mutations in Tat basic domain (Fig. 5D), revealed that arginine residues R55, R56, and R57 (Tat MUT) are important for the interaction with p32, since their mutation to alanine drastically limited Tat/p32 interaction (Fig. 5D lane 3 and SI Appendix, Fig. S5C lane 3). Furthermore, when all five arginine residues in the basic region were altered [R52A, R53A, R55A, R56A, and R57A (Tat BRM)] the interaction with p32 was completely abolished (Fig. 5D lane 2 and SI Appendix, Fig. S5C lane 1). As expected, Tat MUT and Tat BRM lost their transactivation capabilities in TZM-bl cells, supporting these arginine residues in the basic domain as also necessary for Tat’s transactivation (SI Appendix, Fig. S5D).

Fig. 5.

p32 interacts with Tat. A. Tat–p32 interaction in the presence or absence of RNase. Cell lysates from HEK293T cells expressing Flag-Tat were pretreated or not with RNase prior to IP. Tat was immunoprecipitated with Flag antibody, and associated p32 detected by WB. B. Schematic representation of Flag-Tat deletion variants. C. Tat–p32 co-IP with Tat deletion variants. Cell lysate from HEK293T cells expressing Flag-Tat deletion variants were subjected to Flag IP and associated p32 detected by WB. D. Tat-p32 IP with Tat WT, MUT and BRM. Cell lysate from HEK293T cells expressing Flag-Tat and were subjected to Flag IP and p32 detected by WB. E. Schematic representation of GST-p32 truncations expressed in E. coli BL21 (“+” represents Tat binding and “−” no binding). F. In vitro binding assay of GST-p32 and His-Tat WT or BRM protein. One µg purified His-Tat protein were mixed with 10-µg GST–p32 proteins followed by IP with 1-µg anti-His antibody. G. In vitro binding assay of His–Tat and GST–p32 truncations in presence of 0 to 200 nM dCA. Densitometry shown in red below the blot (p32 intensity relative to Tat). H. Cells transiently expressing p32-GFP and Tat-Flag were treated with 200 nM dCA and collected for Flag IP. Right: Densitometric quantification of p32 relative to Tat. Data are n=3 represent mean ± STE, independent Student’s t test, *P < 0.05)

To investigate whether the p32–Tat interaction is direct or mediated by a bridging protein complex, we produced recombinant His-tagged Tat protein (Tat WT and Tat BRM) and GST-tagged p32 (GST-p32) (Fig. S5E lane 2–4). Of note, p32 is a propolypeptide whose cleavage of the first 73 N-terminal amino acids (a.a.) forms the mature protein (50). As such, p32 74–282 was used as the full-length protein (SI Appendix, Fig. S5E lane 4). Next, were performed in vitro binding experiments in which purified Tat–His and GST–p32 were incubated together, followed by His IP. p32 was readily Co-IPed with Tat–WT but not Tat–BRM (Fig. 5F lane 1 and 3), confirming direct interaction between Tat and p32 and Tat’s basic domain mediating this interaction. To further define this interaction, a series of GST–p32 truncations were built (Fig. S5E lane 5-13) and their interaction with Tat investigated. The WT p32 (74-282), truncations 97–282 and 74–263 copurified with Tat (Fig. 5F lane 7, 8 and 13). Very little interaction was observed between Tat and p32 truncations 105–282 and 180–282 (Fig. 5F lane 9–10), while none of the other truncations copurified with Tat (Fig. 5F lanes 11, 12, 14–16), suggesting p32 regions 98–104 a.a. and 247–262 a.a. are necessary for the interaction with Tat.

Since both dCA and p32 bind Tat’s basic domain, we analyzed the p32–Tat interaction in the presence of dCA. In vitro binding experiments showed the p32–Tat interaction was inhibited by 70 to 90% in the presence of 0.2 to 200 nM dCA (Fig. 5G). Co-IP studies also showed that 200 nM dCA inhibits Tat–p32 interaction by 50% in cells transiently expressing Tat and p32 (Fig. 5H). Taken together, these results revealed that Tat interacts with p32 via Tat’s basic domain, which overlaps with the dCA-binding domain, possibly also contributing to the loss of p32 in dCA-treated promoters (Fig. 3A). Furthermore, the interaction of p32 with Tat hinges on two regions, a small 6 a.a. domain between 98–104 a.a., and a 15 a.a. region between 247 and 262 a.a. of p32.

p32 Promotes HIV Tat-Mediated Transcription.

p32 associates with RNAPII and its recruitment to the HIV promoter is further enhanced by Tat (Fig. 3 F and G), likely stabilizing Tat’s interaction with RNAPII. As such, we investigated the ability of p32 to promote transcription by either RNAi depletion (Fig. 6A) or overexpression (OE) in HeLa–LTR-Luc cells (Fig. 6B). In this system, p32 depletion resulted in a 50% reduction in Luc activity at basal level, upon SAHA stimulation, as well as upon Tat transactivation (Tat–Flag, 1–86 a.a.) (Fig. 6A). These results may stem from either a basal role for p32 associating with RNAPII at the TSS-promoting HIV transcription or else from an indirect effect on other cellular genes through p32 depletion. However, the OE of p32 did not alter the HIV promoter’s basal activity nor SAHA-mediated reactivation (Fig. 6B), while it promoted a significant 2-fold increase in transactivation by Tat (Fig. 6B). These results suggest that OE of p32 alone is not sufficient to promote LTR transcription and p32 activation of HIV transcription is Tat specific. To study the activity of p32 on Tat transactivation, a series of GFP-tagged p32 (p32-GFP) truncations were constructed (Fig. 6C), and their expression confirmed by WB (Fig. 6D). These p32 truncations were transfected alongside full-length Tat onto HeLa–LTR–Luc cells (Fig. 6E). Tat transactivation was enhanced to the same degree by the mature p32 74–282 and the WT p32, while p32 deletions of the region 97–104 a.a. and 247–262 a.a. attenuated p32’s Tat transactivation, suggesting both regions are necessary for p32’s activity on Tat transactivation. Concordantly, p32 deletions of these 2 regions abolish p32 binding to Tat (Fig. 6C), these results clearly support the notion that p32 interacts with Tat protein to promote Tat transactivation.

Fig. 6.

p32 promotes HIV Tat-mediated transcription. A. p32 depletion by RNAi inhibits LTR–Luc activity in stable HeLa–LTR–Luc reporter cells. Cells transduced with shRNAs to CD8 or p32 and selected with puromycin (4 µg/mL) for >4 d. Selected cells were either treated with SAHA overnight (16 h) or transfected with Tat (24 h), and LTR activation assessed by quantifying luciferase activity. B. p32 overexpression increases Tat-mediated LTR activation in HeLa-LTR-Luc cells. Cells were transfected with HA–p32 or vector control and treated with SAHA 24 h later for another 16 h before luciferase quantification. C. Schematic representation of GFP–p32 truncations peptides and indication with “+” or “−” of their interaction with Tat or Tat activation ability. “ND” indicates not tested. D. The expression of each p32-truncation in panel C determined by WB with GFP antibody. E. Tat transactivation assay in HeLa–LTR–Luc cells with p32-truncation. Cells were transfected with p32-GFP truncation and Tat or vector control, and luciferase activity measured 48 h later. Results are relative to control (vector), attributed a value of 1. F. p32 increases HIV mRNA production. p32-GFP or vector control were transfected with HIV pNL4-3 in HEK293T cells, 2 d later, HIV mRNA was quantified by RT-qPCR with indicated primers (normalized to GAPDH mRNA). G. Nuclei from cells in panel F were isolated and analyzed by nuclear run-on transcription. Results are relative to vector control attributed a value of 1. Data represent mean ± STE, independent Student’s t test, *P < 0.05.

p32 was originally isolated as a protein tightly associated with the essential cellular splicing factor ASF/SF2 (51). To investigate whether p32 regulates HIV transcription via SF2, we attempted to deplete SF2 by RNAi; however, this reduced cell viability, given SF2’s critical role in mRNA splicing (52). We thus investigated the p32–SF2 interaction and noted that p32 74–246 interacts with SF2 (SI Appendix, Fig. S6A), even though this truncation does not interact with Tat or promote Tat transactivation (Fig. 6C). Further truncations of p32 212–246 a.a. region reduced SF2-binding ability (SI Appendix, Fig. S6A). These results suggest that p32 interacts with Tat and SF2 via different domains. To continue investigating whether p32 regulation of the HIV LTR hinges on its splicing activity, we studied the ability of p32 mutant G108D [p32 (74–282 a.a.) with G35D mutation], unable to inhibit mRNA splicing (40), to promote Tat transactivation. p32–G108D enhanced Tat transactivation to the same degree as p32 WT (SI Appendix, Fig. S6B), suggesting p32 transcriptional regulation is independent of its SF2-binding activity. We also investigated p32–G108D in the context of fully replicative HIV. Either p32, p32–G108D or p32 74–246 (no Tat interaction) OE constructs were cotransfected with HIV pNL4-3 DNA, and total cell mRNA were collected 2 d later (Fig. 6F). Unspliced and spliced HIV mRNAs were analyzed by RT-qPCR, and while WT p32 OE increased HIV mRNA production by two fold, p32 G108D and p32 74–246 had no effect (Fig. 6F). We suspect that p32 G108D did not increase total HIV mRNA levels (Fig. 6F), as it promotes higher multiple splicing events after transcription (40). Furthermore, in nuclear run-on experiments (Fig. 6G), p32 G108D significantly increased HIV nascent transcript synthesis, though to a lesser degree than p32 WT (Fig. 6G), while p32 74–246 (interacts with SF2 but not with Tat), did not significantly increase HIV nascent transcripts (Fig. 6G). Taken together, these results suggest that p32 positively regulates HIV transcription in partnership with Tat, and this activity is independent of p32 antisplicing SF2-regulating functions.

p32 Stablizes Tat Protein.

Tat is well known for having a flexible secondary structure and is cleared from infected cells via multiple degradation pathways (53–56). p32 is a chaperone protein (28, 29) and was shown to stabilize monocyte chemoattractant protein 1 (MCP-1) (57). We thus investigated whether p32 could increase the half-life of Tat to enhance transcriptional activation. Either p32–GFP or vector control (GFP-N1) were transfected alongside Tat into HeLa–CD4 cells, and 24 h after transfection, cells were treated with Cycloheximide (CHX) to inhibit protein synthesis. Tat protein degradation was followed for the next 4 h (Fig. 7A). An obvious stabilization of the Tat protein half-life was observed when p32 was present (Fig. 7 A and C), while no changes in Tat mRNA levels were observed upon p32 or GFP transfection (Fig. 7B). The decay of control Histone H3 was unaffected by p32 (Fig. 7 A and D), suggesting p32 specifically binds and stablizes Tat. Since Tat BRM does not interact with p32, its half-life in the presence of p32 was also investigated (Fig. 7 E and F, Fig. 5F lane 3). p32 did not alter Tat BRM’s half-life, further confirming the specificity of p32-mediated Tat stablization via Tat’s basic domain. Taken together, these results suggest that p32 binds to the basic domain of Tat, stabilizing Tat, and promoting Tat transcriptional transactivation capacity.

Fig. 7.

p32 stabilizes Tat protein. A. WT Tat degradation with/without p32. Hela–CD4 cells were transfected with GFP-p32 and Flag-Tat and 24 h later treated with cycloheximide (CHX) to block protein synthesis. Tat protein degradation was followed for the next 4 h. WB detected Tat with Flag antibody and p32 with GFP antibody. B. Tat mRNA level at 0 h (from panel A) measured by RT-qPCR. C and D. Tat and Histone H3 levels quantified over time by WB (from panel A) and normalized to GAPDH. Results are relative to control (GFP-N1 vector) at 0 h, attributed a value of 1. E and F. Tat–BRM mutant degradation with/without p32, as in A. Data represent mean ± STE, independent Student’s t test, *P < 0.05, **P < 0.01)

p32 Promotes the Association of Tat with Its HIV Transcriptional Enhancing Partners.

p32 interacts with Tat’s basic domain, the domain also responsible for TAR–RNA interaction. To assess how p32 modulates Tat/TAR interaction, we analyzed Tat recruitment to the HIV-5′LTR in the presence or absence of p32 by performing ChIP against Tat in HeLa–TR–Luc cells. We observed a trend towards more Tat occupancy at the HIV–5′LTR when p32 was overexpressed, whereas no changes observed at GAPDH promoter (Fig. 8 A and B). This result suggests that p32 facilitates Tat recruitment to the 5′LTR, which may also be a consequence of increased Tat steady–state levels in the presence of p32. To further understand the upregulation of HIV transcription by p32, the interaction of Tat with CCNT1 and RNAPII in the presence and absence of p32 was investigated. Tat-Flag was transfected onto HEK293T cells stably expressing shRNAs against p32 or negative control CD8, Tat was immunoprecipitated and co-IP of CCNT1 and RNAPII assessed by WB (Fig. 8C). Tat associated RNAPII and CCNT1 was reduced by approximately 50% and 40% respectively after p32 depletion (Fig. 8 C and D). Tat interacts with RNAPII CTD via two binding sites located within the 24–72 a.a. region (58), p32 likely acts as a chaperone stabilizing Tat structure and promoting the Tat–RNAPII interaction. The loss of overall Tat protein levels in p32-depleted cells (clearly observed in the input samples), again highlights Tat’s stability in the presence of p32. Results thus suggest that p32 promotes HIV transcription by enhancing Tat recruitment to the 5′LTR, which may result from an overall increase in Tat protein steady-state levels and the stabilization of protein–protein interactions with required transactivation factors such as CCNT1 and RNAPII.

Fig. 8.

p32 promotes Tat association with its HIV transcriptional enhancing partners. A. HeLa–LTR–Luc reporter cells were transfected with Flag-tagged Tat, p32–GFP or vector control. Tat and p32 expression detected by WB. B. Tat enrichment at the HIV–LTR by ChIP with Flag antibody, using cells from A. Tat recruitment to the GAPDH promoter used as negative control. Results represent percent immunoprecipitated DNA over input after IgG control background subtraction. C. Co-IP of Tat with CCNT1 or RNAPII in the presence/absence of p32. Tat expressed in HEK293T cells with shRNAs to CD8 or p32. Cell lysates used to Tat IP with Flag antibody. Mouse IgG used as negative control. D. Tat-associated CCNT1 and RNAPII levels from panel C normalized to immunoprecipitated Tat. Data represent mean ± STE, independent Student’s t test, *P < 0.05.

Discussion

We used an unbiased approach to identify proteins associating with either actively transcribing HIV promoters, or silenced promoters with the Tat inhibitor dCA. We optimized a ChAPS–MS, previously used in yeast (21) specifically for the HIV promoter loci. ChIP presents some drawbacks for discovery purposes, such as prerequired knowledge of the molecular target, need for specific antibodies to each protein target or post translation modification (PTM) of interest, and availability and avidity of these antibodies. ChAP–MS has the advantage of unbiased detection of multiple and possibly yet unknown proteins associated with a particular DNA region using an affinity purification followed by MS analysis. We developed a catalytically inactive Cas9/gRNA system to enrich purification of the HIV promoter region 123 to 1,097 nt, which encompasses the TSS, Nuc 0, 1, and 2, and DNase hypersensitive regions (DHS) 1 and 2 (Fig. 1B). The best possible signal to noise dynamic range is critical for any locus-specific chromatin purification strategy. The Cas9/gRNA approach outperforms other strategies such as TAL-based and LexA-based ChAP-targeted chromatin enrichment methodologies (21, 24, 59). We tested more than 20 gRNAs spanning the whole HIV–5′LTR and identified 4 gRNAs that provided a maximal 30-fold enrichment over background at DHS-1 (primers 326 and 403), and ~10-fold enrichment at Nuc-1 (primer 550) (Fig. 1C). Interestingly, 3 of the gRNAs providing most efficient enrichment (gRNA B, E, and I) recognized the nuc-free region of the 5’LTR (Fig. 1 B and C), while gRNA-N in Nuc-1 presented the lowest enrichment, suggesting gRNA-binding efficiency is affected by nuc occlusion and DNA accessibility at the HIV-5′LTR. We previously reported that dCA promotes tighter nuc/DNA binding at the HIV–1–5′LTR through maintenance of deacetylated histones (60), which in turn may affect gRNA/dCas9 binding. Nevertheless, we observed the same enrichment level in both DMSO- and dCA-treated 5′LTRs (Fig. 1D), suggesting that the efficiency of dCAS9/gRNA was unaffected by dCA-induced HIV promoter heterochromatinization. Another consideration is the number of integrated HIV copies per cell, which can affect the efficiency of chromatin enrichment. We previously observed higher integration events in long-term dCA-treated NL4-3 chronically infected HeLa cell population, likely due to the survival advantage of HIV transcriptional silenced cells with multiple integrated proviruses, as opposed to ART only treated cells, which may be counter selected due to cytotoxic viral protein expression (8). To ensure similar HIV integration events and uniformity in HIV gene expression per cell, we created a cell clone named M1, from the population of HIV-infected HeLa-CD4 cells, as well as a clone named D6, from a population of HIV-infected Jurkat CD4⁺T cells. Cells were treated with ART to maintain HIV integration levels, which also guarantees the same enrichment efficiency during chromatin purification (Fig. 1D).

This optimized ChAP–MS protocol resulted in the identification of 161 proteins enriched in immunoprecipitates from cells treated with DMSO- versus dCA-treated HIV-5′LTR promoters. Proteins known to associate with actively transcribing HIV promoters were identified, such as Tat-interacting protein AIMP1 (43), PABPC1 (44), and XRCC6 (45), while proteins known to associate with latent promoters, such as histones and TRIM28 were enriched in dCA-treated 5′LTRs, validating our ChAP–MS optimized approach. However, we did not identify well-known HIV–5′LTR-associated proteins, such as transcription factors NF-κB and Sp1, potentially reflecting some ChAP–MS sensitivity limits. As such, protein identification may be biased toward the most abundant or larger proteins with more peptides likely to be detected by MS. We identified p32 (ASF/SF2 splicing factor-associated protein), FUBP3 and PA2G4 in actively transcribing HIV promoters but not in dCA transcriptionally silenced HIV promoters. Knockdown of these factors in chronically infected cells confirmed their role in HIV transcription (Fig. 1 F and G). Here we focused on p32, while FUBP3 and PA2G4 will be reported elsewhere.

p32 is considered an inhibitor of mRNA splicing by associating with splicing factor ASF/SF2 to inhibit ASF/SF2 binding to RNA (61, 62). In the context of HIV, p32 suppresses excessive HIV splicing to promote accumulation of viral genomic transcripts that can be packaged (40). However, p32 has never been directly implicated in HIV-1 transcriptional regulation. The importance of p32 activity in HIV replication was highlighted during infection of primary CD4⁺T cells and Jurkat CD4⁺T cell lines (Fig. 2), with p32 depletion limiting HIV replication to a similar degree as depletion of key host transcriptional players, CyclinT1 and CDK9 (Fig. 2G). Importantly, p32 RNAi reduced total HIV mRNA production, including both unspliced and multiple spliced mRNA, indicating that p32 also regulates HIV transcription, upstream of splicing events. The cooccupancy of p32 and RNAPII on PMA activated HIV promoters in both HeLa-M1 (Fig. 3A and SI Appendix, Fig. S1B) and Jurkat D6 (Fig. 3 B and C) cells suggests that p32 acts as a component of the transcription complex during transcriptional activation. This was demonstrated by the enhanced copurification of RNAPII and p32 in Co-IP experiments in the presence of Tat (Fig. 3F). In the absence of Tat, the presence of p32 at the HIV promoter presence is however very modest (Fig. 3 D and E), for instance a 12-fold increase in RNAPII enrichment on HIV-LTR induced by SAHA results in a two fold nonsignificant enrichment of p32 (Fig. 3E). We speculate that the majority of p32 enrichment on the HIV–LTR is mediated by Tat. Besides, p32 does not seem to be recruited to Tat-independent genes, even when RNAPII is highly abundant, such as GAPDH promoter and ORF (SI Appendix, Fig. S3 C and D).

Tat region 32–48 a.a. has been reported to interact with p32 (49). A later study reported that p32 binds Tat peptide 36–53 a.a., and acetylation of this peptide at lysines K50 and K51 further strengthened the interaction (38). While additional studies may be needed to fully define Tat–p32 interaction, our mutational analysis showed the five arginine residues in Tat’s basic domain, the region responsible for Tat/TAR interaction, are required for Tat–p32 interaction (Fig. 5D). The differences between studies may result from secondary structure differences of peptide fragments versus full-length proteins, as used in our study. Tat’s basic domain also mediates the nuclear localization of Tat, we have reported that Tat WT distributes in the nucleolus while Tat BRM was completely excluded from the nucleolus (6). However, recombinant-purified Tat BRM did not interact with p32 (Fig. 5F, lane 3), suggesting reduced Co-IP of p32 and Tat BRM is not caused by altered cellular distribution (Fig. 5D and SI Appendix, Fig. S5C). Consistent with our results, p32 was reported to interact with the N terminus of the cap protein of Porcine Circovirus via a similar arginine rich motif (24RRR26) (63), and with the arginine rich domain of HIV Rev protein (41). Given that p32 is highly acidic and highly negatively charged (64), it may have high affinity for positively charged motifs. Furthermore, in vitro binding studies using purified p32 truncations showed that p32 regions 97–105 a.a. and 247–263 a.a. are necessary for Tat interaction. This result is somewhat supported by a previous report showing p32 244–255 a.a. required for Tat binding (48). However, this same study showed that p32 208–279 a.a. is the minimal Tat-binding region in the presence of high salt (500 mM NaCl), whereas p32 244–279 a.a. is the minimal region in low salt (150 mM NaCl) in GST pull-down assays (48). This discrepancy might result from differences in methods. We found that purified Tat protein nonspecifically binds GST protein in pull-down assays with Glutathione agarose beads, as such we switched to use anti-His antibody to pulldown Tat. We found that very limited amounts of p32 truncations 105 to 282 and 180 to 282 copurified with Tat, indicating these may bind Tat, however at very low affinity (Fig. 5F). Such a weak interaction does not, however; seem sufficient to promote Tat transactivation (Fig. 6E). The p32 truncations that did not copurified with Tat in in vitro binding assay, also did not promote Tat transactivation, further supporting the specificity of in vitro p32–Tat interactions (Fig. 6C).

We previously reported that the Tat inhibitor dCA blocks Tat-dependent HIV transcriptional activity by binding to Tat’s basic domain (6). Here, we found that dCA inhibits Tat/p32 interaction in a dose-dependent manner (Fig. 5G), further confirming Tat’s basic domain involvement in both dCA and p32 binding, also likely explaining why p32 was not found associating with dCA-treated promoters (Fig. 1E). dCA’s disruption of Tat binding to p32, besides Tat/TAR, is also likely a contributing factor to its incredible potency. Both p32 and dCA bind Tat via its arginine rich basic domain and Tat BRM (R52A, R53A, R55A, R56A, R57A) loses the ability to bind both p32 and dCA (6). Furthermore, both interactions result in Tat stabilization. We previously reported that dCA locks Tat in a stable conformer, with key basic domain residues forming a horseshoe-like patch that accommodates dCA and promotes burying of Trp11 in the optimal hydrophobic environment of Tat’s core (65). Neither p32 OE alone (Fig. 6B) nor dCA treatment affects basal LTR transcriptional activity (6), highlighting p32 activity mainly dependent on Tat. While p32 binds and stabilizes Tat (Fig. 7), the biological outcome of dCA and p32 interaction with Tat is contrasting. dCA blocks Tat/TAR interaction and Tat-dependent HIV transcription, while p32 promotes a twofold increase in the interaction of Tat/TAR (Fig. 8 A and B) and stabilizes Tat’s interaction with RNAPII (Fig. 8 C and D). These results suggest the interaction of p32/Tat and dCA/Tat is somewhat distinct and the contributing roles of involved arginine residues may be different. For instance, Tat acetylation (Ac) at K50 results in Tat liberation from TAR RNA and p-TEFb (66), and Tat Ac-K50 and Ac-K51 promote p32 binding (38).

Although our study does not formally rule out p32 activity on splicing in viral infection, since p32 suppression resulted in an increase in the ratio of multiply spliced mRNA over unspliced mRNA (Fig. 4C), and p32 mutant devoid of SF2-binding ability also did not promote Tat transactivation (p32 truncation 74-211 in SI Appendix, Fig. S6 A and B). We demonstrated that p32 plays a major role as a positive regulator of HIV transcription. First, p32 depletion by RNAi decreases HIV mRNA production including total, unspliced and multiple spliced mRNAs, supported by reduced RNAPII recruitment to the HIV genome (Fig. 4 B and D). Second, p32 RNAi in HeLa–LTR–Luc cells, a system without splicing, resulted in overall decreased LTR activity (Fig. 6A). Furthermore, p32–G108D OE, a variant unable to modulate HIV RNA splicing (40), increases HIV nascent transcript synthesis similarly to p32 WT (Fig. 6G).

We propose that p32 regulates HIV transcription by binding Tat to promote Tat stability and facilitating/stabilizing its association with the transcription machinery at the HIV-LTR. This notion is supported by the following results: 1) p32 is enriched on transcriptionally active HIV genomes (Fig. 3 A and C); 2) p32 associates with RNAPII and this association increases in the presence of Tat (Fig. 3F); and 3) p32 depletion in both HIV chronically infected HeLa M1 cells (Fig. 4B) and stable HeLa–LTR–Luc cells transfected with Tat (Fig. 6A) resulted in reduced LTR transcription. p32 OE alone in stable HeLa–LTR–Luc does not significantly increase the LTR activity (Fig. 6 B and E), suggesting that p32’s Tat-independent transcriptional activity is marginal compared with that in the presence of Tat. The reduction in transcriptional activity observed with p32 depletion in Fig. 6A, in basal or SAHA activated LTR-Luciferase activity, may stem from a low-level presence of p32 associating with RNAPII even in the absence of Tat, as observed in Fig. 3F. As such, p32 is involved in Tat-dependent transcription in two ways: 1) p32 binds and stabilizes Tat increasing Tat’s steady state levels, thus affecting HIV transcription in a positive feedback manner; as Tat fluctuations drive phenotypic bifurcation (67); and 2) p32 recruitment with Tat to the transcription complex (including RNAPII, p-TEFb) at the HIV-LTR, reinforces and stabilizes Tat/TAR and Tat/RNAPII association, enhancing HIV transcription (Fig. 9).

Fig. 9.

Model of p32’s role in HIV transcriptional regulation. A. Summary of Tat and p32 regions necessary for their interaction. B. The chaperone protein p32 may residually engage with RNAPII during basal transcription; however, p32 major effects are observed in the presence of Tat. p32 enhances Tat protein half-life and stabilizes Tat interaction with RNAPII, p-TEFb, TAR, promoting RNAPII transcriptional elongation.

In summary, our HIV-5′–LTR-specific ChAP–MS approach successfully identified host factors, namely p32, directly regulating HIV-1 transcription. Our results suggest that p32 promotes transcription mainly by binding and stabilizing Tat. p32 interacts with Tat via basic domain arginine residues, also involved in TAR interaction. However, p32 binding to Tat does not seem to compete with Tat/TAR binding, instead it promotes HIV transactivation, likely by stabilizing the Tat/TAR and Tat/RNAPII interaction. This study not only furthered our understanding of the mechanisms regulating dCA-mediated latency, but importantly revealed host factors regulating HIV transcription and the p32/Tat interaction as a potential target for block-and-lock approaches for an HIV functional cure.

Materials and Methods

ChAP–MS.

ChAP–MS was performed using NL4-3 chronically infected HeLa–M1 cells treated or not with dCA (100 nM) (SI Appendix, Fig. S1). Cells (3 × 10⁶) were seeded into 10-cm dishes and transfected with 5 µg dCAS9 and gRNA expression plasmids. Cells were collected 48 h later, crosslinked with 1% formaldehyde and chromatin sheared by sonication to approximately 1-kb fragments in buffer (16 mM Tris-HCl, pH 8.1, 1.2 mM EDTA, 167 mM NaCl, 1.1% Triton X-100, 0.01% SDS). The chromatin was cleaned by centrifugation and supernatant (1 mg) precleared with 10 µL Dynabeads™ Protein G (Invitrogen #01009761). Rabbit IgG (8 µg) were conjungated with 30 µL Dynabeads™ Protein G at room temperature (RT) for 15 min followed by 2× washes with PBST (PBS + 0.02% Tween-20). Rabbit IgG-conjugated beads were then incubated with precleared lysate on a rotator overnight at 4 °C, followed by multiple washes (details in SI Appendix, Methods). Mass spectrometry analysis was performed by the Scripps Florida proteomics core facility (additional details in SI Appendix, Methods).

Primary Cell Culture and Transduction.

CD4⁺T cells were isolated by negative selection (EasySep™ Human CD4⁺T Cell Enrichment Kit, Stemcell Technologies, #19052) from four healthy donor PBMCs (OneBlood Florida) and expanded in plates coated with 10 µg/mL anti-IgG (BioLegend, #NC0195353) and 1 µg/mL anti-CD3 OKT (BioLegend, #NC0070918) with soluble anti-CD28.2 (BioLegend, #50-165-915). RPMI 1640 medium was supplemented with 8% human serum (SeraCare, #1830-0005), 1% PSG, 50 ng/µL recominant human IL-2 IS (Milteyni, #130-097-748), and 5 BRMP U/mL Natural Human Interleukin-2 (IL-2)/T Cell Growth Factor (TCGF) (ZeptoMatrix, #0801017). Transduction details in SI Appendix, Methods).

p24 ELISA.

Quantification of HIV p24 capsid production was performed using the antigen capture assay kit from Advanced BioScience Laboratories, Inc. (catalog #5447), according to manufacturer’s protocol.

Nuclear Run-On Analysis.

Run-on transcription was performed as described in https://dx.doi.org/10.17504/protocols.io.khxct7n.

ChIP Assay.

ChIP was performed as previously described with some modifications (4, 6, 7) (SI Appendix, Methods).

Additional Methods.

Detailed protocol provided in SI Appendix.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.