Isoform-Selective Versus Nonselective Histone Deacetylase Inhibitors in HIV Latency Reversal

Abstract

HIV remains incurable due to the persistence of a latent viral reservoir found in HIV-infected cells, primarily resting memory CD4⁺ T cells. Depletion of this reservoir may be the only way to end this deadly epidemic. In latency, the integrated proviral DNA of HIV is transcriptionally silenced partly due to the activity of histone deacetylases (HDACs). One strategy proposed to overcome this challenge is the use of HDAC inhibitors (HDACis) as latency reversal agents to induce viral expression (shock) under the cover of antiretroviral therapy. It is hoped that this will lead to elimination of the reservoir by immunologic and viral cytopathic (kill). However, there are 18 isoforms of HDACs leading to varying selectivity for HDACis. In this study, we review HDACis with emphasis on their selectivity for HIV latency reversal.

Introduction

Since the occurrence of HIV/AIDS in 1981, scientists have worked tirelessly to find a cure for this deadly disease.¹ It was not until 1996 that combination antiretroviral therapy (ART) was introduced, which has led to considerable reduction of morbidity and mortality in HIV-infected individuals.² Even though this marks a great achievement in modern science, it can only control and not cure HIV infection. Combination ART helps reduce plasma HIV RNA to undetectable levels but the virus easily rebinds when treatment is interrupted.² ART is thus limited by lifelong adherence, cumulative side effects, development of drug resistance, and unsustainable costs.^3,4

The major barrier to finding a cure is the ability of HIV to establish an early reservoir of latently infected cells. This latent viral reservoir has been proposed to be made up of different types of cells such as macrophages, follicular dendritic cells, and CD4⁺ subsets of natural killer (NK) cells.^5,6 However, the best-characterized latent reservoirs are the resting memory CD4⁺ T cells.⁶ Resting CD4⁺ T cells harbor proviruses that are silenced through a variety of mechanisms that inhibit replication, therefore making them imperceptible to the host immune response.⁷ Due to this barrier, scientists are looking for strategies to eliminate or incapacitate the latent reservoirs.

One such strategy is the shock and kill; elimination of latently infected cells with the use of latency reversing agents (LRAs).⁸ LRAs work by a variety of mechanisms to stimulate proviruses in latent reservoirs to allow for transcription, translation, and subsequent production of viruses. With ART, these newly expressed HIV will be incapable of reinfection and the latent reservoirs can then be eliminated by immunologic mechanisms or viral cytopathic effect.⁹

The ideal HIV LRA must be able to induce HIV provirus expression without global T cell activation.¹⁰ Several histone deacetylase inhibitors (HDACis) satisfy this condition, making them one of the most studied LRAs.^2,8 Studies have shown that deacetylation of histones by HDACs represses chromatin structure, which plays a vital role in the inhibition of HIV replication.¹¹ By inhibiting HDACs, HDACis promote acetylation and unwrapping of chromatin to induce the expression of HIV.¹²

HDAC Mechanisms and HIV Latency

HDACs (properly called lysine deacetylases) are a class of enzymes that regulate transcription by removing acetyl groups from lysine residues on histones, specifically lysine 9 and 14 from histone 3 (H3K9 or H3K14) and lysine 16 from histone 4 (H4K16).^13,14 This makes transcription factors less accessible to the tightly wrapped DNA.¹⁵ There are four classes (I–IV) based on their size, number of catalytic active sites, cellular localization, and homology to yeast HDAC protein.^16,17 Class I, II, and IV, also referred to as the “classical family,” have zinc-dependent (Zn²⁺) active sites, whereas Class III are structurally different and are nicotinamide adenine dinucleotide (NAD⁺) dependent.¹⁸

Class I HDACs (the most studied in terms of HIV latency) consist of HDACs 1, 2, 3, and 8 and are localized in the nucleus. Class II is grouped into six HDACs, which are further subdivided into Class IIa (HDACs 4, 5, 7, and 9) and Class IIb (HDACs 6 and 10).¹⁹ Class II are mostly localized in the cytoplasm, but shuttle between the cytoplasm and nucleus depending on different signals received.¹⁹ HDAC 11 is the only member of Class IV and it is localized in the nucleus.¹⁹ Class III, also known as sirtuins, has seven members (SIRT 1–7) that have not yet been associated with HIV latency.¹⁶

For latency to be achieved, HIV, like all other retroviruses, must integrate its proviral DNA into the host genome.²⁰ Studies have shown that different transcription factors recruit HDACs to the HIV long terminal repeats (LTR) promoter after integration (Fig. 1), blocking efficient initiation and elongation of transcripts.¹³ Some of these transcription factors include COUP-TF interacting protein (CTIP2), Ying Yang 1 (YY1), late SV40 factor (LSF), Nuclear factor-kappa B (NF-κB p50), activating enhancer binding protein (AP-4), retinoblastoma-family protein 2 (RBF2), c-promoter-binding factor-1 (CBF-1), and specificity protein 1 (Sp1), which are known to primarily recruit HDACs 1 and 2.^2,20

FIG. 1.

Chromatin organization of HIV LTR (promoter) and its association with HDACs and transcription factors in the setting of HIV latency. The HIV promoter is arranged in two nucleosomes (nuc 0 and nuc 1) consisting of two copies each of the core histones H2A (black), H2B (blue), H3 (orange), and H4 (green). The nucleotide +1 denotes transcriptional start site on the viral promoter. Transcription factors like RBF2, p50, CTIP2, Sp1, YY1, AP-4, and LSF recruit HDACs to the HIV LTR.²⁰ These HDACs, in part, promote HIV latency by deacetylating the local histones directly at the provirus integrated site, which in turn brings about repression of chromatin and prevention of RNA polymerase II from binding. HDAC inhibitors block these HDACs and induce histone acetylation, which leads to reactivation of the HIV LTR. AP-4, activating enhancer binding protein; CTIP2, COUP-TF interacting protein; HDACs, histone deacetylases; LSF, late SV40 factor; LTR, long terminal repeats; RBF2, retinoblastoma-family protein 2; YY1, Ying Yang 1.

It is also known that HDAC 3 occupies a position at the HIV promoter and plays a significant role in suppressing transcription.²¹ Also, the HIV LTR has been found to be arranged in two nucleosomes, (nuc 0 and nuc 1), with the nuc 1 present downstream of the promoter, making remodeling of nuc 1 an important step for reactivation from latent HIV (Fig. 1).^2,20 It has been documented that the use of HDACis alone are sufficient to remodel nuc 1.^20,22

HDACis and HIV Reversal

For functionality, HDACis have a cap group, which is used for targeting the surface of HDACs, a linker (aliphatic chain) and functional group that chelates zinc cations.²⁰ Four examples of HDACi family structures are short-chain aliphatic acids (valproic and butyric acids), hydroxamic acids (trichostatin A, vorinostat, givinostat, largazole, and panobinostat), benzamides (entinostat and mocetinostat), and cyclic tetrapeptides and depsipeptides (trapoxin B and romidepsin) (Fig. 2).²⁰ The hydroxamates are known to be the most potent with the short-chain aliphatic acids known to show feeble potency.²³

FIG. 2.

Basic chemical structures of some representative HDAC inhibitors from different families.

Depending on their specificity, HDACis can be termed as selective or nonselective. The selective HDACis can further be divided into class selective or isoform selective based on whether they inhibit different isoforms within the same class or target a single HDAC isoform, respectively.^17,23 HDACis of the hydroxamic acid group such as vorinostat, givinostat, and panobinostat are known to inhibit many of the HDAC class isoforms and have been termed pan-selective or nonselective HDACis because they are nonselective.¹⁸ However, there is evidence that HDACis such as entinostat, mocetinostat, and romidepsin are more selective as they are known to specifically target class I HDACs.^18,20 Another set of class I isoform-selective HDACis are the largazoles SDL256, SDL256, and JMF1080.¹⁶ Tubastatin A and tubacin are typical examples of isoform-selective inhibitors as they act only on HDAC 6.²³

Class I selective HDACis are usually distinguished from the pan-HDACis by their inability to acetylate tubulin (tubulin is only deacetylated by HDAC 6). Hence, to assess the selectivity of HDACis, a Western blot can be used to probe if the inhibitor only promotes acetylation of histone 3 or histone 4 (class I selective) or both histone 3 or 4 and tubulin (Pan HDACi).^16,24,25 It must be noted that HDACi selectivity depends on the concentration of the compound used. Many isoform-selective HDACis become less selective when used at higher concentrations.^16,25,26 Examples of HDACis that have been studied for their effect on HIV infection are shown in Table 1.

Table 1.

List of Some Histone Deacetylase Inhibitors Used to Induce HIV from Latency

HDAC inhibitor	HDAC selectivity	Assay	Refs.
Valproic acids	I, IIa	In vitro	21,27–32
Butyric acids	I	In vitro	2,20
Trichostatin A	I, IIa, IIb	In vitro	20,27,29,31
Vorinostat	I, IIa, IIb	In vitro, in vivo	21,27–29,31,33–35
Givinostat	I, IIa, IIb	In vitro, in vivo	36,37
Panobinostat	I, IIa, IIb	In vitro, in vivo	2,38
Entinostat	HDAC 1, 2, 3	In vitro, in vivo	2,31,33
Mocetinostat	I	In vitro	39
Trapoxin B	HDAC 1, HDAC 6	In vitro	20
Romidepsin	HDAC 1, 2	In vitro, in vivo	40
SDL 148	I	In vitro	16
JMF 1080	I	In vitro	16
SDL 256	I	In vitro	16

HDAC, histone deacetylase.

HDACis in HIV Clinical Trials

Most HDACis have been analyzed in vitro for their ability to reactivate HIV from latency, but very few have been tested in vivo. However, the use of HDACis in vivo is not novel, as HDACs are implicated in other diseases such as cancer. Intensive research has already been done on the pharmacological and toxicological properties of multiple HDACis as anticancer drugs.^41–43 Thus, HDACis are one of the advanced latency reversal agents in terms of clinical studies, examples of which are shown in Table 2.

Table 2.

Summary of Histone Deacetylase Inhibitors in Completed and Ongoing HIV Clinical Trials

HDAC inhibitor	Regimen	Phase	No. of participants	Outcome	Study identifier	Year completed	Refs.
Vorinostat	400 mg; single dose and 10 doses given at 72-h intervals	I/II	16	Increase in cell-associated HIV-RNA within circulating resting CD4+ T cells, but no measurable decrease in frequency of latent infection within resting CD4+ T cells	NCT01319383	2016	44
Romidepsin	0.5, 2, 5 mg/m2 (single dose) or 5 mg (multidose on days 0, 14, 28, and 48)	I/II	59	No significant increase in plasma viremia or cell-associated unspliced HIV-1 RNA	NCT01933594	2018	45
Panobinostat	20 mg; 3 times a week for 8 weeks	I/II	15	Cell-associated unspliced HIV-1 RNA increased significantly at time points where panobinostat were taken	NCT01680094	2014	46
Panobinostat+IFN-alpha 2a	15 mg; on days 0, 2, and 4 of treatment week	I/II	34	Ongoing	NCT02471430	—	47
Chidamide	10 mg; eight oral doses on days 0, 3, 7, 10, 14, 17, 21, and 24	I/II	7	Robust and repeated plasma viral rebound with increased cell associated HIV-1 RNA during treatment and reduction in cell-associated HIV-1 DNA	NCT02513901	2016	48

Are Isoform-Selective Compounds Better Reversal Agents?

Owing to their nonselectivity, pan-HDACis have been shown to come with a degree of toxicity as they interfere with many cellular processes.^16,24,25 In cancer treatment, such side effects include dehydration, thrombocytopenia, diarrhea, fatigue, anorexia, and nausea, among others.^49,50 Due to this shortcoming, the use of more selective HDACis, especially for class I HDACs, has been proposed to reduce off-target effects.^24,25,33,51 However, the question is whether class or isoform-selective compounds will have the same potency as the pan-HDACis. We and others have demonstrated that class I selective inhibitors could be more potent than vorinostat in reactivating HIV in cell lines and primary cell latency models.^16,24,25 However, of the class I selective compounds, only romidepsin has entered clinical studies.⁵²

Although the search for more selective HDACis is of importance to prevent untoward side effects, there is evidence that being too selective may not bring about significant viral reactivation. Barton et al. showed that inhibitors that were isoform selective for either HDAC 1 or 2 alone did not activate latent HIV.¹³ Their data suggested that HDAC 3 was vital for reactivation of quiescent HIV proviruses.¹³ It was also shown that concurrent knockdown of HDAC 1 or 2 with HDAC 3 did not increase expression from the HIV-1 LTR any further than the depletion of HDAC 3 alone. However, it was the combined knockdown of HDACs 1, 2, and 3 that showed the most expression.^13,52 This shows that in terms of selectivity, a class I selective HDACi will be more appropriate than an isoform-selective HDACi for HIV reactivation studies.

So far, studies show that HDACis are not enough to reduce the viral reservoir size when tested in vivo.^16,24,25 Single-cell RNA sequencing and dual-label reporter virus assays show that a minority of latently infected CD⁺ T cells respond to LRAs, including HDACis, likely due to heterogeneity in chromatin organization at different integration sites.^53–56 These data should be taken in the backdrop of the fact that only a handful of HDACis have been tested in vivo. Indeed, all the ones tested were not specifically developed for HIV, but for cancer and other indications.

With in vitro and ex vivo data showing that class I selective HDACis could be more potent LRAs, several groups have been encouraged to develop new HDACis for HIV latency reversal.^57,58 It will be important to advance these compounds through the pipeline to clinical trials. One such compound developed by Merck can produce significant amounts of p24 in resting CD4⁺ T cells isolated from aviremic patients.⁵⁹ Another HDACi, Chidamide, shows strong promise and is currently in clinical trials.⁴⁸

However, given the characteristics of the integrated provirus, successful reactivation of a majority of the latent HIV is likely to come from combinations of different LRAs, including selective HDACis. As such, combination therapy of HDACis and protein kinase C (PKC) agonist such as bryostatin-1 has been proposed. This combination helps lower the concentration of individual agents to clinically acceptable levels, while maintaining robust latency reversal. These studies highlighted the importance of Class I selective HDACis and not pan-HDACis for combination therapy.^16,24,25 In one of these studies, it was shown that, entinostat (Class I HDACi) in combination with bryostatin-1 induced maximal viral outgrowth compared to its pan-HDACi counterparts.²⁴ However, when the pan-HDACis were introduced in that combination, the viral outgrowth seen resembled that of pan-HDACis alone, suggesting that pan-HDACis have the tendency to inhibit robust HIV gene expression.

Probing this inhibitory effect revealed that Pan HDACis like vorinostat inhibit NF-κB and Hsp90, two important HIV-1 transcription factors.^24,25 NF-κB is required for bryostatin-1-induced cell activation, while the activity of Hsp90 is promoted by HDAC 6. Hence, pan-HDACis capable of inhibiting HDAC 6 will acetylate Hsp90 and inhibit its function.²⁴ This makes it imperative to try combinations of class I selective HDACis and PKC analogs. Although PKC agonists are the most potent LRAs ex vivo and in mouse studies, the only human trial done so far failed to increase viral RNA in the serum.^60,61 The doses of bryostatin-1 used in this trial were very conservative for fear of side effects from proinflammatory cytokine release and could account for those results.

In this regard, selective HDACis could come to the rescue as modulators of inflammation caused by the PKC agonists like brostatin-1 and ingenols. Recently, Spivak and colleagues showed that a class I HDACi, suberohydroxamic acid (SBHA), did not reactivate HIV in CD4⁺ T cells isolated from aviremic patients, but could prevent the release of proinflammatory cytokines TNF-α, IL-5, IL-2r, and IL-17.⁶² In addition, entinostat reduced production of proinflammatory cytokines in animal model studies.^63,64

Therefore, it is imperative to try different combinations of PKC analogs and selective HDACis to determine a combination that could synergize in HIV latency reversal and simultaneously reduce cytokine release. For instance, one study showed that prostratin could enhance NK cell killing of CD4⁺ T cells with reactivated HIV. Studies have now demonstrated that PKC analogs, HDACis and bromodomain inhibitors, each tend to reactivate HIV proviruses from distinct unique loci of the chromatin.⁵⁴ While these findings affirm the belief that a singly LRA is unlikely to result in significant latency reversal, it provides hope that a carefully designed cocktail of LRAs could make the shock and kill approach successful.

Perspectives and Conclusion

In light of the shock and kill approach, HDACis have so far not produced encouraging results in vivo. This is because, several mechanisms contribute to the establishment and maintenance of latency. Also, the heterogeneous nature of the viral reservoir as well as the integration site of the provirus have contributed to this limited success seen in HDACis.^5,52 This calls for a better understanding of the silencing mechanisms underlining HIV latency.

While it is important for LRAs such as HDACis to be potent enough to reactivate HIV, it will be even better if they could eliminate the reservoir. This can only be achieved when adequate amounts of viral proteins are expressed within the infected cells. Even after significant amounts of viral proteins are made in latently infected CD4⁺ T cells, several obstacles to their killing remain. First, the HIV regulatory proteins (nef) downregulate expression of MHC class I and limits the presentation of peptides to CD8⁺ T cells. Hence, the most potent HDACi might interfere with reservoir clearance through cytotoxic T cell-mediated killing. Second, latently infected cells have high levels of the antiapoptotic protein Bcl-2. Third, some HDACis may inhibit the function of cytotoxic CD8⁺ T cells.

Finally, chronic T cell exhaustion and immune escape mutations in chronically infected patients also present formidable obstacles.⁶⁵ To overcome these inherent challenges to killing, studies that seek to combine HDACis with therapeutic vaccines, interferon, toll-like receptor agonists, broadly neutralizing antibodies, and proapoptotic compounds have to be intensified.⁶⁶

Due to the heterogeneity of the viral reservoir, it will be of interest to investigate latency reversal agents that can affect the different reservoir types, despite their physiological differences. Also, most of these HIV reservoirs are known to be found in sanctuary sites. Hence, effective latency reversal agents have to be able to penetrate these sites and optimal drug delivery approaches should be considered for this.

Finally, it has been shown that isoform-selective HDACis are better reversal agents than their nonselective counterparts. However, it is becoming clear that the use of a single agent, such as HDACis alone, may not be enough to truly shock and clear the reservoir. Therefore, further experimental and clinical studies are required to assess the effectiveness of new compounds, with different mechanisms of action, and combinations of compounds to reverse HIV latency and eliminate the reservoir.

Authors' Contributions

A.T.B., E.Y.B., and G.B.K. conceived and designed the concepts of the review, and A.A.-M. assisted with the literature search and article writing. All authors were involved in writing and reviewing the article.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This article is part of the EDCTP2 program supported by the European Union (Grant No. TMA2017SF-1955, H-CRIS) to G.B.K. The funder had no role in this publication.