Have HDAC inhibitors fallen from grace?

Two recent papers present conflicting data on the efficacy of histone deacteylase (HDAC) inhibitors, the leading class of latency reversing agents (LRAs).

Replication-competent integrated HIV provirus is able to form a persist reservoir in cellular compartments such as resting memory CD4 T cells by remaining in a long-term latent state that is highly stable.  Activation of latent integrated HIV occurs when host CD4 T cells are activated by normal immunological signals, affecting only a miniscule fraction of the persistent HIV reservoir at any one time, and giving rise to a negligible reservoir decay rate. Thus the latent reservoir presents the single greatest obstacle to treatment-induced viral eradication and the cure of HIV.

Previously, strategies were designed that involved global activation of T cells, as a means of flushing out latent HIV. But these proved to be too toxic for continued investigation. As an alternative, a number of compounds are now being investigated for their ability to induce latent HIV without causing T cell activation. These are known as latency reversing agents (LRAs) and they include agents with a number of different mechanisms of action. One of the most promising classes of LRAs is the HDAC inhibitors, initially developed as anti-cancer drugs. Histone deacetylases (HDACs) are enzymes that get recruited to gene promotors by transcriptional regulators, where they deactylate lysine residues on histones. As a result, chromatin condensation is induced, which packages up DNA into a quiescent “storage” state and represses gene expression. HDACs are thought to be involved in the active repression of HIV expression during viral latency. In vitro tests show that HDAC inhibitors can induce HIV expression from lab-adapted T cells that have been engineered to harbour latent virus. Clinical trials have also shown promising data suggesting that when used in people, the HDAC inhibitor vorinostat can increase HIV RNA expression in resting memory CD4 T cells by an average of 4.8 fold [1].

Despite these promising findings, it is not yet known if this induction of latent HIV leads to the production of mature virions and the death of infected host cells, which might translate into an increased decay-rate of the latent-reservoir. One key concern is whether HDAC inhibitors can even induce expression of latent virus from the resting memory CD4 T cells of HIV-infected patients who are on stable ART. Therefore Korin Bullen, from Robert Siliciano’s group at Johns Hopkins School of Medicine in Baltimore, USA and Datsen Wei, from Gilead and working with John Mellor’s group at the University of Pittsburgh, tested a panel of HDAC inhibitors and other LRAs for their ability to induce latent HIV in patient-derived CD4 T cells (none of the agents tested were Gilead products). The two groups report contradictory findings.

Bullen et al, describe their results in the March 2014 edition of Nature Medicine [2]. Using clinically relevant concentrations of each agent, they assessed the HDAC inhibitors vorinostat, romidepsin and panobinostat, as well as agents from other classes of LRAs: disulfiram, thought to involve NF-kB; JQ1, a bromodomain-containing protein inhibitor and; bryostatin-1, a protein kinase C (PKC) agonist (PKC is a signalling component in T cell activation).  In all experiments PMA plus ionomycin (PMA/I) were used as a positive control, as these agents induce global T cell activation. In order to allow reactivated HIV to sufficiently grow in cultures of patient CD4 T cells (and to maximise viral signal), they first stimulated resting CD4 T cells from 13 ART-treated patients with the panel of LRAs for 18 hours and then added lab-grown T cells to the culture for a further 14 days to allow viral propagation. In previous versions of this assay, the use of T cell blasts for propogation may have caused mixed lymphocyte reactions leading to background T cell activation, which could have caused a false positive signal for viral induction by LRAs. To prevent this they used the lab-adapted T cell line, MOLT-4. Surprisingly, using this new assay, they were not able to measure any induction of latent HIV by p24 ELISA of culture supernatants, following stimulation with HDAC inhibitors. These results were confirmed by assessment of supernatant HIV mRNA in five of these patients, for which only one demonstrated induced virus, and this was in response to the PKC agonist, bryostatin-1. In contrast PMA/I induced HIV p24 expression in CD4 T cells of 11 of 13 patients.

Most PCR assays that laboratories use to detect HIV mRNA are targeted to the gag gene or the joining region of the gag and LTR genes (which are proximal to each other).  Bullen and colleagues were concerned that the sequences detected by these assays do not represent ‘bona fide’ HIV RNA, as ‘HIV integrates into host genes that are actively transcribed in resting CD4 T cells, allowing for the production of chimeric host-HIV transcripts’. These viral transcripts will not be accurate or complete and are not likely to lead to production of functional virus. These host gene-initiated HIV transcripts would contain the HIV gag sequence and so would be indistinguishable from HIV-LTR initiated transcripts when using conventional HIV PCR assays. In order to address this concern, Bullen and colleagues designed a PCR assay to detect a portion of the HIV LTR gene that should not be present in LTR-initiated transcripts, but would be present in host gene-initiated HIV ‘readthroughs’. Using their new PCR assay they treated resting CD4 T cells from five HIV infected patients receiving ART, with vorinostat or PMA/I, and determined that while vorinostat did not induce complete HIV RNA, it did induce a small increase (two fold) in host-initiated ‘readthrough’ transcripts, which were comparable to the induction of conventional PCR gag-containing transcripts. The authors conclude “although not every potential LRA will induce readthrough transcription by activating a host gene, our data show that chimeric host-HIV transcripts can have a confounding effect on the RT-qPCR signal obtained using standard gag primers. Such an effect should be taken into consideration when evaluating LRAs using conventional gag RT-qPCR assays.”

The implication of these results are two-fold. Firstly, HDAC-inhibitors may not be effective at inducing latent HIV in vivo. Secondly, the discrepancy in the effect of LRAs on patients’ T cells and lab-models of latency, suggest that the mechanisms by which HIV latency naturally occurs, are not captured by laboratory latency models.

In contrast to these results, Wei at al describe their findings in the April 2014 edition of PLoS Pathogens [3], in which they found that some HDAC inhibitors were able to induce expression of latent HIV from the CD4 T cells of patients receiving ART. Using an in vitro latency model, with freshly isolated CD4 T cells from healthy donors, they screened the activity of the clinically tested HDAC inhibitors: vorinostat, romidepsin, panobinostat, givinostat, mocetinostat, and pracinostat. Wei et al found that all HDAC inhibitors showed dose-dependent increases in activity, with superior activity for romidepsin. Wei et al then assessed the ability of each of the HDAC inhibitors to induce expression of HIV from purified resting CD4 T cells of HIV infected patients receiving ART. Unlike the viral outgrowth approach taken by Bullen et al, described above, Wei and colleagues treated purified resting CD4 T cells from patients with clinically relevant doses of vorinostat or romidepsin and then assessed HIV RNA expression by PCR with the commercial Roche COBAS HIV test after 6, 12, 24 and 48 hours of culture. This assay would therefore not be so dependent on viral propogation following production of replication-competent infectious virions over two weeks of culture. This is a key difference with the Bullen approach. Using this method Wei et al found that vorinostat induced 2- to 4- fold increases in HIV RNA expression, which peaked at 6 hours and the more potent romidepsin induced 5- to 6- fold increases in HIV RNA expression that peaked after 24 – 48 hours.

In order to demonstrate that extracellular virions were produced from resting CD4 T cells treated with vorinostat or romidepsin, supernatants were tested for HIV RNA after 6 days culture. HIV RNA could be detected in the supernatants of romidepsin-treated cells, but not vorinostat-treated cells. The authors add that HIV RNA released into cell culture supernatants following treatment with romidepsin, could be pelleted by high-speed centrifugation, suggesting that the RNA must be encapsulated in virions, rather than having been released during cell death potentially caused by HDAC inhibitor toxicity.

The authors conclude that “We observed reproducible ex vivo activation of HIV by romidepsin… Given these results and the established clinical safety profile of romidepsin, clinical testing is warranted to assess whether romidepsin can activate latent HIV and potentially reduce the size of the latent reservoir in HIV-infected patients on suppressive combination ART.”

There are two key differences in the approaches taken by the Bullen and Wei groups that could go some way to explaining the discrepancy in their results. First of all Bullen used a viral outgrowth assay that depended on propogation of infectious virus (over 14 days), whereas Wei measured shorter-term viral RNA expression (24 hours – 7 days) that may have represented a partial phase of the viral life cycle. The former method used by Bullen is likely to be more sensitive to the production of replication-competent infectious virus, yet an effect of HDAC inhibitors on viral induction was only found by Wei et al. While there were some differences in the assays used by both groups, the inability of the more robust method used by Bullen to detect HIV activation would seem to suggest that the viral expression Wei detected may not have represented replication competent infectious virus. This is important because without production of mature virions, it is much less likely that cell death would result, which would lead to a corresponding reduction in the size of the reservoir.

The second important difference between the approaches taken by both groups was Bullen’s development of a PCR assay to measure host-initiated HIV ‘readthrough’ transcripts. Conventional HIV PCR assays that target gag and LTR genes are likely to measure viral transcripts that have been initiated by the activation of host genes. These will not reflect complete and properly spliced viral transcripts, from which functional virions can be produced. This is another very important difference between the approach taken by Bullen et al and that of Wei et al. Wei’s detection of an effect of HDAC inhibitors, relies on the commercial COBAS HIV PCR assay produced by Roche Diagnostics. According to Roche [4], this assay uses primers for gag and LTR that would also, presumably, measure incomplete HIV transcripts initiated by host genes. Given that the mechanism of action of HDAC inhibitors is to “release” genes from a “silent” state, enabling their expression, HDAC inhibitors will activate host genes. Therefore HDAC-inhibitors are liable to initiate HIV transcription by activating the promotors of upstream host genes, into which HIV has integrated. The implication of this is that Wei’s detection of HIV mRNA sequences in HDAC-treated cells could conceivably be a “false” signal.

Wei et al say that HIV virions from the supernatants of HDAC-inhibitor treated CD4 cells can be pelleted by high-speed centrifugation, as further support for the claim that supernatant HIV RNA must be encapsulated in released virions (rather than free RNA released from dying cells). However, host cell-derived extracellular RNA occurs in normal biology, in the form of extracellular vesicles or exosomes. These can be the same size and density as HIV virions and so very hard to distinguish by centrifugation. If the centrifuged particles detected by Wei, are extracellular vesicles or cellular debris containing host-initiated viral “read-through” RNA, it is possible that Wei’s data does not confirm viral induction by HDAC inhibitors and Bullen’s conclusion that HDAC inhibitors do not induce reactivation of latent HIV may stand. These questions could be resolved by the use of Bullen’s “readthrough” PCR assay, to measure the RNA in the cells and supernatants that Wei et al have generated.


1)   Archin NM, Liberty AL, Kashuba AD et al. Adminsitration of vorinostat disrupts HIV-1 latency in patients on antiretroviral therapy. Nature. 2012. 487. 482-485.

2)   Bullen CK, Laird GM, Durand CM et al. New ex vivo approach to distinguish effective and ineffective single agents for reversing HIV-1 latency in vivo. Nature Medicine. 2014.

3)   Wei DG, Chiang V, Fyne E et al. Histone Deacetylase inhibitor Romidepsin induces HIV expression in CD4 T cells from patients on suppressive antiretroviral therapy at concentrations achieved by clinical dosing. PLoS Pathogens. 2014. April. Volume 10. Issue 4.

4)   http://molecular.roche.com/assays/Pages/COBASAmpliPrepCOBASTaqManHIV-1Testv20.aspx

A shortened version of this article was first published in HIV Treatment Bulletin May/June 2014 (http://i-base.info/htb/25840)

Leave a Comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.