The potential role for romidepsin as a component in early HIV-1 curative efforts

KEYWORDS:

• HIV persistence
• HIV cure
• Immunotherapy
• Romidepsin
• Histone deacetylase inhibitors

Reversal of latency [1] has been proposed as a first step toward curing HIV-1 infection. By reversing latency, it is believed that cells carrying replication-competent proviruses will be ‘exposed’ to immune-mediated killing and/or apoptosis. This concept has been described as the ‘shock and kill’ approach [2] in which the aim is (1) to activate HIV-1 transcription in latently infected cells using latency reversing agents (LRAs) while preventing new infections by antiretroviral therapy (ART) and (2) to boost the killing of HIV-1-expressing cells either by viral cytopathic and/or by immune-mediated effects. Current immunotherapies are incapable of inducing long-term viral suppression in HIV-1-infected individuals in the absence of ART [reviewed in 3]. LRA is a conceptual term and it should be noted that the in vivo effect of LRAs on infected cells might either be due to augmentation of ongoing viral transcription and/or induction of de novo viral transcription. Regardless, potent latency reversal is vital for the success of ‘shock and kill’ approach; here, we discuss the potential role for the LRA romidepsin as a component in HIV-1 curative efforts.

Identification of LRAs

Over the past decade, extensive in vitro and ex vivo experiments have demonstrated disruption of HIV-1 latency through many different mechanisms using small molecules such as histone deacetylase inhibitors (HDACi), PKC activators, bromodomain inhibitors, and SMAC mimetics (SMACm) (illustrated in Figure 1) [reviewed in 4–6]. The most intensely studied LRAs are HDACi. HDACi counteract HDAC and prevent the removal of acetyl groups from lysine residues causing the histone tails to become more positively charged. This in turn induces chromatin decondensation and gene transcription. HDACi also affect non-histone proteins and cellular processes [7]. There are four classes of HDACs, and class I HDAC are thought to be essential for maintaining HIV-1 latency [8].

Figure 1. Reversal of HIV latency (graphics: Ken Kragsfeldt. Modified from Richman DD. et al. Science 2009 [4] with permission from Richman DD.). Latency reversing agents (LRAs) can disrupt HIV-1 latency through different mechanisms: Protein kinase C (PKC) activators stimulates via kinase signaling and the PKC pathway. Second mitochondria-derived activator of caspases mimetics (SMACm) stimulates the NF-κB pathway. DNA methyltransferase inhibitors (DNMTi) decrease methylation of the HIV-1 promoter in the long terminal repeat (LTR). Histone deacetylase (HDACi) and methyltransferase inhibitors (HMTi) prevent removal of acetyl groups and attachment of methyl groups to the histone, respectively. Bromodomain inhibitors (BRDi) might favor competitive binding of the viral transactivation protein (Tat) to the positive transcription elongation factor b (p-TEFb) complex and/or affect transcription factors. Hexamethylbisacetamide (HMBA) stimulates p-TEFb in the absence of Tat. Hydroxybenzotrialzole and interleukin-7 (IL-7) stimulates the JAK/STAT pathway. Toll-like receptor (TLR) 7/9 agonists stimulate the plasmacytoid dendritic cells (pDC) to produce cytokines that activates the T cell and thus HIV-1 transcription.

Characteristics of romidepsin

The HDACi romidepsin (Istodax^®, Celgene) is a cyclic depsipeptide naturally produced by Chromobacterium violaceum [9]. After intracellular cleavage of a disulfide bond, free sulfhydryl groups on romidepsin interact with ionized zinc in the active site of HDACs [10]. Romidepsin is a pan-HDACi that inhibits class I HDACs at low nM concentrations. Romidepsin is FDA-approved for use in the treatment of cutaneous and peripheral T-cell lymphomas [9]. In this population, the recommended dosing of romidepsin is 14 mg/m² administered intravenously (IV) on days 1, 8, and 15 of a 28-day cycle. Repeat cycles are recommended if the patient continues to benefit from and tolerates the drug. In preclinical experiments, romidepsin (at ~35–40% of the clinical oncology dosing) induced higher levels of HIV-1 expression than any other HDACi in clinical development. This led to virion release from isolated memory and resting CD4+ T cells from individuals on long-term ART [10]. Of note, romidepsin also induced modest dose-dependent CD69 expression on T and B cells overall, but without inducing global activation or affecting cytokine levels [10]. Collectively, these preclinical results prompted our and other groups to design clinical trials aimed at testing the effect of romidepsin on HIV-1 transcription in ART-treated individuals.

The effect of romidepsin on HIV-1 transcription in vivo

In a proof-of-concept phase Ib/IIa trial, we assigned six aviremic HIV-1-infected adults to receive romidepsin 5 mg/m² (~36% of the dosing used in oncology) IV once weekly for 3 weeks while maintaining ART [11]. HIV-1 transcription quantified as copies of cell-associated unspliced HIV-1 RNA increased significantly from baseline during romidepsin treatment (range of fold-increase: 2.4–5.0). Coinciding with elevations in proviral transcription, plasma HIV-1 RNA increased from <20 copies/mL at baseline to readily quantifiable levels at multiple post-infusion time-points in five of six individuals (range 46–119 copies/mL). The cyclic appearance of the observed increases in cell-associated HIV-1 transcription and plasma RNA levels strongly suggested an immediate but transient impact of romidepsin on the epigenetic regulation of HIV transcription. The impact on HIV transcription seemed to increase with repeated romidepsin exposure. While the effect on viral transcription was pronounced, the size of the HIV-1 reservoir remained unaffected by romidepsin. This finding suggests either insufficient magnitude of latency reversal and/or an impaired killing of HIV-1-expressing cells.

To address whether romidepsin in itself disrupted antiviral immunity, we characterized non- and HIV-1-specific cell-mediated immunity and found both to be unaffected. In addition, CD4+ and CD8+ activation increased moderately, while expression of the exhaustion marker PD-1 decreased. These were key observations since previous ex vivo experiments suggest that romidepsin suppresses IFN-γ production and the killing capacity of HIV-1-specific CD8+ T cells [12]. Currently, the AIDS Clinical Trial Group is investigating the effects of ascending romidepsin doses on HIV-1 transcription (www.clinicaltrials.gov, NCT01933594), which is important for optimizing the romidepsin dose as well as for confirming the results observed in the above-mentioned study.

In a non-human primate model (ART-suppressed simian immunodeficiency virus (SIV)mac239-infected Indian-origin rhesus macaques), romidepsin 2.5 mg/m² and 3.75 mg/m² induced a cyclic appearance of histone acetylation in CD4+ T cells and increased plasma SIV-RNA levels [13]. The authors note that plasma SIV-RNA was not contaminated with SIV-DNA (due to cell toxicity), and time to viral rebound was unaffected during ART interruption.

Clinical trials using vorinostat (another HDACi) among ART-treated individuals showed up to fivefold increased HIV-1 transcription following a single dose and daily dosing for 14 days but no increase in plasma HIV-1 RNA was observed [14,15]. A trial investigating panobinostat showed 54% positivity for HIV-1 RNA from samples taken during treatment compared to 30% positivity pretreatment using a nonquantitative transcription-mediated amplification assay [16]. After panobinostat treatment, a subset of four patients showed an association between increased innate immune activity and pronounced reduction in HIV-1 DNA levels [17]. Interestingly, both romidepsin and panobinostat increased T-cell activation in contrast to vorinostat. More trials using different HDACi alone or in combination are underway [6].

Safety of romidepsin in HIV-1-infected individuals

Due to its broad effects on gene regulation, romidepsin treatment is often associated with varying degrees of side effects. In cancer patients, common side effects comprise gastrointestinal (nausea, vomiting, diarrhea, and constipation), hematologic (thrombocytopenia, leukopenia, and anemia), and asthenic (asthenia and fatigue) conditions. To a large extent, these side effects are dose-dependent [9], and at reduced dosing of 5 mg/m², the tolerability of romidepsin has been acceptable among HIV-1-infected individuals. Here, the most common side effects were self-limiting grade 1–2 gastrointestinal symptoms and fatigue [11]. To reduce gastrointestinal discomfort, prophylactic antiemetics were administered prior to romidepsin infusions. In cancer trials, a nonclinically significant, reversible prolongation of corrected QT interval was observed after romidepsin infusion, but without sustained effect on cardiac function [reviewed in 9]. In 17 HIV-1-infected individuals, intensive cardiac monitoring during romidepsin infusions showed no clinically significant abnormalities (http://clinicaltrials.gov, NTC02092116, personal communication). Other potential romidepsin-related side effects include increased risk of infection [12], DNA virus reactivation (described in three cancer patients [reviewed in 18]), and fetal toxicity as well as reduced fertility (observed in animal repeat-dose toxicity studies). Importantly, human endogenous retrovirus expression is not induced by treatment with HDACi in cellular models of HIV-1 latency [19]. Although none of the HIV-1-infected individuals experienced romidepsin-related serious adverse events, careful safety monitoring is required, as the use of romidepsin will expand in HIV-1 cure research in the coming years.

Combinatorial approach

While romidepsin and other LRAs can increase HIV-1 transcription in humans, the magnitude and duration of latency reversal may need to be extended to deplete the reservoir. Whether this is achievable with romidepsin alone is currently unknown. An alternative to increasing the dosing of romidepsin or other LRAs is to combine LRAs targeting HIV-1 latency through different mechanisms. Ex vivo, dual LRA administration is more effective in reversing latency than any LRA administered alone [20]. In addition, targeted drug delivery (e.g. by conjugation to polymers) or increasing the half-life of LRAs may be other ways to enhance their effect in vivo.

Another issue is whether or not reactivated infected cells are eliminated following latency reversal. While active HIV-expressing cells are more likely to undergo apoptosis than non-HIV-expressing cells after HDACi treatment [7], potent HIV-1-specific CD8+ T-cell responses are often needed to eliminate HIV-1-expressing cells after reversal of latency. This was elegantly shown in an experimental study by Shan et al. [21], where priming of HIV-1-specific CD8+ cells was needed to achieve efficient killing of HIV-1-expressing cells.

Thus, combinatorial approaches should be pursued (1) to maximize the proportion of cells reactivated by LRAs and/or (2) to boost the killing of latently infected cells following latency reversal using immunotherapies.

In fact, the first combinatorial clinical trial has recently been completed in chronic HIV-1-infected individuals on ART using romidepsin and Vacc-4x/GM-CSF, a therapeutic HIV-1 vaccination (http://clinicaltrials.gov, NTC02092116). In an extension of an ongoing trial, HIV-1-infected individuals will receive a booster with a MVA.HIVconsv vaccine in combination with romidepsin (http://clinicaltrials.gov, NCT02616874). Moreover, other planned trials using romidepsin in combination with immunotherapy are reportedly underway.

Using romidepsin during ART initiation to limit the size of the reservoir

The HIV-1 reservoir is established within days following primary infection [reviewed in 22]. During untreated infection, newly evolved viruses are continuously integrated and archived as proviruses in CD4+ T cells. Starting ART early during acute phase of the infection diminishes the size of the HIV-1 reservoir [reviewed in 22], and once ART is initiated, viral evolution stops [23]. Following analytical treatment interruptions, however, the first ‘rebounding’ viruses to emerge are often identical to the plasma viruses that were present immediately before ART initiation [reviewed in 24]. This suggests that ART ‘freezes’ the circulating viral population present at treatment initiation and that these latest proviruses may be more important for viral persistence than proviruses from earlier time points [23]. In addition, the HIV-1 reservoir at ART initiation may be more labile and susceptible to interventions due to virus-driven immune activation and natural high turnover rate of the HIV-1 pool [25]. Thus, using romidepsin to target the viral reservoir at ART initiation may be a novel opportunity in HIV-1 research [26]. Rather than waiting for inflammation and the reservoir to settle, it might be a possibility to move ‘shock and kill’ to the time of ART initiation – a concept of early latency reversal. Interestingly, romidepsin inhibited de novo HIV-1 infection in two CD4+ T-cell models [27]. If romidepsin inhibits de novo infection of uninfected CD4+ T cells in vivo, this might act in synergy with ART to prevent viral spreading during latency reversal.

Potential limitations of using romidepsin in HIV-1 infection

While it seems unlikely that romidepsin alone is sufficient to cure HIV-1 infection, some other potential limitations to using romidepsin and other LRAs should also be mentioned. The penetration of LRAs into immune privileged sites (e.g. the brain and genitals) as well as lymphoid tissue may be reduced. For instance, the concentration of romidepsin in the cerebrospinal fluid is estimated to be about 2% of that in plasma [28]. This could potentially be overcome by increasing the dosing of romidepsin, but dose-escalation will also likely result in higher toxicity. Moreover, large interpersonal differences in reservoir size and response to LRA treatment may complicate the interpretation of study results [10,29]. Addressing these temporal (order and duration) and spatial (compartments) administration issues is a key to advancing the HIV-1 cure field [13]. Finally, it should be noted that the current knowledge about romidepsin among HIV-1-infected individuals is still sparse and more data are needed to confirm its safety and effect on HIV-1 latency.

Conclusion

To date, romidepsin is the most potent LRA tested in clinical trials, and thus a ‘prototype’ LRA that can be used both in combinatorial studies with immunotherapy and for benchmarking the potency of new LRAs. Romidepsin may not be the perfect LRA in terms of cost, handling, and side effects – but at these early stages of HIV-1 cure research, romidepsin shows superior effect compared to other LRAs. While moving forward with romidepsin, there should be a continued focus on identifying new and more potent LRAs that are easier to administer and have fewer side effects.

Declaration of interest

This paper has been supported by funding from Lundbeckfonden (R126-2012-12588) and The Danish Research Council (12-133887). The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.