Curaxin CBL0137 has the potential to reverse HIV-1 latency

Abstract

A cure for human immunodeficiency virus type-1 (HIV-1) has been hampered by the limitation of current combination antiretroviral therapy (cART) to address the latent reservoirs in HIV-1 patients. One strategy proposed to eradicate these reservoirs is the “shock and kill” approach, where latency-reversing agents (LRAs) are used to reactivate and promote viral cell death and/or immune killing of reactivated cells. Here, we report that curaxin CBL0137, an anti-tumor compound, can potentiate TNFα-mediated reactivation of latently infected HIV-1cell lines. Additionally, the single use of CBL0137 is sufficient to reactivate HIV-1 latent reservoirs in peripheral mononuclear cells (PBMCs) isolated from HIV-1 positive, cART-treated, aviremic patients. Thus, CBL0137 possesses capabilities as a LRA and could be considered for the “shock and kill” approach.

Introduction

Since the implementation of cART, rates of HIV-1 mortality, morbidity and newly acquired infections have decrease drastically¹. However, long-term therapy is still required because the virus established a reversible state of latency in memory CD4+ T cells. To address these reservoirs, two main strategies are being investigated. First, the “shock and kill” approach aims to purge the latent reservoirs by inducing renewed viral transcription/reactivation using latency-reversing agents (LRAs) to favor viral cytopathic cell death and immune clearance. Second, another strategy coined the “block and lock” approach focuses on the development of latency-promoting agents (LPAs) to instead induce a deep state of HIV-1 resulting in permanent silencing of latently infected cells and lead to their potential decay overtime. In that regard, we have previously implicated the facilitates chromatin transcription (FACT) complex as a negative regulator of HIV-1 latency ². Furthermore, we showed in an earlier publication that curaxin 100 (CBL0100), a compound first described as an anti-tumor drug that targets FACT, is a promising LPA by targeting HIV-1 transcriptional elongation^{3, 4}. Interestingly, another curaxin, CBL0137, has been implicated in promoting HIV integration. Its mechanism of action involves FACT-mediated destabilization of chromatin nucleosomes to favor viral genome entry into newly accessible DNA⁵. However, there are no studies looking at the effects of CBL0137 in the context of HIV-1 latency. Thus, in the present work we decided to investigate the potential of CBL0137 as an anti-latency agent.

Material and Methods

Flow Cytometry.

J-LAT6.3 or CA5 cells were treated with DMSO or stimulated with 10 ng/ml (JLAT 6.3) or 2ng/ml (CA5) TNFα, 10 nM bryostratin-1 (Bryo-1), 0.335 μM SAHA, or 1 μM JQ1. 2.5×10^6 of aforementioned cells were also co-treated with DMSO or CBL0137 (0.5 μM) for 24hrs. Viable cells were then sorted using an Accuri C6 Plus flow cytometer (BD Biosciences). Percentages of GFP+ cells were analyzed using Flow Jo.

qPCR assay to measure HIV-1 transcription.

5×10^5 U1/HIV-1 cells were activated with 10 ng/ml TNFα, or mock activated with PBS, and co-treated with either DMSO (0.1%) or CBL0137 (0.5 μM). Cells were then collected and subjected to mRNA extraction, reverse transcription and qPCR⁴. The following qPCR primers of HIV-1 transcripts were used: initiation (10–59 bp), elongation-1 (29–180 bp), elongation-2 (836–1015), and gag mRNA (Table S1). GAPDH was used as a reference gene. The same qPCR method was used for DHIV-nef infected Jurkat cells.

PBMCs from HIV-1 positive subjects.

consented HIV-1 positive, cART-treated, aviremic patients (<20 copies/ml) were recruited through the AIDS clinic at the Strong Memorial Hospital of URMC to donate whole blood via leukapheresis. This study was approved by the University of Rochester Research Subjects Review Board (#RSRB00053667). Peripheral blood mononuclear cells (PBMCs) were then isolated and cultured in complete medium supplemented with 20 U/ml IL2 and in the presence of 600 nM of Nevirapine for 3 days. Next, PBMCs were subjected to CD8+ T cells depletion as previously described ⁴. 3–5×10^6 CD8-depleted PBMCs were washed and resuspended in fresh medium in the presence DMSO (0.1%), CBL0137 (0.5 μM) or anti-CD3/CD28 Dynabeads (1 bead/2 cells) + DMSO for 3 more days. Supernatant was collected and viral RNA copies/ml was assessed by ultrasensitive qPCR⁶. The same experimental setup was used for determination of CBL0137’s potential additive effects with TNFα in the context of patient-derived cells with added conditions of TNFα (10 ng/ml) and CBL0137 (0.5 μM) +TNFα (10 ng/ml).

Results and discussion

We first determined a suitable concentration of CBL0137 to investigate its effects on HIV-1 replication and latency (Fig. S1). Jurkat, J-LAT6.3, CA5 and U1/HIV-1 cells were treated with CBL0137 at increasing concentrations (0–4 µM) and cell viability was determined at 24hrs post-treatment (pt). CC₅₀ was calculated for each cell line: Jurkat (CC₅₀ = 1.7 µM); J-LAT6.3 (CC₅₀ = 2.1 µM); CA5 (CC₅₀ =1.9 µM), and U1/HIV-1 (CC₅₀ = 2.0 µM) (Fig. S1b). Using Spina et al’s ⁷ article on the evaluation of anti-latency drugs (where viability of compounds ranged from 69–96%), a concentration of CBL0137 maintaining cell death ≤10 percent (CC₁₀) across all cell lines was selected 0.5 µM (Table S2). Interestingly, this concentration was also used by another group for CBL0137 in a report investigating the importance of FACT complex in HIV-1 integration⁵.

Next, CBL0137’s effect on acute HIV-1 replication was evaluated since CBL0100 has been shown to inhibit acute HIV-1 replication⁴. Specifically, Jurkat cells were pre-treated with DMSO (0.1 %), CBL0137 (0.5 μM) or CBL0100 (0.1 μM) and then infected cells with DHIV-nef⁸. CBL0137 blocked viral replication to the same extent as CBL0100 without any cytotoxicity (Fig. S2). However, these results contrast with Matysiak et al.’s paper, which showed that CBL0137 targeting of the FACT complex facilitated HIV-1 integration by 1.5 fold and enhanced early steps of HIV-1 replication ⁵. Nevertheless, emphasis was strictly put on studying FACT at the integration step and these experiments were limited to HeLa or HEK293 cells, which do not recapitulate the cellular tropism of HIV-1. In the present study, Jurkat T cells were chosen to evaluate the overall effects of CBL0137 on a single cycle viral infection and showed that CBL0137 predominantly inhibits acute HIV-1 replication in this more physiological relevant system. This indicates that CBL0137 might have additional effects on other steps of the viral life cycle during acute HIV-1 replication.

Since curaxins are described as FACT inhibitors and previous work reported that depletion of FACT reactivates HIV-1^{2, 3}, the effects of CBL0137 in post-integrated latency models were further investigated⁴. Two latency T cell lines were used: JLAT6.3 and CA5. Importantly, the JLAT-6.3 model is relatively more difficult to reverse from latency; whereas, CA5 one is easier ^{7, 9, 10}. This allowed examining CBL0137’s capacity to either enhance or suppress reactivation in different latency backgrounds. At 24hrs pt, CBL0137 or CBL0100 alone did not show any significant effect as compared to DMSO at the basal level (Fig. 1a–b). Interestingly, CBL0137+TNFα significantly increased GFP+ cells by approximately 50% for J-LAT6.3 cells and by 25% for CA5 cells with minimal cytoxicity (≤20%) (Figs. 1a–b and S3). Surprisingly, CBL0100 did not block reactivation in both cell lines. This is in contrast to what we have seen in primary CD4+ T cell model and ex vivo models in a previous report⁴. This is potentially due to alternative latency mechanism in these cell lines that CBL0100’s effect is refractory to, which shall be included for future studies.

Fig. 1. CBL0137 enhances TNF-induced HIV-1 reactivation in latency cell lines but not other LRAs.

(a-b) GFP+ cell population was quantitified by flow cytometry for J-LAT 6.3 (a) or CA5 (b) cells that were treated with TNFα (10 and 2 ng/ml, respectively); 10 nM bryostratin-1 (Bryo); 1 μM JQ1; 0.335 μM SAHA; or mock-stimulated in the presence of DMSO or CBL0137 for 24 hours. Bryo+JQ1 was also used as representative control for LRA combination. Data is obtained from at least four independent experiments. Data is represented as the mean ± s.d, **p < 0.01, ***p < 0.001, ****p < 0.0001, using 1-way ANOVA followed by Tukey’s post test. Not significant (N.S.). (c) U1/HIV-1 cells were treated similarly as in (a-b) and RT-qPCR assay was performed to measure the mRNA level of HIV-1 gag (left panel), or HIV-1 initiated and elongated transcripts in TNF-stimulated cells (right panel). GAPDH was used as the reference. Data is normalized to DMSO control. Data was obtained from three independent experiments and represented as mean ± s.d., **p < 0.01, using 1-way ANOVA followed by Tukey’s post test. Not significant (N.S.).

Given that CBL0137 enhanced TNFα reactivation, CBL0137 was tested in combination with three representative LRAs from major LRA classes at clinically relevant concentrations: Bryo-1 as a PKC agonist (10 nM), JQ1 as a bromodomain and extra terminal motif inhibitor (1 μM) and SAHA/vorinostat as a histone deacetylase inhibitor (0.335 μM)¹¹. In the case of J-LAT6.3, neither Bryo-1, JQ1, SAHA alone or in combination with CBL0137 could reactivate cells. Of note, a small but significant increase in GFP+ population (~2.5%) was seen for the Bryo-1+JQ1 regimen, which is arguably one of the better LRA combinations for reactivation of HIV-1 based on previous reports (Fig 1a–b) ¹². Interestingly, CBL0137+TNFα had the highest rate of reactivation in J-LAT6.3 cells reaching up to 32% of cells (Fig. 1a).

In the case of CA5, Bryo-1 and JQ1 increased the number of GFP+ cells to ~70% and ~18%, respectively. Additionally, the combination of Bryo+JQ1 was even more effective reaching 83 % of cells, a result that was statistically significant than either compound alone (Fig. 1b). In contrast, SAHA did not reactivate HIV-1 in CA5 cells. Combinatory treatment with CBL0137 did not enhance reactivation of either Bryo-1 or JQ1 significantly. Moreover, CBL0137 only mildly increased SAHA reactivation by approximately ~1.7 fold as compared to SAHA alone but this result was not statistically significant. Both Bryo+JQ1 and CBL0137+TNFα reached a similar reactivation rate of 83% in CA5 (Fig. 1b). CBL0100 had no effect on CA5 and J-LAT6.3 cells either alone or combination with the other LRAs except for a mild suppression of Bryo-1 in CA5 (Fig. 1a–b). Taken together, these results indicate that single treatment of CBL0137 fails to reactivate latent HIV-1 in the J-LAT6.3 and CA5 cell lines but the CBL0137+TNFα showed reactivation potency in both cell lines to the same extent or even greater than the optimized LRA combination Bryo+JQ1 with a better toxicity profile (≤20% cell death for CBL0137+TNFα vs. 10%–40% for Bryo+JQ1) (Fig. S3) ¹².

It is interesting that CBL0137 could potentiate TNFα but not Bryo-1, since both act through the induction of NF-κB signaling, albeit by different mechanisms. TNFα binds to its receptor and activates the TRAF pathway that leads to NF-κB translocation into the nucleus; whereas, Bryo-1 induces the PKC pathway to promote the same end result ^{13, 14}. Thus, future studies targeting components in these pathways could provide mechanistic insights for CBL0137. Alternatively, CBL0137 was reported to increase chromatin accessibility by intercalating into DNA and disrupting proper DNA/protein interactions by a process termed “c-trapping” ^{5, 15}. Therefore, hypothetically CBL0137 could further enhance HIV-1 reactivation by increasing chromatin accessibility, which could allow for easier recruitment/processivity of the general transcriptional machinery, at an already “sensitized” HIV-1 locus due to TNFα ^{5, 13}. Additionally, CBL0137 fails to considerably enhance either JQ1 or SAHA mediated latency reversal. One explanation could be due to the poor reactivation potential of JQ1 or SAHA as compared to Bryo-1 or TNFα in these latency cell lines. The suboptimal reactivation in HIV-1 latency cells could lead to less accessible chromatin for CBL0137 to intercalate into, thereby effectively preventing CBL0137-mediated increase chromatin accessibility.

A recent study suggested that myeloid cells could represent a source for HIV-1 viral persistence; therefore, CBL0137’s effect on U1/HIV-1 cells was evaluated (Fig. 1c) ¹⁶. Indeed, CBL0137 treatment further enhanced the TNFα-induced gag mRNA expression in U1/HIV-1 cells with cell cytoxicity ≤20% (Figs. 1c left panel and S3). Additionally, we further determined CBL0137’s effect on the level of viral transcripts with different lengths, which showed that all transcripts were up-regulated by CBL0137+TNFα (Fig. 1c right panel). Interestingly, there is a renewed interest in the use of TNFα in combination with other LRAs as a means to reactivate latency in the “shock and kill” cure approach¹³. Thus, CBL0137 could potentially be used with TNFα for this approach.

Despite limited potency at the basal level in cell lines, CBL0137’s potential as a LRA ex vivo was examined. Specifically, CD8-depleted PBMCs from four HIV-1 positive, cART-treated, aviremic individuals were cultured in the presence of CBL0137 (0.5 µM) or anti-CD3/CD28 beads (αCD3/CD28), or DMSO (Figs. 2 and S4). CBL0137 treatment was able to reactivate latent viral reservoirs from three donors; in contrast, αCD3/CD28 reactivated all donors albeit to varied degrees. Specifically, CBL0137 had the most dramatic effect on viral production for donors 1 and 2 and achieved viral copies numbers higher than αCD3/CD28 for these donors. Furthermore, for donor 3 CBL0137 treatment resulted in a moderate response; whereas, no effect was observed for donor 4. In contrast, these donors had higher responses when treated with αCD3/CD28 (Fig. 2a and S4a). As previously reported, CBL0100 alone does not lead to HIV reactivation in patient-derived reservoir cells, as seen in Fig. S4b ⁴. We also briefly investigated as a proof of principle whether a combination of TNFα with CBL0137 would show further enhancement of reactivation in cells isolated from three additional aviremic patients: donors 5, 6 and 7 (Fig S5). CBL0137 alone enhanced reactivation as compared to DMSO for donors 5 and 6, supporting what we previously observed for donors 1–3. On the other hand, single TNFα only reactivated latent HIV-1 for donors 5 and 7. Interestingly, in the case of donor 5 CBL0137+TNFα demonstrated at least an additive effect than either CBL0137 or TNFα alone (Fig S5a). In contrast, no additional effect was seen for donors 6 or 7, probably due to genetic heterogeneity across subjects. However, the results from donor 5 are promising because they inform us that CBL0137 can enhance TNFα mediated latency reversal in patient-derived cells and encourage further validation of this combination in future studies.

Fig. 2. CBL0137 reactivates HIV-1 latent reservoirs in PBMCs of HIV-infected, cART-treated, aviremic donors.

(a) CD8-depleted PBMCs from four aviremic HIV-1 patients were incubated with DMSO, CBL0137 (0.5 μM), or anti-CD3/CD28 beads (αCD3/CD28) for a period of three days, followed by the assessment of viral production using the ultra-sensitive qPCR assay (the limit of detection of this assay is 1 viral RNA copy/ml). Results are shown for all four donors. Data is represented as the grand mean, *p<0.05, using the Kruskal-Wallis ANOVA test. (b) Cell viability of CBL0137-treated PBMCs from donors in (a). Data is normalized to DMSO control and represented as the grand mean.

The observation that CBL0137 lacks responsiveness in cell lines but competent in patient-derived cells is reminiscent of another LRA, hexamethylbisacetamide, which showed a similar pattern between these two cell systems of HIV-1 latency⁷. Therefore, it is not uncommon for LRAs to elicit different responses across various HIV-1 latency models as demonstrated by Spina et al⁷. In the case of CBL0137, the underlying mechanism to explain its discrepant effects is still unknown. A possible explanation is that there may be a transcriptional block that has to be removed by TNFα as one of prerequisites for CBL0137 to elicit its LRA potency in cell lines, which does not exist in patient cells harboring latent HIV-1. Such transcriptional block could be at the initiation step due to transcriptional interference that occurs readily at the 5’ LTR of T cell lines as described by Lenasi et al.¹⁷ Alternatively, this discrepancy could due to the p53 pathway^{3, 15}. Jurkat cells are the parental cell lines for J-LAT6.3 and CA5, which all lack functional p53 ¹⁸. In addition, U1/HIV-1 cells have been proposed to lack endogenous p53¹⁹. In contrast, primary cells such as patient-cells have been shown to have functional p53 and its inhibition leads to reduce reactivation potential²⁰. Interestingly, CBL0137 is known to increase chromatin accessibility via trapping of FACT on the destabilized nucleosomes to further mediate nucleosome disassembly and cause activation of p53 by phosphorylation at serine 392³. Thus, the difference in the effects of CBL0137 on HIV-1 reactivation between models could also link to availability of an intact p53 protein and CBL0137’s ability to induce it through post-translational modifications.

An initial step in delineating the mechanism of CBL0137’s activity was to determine whether it depends viral transactivator Tat as it is essential for efficient reactivation and/or transcription elongation. One of the downstream effects of TNFα is the induction of Tat protein expression, which is crucial for proper HIV-1 transcription elongation¹³. To evaluate if CBL0137 acts through Tat to exert its effect on transcription, we used the JLTRG system, which expresses GFP under the control of the HIV promoter, HIV long terminal repeat (HIV-LTR), when Tat protein is introduced. The retroviral pQCXIP vector expressing Tat or empty vector was transduced into these cells treated with CBL0137 or DMSO (Fig S6). Our results indicate no difference in the percentage of GFP-positive cells between CBL0137 and DMSO in the absence of Tat. In contrast, CBL0137 treatment leads to a moderate increase (~1.5 fold) as compared to DMSO in the presence of Tat. Thus, these results suggest that CBL0137’s effect on transcription might act through Tat to enhance transcription/reactivation.

Overall, this study identifies CBL0137 as a novel LRA. CBL0137 enhances the TNFα-stimulated HIV-1 transcriptional activity in cell line models of HIV-1 latency and is able to induce the reactivation of HIV-1 proviruses in five out of seven patient-derived reservoir cells. Moreover, CBL0137 did not exhibit serious cytotoxicity at the tested concentration in vitro and ex vivo. Thus, these results indicate that CBL0137 is an appealing LRA candidate for future clinical investigation as an anti-latency therapeutic, especially considering that a phase I clinical trial is currently underway to evaluate this drug’s safety and effect as a novel anti-tumor agent (NTC01905228).

Supplementary Material

Supp figS1-6

Fig. S1. Chemical structure and cell viability of curaxin CBL0137. (a) Chemical structure of CBL0137. (b) CC₅₀ of CBL0137 was measured for Jurkat, J-LAT 6.3, CA5, and U1/HIV-1 cell lines following 24-hour treatment with increasing concentration (0–4 μM). Data is normalized to DMSO and represented as an average of at least two independent experiments for each cell line. Error bars represent mean ± s.d.

Fig. S2. CBL0137 inhibits acute HIV-1 replication in Jurkat T cells. (a) Relative HIV-1 gag mRNA expression was measured in DHIV-nef infected Jurkat T cells following the treatment of DMSO (0.1%), CBL0137 (0.5 μM), or CBL0100 (0.1 μM). GAPDH was used as the reference gene. Data is normalized to DMSO and represented as the mean ± s.d, ***p < 0.001, using 1-way ANOVA followed by tukey post test. (b) Cell viability of curaxin-treated, HIV-infected of Jurkat T cells in (c). Data is normalized to DMSO control and represented as the mean ± s.d.

Fig. S3. Viability of J-LAT 6.3, CA5, and U1/HIV-1 cell lines following curaxin and LRA treatment. Cell viability of latency cell lines following 24-hour treatment of indicated compound(s). Data is normalized to DMSO control. Data is obtained from at least three independent experiments. Data is represented as the mean ± s.d.

Fig. S4. Effect of curaxin CBL0137 on HIV-1 reactivation in PBMCs of HIV-infected, cART-treated, aviremic donors. (a) CD8-depleted PBMCs from four aviremic HIV-1 donors were incubated with DMSO, CBL0137 (0.5 μM), or anti-CD3/CD28 beads (αCD3/CD28) for a period of three days, followed by the assement of viral production using the ultra-sensitive qPCR assay (the limit of detection of this assay is 1 viral RNA copy/ml). Data is shown for each individual donor. Data is the representative of two independent qPCR run with technical triplicates. Data is represented as mean ± s.d. (b) A representative repeat of the effect of CBL0137 (0.5 μM), CBL0100 (0.1 μM), or anti-CD3/CD28 on viral load of CD8-depleted PBMCs from four aviremic HIV-1 donors after three day treatment. the ultra-sensitive qPCR assay was used (the limit of detection of this assay is 1 viral RNA copy/ml).

Fig. S5. Effect of TNFα in combination with CBL0137 on HIV-1 reactivaiton in PBMCs of three HIV-infected, cART-treated, aviremic donors. (a) The effect of CBL0137 (0.5 μM), TNFα 10ng/ml), CBL0137 (0.5 μM)+TNFα 10ng/ml), or anti-CD3/CD28 on viral load of CD8-depleted PBMCs from three aviremic HIV-1 donors after 3-day treatment. The ultra-sensitive qPCR assay was used (the limit of detection of this assay is 1 viral RNA copy/ml). (b) Cell viability of averimic donors following 3-day treatment of indicated compound(s). Data is normalized to DMSO control. Data is represented as the mean ± s.d.

Fig S6. CBL0137 works in a Tat-dependent manner. JLTRG cells were tranduced with either pQCXIP-Tat (+Tat) or empty vector (-Tat) for 48 hours. Cells were treated with CBL0137 (0.5 μM) or DMSO for additional 24 hours. Cell were sorted for GFP-positive population. Data was obtained from three independent experiments and represented as mean ± s.d., *p < 0.05, using student t-test. Not significant (N.S.).