Upon HIV-1 infection, a reservoir of latently infected resting T cells prevents the eradication of the virus from patients. To achieve complete depletion, the existing virus-suppressing antiretroviral therapy must be combined with drugs that reactivate the dormant viruses. We previously described a novel chemical scaffold compound, MMQO (8-methoxy-6-methylquinolin-4-ol), that is able to reactivate viral transcription in several models of HIV latency, including J-Lat cells, through an unknown mechanism. MMQO potentiates the activity of known latency-reversing agents (LRAs) or “shock” drugs, such as protein kinase C (PKC) agonists or histone deacetylase (HDAC) inhibitors. Here, we demonstrate that MMQO activates HIV-1 independently of the Tat transactivator. Gene expression microarrays in Jurkat cells indicated that MMQO treatment results in robust immunosuppression, diminishes expression of c-Myc, and causes the dysregulation of acetylation-sensitive genes. These hallmarks indicated that MMQO mimics acetylated lysines of core histones and might function as a bromodomain and extraterminal domain protein family inhibitor (BETi). MMQO functionally mimics the effects of JQ1, a well-known BETi. We confirmed that MMQO interacts with the BET family protein BRD4. Utilizing MMQO and JQ1, we demonstrate how the inhibition of BRD4 targets a subset of latently integrated barcoded proviruses distinct from those targeted by HDAC inhibitors or PKC pathway agonists. Thus, the quinoline-based compound MMQO represents a new class of BET bromodomain inhibitors that, due to its minimalistic structure, holds promise for further optimization for increased affinity and specificity for distinct bromodomain family members and could potentially be of use against a variety of diseases, including HIV infection.

IMPORTANCE The suggested “shock and kill” therapy aims to eradicate the latent functional proportion of HIV-1 proviruses in a patient. However, to this day, clinical studies investigating the “shocking” element of this strategy have proven it to be considerably more difficult than anticipated. While the proportion of intracellular viral RNA production and general plasma viral load have been shown to increase upon a shock regimen, the global viral reservoir remains unaffected, highlighting both the inefficiency of the treatments used and the gap in our understanding of viral reactivation in vivo. Utilizing a new BRD4 inhibitor and barcoded HIV-1 minigenomes, we demonstrate that PKC pathway activators and HDAC and bromodomain inhibitors all target different subsets of proviral integration. Considering the fundamental differences of these compounds and the synergies displayed between them, we propose that the field should concentrate on investigating the development of combinatory shock cocktail therapies for improved reservoir reactivation.

Upon initial HIV-1 contagion, most of the afflicted CD4+ T lymphocytes die rapidly in response to viral infection. A major breakthrough in controlling HIV came in 1996 with the introduction of highly active antiretroviral therapy (ART), which is able to sustain the population of T lymphocytes in patients. However, a small but significant number of latent infected cells survive the antiviral therapy, and upon cessation of treatment, the latent proviruses become capable of expression without obstruction, and the viremia reemerges.

Active CD4+ T lymphocytes possess the capacity to revert to a quiescent state with minimal HIV-1 gene expression following the proviral integration and persist as long-lived central or transitional memory T lymphocytes sheltering latent HIV-1 genomes. These T lymphocyte populations harboring the latent reservoirs cannot be detected by the immune surveillance, since viral antigens are not presented to immune effector cells, and ART remains ineffective against an already integrated provirus. Unlike active T lymphocytes, which have a short half-life, in a dormant state the memory T lymphocytes possess an estimated half-life of approximately 44 months and are thus considered to be the primary reason why the disease remains a chronic affliction (reviewed in reference 1). Mathematical models predict that the eradication of a reservoir consisting of 106 cells would take 73 years in vivo (2). Extensive efforts have been carried out within the last 25 years to characterize these cells and to understand how HIV-1 is regulated after integration and why it can remain transcriptionally latent.

In order to cure a patient, the viral reservoir must be either completely eradicated or at least depleted to a level at which viral rebound is deemed unlikely (3). To achieve HIV eradication from infected patients, it has been suggested that ART be combined with drugs that “shock” the proviral transcription into activity and flush out the dormant viruses (4). Following the reactivation of latent proviruses, the immune system and cytopathogenicity are responsible for killing the infected cells, while the continuous ART guarantees protection against further infection.

Small-molecule inhibitors are commonly regarded as the preferred method in forcing molecular regulation. Due to technical reasons, like membrane penetration, mechanical simplicity, rapid function, cost-effectiveness, and in vivo stability, the “shock and kill” field is currently engaged in the identification and development of small-molecule latency-reversing agents (LRAs). It has been proposed that HIV gene expression reactivators can be grouped into two categories: direct activators and noise enhancers (5). The reasoning for this type of categorization is that the two groups of drugs possess conceptually contrasting mechanisms on the latent viral promoter, allowing them to synergize when combined (6).

Direct activators, such as protein kinase C (PKC) agonists, tumor necrosis factor alpha (TNF-α), and T cell receptor agonists, are responsible for introducing stimulatory transcription factors to the promoter (such as NF-κB and nuclear factor of activated T cells [NFAT]) and stimulate the transcription process. Although these agents present highly efficient rates of reactivation of proviral transcription, the downside of the modulators is their aggressiveness. The highly potent compounds are incapable of discriminating between infected and uninfected cells, leading to massive T lymphocyte activation, a decrease in the patient's immunological memory, and oftentimes a cytokine storm. On the other hand, noise enhancers are responsible for modulating the chromatin state, easing the access of transcription factors to the viral promoter, and ultimately assisting the elongation process. This class of agents includes histone deacetylase (HDAC), methyltransferase, and bromodomain inhibitors. HDAC inhibitors (HDACi) have already been approved for clinical use against T cell lymphomas; thus, due to patient safety reasons, these drugs are considered primary candidates in terms of viral reactivation. Though the reported pilot studies utilizing HDAC inhibitors so far have proven them to be less efficient than expected, there still is potential—most of the completed clinical trials have shown an increase in intracellular viral transcription and occasionally also a higher viral load, but none of the trials have yet reported a decrease of viral reservoir size (reviewed in reference 7).

In the last 4 years, numerous studies have substantiated the notion that BET bromodomain inhibitors (BETi) can trigger HIV transcription in latently infected cells, thus activating viral replication (8–10). JQ1 was described as the first of its class as a small-molecule inhibitor of bromodomain-containing protein 4 (BRD4), displaying the highest affinity for the first bromodomain (BD1) of BRD4, and it has received much attention for its therapeutic potential against multiple myeloma and other cancer types related to the c-Myc oncogene (11–13). The effect of JQ1 in viral reactivation can be explained by tilting the competition between BRD4 and Tat for association with the P-TEFb kinase complex in Tat's favor (14). A larger pool of P-TEFb becomes vacant to associate with Tat and activate transcription elongation of the HIV genome. In line with the activator-enhancer drug type classification, bromodomain inhibition has been shown to confirm the premise of chromatin-loosening drugs synergizing with direct NF-κB activators to reactivate viral transcription, both in laboratory Jurkat models and under patient-derived ex vivo conditions (15, 16).

An excellent approach for discovering new antiretrovirus-activating small molecules is using cell lines that harbor a minimal provirus genome. An example of a viral latency model is the J-Lat A2 clone, described previously (17). A 4-kbp-long sequence consisting of various elements from the full-length HIV-1 genome and a green fluorescent protein (GFP) sequence (5′ long terminal repeat [LTR]-Tat-internal ribosome entry site [IRES]-GFP-3′ LTR) was integrated into an intron of the UTX gene of the Jurkat T cell line. In search of new LRAs, our group previously screened a library of 6,000 small molecules to identify basic agents capable of reactivating the latent HIV-1 minigenome from A2 cells (18). Virtual screening for further similar chemicals and additional substitutions of the functional groups in these compounds led to the identification of 8-methoxy-6-methylquinolin-4-ol (MMQO). While the molecular mechanism of MMQO remained elusive, it was shown that the molecule alone does not induce the transcriptional activity of minimal promoters containing binding sites for the typical HIV-1-activating transcription factors NF-κB, NFAT, AP-1, and Sp1. Furthermore, we observed that, in addition to inducing HIV LTR transactivation, MMQO is also able to display immunosuppressive activity by repressing CD3-induced interleukin 2 (IL-2) and TNF-α promoter activation (18).

Based on the previously published results with anti-CD3 (α-CD3) antibody treatments, we hypothesized that MMQO functions through a pathway that exhibits immunosuppressive properties. In addition, taking into consideration that MMQO affects proviral transcription, a process that is highly dependent on cellular host factors, we assumed it to be plausible that the drug would also affect the cell transcriptome on a global scale. Therefore, we utilized gene expression microarrays to characterize and identify the cellular pathways affected by MMQO. We reconfirmed the immunosuppressive behavior of the reagent and showed it to dysregulate genes known to be sensitive to chromatin acetylation. By computational and biochemical means, we determined that MMQO reverts HIV-1 latency by inhibiting the bromodomain protein BRD4. Furthermore, utilizing the barcoded HIV ensembles (B-HIVE) methodology (19), we were able to demonstrate how the inhibition of BRD4 influences a different subset of latently integrated proviruses than does the inhibition of HDACs or the activation of PKC-dependent pathways. The present study documents a method to identify the molecular mechanisms of small molecules with unknown targets and demonstrates that a scaffold bromodomain-inhibiting compound functions on a population of HIV-1 integration sites different from those of other, more commonly used LRAs.

MMQO exerts its effect on HIV minigenome expression independently of viral Tat protein.In order to determine the mechanism of action of MMQO, we sought to determine whether MMQO could exert its effect independently of the viral Tat protein. The discovery of MMQO was spearheaded by utilizing the latent HIV-1 J-Lat A2 clone, which contains a minimal sequence of HIV-1, encoding only GFP and the viral Tat protein (18). Although the Tat protein is considered a crucial component for efficient HIV transcription, models containing latent minigenomes lacking Tat are known to respond to external signals, albeit at lower efficiencies. Utilizing the GFP-expressing J-Lat model E89 (20), which contains a latent Tat-negative construct, we found that MMQO functioned similarly to our previous observations with Tat-expressing constructs (Fig. 1A). As a signal of viral expression, we monitored how cells expressed GFP after 24-hour treatments with four different HIV-reactivating compounds alone or in combination with MMQO via flow cytometry: the enhancer drugs hexamethylene bisacetamide (HMBA) and the HDAC inhibitor SAHA (also known as vorinostat) and the PKC pathway agonists phorbol 12-myristate 13-acetate (PMA) and prostratin. While MMQO alone exhibited a moderate response, it nevertheless strongly potentiated the reactivating ability of the other drugs, as previously reported with Tat-expressing viruses (18). As expected, PMA and prostratin were able to induce viral reactivation at notably higher rates than the other three compounds, due to their roles as highly potent NF-κB-stimulatory agents and because the latent cells were originally selected as PMA responsive (20).

MMQO reactivates latent HIV independently of Tat and potentiates other LRAs. (A) MMQO exerts its effect on the minigenome independently of viral Tat protein. Shown is flow cytometry analysis of a previously established latent Jurkat E89 clone, which was infected with a Tat-negative GFP-expressing minigenome as described previously (20). The cells were treated for 24 h with HMBA (0.5, 2.5, or 5 mM), SAHA (0.1, 0.5, or 2.5 μM), PMA (1 to 4 or 10 nM), or prostratin (Prost.; 0.25, 0.5, or 1 μM) in combination with MMQO (160 μM) or an equivalent volume of vehicle (DMSO). GFP expression intensity was measured by FACS and expressed as a percentage of GFP-positive cells. Means and standard deviations (SD) from an experiment performed in triplicate are shown. (B) MMQO potentiates LRAs rapidly. RT-qPCR results show rapid activation of latent HIV by a PKC agonist and MMQO. E89 and A2 Jurkat clones were treated for 1 h with MMQO (160 μM), PMA (10 nM), a combination of the two, or an equivalent volume of DMSO as a vehicle. GAPDH (glyceraldehyde-3-phosphate dehydrogenase) was measured for normalization, and the results are represented relative to results from the DMSO-treated cells. Primers to detect the 5′ LTR were used. Means and SD from 4 (E89) and 2 (A2) independent experiments measured in duplicate are shown.

The effect of MMQO by itself or potentiating the effect of PKC pathway activators on HIV reactivation was also observed upon 1-h drug treatments in both Tat-negative (E89) and Tat-expressing (A2) J-Lat cells, as tested by reverse transcription-quantitative PCR (RT-qPCR) (Fig. 1B). The rapid response to these drugs suggests that they function directly on the HIV minigenome, as opposed to through a secondary wave of transcriptional products. Similar to HDAC inhibitors, MMQO may function as a noise enhancer type of drug. As expected, cells with the ability to express Tat upon initial HIV promoter activation (A2 clone) were reactivated at considerably higher rates.

MMQO potentiates latency-reversing agents in ex vivo latently infected primary CD4 T cells.We also tested the activity of MMQO in an ex vivo infection primary model of latency, which relies on direct spin infection of primary CD4+ T cells with full-length replication-incompetent HIV driving the luciferase reporter gene. We observed a dose-dependent increase in luciferase activity after 24 h of stimulation with MMQO (Fig. 2A), confirming the data obtained in T cell line latency models. We then examined whether MMQO is capable of enhancing HIV latency reversal when used in combination with other LRA class molecules, proposed to also function via distinct mechanisms in primary latently infected cells. We costimulated latently infected primary CD4+ T cells with MMQO alone or together with SAHA or prostratin, as indicated (Fig. 2B). In accordance with our results from latent T cell lines, the activities of both SAHA and prostratin were enhanced when the cells were cotreated with MMQO, with MMQO-prostratin cotreatment reaching latency reversal over 40% of that obtained with our positive-control PMA-ionomycin. Furthermore, the concentrations of MMQO used alone or in combination treatment did not result in a significant decrease in viability of primary CD4+ T cells (Fig. 2C). Neither was there a significant increase in apoptosis after 24 h or 72 h of stimulation with increasing concentrations of MMQO, as determined by annexin V staining (Fig. 2D).

MMQO potentiates LRAs in ex vivo latently infected primary CD4 T cells. (A and B) Dot plots showing latency reversal in ex vivo-infected primary CD4+ T cells. The data are presented as the fold increase in luciferase activity after 24-h treatment with an MMQO concentration gradient (A) and MMQO (80 μM) cotreatments with SAHA (350 nM) or prostratin (400 nM) (B); each point represents a single measurement. The experiments were performed in duplicate using cells isolated from 6 (A) or 4 (B) healthy blood donors. Student's t test was used to confirm significant differences between treatments. *, P < 0.05. (C and D) MMQO does not reduce viability and does not induce apoptosis in primary CD4+ T cells. (C) Latently infected primary cells were left untreated (Untr) or treated as indicated for 24 h, and viability was assessed by flow cytometry. (D) Uninfected primary CD4+ T cells were left untreated or treated as indicated for 24 h or 72 h, followed by annexin V staining and flow cytometry analysis. The data presented are the means of at least 3 independent healthy blood donors ± standard deviations. Gliotoxin (GTX) was used as a positive control for annexin V staining experiments.

Global expression profile of MMQO treatment.Recognizing that the final output of a drug's functionality can be efficiently measured in mRNA production, we decided to follow up the characterization of MMQO by utilizing a genome-wide Agilent mRNA expression microarray platform with transcripts extracted from Jurkat cells treated or not with MMQO. We opted to use native Jurkat cells for the microarray experiment in order to identify MMQO's mechanism of action without the possibility of interference from the HIV-1 minigenome, specifically by the Tat protein, which has been reported to reprogram the cellular epigenetic landscape (21). Considering how MMQO induces minigenome expression to about an 8- to 10-fold increase after 8 h of treatment at 50% effective concentration (EC50) (18), we concluded that 8 h was an appropriate treatment time for the microarray experiments, since the specific target genes of MMQO might respond similarly to HIV-1 promoter stimulation and be differentially regulated. In addition, this relatively short treatment time should minimize excessive changes on the protein level, which could otherwise cause unwanted secondary responses in the transcriptome. The change of all transcripts between the untreated control groups and the MMQO-treated groups was considered to show differential expression if the genes presented a fold change of at least 1.5 with a P value lower than 0.05 after adjustment for multiple tests (22). We determined that, in total, MMQO regulated 2,193 transcripts at a fold change cutoff of 1.5 and 549 transcripts with a fold change of 2.0. As shown in the volcano plot in Fig. 3A, it can be concluded that MMQO causes substantially more potent downregulation as opposed to upregulation of genes. This trait is better illustrated in the bar graph in Fig. 3B, where the percentages of differentially expressed transcripts are sorted by fold change cutoffs. For a small collection of genes (n = 9), we performed an independent validation by qPCR to confirm the reproducibility of the microarray (Fig. 3C).

Global expression profile of MMQO treatment. Expression microarray (Agilent) analysis of Jurkat cells treated or not with MMQO for 8 h was performed in duplicate. (A) Volcano plot of gene expression differences between treated and nontreated samples. The blue and red dots highlight all the statistically significant downregulated/upregulated transcripts (fold change [FC] ≥ 1.5; P < 0,05). The Myc gene and other genes that are regulated by acetylation-dependent networks are highlighted. (B) The total number of protein-coding genes significantly up- or downregulated by 8 h MMQO were categorized into four groups based on their mean fold changes compared to the untreated samples. The number of upregulated genes was divided by the number of downregulated genes in each expression group, which is displayed as a percentage. (C) For a subset of genes, we performed an independent RT-qPCR validation of microarray results from Jurkat cells treated with MMQO (80 μM) or DMSO for 8 h or left untreated. GAPDH was measured for normalization, and the results are presented relative to untreated cells. The values obtained from the 8-h MMQO treatment microarray are displayed as gray bars. (D and E) Downregulation of c-Myc by MMQO in Jurkat cells. (D) RT-qPCR results depicting MYC downregulation in Jurkat cells treated with MMQO (160 μM) at different time points (0 to 8 h). (E) RT-qPCR results showing MYC downregulation in native Jurkat cells treated with various doses of MMQO for 1 h. GAPDH was measured for normalization, and the results are presented relative to untreated cells. The means and SD from a representative experiment measured in duplicate are shown. (F) Western blot analysis of c-Myc protein expression. Jurkat cells were incubated for 12 h with MMQO (160 μM) or SAHA (5 μM) or left untreated. Total protein was extracted with RIPA buffer and analyzed by immunoblotting against c-Myc and α-tubulin as a loading control. The arrowhead indicates the specific c-Myc band.

The global nonspecific transcriptional response was unexpected, since we initially anticipated identifying precise pathways with a low number of specifically regulated genes, allowing us to predict the factors involved. Nevertheless, the strong downregulation pattern of MMQO could partially be traced back to its immunosuppressive nature—a substantial fraction of genes from this data set are known to be direct target genes of NF-κB (see Table S1 in the supplemental material), while pathway analysis suggested that at least nine immunogenic pathways are downregulated (see Table S2 in the supplemental material).

Utilizing the Gene Set Enrichment Analysis (GSEA) toolkit, we observed the most significant MMQO-downregulated genes to be enriched in the transcription factor c-Myc-related gene sets; of the top 10 gene sets that correlated most closely with the genes downregulated by MMQO, 6 were canonically c-Myc dependent (see Table S3 in the supplemental material). In marked contrast, MMQO treatment did not significantly induce any gene set for transcription factors related to HIV-1 reactivation (e.g., NF-κB, STATs, IRFs, or AP-1).

MMQO targets acetylation-sensitive genes.Other than the dysregulation of immunosuppressive and c-Myc-dependent pathways among the GSEA results, HDAC-related gene sets were largely enriched, with nine gene sets correlating positively and three gene sets correlating negatively with MMQO treatment (see Table S4 in the supplemental material). It should be noted that these HDAC data sets contain nearly 1,000 transcripts, coinciding with the broad response produced by MMQO and suggesting the plausibility of MMQO being an HDAC inhibitor. A large proportion of the genes encompassed in the HDAC data sets are additionally related to the immune system, thus providing a possible explanation for the broad immunosuppression that has been witnessed with MMQO (18). This observation is also in agreement with previously published data showing that HDAC inhibitors have an immunosuppressive impact on both Jurkat cells and regulatory immune cells in general (23). A significant hallmark of HDAC inhibition in T cell leukemia cell lines like Jurkat is also the dysregulation of c-Myc. The downregulation of MYC mRNA by MMQO was confirmed to be dose dependent, extremely rapid, and persistent (Fig. 3D and E). The drastic decline of MYC gene expression due to MMQO further translates into the protein level (Fig. 3F), similar to the previously reported decrease caused by histone deacetylase inhibitors in Jurkat cells (24).

We performed another microarray experiment to compare the change in the gene expression profile between MMQO and the pan-HDAC inhibitor trichostatin A (TSA). While the overlap of the two compounds is significant (Fig. 4A), noteworthy divergences could also be highlighted. In total, 892 genes were differentially regulated by MMQO, while TSA treatment affected 1,594 genes (fold change cutoff, ±1.5); the overlap was 619 genes. Even though most of the MMQO-regulated genes are also dysregulated by TSA, only a minority of TSA-regulated genes are dependent on MMQO. A few of the prominent overlapping genes are highlighted in the volcano plot in Fig. 3A. The overlap between the two data sets is depicted in Venn diagrams, with up- and downregulated genes depicted separately (Fig. 4B).

MMQO targets acetylation-sensitive genes. (A and B) Microarray analysis comparing 3-h MMQO (160 μM) and TSA (200 nM) treatments in Jurkat cells. Expression data were obtained by hybridization with an Agilent Human microarray platform. (A) Scatterplot depicting the fold changes for the 3,376 significant transcripts (q < 0.05) from the TSA and MMQO 3-h data sets. The Pearson correlation coefficient and the number of genes in each quadrant are shown. (B) Venn diagrams of genes mutually upregulated or downregulated by MMQO or TSA (FC ≥ 1.5; q < 0.05). The sizes of the circles are proportional to the gene numbers. (C) MMQO does not cause global hyperacetylation. Jurkat cells were incubated for 24 or 48 h with MMQO (160 μM), SAHA (5 μM), or DMSO or left untreated. Total protein was extracted with RIPA buffer and analyzed by immunoblotting against acetylated histone H4 (H4-Ac), H3, or H4K12 or total H3 as a loading control.

Other than the differing effects of the chemicals on the transcriptome, we did not observe any global hyperacetylation of the chromatin in response to MMQO treatment (Fig. 4C). However, it is noteworthy that both MMQO and HDAC inhibitors presented potent immunosuppressive traits, hinting at coinciding mechanistic functions.

MMQO functions as a bromodomain inhibitor.A thorough review of the existing literature based on our experimental data led us to hypothesize that MMQO might function as a bromodomain-inhibiting protein, specifically by directly targeting the BET protein family member BRD4. The BET family proteins are responsible for recognizing acetylated chromatin via their bromodomains and for mediating transcriptional processes globally. Besides reactivating HIV transcription both Tat dependently and independently, downregulating c-Myc, and playing roles in proliferation and apoptosis, bromodomain inhibitors like JQ1 have also been reported to mimic HDAC inhibitors in their functions on the transcriptomes of cells from lymphoma lineages (25). The similarities between HDAC and bromodomain inhibition stem from their mutual protein targets. While HDAC inhibitors promote global chromatin hyperacetylation, inhibition of BET family proteins releases these proteins from the core histones, thus exposing the already acetylated lysine tails of the nucleosomes.

In addition to targeting a similar set of genes, BET family inhibitors have been shown to cause severe immunosuppression by disabling NF-κB's ability to mobilize to inflammatory superenhancer regions by disabling the direct interaction between BRD4 and the RelA subunit (26–29). We therefore performed another analysis comparing our microarray data with previously published data sets from microarrays performed with JQ1-treated J-Lat 10.6 HIV latency model cells (8). Even though there were crucial differences between the setups of our MMQO experiment and the published JQ1 experiments (24 h for JQ1 versus 8 h for MMQO, native Jurkat versus Tat-expressing J-Lat cells, and Agilent versus Affymetrix platforms), a significant overlap could be observed between the transcriptomes of the two drugs (Fig. 5A and B). Of note, the Pearson correlation coefficient is also higher than that for the previous comparison with TSA and MMQO (0.841 versus 0.695) (Fig. 4A).

MMQO functions as a bromodomain inhibitor. (A) Correlation between the transcriptome responses to MMQO and JQ1 treatments with RNA expression microarrays. Shown is a scatterplot of the fold changes for the 1,773 significant genes (q < 0,05) from the JQ1 24-h (8) and MMQO 8-h (this work) data sets. The Pearson correlation coefficient and the number of genes in each quadrant are shown. (B) Venn diagrams of genes mutually upregulated or downregulated by MMQO or JQ1 (FC ≥ 1.5; q < 0.05). The sizes of the circles are proportional to the gene numbers. (C) MMQO and JQ1 affect expression of the latent HIV-1 minigenome similarly. The indicated latently infected Jurkat clones were treated with MMQO (160 μM) or JQ1 (1 μM) for 24 h, and HIV expression was analyzed by FACS and expressed as a percentage of GFP-positive cells. The experiment was performed in triplicate. (D) MMQO and JQ1 do not synergize on HIV reactivation. J-Lat A2 latently infected cells were treated with various doses of MMQO (40 to 160 μM) or JQ1 (0.1 to 2 μM) for 24 h, and HIV expression was analyzed by FACS and expressed as a percentage of GFP-positive cells. Calculation of synergy for the different combinations was carried out according to the Bliss independence model (16) and is represented as a heat map on the right. The experiment was performed in triplicate. (E) Ex vivo-infected primary CD4+ T cells were left untreated or treated with 80 μM MMQO alone or in combination with 1 μM OTX-015 or 150 nM JQ1 for 24 h, followed by luciferase assay; each point represents a single measurement. Experiments were performed in duplicate using cells isolated from at least 3 healthy blood donors. Student's t test was used to confirm significant differences between treatments. *, P < 0.05. A P value of 0.08 is also shown. n.s., nonsignificant. (F) RT-qPCR analysis of the effects of BETi and HDACi on genes selected for their opposite expression in the 3-h MMQO/TSA arrays. Jurkat cells were treated for 3 h with MMQO (160 μM), RVX-208 (80 μM), JQ1 (1 μM), SAHA (5 μM), or TSA (200 nM). GAPDH was measured for normalization, and the results are represented relative to untreated cells. The heat map color coding represents the fold change to untreated cells. The experiment was performed in duplicate, and only genes that had reliable SD values are depicted.

To validate the microarray results, we compared the behaviors of MMQO and the gold standard BRD4 inhibitor, JQ1, alone or in combination, on HIV reactivation. The effects of JQ1 (1 μM) and MMQO (160 μM) on the latent HIV minigenome were similar in both Tat-positive (A2) and Tat-deficient (E89) minigenome-containing cells (Fig. 5C). To quantitate the interaction of JQ1 and MMQO, we treated A2 J-Lat cells with nontoxic increasing concentrations of both drugs (Fig. 5D). The ability of one drug to stimulate HIV reactivation was consistently decreased in the presence of increasing concentrations of the other drug, suggesting that the two drugs were targeting the same repressive mechanism. At high concentrations, both drugs behaved with neither a synergistic nor an additive pattern. This was further addressed using the Bliss independence model for combined drug effects. This model assumes that the stimulants act through separate mechanisms so that their effects multiply when administered in combination. A drug combination with an effect significantly exceeding the value predicted by the Bliss model can be said to exhibit synergy. We confirmed that JQ1 and MMQO cotreatments did not exhibit synergy at medium and high doses, but rather conformed to the predictions of the Bliss independence model (Fig. 5D, right) (>0, synergy; <0, antagonism). However, synergy was apparent at the lowest doses, leaving the Bliss testing partly inconclusive (see Discussion below).

Furthermore, we also examined the effect of MMQO cotreatment with JQ1 and another BET inhibitor compound, OTX-015, on ex vivo-infected primary CD4+ T cells. Similar to what was observed in our T cell line models of latency, neither OTX-015 nor JQ1 was able to elicit synergistic latency reversal when used in combination with MMQO. Cotreatment resulted in a moderate additive-effect increase in luciferase activity, supporting the notion that MMQO targets the same repressive mechanism as OTX-015 and JQ1 (Fig. 5E).

To fully confirm that MMQO preferably displays BET-inhibitory behavior, we compared the effects of two known BRD4 inhibitors (JQ1 and RVX-208) and two HDAC family inhibitors (SAHA and TSA) on the MMQO-specific genes identified in the preceding microarrays. For this assay, we specifically concentrated on genes that showed opposite regulation by TSA in the 3-hour microarray (Fig. 4A, top left and bottom right quadrants). As a control for the proper functioning of the compounds, we additionally included genes universally dysregulated by both classes of drugs, like IRF7, MYC, RAG1, and CXCR7. Indeed, following a 3-h treatment with the five reagents, we witnessed MMQO displaying BET inhibitor-like behavior (Fig. 5F). As anticipated, we observed that HDAC inhibitors showed expression patterns similar to those of BET bromodomain inhibitors on the universal target genes. IRF7 was upregulated by both classes of drugs, while MYC, CXCR7, and RAG1 showed strong downregulation by all the drugs, further confirming the similarities between the two classes of compounds. However, MMQO-specific genes, like ADM, IFIT1, CCR7, ICOS, and LRIG1, were downregulated by bromodomain inhibitors but were upregulated by HDAC inhibitors. ZBTB1 and MEPCE displayed potent downregulation by TSA and SAHA while being upregulated and unregulated by bromodomain inhibitors, respectively. According to this assay, all the chosen MMQO target genes displayed the expected differential regulation by the corresponding classes of drugs.

Furthermore, similar conclusions could be drawn by looking at the microarray data. Of the 295 genes upregulated with MMQO and downregulated with TSA in the scatterplot in Fig. 4A (3 h; q value < 0.05), 55 genes were upregulated with JQ1 and none was downregulated, according to the JQ1 microarray data shown in Fig. 5A (24 h; q < 0.05). Of the 309 genes downregulated with MMQO and upregulated with TSA, 32 genes were downregulated with JQ1 and only 3 were upregulated (see Fig. S1 in the supplemental material).

In conclusion, consistent with the hypothesis that MMQO is a BRD4 inhibitor, we observed that genes that were differentially regulated by MMQO were also specifically dysregulated by the bromodomain inhibitors JQ1 and RVX-208. Furthermore, MMQO and bromodomain inhibitors acted nonsynergistically on the reactivation of latent HIV.

Structural basis of MMQO interactions with the bromodomains of BRD4.To understand MMQO recognition by the bromodomains (BD1 and BD2) of BRD4, we next performed a nuclear magnetic resonance (NMR) binding study using 1H/15N heteronuclear single quantum coherence (HSQC) spectra to assess MMQO binding to BD1 or BD2 of BRD4. As shown in Fig. 6A, both 15N HSQC spectra exhibited major chemical shift perturbation of key residues upon addition of MMQO to the protein solution, confirming that MMQO is able to bind to both BD1 and BD2. We determined the binding affinities of MMQO to BD1 and BD2 of BRD4 with a Kd (dissociation constant) of 5.45 μM for BD1 and a Kd of 3.44 μM for BD2 using isothermal titration calorimetry (ITC) (Fig. 6B).

Structural analysis of MMQO interaction with BRD4 bromodomains. (A) Two-dimensional (2D) 15N-HSQC spectrum of BRD4 BD1 (left) or BD2 (right) in the free form (black) and in complex with MMQO compound (red). The protein concentration was 0.1 mM, and the molar ratio of the protein to the compound was 1:5. (B) ITC measurement of BRD4 BD1 or BD2 binding to MMQO. (C) Ribbon depiction of the lowest-energy NMR structure of BRD4 BD1 (light blue) in complex with MMQO (green). (D) Ribbon and stick diagram of BRD4 BD1 binding pocket showing side chain interactions of protein residues in BRD4 BD1 with MMQO. BD1 residues involved in ligand binding are colored light blue, and MMQO is green. The orientation is the same as in panel C. (E) Electrostatic potential surface representation of BRD4 BD1 bound to MMQO (green). The quinoline pyridine ring is exposed to the solvent outside the pocket. The orientation is the same as in panel C.

To determine the detailed molecular basis of MMQO recognition by BRD4 bromodomains, we solved the three-dimensional structure of BRD4 BD1 bound to MMQO using heteronuclear multidimensional NMR spectroscopy (Fig. 6C). Analysis of the 20 final NMR structures of the complex revealed root mean square deviations (RMSDs) of 0.16 Å and 0.50 Å for secondary backbone and heavy atoms, respectively (data not shown; see Table S5 in the supplemental material), signifying well-resolved NMR structures with highly refined conformations. Superimposing backbone atoms between the NMR structure of the BRD4 BD1-MMQO complex and the crystal structure of the BRD4 BD1-MS417 complex (Protein Data Bank [PDB] accession no. 5Z9C and 4F3I, respectively) yielded an RMSD at 0.94 Å, indicating that the protein structures are similar in different ligand-bound states (data not shown).

Although MMQO and MS417 bind similarly into the BRD4 BD1 pocket, MMQO does not form a hydrogen bond interaction with Asn140 of BD1, which is key for acetyl-lysine, as well as MS417, binding (30, 31). Nevertheless, the carbonyl group of the quinoline ring in MMQO is close to the side chain amide of Asn140. The aromatic ring with two methyl groups (ring A) intercalates into the acetyl-lysine binding pocket in BRD4 BD1, and the methyl group was observed to have nuclear Overhauser effect (NOE) signals in the NMR spectra to side chain atoms of Val87, Leu92, Leu94, Tyr139, and Ile146, indicating ligand recognition involving hydrophobic interactions (Fig. 6D and E). The methoxy group of ring A is bound in the WPF (Trp81, Pro82, and Phe83) groove and shows hydrophobic interactions with Trp81, Pro82, Val87, Leu92, and Ile146. Also, an aromatic orthoproton with respect to the methyl and methoxy groups of ring A displays hydrophobic interactions with Phe83, Val87, Cys136, Ile146, and Ala150, while the paraproton to the methoxy group interacts with Leu92, Leu94, and Ile146. The quinoline pyridine ring is exposed to the outside of the binding pocket, and protons on the pyridine ring did not show any intermolecular NOEs to the protein. Because the residues of BRD4 BD1 involved in MMQO binding are conserved in BD2, this explains why BD2 binds to MMQO with similar affinity.

Altogether, the RT-PCR, microarray, and in vitro experiments convincingly demonstrated that MMQO interacts with the two bromodomains of the BRD4 protein, thus forcing the dysregulation of the acetylation-sensitive transcriptome.

BET inhibitors reactivate latent HIV-1 integrations distinct from those of HDAC inhibitors and PKC pathway agonists.Understanding how MMQO functions allowed us to investigate its effect on the latent minigenome more adequately. Recently, a method called B-HIVE was proven to function as an effective tool to map HIV inserts across different integration loci throughout the genome and to correlate their responsiveness to LRAs (19). Utilizing this new method, we intended to characterize the specific subsets of proviral integrations capable of being reactivated by BRD4 inhibition with MMQO and JQ1. As mechanisms for comparison to bromodomain inhibition, we also included SAHA as an HDAC inhibitor and prostratin as a PKC pathway-inducing agent.

Native Jurkat cells were first infected with a barcoded library of GFP-expressing minigenomes. Four days later, the GFP-positive cells were sorted by fluorescence-activated cell sorting (FACS) and left to expand for 19 days. The GFP-negative cells were then isolated, cultured for 2 weeks, and used for treatments. Following a 24-hour treatment in duplicate with either MMQO, JQ1, SAHA, prostratin, or an equivalent volume of dimethyl sulfoxide (DMSO) as a vehicle, we extracted RNA from the treated cell pool and amplified the reverse-transcribed viral cDNA with HIV-specific primers. Each reactivated provirus produced a barcoded transcript, allowing us to distinguish it from the others. The amplified barcodes were then submitted to sequencing and bioinformatics analysis to determine specific proviruses that were activated by each treatment. Hierarchical clustering carried out on the reactivated barcodes confirmed the tight clustering of the duplicate controls (Fig. 7A). More importantly, the MMQO-treated samples clustered remarkably well with the JQ1-treated samples, confirming MMQO specificity as a bromodomain-inhibiting reagent. To our surprise, the responding barcodes from prostratin and SAHA treatments grouped together but separately from the bromodomain-inhibited samples. This result demonstrates how different subsets of HIV-1 minigenome integrations are activated by separate canonical reactivating mechanisms. The dendrogram denoting expression of all sequenced integrations shown in Fig. 7A demonstrates how prostratin activates a larger number of HIV integrations, and to a greater extent than the other drugs, in agreement with the data shown in Fig. 1. The integrants responded almost identically to MMQO and JQ1 but differently to SAHA, confirming once more the functionality of MMQO.

Effects of bromodomain inhibitors on individual proviruses. (A) Cells infected with a barcoded library of HIV minigenomes and sorted for silenced HIV were treated with either MMQO, JQ1, SAHA, prostratin (PRO), or DMSO for 24 h. RNA was extracted and subjected to B-HIVE to determine the specific proviruses that were activated with each treatment. Shown is a heat map and its corresponding dendrogram of the mRNA tag counts of proviruses under different conditions. The experiment was performed in duplicate with each sample sequenced twice. (B) Scatterplots of the mRNA tag counts of proviruses for MMQO compared to other drugs. (C) Pearson correlation coefficients corresponding to the scatterplots of the mRNA tag counts of proviruses under different conditions.

Next, in scatterplots, we compared the mRNA barcode tag counts of proviruses under each condition (Fig. 7B and C). Again, MMQO response was strongly correlated with JQ1, similar to the duplicates, and the lowest correlation was with prostratin. This indicated that the responses of each provirus were very similar between treatments with drugs from the same functional family and reproducible with the same drug. However, there was substantial variation in the expression of the same provirus treated with different drugs. Thus, bromodomain inhibitors stimulate different subsets of latent proviruses than HDAC inhibitors or PKC agonists. The high correlation (0.92) between the relative expression of the proviruses upon MMQO and DMSO treatment indicates that MMQO has similar effects on most of the proviruses. If MMQO acted strongly on only a subset of the proviruses, the overall distribution of expression would differ and the correlation with DMSO treatment would decline further (as in the case of prostratin).

Next, we ranked the proviral integrations as a function of their response to each drug and focused our analysis on the top 15% and 50% of responders to each drug. Most of the MMQO and JQ1 top responsive proviruses were common between the two drugs (see Fig. S2 in the supplemental material). Approximately 75% of these proviruses were inserted within active genes (the rest consisted of 6% in enhancers and 19% in intergenic regions). Among the 38 genes that hosted an HIV integration that responded to MMQO and JQ1 (from the list of the top 50% of responders), 9 were dysregulated upon MMQO treatment in our microarray experiment (4 genes were upregulated, and 5 were downregulated). All of these genes contain at least a region enriched in BRD4 according to chromatin immunoprecipitation sequencing (ChIP-seq) data in Jurkat cells published previously (GEO accession number GSE83777). In Fig. S3 in the supplemental material, we present snapshots of several of these genes. Therefore, proviruses that respond to BET inhibitors do not need to be inserted in genes that respond to the drug. This is in line with the finding that there is no correlation between the expression of the HIV provirus and its host gene (19). In fact, BRD4 is present at the promoters of a vast majority of host genes, colocalizing with the RNA polymerase (see Fig. S3 in the supplemental material; data from GSE83777). According to the MMQO and JQ1 microarrays mentioned above, only a minor proportion of genes showing BRD4 at the promoter are affected by treatment with BETi.

In conclusion, we report that MMQO, despite having a very different chemical structure, behaves like JQ1 and other BETi on HIV and global host gene expression, targeting BRD4 directly. BETi reactivate a subset of HIV proviruses with limited overlap with HDACi and PKC agonists. Although most proviruses are hosted within active genes and genes may present BRD4 at their promoters, HIV activation by BETi does not correlate with the response of the corresponding host genes to these drugs.

In this work, we describe a workflow to characterize and determine the mechanism of a novel HIV-1-reactivating scaffold compound. Combining simple HIV-1 latency models with basic biochemical methods, RNA expression microarray technologies, and in vitro ligand identification methods, we demonstrate that MMQO interacts with bromodomains of BRD4 and functions as an inhibitor of the protein. BRD4 inhibition as a mechanism provides potential not only in anti-HIV settings, but also for suppression of tumorigenesis and as a regulator of immunomodulatory clinical applications. As indicated in the comparative microarrays between HDAC and bromodomain inhibition, the latter mechanism targets a more limited set of genes, which could possibly highlight the possibility that BET inhibitors have fewer side effects in clinical applications than HDAC inhibitors. Nevertheless, bearing in mind the simplistic structure of MMQO and the molecular promiscuity of BET family inhibitors in general (reviewed in reference 32), it is possible that MMQO additionally targets other bromodomains that we were not able to identify due to the constrained mRNA and protein expression detection-based methods.

Functional evaluation of MMQO.The quinoline scaffolds that MMQO is composed of have previously been called “privileged structures” with rich diversity in biological properties (33). Numerous quinoline-structured chemotherapeutics are currently available on the market (e.g., lenvatinib, topotecan, and irinotecan), and they have been found to be applicable in research settings. Quinolines are applied as DNA intercalators, G quadruplex structure stabilizers, androgen receptor antagonists, metal ion chelators, and antimitotic agents. A large variety of inhibitors of tubulin polymerization, histone acetyltransferases, topoisomerases, kinesins, mTOR, PARP, proteasomes, and mitogen-activated protein kinases (MAPKs) have also been developed. Immunomodulatory effects of quinolones have been described with HDAC-, sirtuin-, STAT3-, and NF-κB-inhibitory chemicals (reviewed in reference 34). The 8-hydroxyquinoline structures, which served as lead compounds for designing MMQO, are specifically known to present excellent scaffold compound characteristics and have shown promise in the development of anticancer, antifungal, and antiparasitic agents and even as HIV integrase inhibitors (35). Although MMQO displays potential as a new class of bromodomain inhibitor, broader in vitro assays targeting the whole BET family should be carried out in order to evaluate its specificity toward their different bromodomains.

During the complex process of lead compound design, the drug-related factors that have to be taken into account are its potency, half-life in physiological settings, ease of extraction or synthesis, molecular weight (smaller molecules diffuse better), and number of reactive hydrogen bonds, among other requirements (reviewed in reference 36). Following the discovery of the first BRD4 inhibitor in 2010, various small molecules besides JQ1 have been designed. The most successful reagents are based on the triazolodiazepine structure (such as JQ1, CPI-203, OTX-015, and I-BET-762), but alternative scaffolds have also been proposed (37). Although quinoline structures have been the basis for BRD4 inhibition, as well (38), to our knowledge, none of these preliminary compounds has been developed into a final commercially available product.

MMQO provides a new minimalistic yet optimizable platform to design inhibitors of the BET family proteins, whether for clinical applications or simply as biochemical tools for research use. With the existing structure, we demonstrated that MMQO inhibits both BD1 and BD2 of BRD4 with similar potencies. To our surprise, the affinity between BRD4 and MMQO exists independently of interaction with the asparagine residues, normally a compulsory feature for bromodomain inhibitors. As a starting point for modification, improvements could be carried out on the benzene ring of MMQO to improve its affinity for the asparagine residues of bromodomains. Additionally, enhancement of MMQO could be achieved by designing additional functional groups to interact with the solvent-exposed WPF shelf of the BET family proteins, which are lacking in the existing MMQO structure but are necessary to determine the drug's higher affinity and specificity.

Along with our studies, we tested another bromodomain inhibitor for HIV reactivation, RVX-208, known to bind preferentially to BD2 domains of BET family proteins (39). Reactivation was achieved with doses ∼20 times higher than those normally used for BD2 inhibition (data not shown), suggesting that the compound functions on the HIV-1 minigenome through nonspecific mechanisms, most probably by inhibiting the BD1 domain of BRD4. Our results suggest that inhibition of BRD4 BD2 alone is not sufficient to rescue the HIV-1 minigenome from latency. BD2 has been previously shown to bind the triple-acetylated cyclin T1 subunit of the P-TEFb complex, thus regulating its transcriptional activity (40). The lack of HIV-1 reactivation in this context suggests that the chemical inhibition of BD2 is not a sufficiently potent mechanism to increase the global free P-TEFb pool that can be hijacked by the viral Tat protein. Interestingly, our results with Tat-negative minigenomes highlight the fact that bromodomain inhibitors function independently of the Tat protein, suggesting that BRD4 could potentially harbor an alternative suppressive mechanism in HIV-1 transcription, aside from its canonical competition with the P-TEFb complex. Analogous observations were also made with similar HIV-1 minigenomes (9, 41). The exact repressive roles of BRD4 on the HIV-1 promoter remain elusive and should be further studied.

JQ1 is known to additionally inhibit the second bromodomain of BRD4 (BRD4 BD2), as well as other members of the BET family (42). If the drug can target multiple sites simultaneously, even on the same protein, then according to the Bliss model, it can be considered to have different mechanisms. This could explain the synergism we observed between MMQO and JQ1 at low doses. Nevertheless, at higher doses, this synergy dissipates, suggesting that the effect on the latent minigenome can easily be saturated and the alternative BET-inhibitory mechanisms can be overwhelmed.

Reimagining the shock.CD4+ T lymphocytes are known to migrate and reside in various organs besides the blood, including the brain, lymph nodes, gut, lungs, and female reproductive tract, to name a few (43). Since only 2 to 5% of the total CD4+ T lymphocyte population resides in the blood, it is crucial that the proposed new LRAs possess the ability to diffuse to and function in different tissues. In addition to participating in various physiological functions, the migrated T lymphocytes might also possess diverging phenotypes, which could further interfere with drug responsiveness. The uneven outcomes due to the differences of individual compounds in the case of shock and kill therapies have already been observed: even though the HDAC inhibitors romidepsin and panobinostat increased plasma viremia and T lymphocyte activation, SAHA failed to do so (reviewed in reference 7). Furthermore, prior experiments utilizing the B-HIVE method demonstrated how SAHA targeted proviral integration sites separate from those targeted by phytohemagglutinin (PHA), primarily by reactivating minigenomes in the vicinity of gene enhancers (19). Due to the complex nature of the infected cells, a large variety of LRAs have to be identified and developed for tissue penetration, cell type, and provirus integration locus specificity.

Employing the same B-HIVE technology, we illustrated how bromodomain inhibition additionally targets separate subsets of integration events. Based on the distinct targeting mechanisms and the synergies witnessed among the tested compounds (MMQO, prostratin, and SAHA), these different agents might in the future have to be administered as cocktail regimens for an efficient shock and kill therapy. Ideally this shock cocktail should have minimal effect on T lymphocyte activation and global chromatin stability yet be able to maximally induce proviral reactivation. Even though small-molecule PKC pathway agonists like PMA, prostratin, and bryostatin 1 are known to lead to a cytokine storm in clinical trials against cancer, the doses needed for viral reactivation result in minimal cytokine release from T lymphocytes under ex vivo conditions (16). Additionally, the inhibitors of histone deacetylases, like valproic acid, TSA, and SAHA, have been shown to maintain the activation of resting T lymphocytes at a minimal level (reviewed in reference 44).

According to recent observations, once HIV-1 transcription passes the first phase of Tat production, HIV-1 expression can function autonomously of cellular relaxation due to the potency of the positive-feedback loop of Tat (45). This fact suggests that HIV-1 reactivation should be possible in resting CD4+ T lymphocytes in a clinical setting, as well. More importantly, it implies that for an efficient therapy, activating only the first block of viral transcription (Tat protein production) should be enough to initiate the shock cascade. We propose that further assays be carried out to fully comprehend the effects of mixtures of reagents with distinct mechanisms of action in combination with the B-HIVE technology. Furthermore, these experiments should be carried out in Tat-negative models for added benefit, since in a physiological latency setting, the viral Tat protein is not sufficiently abundant. Alternatively, CD32a was recently identified as a specific marker of quiescent and latently infected HIV-harboring T lymphocytes (46). For additional clinical significance, the optimal synergies between the different classes of drugs for clinical pilot studies could be identified by utilizing quantitative viral outgrowth assays of sorted CD32a-expressing T lymphocytes isolated from HIV-positive patients. The clarification of the full potential of the synergies these compounds will generate could lead to a clinically acceptable regimen that is also capable of delivering the needed reservoir reactivation within patients.

Currently the HIV research field is still struggling with crucial unanswered questions on latency, such as the location of viral reservoirs within the patient, the cell types harboring these proviruses, and their responsiveness to LRA (reviewed in reference 47). We think that no single compound is diverse or potent enough to qualify as a universal solution and to reactivate the whole quiescent viral reservoir; thus, a combination of optimally synergizing pharmaceuticals administered at safe doses would be needed for a maximal proviral response. As an added benefit, the cocktail compounds could theoretically be administered at lower doses than their individual components, since ideally the synergy between the reagents should exceed the potential of any single agent. With the recent development of new chemicals to reactivate latent HIV via alternate pathways, like Toll-like receptors (48, 49) and the inhibition of the SWI/SNF complex (50), a broad range of possible cocktails can be assayed for efficacy. Keeping in mind that MMQO targets a unique subset of proviral integrations, bromodomain inhibitors could play a decisive role in the design of efficient shock remedies.

Cell lines, culturing conditions, and cell treatments.Jurkat cells or latently infected derivatives carrying the pEV731 or pEV658 minigenome (LTR-Tat-IRES-GFP-LTR) were grown at 37°C with 5% CO2 in RPMI 1640 medium (Sigma; R8758) supplemented with 10% fetal bovine serum (FBS) without additional antibiotics. Jurkat clones and heterogeneous Jurkat populations were described previously (17, 18). Heterogeneous HeLa cell populations were created according to a previously described protocol (17). HEK293T and HeLa cell lines were grown at 37°C with 5% CO2 in Dulbecco's modified Eagle's medium and GlutaMax medium containing 10% FBS supplemented with 100 U/ml penicillin and 100 μg/ml streptomycin.

When indicated, cells were treated with reagents at the concentrations and for the durations provided in the figure legends. All the reagents were dissolved in DMSO, and when possible, the concentration of DMSO was kept at 1 μl/ml or below. DMSO, PMA, HMBA, prostratin, and TSA were purchased from Sigma and SAHA, JQ1, and RVX-208 from Selleck Chemicals. All the reagents were diluted in DMSO.

Primary CD4+ T cell isolation and infection.Primary CD4+ T cells were isolated from buffy coats from healthy donors by Ficoll gradient, followed by density-based negative selection of CD4+ T cells using RosetteSep (Stem Cell Technologies, Cambridge, United Kingdom). After isolation, the cells were left for 24 h to rest, followed by infection as previously described, with minor modifications (51). Briefly, CD4+ T cells were infected with the pNL4-3.Luc.R− E− virus by spinoculation at 1,200 × g for 120 min. The spin-infected cells were incubated overnight at 37°C in a humidified 95% air-5% CO2 atmosphere. The next day, the cells were washed and cultured in RPMI 1640 medium (Sigma) supplemented with 10% FBS, 100 μg/ml penicillin-streptomycin, and 5 μM saquinavir mesylate. Three days after infection, the cells were left untreated or treated as indicated in the presence of 30 μM raltegravir. Cells were harvested 24 h after stimulation, and luciferase activity was measured with a luciferase assay system (Promega, Leiden, Netherlands). Viral pseudotyped particles were generated as previously described (50). The data were normalized either as the fold increase over the untreated control or as percent PMA. The following compounds were used to stimulate cells: PMA (Sigma-Aldrich), ionomycin (Sigma-Aldrich), SAHA/vorinostat (Selleck Chemicals), prostratin (Sigma-Aldrich), pyrimethamine (Sigma-Aldrich), CAPE (MP Biomedicals), OTX-015 (ApexBio), and JQ1 (Sanbio).

Apoptosis and viability of primary CD4+ T cells.In order to determine apoptosis, primary CD4+ T cells were left untreated or stimulated in duplicate for 24 and 72 h, as indicated. The cells were washed with Dulbecco’s phosphate-buffered saline (DPBS) (Lonza) supplemented with 3% serum, 2.5 mM CaCl2 and stained with anti-annexin V-phycoerythrin (PE) (BD Biosciences; catalog no. 556454). The viability of the ex vivo-infected primary CD4+ T cells was measured by flow cytometry on the basis of forward versus side scatter. Samples were analyzed with a Becton Dickinson Fortessa flow cytometer. The data represent cells isolated from at least 3 independent buffy coats.

Microarrays.Total RNA was extracted using a High Pure RNA isolation kit (Roche) according to the manufacturer's instructions. High RNA integrity was assessed by Bioanalyzer nano-6000 assay (Agilent Technologies). Sample preparation was described previously (52). For each sample, 100 ng of total RNA was reverse transcribed into cDNA with a T7 promoter, and the cDNA was in vitro transcribed into cRNA in the presence of Cy3-CTP using a low-input Quick Amp kit (Agilent). Labeled samples were purified using RNeasy mini-spin columns (Qiagen). Then, 600 ng of cRNA was preblocked and fragmented in Agilent fragmentation buffer and mixed with Agilent GEx hybridization mix. The hybridization mixture was laid onto each sector of a subarray gasket slide and sandwiched against an 8 by 65,000 format oligonucleotide microarray (Human v1 Sureprint G3 Human GE 8×60K microarray; Agilent design 028004) inside a hybridization chamber, which was hybridized overnight at 65°C. Subsequently, the array chambers were disassembled, submerged in Agilent Gene Expression Buffer 1, and washed for 1 min in another dish with the same solution using a magnetic stirrer at 200 rpm at room temperature, followed by 1 min in Agilent Gene Expression Buffer 2 with a magnetic stirrer at 200 rpm at 37°C, immediate withdrawal from the solution, and air drying. The fluorescent signal was captured as TIF images with an Agilent scanner using the recommended settings with Scan Control software (Agilent). The signal intensities were extracted into a tabulated text file using Feature Extraction software (Agilent) with the appropriate array configuration and annotation files. The normalized log2 intensities were obtained by the quantile method with normalized expression background correction using the Bioconductor Limma package in R.

Microarray analysis.Genes were sorted and organized for further analysis based on the normalized log2 intensities obtained. Transcripts with a false-discovery rate (q value) of <0.05 were considered significantly differentially expressed, and transcripts represented in multiples in several probes had their fold changes calculated to a mean value. The differentially expressed genes were then analyzed utilizing the g:Profiler toolkit (https://biit.cs.ut.ee/gprofiler/), GSEA software (http://software.broadinstitute.org/gsea/index.jsp), and R. Scatterplots and Pearson's correlation coefficients were produced using in-house R scripts.

RNA extraction, reverse transcription, and real-time PCR.Total RNA was extracted using a TRIzol kit (Ambion). cDNA was generated from 50 to 100 ng of RNA using a Superscript Vilo cDNA synthesis kit (Invitrogen). Gene products were analyzed by qPCR using SYBR green master mix (Invitrogen) and specific oligonucleotides in a Roche Applied Science 480 light cycler machine on 96-well plates. The primers used for qPCR are listed in Table S6 in the supplemental material.

Protein extraction, gel electrophoresis, and immunoblotting.Cells were washed once with PBS, and proteins were extracted in the lysis buffers indicated in the figure legends. Radioimmunoprecipitation assay (RIPA) lysis buffer was supplemented with 1× protease inhibitor cocktail (Roche), 1 mM Na3VO4, 5 mM NaF, 1 μg/ml leupeptin, 0.5 μg/ml pepstatin, 0.5 μg/ml aprotinin, 20 mM β-glycerophosphate, and 1 mM phenylmethylsulfonyl fluoride (PMSF) to block product degradation. The protein concentration was determined by bicinchoninic acid (BCA) assay (Pierce), and 10 to 30 μg of protein was boiled in Laemmli buffer and electrophoresed in 7.5 to 15% SDS-polyacrylamide gels. The separated proteins were transferred to nitrocellulose or polyvinylidene difluoride (PVDF) membranes (constant 400 mA; 4°C) for 1.5 h. Blots were blocked in Tris-buffered saline (TBS) solution containing 0.1% Tween 20 (TBST) and either 5% nonfat dry milk, 3% bovine serum albumin, or 1:1 Odyssey blocking buffer for 1 h and incubated with primary antibodies at room temperature for 1 h or overnight at 4°C, followed by 3 10-min washes with TBST, and incubated with secondary antibodies for 1 h at room temperature. Following 3 washes of the secondary antibodies, the immunodetection of specific proteins was carried out with primary antibodies using the Odyssey infrared imaging system (Li-Cor).

Flow cytometry.GFP fluorescence was measured in a Cytomics FC500 MPL flow cytometer or a CytoFlex system (Beckman Coulter). A two-parameter analysis was used to distinguish viable cells (identified by forward and side scatter) containing GFP-derived fluorescence (525 nm) from the background. Fluorescence was represented in a logarithmic scale, and on average, >10,000 events were observed per sample. Optical calibration was carried out using 10-nm fluorescent beads (Flow-Check fluorospheres; Beckman Coulter). Cell sorting was carried out with a FACS Aria cell sorter (BD Biosciences).

Bliss independence model.Lack of synergy between JQ1 and MMQO was calculated according to the Bliss independence model as described previously (16). The data are presented as the difference between the observed and predicted fractional responses relative to a 24-hour PMA (10 nM) treatment. The Bliss independence model assumes that two different drugs act through separate molecular mechanisms and therefore in an additive manner.

Sample preparation for NMR.The first bromodomain (BD1) of human BRD4 (residues 53 to 169) in the pNIC28 vector was expressed and purified using a procedure described previously (53). Briefly, the His-tagged BRD4 BD1 domain was overexpressed in Escherichia coli pRIL plasmid BL21-CodonPlus cells and induced with 0.3 mM isopropyl-β-d-thiogalactopyranoside at 18°C. His-tagged BRD4 BD1 was purified by HiTrap immobilized-metal affinity chromatography (IMAC) (GE Healthcare). After removing the His tag with thrombin or tobacco etch virus (TEV) protease treatment, protein samples were further purified on a Superdex 75 or Superdex 200 column (GE Healthcare). Uniformly 15N- or 15N/13C-labeled proteins were prepared as unlabeled protein by growing the bacteria in M9 minimal medium containing 15NH4Cl with or without [13C]glucose.

Nuclear magnetic resonance spectroscopy.The BRD4 BD1 domain-MMQO compound complex was used for structure determination. NMR samples of the BD1 domain (0.5 mM) in complex with 1.5 mM MMQO compound were prepared in phosphate-buffered saline (PBS) (pH 7.4) in H2O-D2O (9/1) or D2O. All NMR spectra were acquired at 25°C on Bruker 500-, 700-, 800-, and 900-MHz spectrometers equipped with z-axis gradient triple-resonance cryoprobes. The backbone 1H, 13C, and 15N resonances were assigned using standard three-dimensional triple-resonance HNCA, HN(CO)CA, HN(CA)CB, and HN(COCA)CB experiments (54). The side chain atoms were assigned from three-dimensional HCCH-total correlation spectroscopy (TOCSY), HCCH-correlation spectroscopy (COSY), and (H)C(CO)NH-TOCSY data (55). The NOE-derived distance restraints were obtained from 15N- or 13C-edited three-dimensional nuclear Overhauser effect spectroscopy (NOESY) spectra. The MMQO was assigned from a one-dimensional 1H spectrum and two-dimensional TOCSY and NOESY and 13C/15N-filtered TOCSY and NOESY spectra. The intermolecular NOEs used in defining the structure of the complex were detected in 13C-edited (F1), 13C/15N-filtered (F3) three-dimensional NOESY spectra (unlabeled MMQO compound bound to 13C/15N-labeled BD1 protein) (56). Spectra were processed with NMR Pipe and analyzed using NMR View (57, 58).

Structure calculations.The structures of the BRD4 BD1 domain/MMQO were calculated with a distance geometry simulated annealing protocol with CNS (59). Initial protein structure calculations were performed with manually assigned NOE-derived distance constraints. Hydrogen bond distance and ϕ and ψ dihedral-angle restraints from the TALOS-N prediction were added at a later stage of structure calculations for residues with characteristic NOE patterns (60, 61). The converged structures were used for the iterative automated NOE assignment by ARIA refinement. Structure quality was assessed by CNS, ARIA, and PROCHECK analysis (61, 62). A total of 30 intermolecular NOE-derived distance restraints were added in the structure determination of the BRD4 BD1-MMQO complex. A family of 200 structures was generated, and the 20 structures with the lowest energies were selected for the final analysis.

Isothermal titration calorimetry.Experiments were carried out on a MicroCal auto-ITC200 instrument at 25°C while stirring at 750 rpm in PBS buffer (pH 7.4). The MMQO sample (0.04 mM) in the PBS buffer was placed in the cell, whereas the microsyringe was loaded with the protein sample (0.5 mM). The titrations were conducted using 20 successive injections of 2.0 μl (the first at 1.0 μl and the remaining 19 at 2.0 μl) with a duration of 4 s per injection and 150 s between injections. The collected data were processed using the Origin 7.0 software program (Origin Lab) supplied with the instrument according to the “one set of sites” fitting model.

B-HIVE.Native Jurkat cells were infected at a multiplicity of infection (MOI) of 0.1 with a barcoded library of GFP-expressing minigenomes described previously (19). Four days later, 10,000 GFP-positive cells were sorted by FACS and left to expand for 19 days. At that time, a sample of the cell pool was collected, the DNA was extracted, and barcodes were mapped by inverse PCR as described previously (19). The GFP-negative cells were thereafter isolated and cultured for 2 weeks. Subsequently, the cells were grown to 70% confluence, separated into 10 pools, and treated in duplicate with MMQO (160 μM), JQ1 (1 μM), prostratin (10 μM), SAHA (5 μM), or an equivalent volume of DMSO (1 μl/ml). For each sample, two independent reverse transcription reactions were performed, each on 10 μl purified mRNA, to which was added 1 μl 20 μM reverse primer (TTTCGCTTTTAATACGACTCACTAT) and 1 μl 10 mM deoxynucleoside triphosphates (dNTPs) (Thermo Fisher Scientific; R0181). The RNA was denatured at 95°C for 1 min and incubated on ice Then, 8 μl master mix containing 4 μl 5× cDNA synthesis buffer, 1 μl 0.1 M dithiothreitol (DTT), 1 μl 40-U/μl RNase Out, 1 μl diethyl pyrocarbonate (DEPC)-treated water, and 1 μl 15-U/μl Thermo Script (reagents included in the ThermoScript RT-PCR system; Invitrogen; 11145-024) was added to the denatured RNA, and the mixture was incubated at 65°C for 1 h. The reaction mixture was heat inactivated at 85°C for 5 min, and 5 μl RT product was used as the template in 50 μl standard Phusion polymerase reaction mix (Thermo Fisher Scientific; F530S) in GC buffer with 1 μM primers (AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCT and CAAGCAGAAGACGGCATACGAGAT-index-TTTTAATACGACTCACTATA, where “index” is a 4-nucleotide sequence that identifies the experiment). The cycling conditions were as follows: 98°C for 1 min; 98°C for 20 s, 60°C for 30 s, and 72°C for 1 min (27 cycles); 72°C for 5 min. The samples were sequenced as 50-bp single reads on a HiSeq 2000 sequencer (lllumina) using v3 sequencing chemistry.

HIV barcodes were extracted from paired-end reads through an inexact search of the T7 promoter sequence (TATAGTGAGTCGTA) allowing up to 3 errors (mismatches, insertions, and deletions) using Seeq v1.1.2 (http://github.com/ezorita/seeq). The barcodes were clustered with Starcode v1.0 (63), allowing one mismatch and using the “message passing” clustering algorithm. We considered only barcodes that had more than 1 read under at least 17 of the 20 replicate conditions. We then computed the pairwise Pearson coefficient of correlation, r, between the replicate conditions and used 1 − r as a dissimilarity metric between them. Clustering and dendrogram representations were performed with the hclust() function of R with default parameters.

The total number of reads per sample is fixed at pooling time, so the measurements indicate the relative expression of the provirus within a given treatment, but they cannot be compared between treatments. To that end, the number of reads was divided by the total amount of GFP fluorescence, defined as the percentage of GFP-positive cells multiplied by the average fluorescence of the GFP-positive cells. This score is a proxy for the total production of virus mRNA, establishing a baseline to compare the expression of a provirus across treatments.

Accession number(s).The solution structure of the BRD4 BD1-MMQO complex and the NMR spectral data are available in Protein Data Bank (PDB) under PDB code 5Z9C and BioMagResBank (BMRB) code 36163 (http://www.bmrb.wisc.edu/search/instant.php?term=36163), respectively.

This work was supported by funding from the Spanish Ministry of Economy and Competitiveness (MINECO); the European Regional Development Fund (grant BFU2014-52237-P); Fundación para la Investigación y Prevención del SIDA en España (FIPSE 360946/10; to A.J.); ERC Synergy Grant 609989; Centro de Excelencia Severo Ochoa 2013-2017, SEV-2012-0208; and Plan Nacional BFU2012-37168 (to G.J.F.). The work was also supported in part by the research fund of the First Hospital of Jilin University (Changchun, China), grants from the National Institutes of Health (to M.-M.Z.) and the European Research Council under the European Union's Seventh Framework Programme (FP/2007-2013)/ERC STG 337116 Trxn-PURGE, and Dutch AIDS Fonds grants 2014021 and an ErasmusMC research grant (to T.M.).

NOTE: We request your email address only to inform the recipient that it was you who recommended this article, and that it is not junk mail. We do not retain these email addresses.

ASM journals are the most prominent publications in the field, delivering up-to-date and authoritative coverage of both basic and clinical microbiology.

Agitated Tank Reactor

Anionic Polyacrylamide, Cationic polyacrylamide, Nonionic Polyacrylamide - Oubo,https://www.oubochem.com/