Didehydro-Cortistatin A Inhibits HIV-1 by Specifically Binding to the Unstructured Basic Region of Tat

dCA directly interacts with Tat.The production of functional recombinant HIV-1 Tat in Escherichia coli is not trivial as the proper folding of Tat is extremely important for activity, and when expressed by itself, Tat agglomerates mostly because of oxidation of the cysteine residues in the cysteine region. Tat can also adhere to different surfaces and anionic polymers via its charged and hydrophobic residues (21, 22). We produced and purified recombinant HIV Tat HXB3 (residues 1 to 89) with a 6×His tag in the C-terminus in E. coli. For this, we have partially optimized the arginine codon usage and changed the rare AGA into codon CGT at arginine positions 7, 53, 56, and 87 of Tat (Fig. 1A). These changes in combination with the use of M9 medium resulted in the production of approximately 5 mg/liter of purified Tat. After Tat purification using nickel affinity chromatography followed by size exclusion filtration, Tat protein size was analyzed by SDS-PAGE followed by Coomassie blue staining (Fig. 1B). The specificity of the bands was confirmed by Western blotting using anti-Tat serum and anti-His antibody (Fig. 1B). The purity of Tat was further verified by size exclusion chromatography and high-performance liquid chromatography (HPLC) (Fig. 1C and D). We confirmed the activity of Tat by performing transactivation assays in both HeLa-CD4 cells stably expressing the HIV-1 5′ long terminal repeat (LTR) promoter driving luciferase expression (HeLa-CD4-LTR-Luc cells) (10) and the OM10.1 cell line model of HIV-1 latency (23) (Fig. 1E and F; see Fig. S2A and S2B in the supplemental material). In HeLa-CD4-LTR-Luc cells, the activity of our purified Tat was similar to the published activity of Tat obtained from the AIDS Reagent Program (10). In addition, denatured Tat and Tat mutated in the basic domain (Tat Mut [arginine residues mutated to alanine residues]), used as negative controls, were unable to transactivate the reporter (Fig. S2A). As expected, dCA blocked Tat transcriptional activity in a dose-dependent manner (Fig. 1E). Controls such as raltegravir, an HIV-1 integrase inhibitor, and SCM [S-(−)-carbidopamonohydrate], a nonspecific compound (which inhibits l-amino acid decarboxylase) not known to block HIV replication, had no inhibitory activity. In OM10.1 cells, Tat significantly activated HIV-1 transcription, while Tat Mut did not (Fig. 1F). The addition of dCA (250 nM) inhibited transcriptional activity of Tat (1 μg) to the same extent as dCA alone (Fig. 1F), while SCM, used as a negative control, had no activity.

• Open in new tab
• Download powerpoint

FIG 1

Purity and activity of the monomeric His-tagged Tat HXB3 and analysis of its direct binding to dCA. (A) Amino acid sequence of His-tagged-Tat HXB3 protein. (B) Analysis of recombinant Tat purity and monomeric status upon Coomassie staining and Western blotting with an anti-Tat serum or anti-His antibody. Data are representative of n = 3 independent experiments. (C) Size exclusion chromatography of the recombinant Tat. (D) HPLC of the recombinant Tat. Data in panels C and D are representative of n = 2 independent experiments. (E) dCA inhibits the transactivation activity of recombinant Tat in HeLa-CD4-LTR-Luc cells. Raltegravir (Ralt.) and compound S-(−)-carbidopamonohydrate (SCM) were used as negative controls. Data are the mean ± standard error of the mean (SEM) from n = 3 independent experiments. (F) dCA inhibits the transactivation activity of recombinant Tat in the OM10.1 cell model of HIV-1 latency. Compound SCM and transactivation with recombinant Tat Mut were used as negative controls. Data are the mean ± standard deviation (SD) from n = 3 independent experiments. (G) ITC of dCA binding to Tat. The top plot shows the baseline corrected experimental data. In the bottom plot, results were converted to molar heat and plotted against the molar ratio of dCA to Tat protein. Data are representative of n = 2 independent experiments. Statistical significance was determined using one-way ANOVA with post hoc Tukey’s test. ns, not significant; ***, P < 0.0001; *, P < 0.01.

Next, we studied the direct interaction of Tat with dCA using ITC. After gradual titration of Tat with dCA in the ITC cell, the integrated heat data of the dCA-Tat interaction were collected (Fig. 1G). This ITC titration curve fitted well with a single-site binding process with a nonlinear least-squares regression, allowing the determination of the association constant (K_a) and entropy (ΔS) and enthalpy (ΔH) changes, as well as the average number of binding sites (N). We determined a 1:1 binding stoichiometry with a K_a of ∼(9.39 ± 1.7) × 10⁶ M⁻¹ (dissociation constant [K_d] = 100 nM). Moreover, dCA binds to Tat at 25°C with a negative (favorable) enthalpy of −3.5 ± 0.1 kcal mol⁻¹ and positive (favorable) entropy of +20.0 ± 0.5 cal mol⁻¹ K⁻¹. The favorable values of ΔH and ΔS are consistent with the characteristics of a combination of van der Waals, hydrophobic, and electrostatic interactions in the binding process. The difference between Tat/dCA affinity (ITC, 100 nM) and dCA inhibitory activity in cell-based assays (0.6 to 2 nM) (Table 2) (10) could be the result of the nature of the Tat-TAR feedback loop that amplifies activity in vivo; especially since the amount of Tat in vivo after infection is very low (24).

Blind modeling of dCA binding to Tat’s active conformation of the activation and basic domains.We previously showed that dCA inhibits Tat from HIV clades A, B, C, D, and E (13); thus, molecular docking of dCA to Tat was performed using AutoDock Vina and the HIV-1 Tat Mal protein (D clade) as a template (PDB entry 1K5K) (Table 1 and Fig. 2). PDB entry 1K5K was chosen based on the following: (i) all its NMR ensembles are converged and energetically minimized, (ii) its structure and sequence superimpose with those of other Tat proteins (Table 1, Fig. 2; see Fig. S3 in the supplemental material), (iii) the basic patch residues are docked against the activation domain and are solvent exposed (Table 1 and Fig. 2), (iv) its structure was validated by biochemical assays and by mutagenesis (14, 15, 17, 25 – 29), (v) its NMR was performed under the active/functional conditions of the protein, and (vi) it has a low clashscore and low percentages of Ramachandran and side-chain outliers (Table 1). Initially, blind docking with dCA and similar scaffolds were done with a cubic grid size of 45 Å in AutoDock Vina, the exhaustiveness level was set at 10, and default options were used for the remaining parameters. Consistent binding poses were obtained, and the top 10 best-docked ligand poses were selected based on the free energy predictions (from −7.9 to −6.5 kcal/mol) (Fig. 2A). In most cases, the ligands were oriented in similar orientations, and those poses that were in opposite orientation differed in the free energies by <0.2 kcal/mol. We obtained the basic patch as the binding interface as the top two hits by blind docking. We further targeted the basic domain binding site for clear orientation of the ligands by reducing the grid size to 28 Å, approximately half the size of the blind docking grid. We obtained consistent docking sites with the orientation of dCA and other analogs in the basic patch (residues 49 to 59) with small changes in the free energies (ΔG of less than −0.1 kcal/mol). Only the top solution out of the 10 best hits of dCA was used for pharmacophore analysis of the Tat binding site. Our docking analysis showed that the small molecule dCA binds preferentially to the basic patch of Tat protein. On one side of the dCA molecule, basic patch residue Arg⁵⁵ directly interacts with its nitrogen atom (N²⁷) on the isoquinoline group (Fig. 2B). Furthermore, we performed similar docking experiments with additional Tat structure PDB entries 1JFW, 1TIV and 1TBC (1TAC was not included given its similarities to 1TBC). We found the binding of dCA to Tat to be similar, with the isoquinoline and intact cycloheptene groups dictating dCA’s ability to bind to the basic patch of the HIV-1 Tat proteins (see Fig. S3 and S4 in the supplemental material).

• Open in new tab
• Download powerpoint

FIG 2

Molecular modeling of the interaction of dCA with HIV-1 Tat protein PDB entry 1K5K. (A) The small molecule dCA and analogs with similar scaffold preferentially docked to the basic patch of the HIV-1 Tat protein. (B) Close-up view of the best pose of dCA binding to Tat in the docking analysis. Basic patch residues of the NMR ensemble are shown in stick representation, and the ligand dCA is shown in yellow. (C and D) Close-up view of the binding site of two inactive analogs of dCA (analogs 2 and 8) and the interacting residues in HIV-1 Tat protein. (E) Overview of dCA binding to Tat protein—a horseshoe-shaped dCA binding site. The basic patch of Tat is shown as a gray surface, and dCA compound is shown in stick representation with a transparent blue surface.

Structure-function relationship of dCA.To understand the structure-function relationship of the dCA molecule with respect to our structural model, we synthesized and characterized the ability of 11 analogs of dCA to inhibit HIV infection (Table 2). Figure S5 in the supplemental material shows their NMR profiles. We used HeLa-CD4 cells stably expressing the HIV-1 5′ LTR driving β-galactosidase production (HeLa-CD4-LTR-LacZ cells) as a reporter for the infection with isolate NL4-3. Table 2 summarizes the IC₅₀, 50% cytotoxic concentration (CC₅₀), and therapeutic index (TI) for each compound. The loss of dCA activity was directly associated with the positioning and the chemical environment of the azote on the F ring and the presence of oxygen in ring B. Moreover, the E-F “skeleton angle” and the distance between the A-B-C-D and F rings affected the activity by more than 50-fold (analogs 7 and 10). Interestingly, the oxidation at position 16 in ring D improved the IC₅₀ (from 0.6 nM to 0.5 nM for dCA versus analog 9), as well as the CC₅₀ (from 20 μM to >50 μM) and the TI (from 33,333 to >100,000).

As we mentioned above, dCA is an analog of the natural product CA. CA has been reported to display antiproliferative properties toward HUVECs and antileukemic activity in CA-sensitive AML cell lines by selectively inhibiting CDK8 activity (7 – 9). Importantly, the characteristics necessary for the antiretroviral activity of dCA do not seem to overlap features needed for HUVECs’ antiangiogenic activity, which requires at least one OH or NMe₂ in ring A, a certain A-B-C skeleton angle, no oxidation of C₁₆ and C₁₇ in ring D, and an intact stereochemistry of ring E-F (30).

In our structure-function relationship analysis, dCA and analogs 1, 3, 4, 8, and 9 (all active except 8) have the same orientation and position of nitrogen atom, and they are within hydrogen bond distance with the guanidinium group of Arg⁵⁵ (Fig. 2A, B, and D). Importantly, all these molecules show higher affinities than the molecules with nitrogen atoms in different positions (Table 2). On the other side of dCA molecule, the backbone atoms of the basic residues Lys⁵¹ and Arg⁵² interact with the hydroxyl groups on the estrone group and stabilize the dCA molecule (Fig. 2B). However, even when the nitrogen atom of the analogs is in an optimal orientation and the cycloheptene group on the cortistatin scaffold is perturbed—for example, analog 8 (Fig. 2D)—these interactions are not stabilized as the hydroxyl groups are not interacting with the backbone atoms of the basic patch. In our structure-function relationship analysis, though the phthalazine derivative (analog 5) has a similar orientation to dCA and has one of the nitrogens in the same position as isoquinoline, it lacks the C-H group and is replaced by a nitrogen atom, reducing its affinity to Tat. The isoquinoline group of dCA readily forms a hydrogen bond with Arg⁵⁵, and the adjacent C-H group readily forms a hydrogen bond with backbone carbonyl of Pro³ from the N-terminal region of Tat protein (see Fig. S6 in the supplemental material). Hence, we propose that the position and orientation of this nitrogen atom in the isoquinoline group and the adjacent C-H group are important. The heterocyclic C-H bond donors are typically stronger when the carbon atom carrying the donor is adjacent to a heteroatom in the ring, with electron-withdrawing elements or substituents amplifying the effect. Thus, we suggest that an isoquinoline group and an intact cycloheptene ring in these analogs are important for their potent inhibitor activity (Fig. 2B to D). These compounds are also stabilized by the hydrophobic noncharged carbon side chain atoms (-CH₂-) tails of the basic patch residues Lys⁵⁰, Lys⁵¹, and Arg⁵³. Moreover, the N-terminal residue Glu² interacts with Arg⁵³, and the residue Pro³ interacts with the isoquinoline group, further stabilizing the binding surface for the dCA molecule. In sum, Lys⁵¹ and Arg⁵² interact with the hydroxyl groups of dCA, and the isoquinoline group of dCA forms a hydrogen bond with Arg⁵⁵ and the backbone carbonyl of Pro³. The interaction of Glu² with Arg⁵³ may further stabilize the binding of dCA to Tat by bringing both domains (N-terminal and basic domains) into close proximity. Concurrently, the addition of dCA to the strong intramolecular interactions between both domains locks the protein in a more stable structure, further connecting both domains of the protein. All these residues in the basic patch form a horseshoe-like surface (Fig. 2E) for the bulky dCA ligand and similar scaffolds and enable their potent inhibitor properties. We observed similar binding of dCA and analogs to Tat using other known NMR structures (1JFW, 1TIV and 1TBC), and essentially the binding features to the basic patch residues are preserved (Fig. S3 and S4).

dCA specifically inhibits the functions of the basic domain of Tat but not of Rev or Hexim-1.If dCA were to bind the basic domain of Tat in its disordered form, then one would predict it might also bind to other basic peptide sequences and block their functions. The two well-known proteins that share a similar basic domain with Tat are HIV-1 Rev protein and the cellular protein Hexim-1 (Fig. 3A). As such, we investigated whether dCA could inhibit these similar basic domain functions. The basic domain of Rev is responsible for its nuclear localization (31); therefore, we assessed whether dCA could impact the subcellular localization of Rev (Fig. 3B) as previously shown for Tat (10). We transfected HeLa-CD4-LTR-Luc cells with a plasmid expressing FLAG-tagged Rev and assessed Rev localization in the presence of 100 nM dCA. As expected for Tat used as the positive control, dCA caused a significant redistribution of Tat to the periphery of the nucleolus (Fig. 3B and C), forming a distinctive ring-like structure; however, the localization of Rev was left unchanged (Fig. 3B and C). Next, we investigated whether dCA was affecting the viral mRNA export function of Rev (Fig. 3D; Fig. S7A in the supplemental material). As such, we infected HeLa-CD4-LTR-LacZ cells with the NL4-3 strain and treated the cells with dCA or control compounds (KPT or SCM). KPT is an inhibitor of CRM1-mediated nuclear export. Twenty-four hours later, cells were collected, and total, nuclear, and cytoplasmic mRNAs were extracted, reverse transcribed, and quantified by quantitative real-time PCR (qRT-PCR). In whole-cell lysates (Fig. 3D, left), dCA treatment reduced all viral transcripts produced, as expected from a specific HIV-1 transcriptional inhibitor. In the presence of KPT, unspliced and singly spliced viral mRNAs were reduced, while the multiply spliced 2-kb viral mRNA species were increased (as previously shown in HeLa cells [32]). SCM, an unrelated compound, had no effect on HIV mRNAs. When analyzing the ratio of nuclear versus cytoplasmic viral mRNAs (Fig. 3D, right), KPT treatment resulted in a drastic accumulation of singly spliced and unspliced viral mRNAs in the nucleus compared to the dimethyl sulfoxide (DMSO) control, while SCM had no activity. dCA treatment did not mediate nuclear retention of singly or unspliced viral mRNAs as observed for KPT. We confirmed these results using the human T4 lymphoblastoid cell line CEM-SS (Fig. S7A). Together our results demonstrate that dCA does not impact Rev’s basic domain functions.

• Open in new tab
• Download powerpoint

FIG 3

dCA does not interfere with Rev and Hexim-1 functions. (A) Sequence of the basic domains of HIV Tat HXB3, HIV Rev Bru and intracellular Hexim-1. (B) dCA does not exclude Rev from the nucleolus. Shown is confocal microscopy analysis of the subcellular localization of transfected FLAG-Tat or FLAG-Rev in the presence of DMSO or dCA in HeLa-CD4-LTR-Luc cells. Proteins were recognized with anti-FLAG and Alexa Fluor 568-tagged anti-IgG. Magnification 60×. Data are representative of n = 2 independent experiments. (C) Quantification of the nucleolar fluorescence of panel B using ImageJ. Data are the mean ± SEM from 3 to 4 different cells. (D) dCA does not perturb the HIV RNA cytoplasmic export function of HIV Rev. HeLa-CD4-LTR-LacZ cells were infected with strain NL4-3 for 24 hours. The next day, compounds (dCA, 10 nM; KPT, 400 nM; and SCM, 10 nM) were added for 24 hours. Total, nuclear, and cytoplasmic mRNAs were extracted, and viral messages were determined by qRT-PCR. GAPDH was used for normalization. Data are the mean ± SD from n = 5 independent experiments. (E) dCA does not interfere with the binding of Hexim-1 with 7SK snRNP. HeLa-CD4-LTR-LacZ cells were incubated with the indicated compounds. Forty-eight hours later, ChIP RNA with anti-Hexim antibody was performed. U1 snRNP, an unrelated RNA, and RNase treatment were used as controls. Data are the mean ± SD from n = 3 independent experiments. (F) dCA does not exclude intracellular Hexim-1 from the nucleus. Shown is confocal microscopy analysis of the subcellular localization of Hexim-1 in the presence of DMSO or dCA in HeLa-CD4-LTR-Luc cells. Magnification 60×. (G) Quantification of nuclear fluorescence of panel F using ImageJ. Data are the mean ± SEM from 3 to 6 different cells. Statistical significance was determined using one-way ANOVA with post hoc Tukey’s test. ns, not significant; ***, P < 0.0001; **, P < 0.001.

Most cellular P-TEFb is found in an inactive form in complex with the 7SK small nuclear ribonucleoprotein (snRNP) (33, 34), a tridimensional 4-stem-loop structure that serves as an RNA scaffold, among other proteins, for the homodimer of the CDK9-inhibitory protein Hexim-1/2. Hexim-1 directly binds to cyclin T1 and is responsible for the inactivation of the kinase activity of P-TEFb (35, 36). One proposed mechanism for P-TEFb release from 7SK snRNP is the direct competition between Tat and Hexim-1 for binding cyclin T1 and/or 7SK snRNP (reviewed in references 37 to 39). Given the high homology between Tat’s basic domain and Hexim-1, this basic region could also interfere with Hexim-1 association with 7SK snRNP by binding to a TAR-like motif within stem-loop 1 of 7SK snRNP (34, 40, 41). As such, we investigated whether dCA could interfere with the association of the basic domain of Hexim-1 with 7SK snRNP. Intracellular Hexim-1 was immunoprecipitated, and 7SK snRNP bound to Hexim-1 was extracted, reverse transcribed, and quantified by qRT-PCR as previously described (42) (Fig. 3E). HeLa-CD4-LTR-LacZ cells were treated with DMSO, dCA, or SCM (used as a negative control). The interaction between Hexim-1 and 7SK snRNP was similarly detected in the presence of DMSO, dCA, or SCM, but not when samples were subjected to RNase treatment, which was used as a positive control. The specificity of the binding of 7SK snRNP to Hexim-1 was also confirmed by the absence of U1 snRNP in the samples, used as a control. In parallel, we assessed the activity of dCA on Hexim-1 subcellular localization, but as expected, dCA did not have any effect on Hexim-1 nuclear localization (Fig. 3F and G). Altogether, our results suggest that dCA does not interfere with the activity of Rev or the binding of Hexim-1 to 7SK snRNP, reinforcing dCA’s specificity for the basic domain of Tat. This further supports our model that dCA is not affecting any stretch of poly-arginine residues, but rather interacts with the Tat basic patch in the context of the Tat structure.

dCA activity on HIV-1 transcription is independent of CDK8.The Mediator complexes are considered to be an integral part of the preinitiation complex by interacting with proteins, including RNAPII, general transcription factors, and transcription elongation factors, thus, serving as a signaling hub capable of integrating multiple inputs for precise control of RNAPII (43). In mammals, the Mediator is a large complex composed of 25 to 30 subunits organized into four distinct modules referred as the head, the middle, the tail, and the kinase module. The kinase module, consisting of CDK8, or its paralog CDK19, cyclin C, and two additional subunits, Med12 and Med13, carries the only known enzymatic activity in the complex and associates with the rest of the complex in a facultative fashion (44). Although the kinase module was initially characterized as a repressor, recent work has shown that it is also required for gene activation across multiple transcriptional programs (45). Several phosphorylation targets have been identified, including the RNAPII C-terminal domain (CTD) (45).

The only other known ligand for dCA is CDK8. Although the characteristics necessary for the antiretroviral activity of dCA do not overlap features needed for CDK8 inhibitory activity, it was important to verify whether the activity of dCA on HIV transcription involved CDK8 activity. As such, virus-like particles (VLPs) expressing short hairpin RNAs (shRNAs) against CDK8, CDK19, CDK8/CDK19, or the shRNA control were transduced into HeLa CD4 cells. Stabilization of the knockdown was performed by puromycin selection of the populations over a 2-week period. Both RT-qPCR and Western blotting confirmed successful knockdown of CDK8/CDK19 proteins (Fig. 4A and B). These cells were then infected with NL4-3 and treated or not with dCA, and production of viral RNA and p24 in the supernatant were assessed 72 hours later. While the viability of the cells remained unchanged (Fig. S7B), reduced expression of CDK8 and/or CDK19 did not significantly affect viral mRNA expression (Fig. 4D) and p24 capsid production, nor interfered with the antiviral activity of dCA (Fig. 4D and E). Our results are supported by prior report by Riuz and colleagues (46), which did not observe differences in viral replication upon knockdown of CDK8-associated Med12 and Med13 proteins. Moreover, work performed by the Karn group (47) demonstrated the mediator/CDK8 module is associated with the promoter during latency. Activation of NF-κB by tumor necrosis factor alpha (TNF-α) causes CDK8 to dissociate from the mediator, resulting in recruitment of TFIIH followed by RNAPII CTD phosphorylation and elongation. The mediator/CDK8 is lost during transcriptional activation, and thus, seemingly a marker of latency. As such, if dCA was to inhibit CDK8 in the context of HIV, one would expect activation of transcription rather than inhibition. Altogether, dCA does not mediate its anti-HIV activity through CDK8 inhibition, but rather through blocking Tat activity.

• Open in new tab
• Download powerpoint

FIG 4

CDK8 does not regulate HIV-1 infection, and the activity of dCA is independent of CDK8. HeLa CD4 cells were transduced with VLPs expressing shRNAs against CDK8, CDK19, or both and puromycin selected to stabilize knockdown (KD). These cells were then infected with the NL4-3 strain for 24 h, in the presence of DMSO or dCA (200 nM). After 24 hours, cells were washed and fresh medium with compounds was added. Seventy hours later, mRNA, DNA, and p24 capsid in supernatant were collected and measured. (A and B) CDK8 mRNA and CDK19 mRNA expression was assessed by qRT-PCR and normalized to GAPDH mRNA. Shown is the mean ± SEM from n = 3 independent experiments. (C) Protein expression of CDK8 and CDK19 was analyzed 72 hours postinfection by Western blotting. GAPDH was used as an internal control. The image shown is representative of n = 3 independent experiments. (D) Viral mRNA production measured by qRT-PCR using primers against the Nef region was measured as a function of total integrated proviral DNA. Data represent the mean ± SEM from n = 3 independent experiments. (E) p24 capsid production in the supernatant relative to total HIV DNA and normalized to the DMSO of each condition is represented. Data represent the mean ± SEM from n = 3 independent experiments. Statistical significance was determined using one-way ANOVA with post hoc Tukey’s test. ns, not significant; ns^a, DMSO KD CDK19 versus dCA KD CDK19. *, P < 0.01; ***, P < 0.0001; **, P < 0.001.

dCA binding stabilizes Tat structure.Our results suggest that dCA strengthens the intramolecular interactions between the basic and the N-terminal domains of Tat, promoting a more stable structure. As such, we assessed whether dCA increased Tat stability. First, we verified whether Tat when bound to dCA maintained proper folding by confirming that it was recognized by the conformational monoclonal antibody (MAb) 7G12 (48). Using competitive enzyme-linked immunosorbent assays (ELISAs), we incubated different ratios of dCA with Tat protein on an ELISA plate. In parallel, the plate was coated with Tat first, and dCA was added the following day. In both cases, the antibody at an equimolar ratio to Tat was added afterwards, followed by absorbance measurement. The MAb 7G12 was at limiting concentrations when dCA was added at higher molar ratios than Tat. Similar binding of MAb 7G12 to Tat protein was observed in the absence or presence of increasing concentrations of dCA (Fig. S8A). In pulldown assays followed by high-performance liquid chromatography (HPLC), Tat was first incubated with MAb 7G12, and then dCA was added to the complex before glycine HCl buffer elution and HPLC analysis with a C₈ reverse-phase column (Fig. S8B). dCA’s peak was observed only when MAb 7G12-labeled Tat was present in the sample. We did not observe a peak of Tat because of the 10- to 20-µg detection threshold of the C₈ HPLC column. The MAb 7G12’s peak was also not visible due to the cutoff of the C₈ column or to the precipitation of the antibody in the salt-free elution buffer. Collectively, these results suggest that (i) Tat maintains proper folding since it binds a conformational epitope-recognizing antibody, and (ii) dCA bound to Tat does not occlude surface-exposed residues of the three-dimensional Tat, believed to be targeted by MAb 7G12 (¹MEPV⁴, ⁴⁴GISYGRKK⁵¹, and ⁹⁹PED¹⁰¹) (27). In addition, since MAb 7G12 recognizes and neutralizes Tat from several variants, presenting small local structure variations (2), the changes of Tat structure promoted by dCA do not seem significantly important to prevent Tat recognition by MAb 7G12.

Next, we investigated whether dCA mediated conformational changes in Tat. NMR studies have reported the fluorescence of Tat’s unique Trp¹¹ to be highly sensitive to conformational modifications (2, 15). As such, we assessed Tat fluorescence between wavelengths of 310 nm and 410 nm (Fig. 5A, left). As reported, a λ_max was observed at 360 nm, and this fluorescence was quenched with acrylamide, used as control. The addition of dCA to Tat (dCA/Tat ratio of 7.5/1) resulted in the following changes: (i) decreased intensity of the Trp fluorescence, (ii) a reduction in λ_max (from 360 to 350 nm), and (iii) an increased I_330/350 ratio with dCA (I_330/350 = 0.893) compared to Tat alone (I_330/350 = 0.73). Altogether these changes suggest that the presence of dCA amplified the optimal hydrophobic environment for Trp¹¹. Raltegravir (raltegravir/Tat ratio of 7.5/1), used as a negative control, had no effect on the intrinsic fluorescence of Tat. To confirm these results, λ_max was measured at additional ratios of dCA to Tat protein and similar results were observed (Fig. 5A, right). In sum, these results suggest that dCA promotes burying of Trp¹¹ in the optimal hydrophobic environment in the core of Tat, resulting in the stabilization of the protein.

• Open in new tab
• Download powerpoint

FIG 5

dCA stabilizes Tat structure. (A) Trp fluorescence assays. (Left) Schematic of the assay. (Center) DMSO, dCA (7.5× Tat concentration) or raltegravir (Ralt.; 7.5× Tat concentration) was incubated with Tat. The fluorescence of the Trp¹¹ of Tat protein was measured at the indicated wavelengths. Acrylamide (250 mM) was used as a quenching control of Tat fluorescence. (Right) Peak absorbance (λ_max) versus increasing fold ratios of compounds to Tat protein. Data are representative of n = 3 independent experiments. (B) DARTS assays. Tat was incubated with DMSO or dCA, at increasing concentrations, in the presence of increasing concentrations of proteinase K (PK). The protection was revealed by Western blotting. Analog 2 (Ana. # 2), an inactive analog of dCA (Table 2), was used as a negative control. Data are representative of n = 2 independent experiments. (C) Steady state of Tat in the presence of dCA in vivo. HeLa-CD4-LTR-Luc cells were incubated with recombinant Tat and then washed and incubated with the indicated compounds for 6 hours. The protection of Tat from degradation was revealed by assessing the amount of protein left by Western blotting. (Left) Western blot representative of n = 4 independent experiments. MG132 was used as a positive control of Tat protection from degradation, and raltegravir was used as a negative control. (Right) Densitometry of the Tat band normalized to the actin band. Results represent the mean ± SEM from n = 4 independent experiments. Statistical significance was determined using one-way ANOVA with post hoc Tukey’s test. ns, not significant; ***, P < 0.0001; **, P < 0.001.

To confirm that dCA stabilizes Tat structure, we performed drug affinity responsive target stability (DARTS) assays to measure the protection against proteolysis conferred on the target protein by interaction with a small molecule. For this purpose, we incubated Tat with several concentrations of dCA and different concentrations of proteinase K (PK) for 20 min at room temperature. PK digests proteins preferably after hydrophobic, aliphatic, and aromatic residues. The protection was determined by Western blotting with an anti-His antibody (Fig. 5B). Tat alone was proportionally degraded with increasing concentrations of PK. When the concentration of dCA was increased, Tat degradation by PK was significantly reduced. The inactive analog, analog 2 (Table 2), used as a negative control, did not protect Tat from degradation by PK, further supporting the molecular modeling results. Both the fluorescence assays and the DARTS assays support the notion that dCA stabilizes Tat protein.

Next, we determined the half-life of recombinant Tat in HeLa-CD4-LTR-Luc cells in the presence of dCA (Fig. 5C). We incubated recombinant Tat with cells for 4 hours, and after several washes to remove extracellular Tat protein, we incubated cells with DMSO, dCA, or raltegravir for an additional 6 hours. Western blotting with anti-His antibody was performed to reveal intracellular Tat. Tat protein levels after 6 hours were reduced by more than 80%, which is in agreement with previous reports (49). dCA significantly protected Tat from degradation, increasing the percentage of remaining Tat from 18% in DMSO to 40% with dCA. Raltegravir, used as a negative control, did not significantly alter Tat intracellular concentration (28%). The proteasome inhibitor MG132, used as a positive control, confirmed the previously reported involvement of the proteasome in Tat degradation (49, 50). dCA protected Tat from degradation to the same extent as MG132 (40%). Altogether, these results suggest that dCA may increase the steady-state level of Tat in cells by stabilizing Tat structure.

dCA blocks Tat-TAR interaction.Tat amplifies viral mRNA expression by binding through its basic domain to TAR, the 5′-terminal region stem-bulge-loop structure of HIV mRNAs, and recruits several cofactors to the viral promoter to promote transcriptional elongation. We investigated whether dCA could compete with TAR for Tat binding. For this purpose, we performed EMSA with radiolabeled TAR and recombinant Tat in the presence or absence of dCA (Fig. 6A). Raltegravir and a version of TAR with the Tat binding bulge deleted (ΔTAR) were used as negative controls. dCA inhibited the interaction of Tat with TAR in a dose-responsive manner, up to the same degree as the background determined by the nonspecific interaction of Tat with ΔTAR. This nonspecific interaction was previously reported, revealing two binding sites of Tat on TAR: a specific and high-affinity binding site and a nonspecific and low-affinity binding site (51). Raltegravir, used as a negative control, had no inhibitory activity.

• Open in new tab
• Download powerpoint

FIG 6

dCA binding to Tat disrupts Tat-TAR interaction. (A) EMSA of full-length Tat with TAR in the presence of dCA. Raltegravir (Ralt.) and deleted bulge TAR (ΔTAR) were used as negative controls. Data are representative of n = 3 independent experiments. (B) EMSA of Tat’s basic domain with TAR in the presence of increasing concentrations of dCA to Tat’s basic peptide. Pep B is a peptide encompassing the basic domain of HIV Tat (⁴⁵ISYGRKKRRQRRRAP⁵⁹). Pep C (⁶⁹LSKQPASQPRGDPTG⁸³) and unlabeled or cold TAR (TARc) were used as negative controls. Neomycin, which binds to Tat and TAR, was used as a positive control. Ratios of compounds (or TARc) concentrations to peptide are shown. Data are representative of n = 3 independent experiments. (C) dCA does not bind to TAR in ITC assays. Shown are ITC results from binding of dCA (left) or Tat peptide (right) to TAR at 25°C. In each panel, the top plot shows the baseline corrected experimental data. For the lower plots, results were converted to molar heats and plotted against the molar ratio of dCA or peptide to RNA. Data are representative of n = 2 independent experiments. (D) Tat-TAR interaction blocked by dCA in HeLa-CD4-LTR-Luc cells in vivo. Cells transfected with Tat or Tat Mut (arginine residues mutated to alanine in Tat’s basic domain) were incubated with DMSO or the indicated compounds (25 nM). Forty-eight hours later, cells were either lysed to determine the luciferase activity or cross-linked for ChIP assays. Luciferase activity was determined relative to protein concentration (Bradford assay) (Fig. S9A). Western blotting was performed to verify the amount of Tat under each condition (Fig. S9B). ChIP of Tat was followed by qRT-PCR of TAR DNA. The results are presented as percentage of input. The mean ± SEM is represented. Data are representative of n = 3 independent experiments. Statistical significance was determined using one-way ANOVA with post hoc Tukey’s test with 3 technical replicates. ns, not significant; ***, P < 0.0001.

We also performed the same experiment with a peptide encompassing the basic domain of Tat (Pep B; residues 45 to 59) and dCA. In this experiment, as negative controls we used a peptide of the C terminus of Tat (Pep C; residues 69 to 83), neomycin (a nonspecific inhibitor of Tat and TAR) (52), and unlabeled cold TAR (TARc) (Fig. 6B). dCA inhibited the interaction of Pep B with TAR in a dose-dependent manner, while Pep C did not interact with TAR and raltegravir did not compete with the Pep B-TAR interaction. As expected, both neomycin and TARc inhibited the interaction of Pep B with TAR. Using ITC, we also confirmed that dCA does not bind to TAR (Fig. 6C, left) as opposed to the basic peptide of Tat (Fig. 6C, right).

To confirm these results, we carried out a chromatin immunoprecipitation (ChIP) assay of TAR RNA on HeLa-CD4-LTR-Luc cells, upon transfection with FLAG-Tat or FLAG-Tat mutant (Mut) proteins in the presence or absence of dCA. We confirmed that dCA inhibited Tat transactivation activity by measuring the luciferase activity, and FLAG-Tat and FLAG-Tat Mut were well expressed under the different conditions (Fig. S9). Chromatin was immunoprecipitated with an anti-FLAG antibody, and qRT-PCR was performed to detect the presence of chromatin-coding TAR (Fig. 6D). A Tat-TAR interaction was detected in the presence of DMSO or raltegravir used as a negative control. However, dCA reduced this Tat-TAR interaction to similar levels as FLAG-Tat Mut, which aspecifically interacts with TAR. Altogether, our in vitro and in vivo results demonstrate that dCA blocks Tat transactivation by inhibiting Tat binding to TAR, completing our previous reports demonstrating dCA activity only on viruses with a competent Tat/TAR feedback loop (11).