Getting the “kill” into “shock and kill”: strategies to eliminate latent HIV

Abstract

Despite the success of antiretroviral therapy (ART), there is currently no HIV cure and treatment is lifelong. HIV persists during ART due to long lived and proliferating latently infected CD4+ T-cells. One strategy to eliminate latency is to activate virus production using latency reversing agents (LRAs) with the goal of triggering cell death through virus-induced cytolysis or immune-mediated clearance. However, multiple studies have demonstrated that activation of viral transcription alone is insufficient to induce cell death and some LRAs may counteract cell death by promoting cell survival. Here, we review new approaches to induce death of latently infected cells through apoptosis and inhibition of pathways critical for cell survival, which are often hijacked by HIV proteins. Given advances in the commercial development of compounds that induce apoptosis in cancer chemotherapy, these agents could move rapidly into clinical trials, either alone or in combination with LRAs, to eliminate latent HIV infection.

Introduction

Antiretroviral therapy (ART) effectively suppresses HIV replication, but treatment is life long and there is no HIV cure (Chun et al, 1997a; Chun et al, 1999; Finzi et al, 1997; Rathbun et al, 2006). The major obstacle that impedes HIV eradication in ART-treated individuals is the establishment of long lived and proliferating, latently infected cells (Finzi et al, 1999; Pierson et al, 2000; Siliciano et al, 2003). HIV latency is defined as the persistence of integrated HIV DNA (Koup et al, 1994; Pedersen et al, 1989) that is transcriptionally silent but replication competent (Chun et al, 1995; Chun et al, 1997b). Because there is little or no expression of viral proteins, latently infected cells escape immune recognition and clearance. Latently infected cells may also have a potential survival advantage, but this mechanism remains unknown.

CD4+ T cells are the main target of HIV infection and are thought to be the predominant cell type harbouring latent virus. Macrophages are also permissive to HIV infection in vitro and infected macrophages are clearly detected in HIV-infected individuals but whether HIV-infected macrophages persist on ART in vivo remains controversial (Clayton et al, 2017). Additionally, other cell types have been shown to harbour latent provirus, although their contribution to the long term persistence and viral rebound off ART remains less well understood. These include naïve CD4+ T-cells (Wightman et al, 2010), γδ T-cells (Soriano-Sarabia et al, 2013) and hematopoietic progenitor cells (Carter et al, 2010).

Within these infected cells, multiple mechanisms maintain HIV latency (Mbonye & Karn, 2017). These include: the site of proviral DNA integration, low levels of cellular transcription factors in a resting cell that are needed for virus production, epigenetic modifications such as histone acetylation/methylation and chromatin remodelling, impaired RNA splicing and nuclear export, and impaired translation (reviewed in (Ruelas & Greens, 2014)). One strategy to eliminate latently infected cells is to activate virus transcription, protein expression and virion production using latency reversing agents (LRA), potentially triggering cytolysis or immune-mediated clearance (Deeks, 2012). This approach is termed “shock and kill” (Deeks, 2012) and is performed in individuals on ART to prevent new rounds of infection (Figure 1).

Figure 1. Shock and Kill strategy to eliminate HIV latently infected cells.

The “shock and kill” strategy uses latency reversing agents (LRAs) to increase HIV transcription, protein expression and virion production. The cell may potentially die through virus-mediated cytopathic events or immune-mediated clearance.

LRAs: Latency reversing agents, ART: antiretroviral therapy

To date, clinical trials with LRAs have demonstrated that activation of viral gene expression is possible in vivo, but there is limited or no clearance of the reactivated cells (Archin et al, 2017; Archin et al, 2012; Elliott et al, 2015; Elliott et al, 2014; Gutierrez et al, 2016; Rasmussen et al, 2014; Sogaard et al, 2015). To promote the turnover of reactivated cells, immune based strategies including antibodies, T-cell vaccines and immunotherapeutics that enhance T-cell function, are being investigated to kill reactivated cells following latency reversal (reviewed elsewhere (Brockman et al, 2015; Marsden & Zack, 2015; Wykes & Lewin, 2017)). Alternatively, multiple lines of evidence reveal that HIV infection can impact both pro-apoptotic and anti-apoptotic pathways, potentially favouring cell death or cell survival respectively. Consequently, manipulation of the pro-apoptotic functions of HIV to favour apoptosis and cell death may lead to the clearance of latently infected cells after reactivation. Therefore, strategies that could potentially eliminate latently infected cells after reactivation through the induction of apoptosis are reviewed here.

Latency Reversing Agents

Various LRAs can increase viral gene expression from latency in vitro, including: histone deacetylase inhibitor/s (HDACi/s), histone methyltransferase (HMT) inhibitors, DNA methyltransferase inhibitors, bromodomain inhibitors, protein kinase C (PKC) agonists (reviewed in (Rasmussen & Lewin, 2016), as well as PI3K/Akt pathway inhibitors that affect cell survival and agonists for the innate immune receptors TLR7 or TLR9 (Offersen et al, 2016; Tsai et al, 2017) (Figure 2). Only a limited number of these agents have been studied in clinical trials. These include the HDACis vorinostat (Archin et al, 2017; Archin et al, 2012; Elliott et al, 2014), panobinostat (Rasmussen et al, 2014) and romidepsin (Sogaard et al, 2015), the disulfiram drug shown to affect PI3K/Akt and more commonly used to treat alcoholism (Elliott et al, 2015; Spivak et al, 2014; Xing et al, 2011), and the TLR9 agonist MGN1703 (Vibholm et al, 2017). The PKC agonist bryostatin was also evaluated in vivo but had no effect on the transcription of latent HIV (Gutierrez et al, 2016). In these clinical studies, all LRAs except bryostatin increased HIV RNA transcription when administered alone, but there was no decline in the number of infected cells (Archin et al, 2012; Elliott et al, 2015; Elliott et al, 2014; Gutierrez et al, 2016; Rasmussen et al, 2014; Sogaard et al, 2015). In one clinical trial, the combination of the HDACi romidepsin with a T-cell vaccine led to a modest reduction in HIV DNA, although the mechanism for this reduction was not explored (Leth et al, 2016). More recently, in studies of SIV-infected rhesus macaques on ART and HIV-infected individuals on ART, the administration of TLR7 and TLR9 agonists led to significant increases in plasma SIV and HIV RNA respectively, consistent with latency reversal (Vibholm et al, 2017; Whitney et al, 2016; Whitney et al, 2015).

Figure 2. Classes of Latency reversing agents (LRAs).

Latency reversing agents (LRAs) can act on different pathways resulting in an increase in HIV transcription and/or virion production. P-TEFb: Positive transcription elongation factor b, TLR: Toll-like Receptor, mTOR: Mechanistic target of rapamycin, STAT5: Signal transducer and activator of transcription 5, IL-15: Interleukin-15

There may be multiple reasons why HDACis alone are unable to induce clearance of infected cells. First, at least in an artificial in vitro model of HIV latency, where there is overexpression of the pro-survival protein Bcl-2, HDACis did not induce virus-mediated cytolysis (Shan et al, 2012). Second, the more potent HDACi have been shown to directly inhibit CD8+ T-cell and natural killer (NK) cell function in vitro (Pace et al, 2016), although this was not observed with clinical trials of HDACis (Elliott et al, 2014; Sogaard et al, 2015). Third, in HIV+ individuals treated with ART during chronic infection, immune escape mutations have been demonstrated in latently infected cells (Deng et al, 2015). Fourth, there is persisting immune dysfunction as demonstrated by increased expression of T-cell exhaustion markers and this may impair T-cell mediated clearance of infected cells (Chew et al, 2016; Day et al, 2006). Collectively, these data indicate that reactivation of latently infected cells alone with LRAs is unlikely to lead to elimination of the infected cells and an additional intervention for the “kill” will be needed.

Apoptosis and HIV latency

Apoptosis is a cellular process that results in programmed cell death of unwanted or damaged cells (Bai et al, 2014). Apoptosis is tightly controlled and defective regulation of apoptosis is implicated in many human diseases such as cancer, inflammation and autoimmune disease (Bai et al, 2014). There are two main apoptotic pathways, the intrinsic and extrinsic pathways.

The extrinsic pathway is mediated by death receptors on the surface of the cell, including members of the tumour necrosis factor (TNF) receptor family. Binding of ligands to their respective death receptors results in the recruitment and activation of members of the caspase protease family that induce apoptosis (Igney & Krammer, 2002). Well characterised death receptors and their corresponding ligands include Fas/Fas ligand (FasL) and tumour necrosis factor receptor 1 (TNFR1)/ TNF-α (Elmore, 2007). When Fas binds to Fas ligand, the receptors recruit Fas-associated protein with death domain (FADD). Dimerization of the death effector domain results in association of FADD with procaspase-8 and the formation of a death-induced signalling complex (DISC) that cleaves procaspase-8 into active caspase-8 (Kischkel et al, 1995). Active caspase-8 promotes activation of the caspase cascade including cleavage and activation of caspase-3 leading to apoptosis (Green & Llambi, 2015).

In contrast, the intrinsic pathway is often referred to as the mitochondrial pathway and is initiated by cellular stress, DNA damage, radiation and other stress signals. These stress signals lead to the formation of a pore in the mitochondria membrane. This leads to permeabilization of the mitochondrial membrane and the release of cytochrome c and second mitochondrial-derived activator of caspases (Smac) and the protease Omi into the cytosol (Estornes & Bertrand, 2015). Once in the cytosol, these mediators trigger apoptosome formation and activate the caspase cascade to induce apoptosis (Green & Llambi, 2015).

Apoptosis is regulated by members of the Bcl-2 protein family (Adams & Cory, 1998) and these are summarised in Table 1. Bcl-2 family members that tightly regulate apoptosis at the mitochondrial level include; the pro-apoptotic molecules Bak and Bax; anti-apoptotic molecules Bcl-X_L, Bcl-2, Bcl-W, A1 and Mcl-1; and BH3-only proteins that act as sensors of apoptotic stimuli including Bad, Bid, Bim, Bik, Bmf, Hrk, Puma and Noxa (Czabotar et al, 2014; Happo et al, 2012) (Table 1). Members of the inhibitor of apoptosis (IAP) protein family including X-linked inhibitor of apoptosis protein (XIAP) and cIAP can also regulate apoptosis by preventing caspase activation to suppress apoptosis (Czabotar et al, 2014).

Table 1.

Bcl-2 family proteins and their main mechanism of action

Pro-/Anti-Apoptotic	Protein	Mechanism of Action	References
Anti-apoptotic	Bcl-2	Binds to and sequesters pro-apoptotic proteins Bax, Bim, Puma, Bmf and Bad	(Chen et al, 2005; Letai et al, 2002)
	Bcl-XL Bcl-W	Binds to and sequesters pro-apoptotic proteins Bax, Bak, Bim, Puma, Bmf, Bik, Hrk and Bad
	A1 Mcl-1	Binds to and sequesters pro-apoptotic proteins Bak, Bim, Puma, Noxa and Hrk	(Dai et al, 2016)
Pro-apoptotic	Bax/ Bak	Once activated, Bax and Bak undergo conformational changes, oligomerize and permeabilize the mitochondrial outer membrane, resulting in the release of cytochrome c and Smac/DIABLO	(Annis et al, 2005; Kim et al, 2009; Kuwana et al, 2002)
	Activator BH3-only proteins Bid Bim Puma	High affinity for Bax/Bak Binds directly to and activates Bax/Bak Sequestered by all anti-apoptotic proteins	(Chen et al, 2005; Kim et al, 2006; Letai et al, 2002)
	Sensitizer BH3-only proteins Bik Bmf Bad Hrk Noxa	Binds to anti-apoptotic molecules, preventing these molecules from sequestering Bax/Bak. Displaces activator BH3-only proteins from anti-apoptotic proteins allowing the activator proteins to bind and directly activate Bax/Bak.	(Certo et al, 2006; Kuwana et al, 2005; Letai et al, 2002)

Bcl-2: B cell lymphoma-2, Bcl-X_L: B cell lymphoma-extra large, Bcl-W: B-cell lymphoma-W, A1: Bcl-2-related gene A1, Mcl-1: myeloid cell leukemia 1, Bax: Bcl-2-associated X protein, Bak: Bcl-2 antagonist killer 1, Bid: Bcl-2 interacting domain death agonist, Bim: Bcl-2 interacting mediator of cell death, Puma: p53-upregulated modulator of apoptosis, Bik: Bcl-2 interacting killer, Bad: Bcl-2 antagonist of cell death, Bmf: Bcl-2 modifying factor, Hrk: Harakiri

Both the intrinsic and extrinsic pathways converge at the stage of caspase-8 activation. When the extrinsic apoptotic pathway is triggered, the formation of active caspase-8 can lead to apoptosis in a mitochondrial-independent pathway through the activation of the effector caspases (caspase-3 and -7) (Elmore, 2007). Additionally, active caspase-8 can cleave the BH3-only protein Bid producing truncated Bid (tBid) (Li et al, 1998). tBid can induce mitochondrial outer membrane permeabilization leading to the activation of mitochondrial-dependent apoptosis (Luo et al, 1998). Therefore caspase-8 provides a link between the extrinsic “death-receptor” and intrinsic “mitochondrial” apoptotic pathways.

HIV proteins can interfere with apoptosis pathways leading to a range of outcomes including activation, inhibition or delay in cell death (Timilsina & Gaur, 2016) (summarised in Figure 3). During the early stages of the viral life cycle, the viral proteins Nef, Tat and Vpr can impair apoptosis to promote cell survival, allowing for viral replication (Timilsina & Gaur, 2016). In contrast, in the late stages of the viral life cycle, the HIV envelope (Env) and Vpu proteins can promote apoptosis of infected cells by a variety of mechanisms (Timilsina & Gaur, 2016). Expression of the viral protease later in viral replication can also lead to the generation of a pro-apoptotic Casp8p41 peptide that promotes apoptosis (Cummins et al, 2016b). HIV Tat, Vpr and Nef also transition to pro-apoptotic roles later in the viral lifecycle when Tat and Vpr expression increases (Timilsina & Gaur, 2016) (Figure 3). It is likely that the balance between anti-apoptotic and pro-apoptotic viral and cellular proteins determine the fate of infected cells between survival or apoptosis.

Figure 3. Effects of HIV proteins on apoptosis and compounds to foster apoptosis of cells.

When stress stimuli induce the intrinsic mitochondrial pathway, the Bak or Bax members of the B-cell lymphoma 2 (Bcl-2) family are activated to promote apoptosis. However, activation of the PI3K pathway promotes cell survival by preventing steps in the apoptosis pathway. PI3K activation leads to conversion of phosphatidylinositol 4,5-biphosphate (PIP2) to phosphatidylinositol (3,4,5)-triphosphate (PIP3). This step is also negatively regulated by phosphatase and tensin homolog (PTEN). The conversion of PIP2 to PIP3 leads to the activation of Akt protein. Akt can inhibit the forkhead box protein O1 (FOXO1) transcription factor from translocating into the nucleus to induce pro-apoptotic target genes, preventing apoptosis. Additionally, Akt inhibits the pro-apoptotic protein Bad from inhibiting anti-apoptotic Bcl-2 members like Bcl-2 and Bcl-X_L, which in turn is inhibited from activating Bak or Bax thereby also preventing apoptosis.

A) Pathways leading to cell survival. Anti-apoptotic HIV proteins (shown in red) expressed early in the virus life cycle that can promote cell survival include Nef, Tat and Vpr. Nef can bind PI3K proteins to increase the conversion of PIP₂ to PIP₃. This leads to an increase in Akt protein that subsequently leads to inhibition of pro-apoptotic molecules such as Bad from activating apoptosis thereby promoting cell survival. Additionally, Nef can prevent p53 from activating the pro-apoptotic molecule PUMA which directly inhibits Bcl-2 to prevent apoptosis. HIV Tat protein can inhibit PTEN, thereby promoting the PI3K/Akt pathway to promote cell survival. Tat can also directly inhibit FOXO1 transcription factors to prevent the transcription of pro-apoptotic genes, preventing apoptosis. Tat also prevents the pro-apoptotic molecule PUMA from inhibiting Bcl-2. HIV Vpr can lead to the increase of anti-apoptotic molecule Bcl-2 in addition to inhibiting the pro-apoptotic Bak molecule.

B) Pathways leading to cell death. When activated, Bax proteins form pores in the mitochondrial membrane, leading to the release of cytochrome C and second mitochondria-derived activator of caspase (Smac) into the cytosol. Cytosolic cytochrome C leads to the formation of the apoptosome that activates caspase-9. Caspase-9 then cleaves pro-caspase -3, -7 into their active caspase-3 and caspase-7 forms, which lead to apoptosis. Multiple HIV proteins (shown in green) can interact with the members of the apoptosis pathway later in the viral life cycle, leading to increased apoptosis. Nef can inhibit both the anti-apoptotic molecules Bcl-2 and Bcl-X_L. Higher expression of HIV Vpr can bind and lead to the upregulation of Bak to initiate apoptosis. Higher expression of Tat also leads to the upregulation of Bax, resulting in the release of cytochrome C from the mitochondria and subsequent activation of the caspase cascade. HIV Env protein can inhibit the anti-apoptotic protein Bcl-2 to favour apoptosis. The HIV protease protein also cleaves caspase-8 to the pro-apoptotic Casp8p41 peptide to promote apoptosis. It is likely that the balance of pro-apoptotic versus anti-apoptotic cellular and viral proteins decide the fate of infected cells.

C) Modulation of apoptosis pathways using compounds. Compounds (in light blue) that act on different cellular proteins in these pathways could potentially be used to tip the balance toward apoptosis of HIV infected cells. PI3K inhibitors block the conversion of PIP₂ to PIP₃, decreasing active Akt within a cell to stop Akt from inhibiting apoptosis. Akt inhibitors directly decrease active Akt to prevent Akt from inhibiting apoptosis to ultimately induce apoptosis. Bcl-2 inhibitors such as Venetoclax inhibit anti-apoptotic Bcl-2 to sensitize the cells towards apoptosis. Smac mimetics competitively bind inhibitors of apoptosis proteins like X-linked inhibitor of apoptosis (XIAP) to promote apoptosis. Retinoic acid-inducible gene I (RIG-1) detects viral RNA as well as activating pro-apoptotic Bak/Bax proteins, leading to apoptosis. In addition to viral RNA, RIG-1 inducer compounds like the retinoic acid derivative acitretin can also trigger apoptosis.

While productive HIV infection typically leads to the death of infected cells (Petit et al, 2002), HIV latently infected CD4+ T-cells are long lived (Finzi et al, 1999). These latently infected T-cells can also proliferate (Maldarelli et al, 2014; Simonetti et al, 2016; Wagner et al, 2014) leading to clonally expanded cells that may contain identical and intact provirus (Bui et al, 2017; Simonetti et al, 2016). It is possible that latently infected cells avoid death from the virus integrating into genes like BACH2 encoding the transcriptional regulator that may promote cell survival (Maldarelli et al, 2014), or because there is no expression of pro-apoptotic viral proteins like HIV Env, protease and Vpr (Abbas & Herbein, 2012), or there is an absence of viral RNA detection through the RIG-I pattern recognition receptor (PRR) (Li et al, 2016). Additionally, increases in the anti-apoptotic protein Bcl-2 results in elevated levels of anti-oxidant molecules like glutathione and thioredoxin, thereby protecting latently infected cells from oxidative stress-induced apoptosis (Aillet et al, 1998). The XIAP protein, which inhibits the protease activity of caspase-3, -7 and -9 (Berro et al, 2007; Neumann et al, 2015), is also increased in latently infected cells compared to uninfected cells (Berro et al, 2007) to favour cell survival. Finally, inhibition of other pro-apoptotic cellular proteins in latently infected cells may also be involved. For example, knockdown of pro-apoptotic proteins such as Bax and FADD have been shown to prolong cell survival in Jurkat T cells infected with HIV (Wang et al, 2011).

These data collectively suggest that in latently infected cells, the absence of pro-apoptotic viral proteins, together with elevated anti-apoptotic and decreased pro-apoptotic cellular proteins, together promote cell survival.

Non-apoptotic cell death mediated by HIV

HIV infection of CD4+ T cells can lead to cell death through numerous mechanisms. The main proposed form of programmed cell death is apoptosis. However other forms of programmed cell death may contribute. Pyroptosis is a highly inflammatory form of cell death that features cellular swelling, rupture of the plasma membrane and the release of the cells cytoplasmic contents and pro-inflammatory cytokines into the extracellular milieu (Doitsh et al, 2014). During active HIV infection, abortive reverse transcription of the viral RNA genome to DNA can result in incomplete viral DNA reverse transcripts that are sensed by protein recognition receptors such as interferon-gamma-inducible protein 16 (IFI16). This results in assembly of the inflammasome by capsase-1 activation and ultimately leads to pyroptosis. (Monroe et al, 2014). Pyroptosis is mediated by caspase-1 and works independently of the other pro-apoptotic caspases. It is unclear if pyroptosis plays any role in a latently infected cell.

Another mechanism of cell death is necrosis, which is non-apoptotic, passive, accidental and unregulated cell death. Necrosis does not require activation of specific pathways and can occur as a consequence of extensive cell damage. However, necrosis is not always passive and can result from a signal cascade in a regulated manner, much like apoptosis, aptly termed necroptosis (Degterev et al, 2005). Necroptosis can be triggered by death receptors tumour necrosis factor (TNF) and TNFR1, which requires the activity of receptor interacting protein kinase-1 (RIPK1) and -3 (RIPK3) (Holler et al, 2000). When TNF-α binds to TNFR1, TNF receptor-associated death domain (TRADD) recruits RIP1. In the absence of caspase-8, RIPK1 assembles with RIPK3 to form a complex termed the necrosome, acting as the signal transducer for necroptosis (Vanden Berghe et al, 2014). Mixed lineage kinase domain like (MLKL) protein is then activated through phosphorylation by the necrosome resulting in the insertion of this protein into the plasma membrane and subsequent expulsion of intracellular contents (Wang et al, 2014). Alternatively, other death receptors such as the toll like receptors (TLRs) and the cytosolic DNA sensor DNA-dependent activator of IFN regulator genes (DAI) have also been shown to induce necroptosis (Kim et al, 2001; Upton et al, 2012). There is accumulating evidence that pyroptosis and necroptosis play a role in untreated, active HIV infection leading to the loss of CD4+ T cells (Doitsh et al, 2014; Pan et al, 2014). However, there is little evidence that pyroptosis or necroptosis are activated following latency reversal. Although, further work is needed to determine whether these pathways could be exploited to preferentially eliminate latently infected cells.

Pro-apoptotic compounds to clear HIV latently infected T-cells

Multiple pro-apoptotic compounds have been developed, typically for the cancer field (reviewed in (Hassan et al, 2014; Kalimuthu & Se-Kwon, 2013)). Compounds that maybe useful in clearing HIV latently infected cells include: Bcl-2 antagonists, PI3K/Akt inhibitors, Smac mimetics/XIAP inhibitors and RIG-I inducers (summarised in Figure 3C and Table 2). The potential activity of these compounds in eliminating HIV latently infected cells is discussed below.

Table 2.

Summary of compounds that induce apoptosis of HIV latently infected cells

Inducers of Apoptosis	Pathway Description	Compounds to induce apoptosis	Effects on HIV latency in vitro	Clinical Development for non-HIV indications	References
Bcl-2 antagonists	Anti-apoptotic cellular Bcl-2 sequesters the pro-apoptotic Casp8p41 fragment generated by HIV protease. Inhibiting Bcl-2 liberates pro-apoptotic Casp8p41 generated by HIV protease to promote apoptosis.	Venetoclax	Depletion of latently infected cells using CD4+ T-cells from HIV-infected individuals on ART ex vivo	Licensed- chronic lymphocytic leukemia	(Cummins et al, 2016a)
		Navitoclax	Similar results to Venetoclax, but resulted in increased non-specific toxicity	Phase II- chronic lymphocytic leukemia	(Cummins et al, 2016a)
PI3K/Akt Inhibitors	PI3K leads to activated Akt, which inhibits pro-apoptotic Bad and the transcription of pro-apoptotic genes. Inhibiting PI3K/Akt blocks this process, allowing pro-apoptotic Bad to function and transcription of pro-apoptotic genes, favouring apoptosis.	Edelfosine Perifosine Miltefosine	Inhibited Akt activation induced by HIV infection in macrophages, leading to death upon cellular stress	Phase II- non-small-cell lung cancer Phase II- non-small-cell lung cancer Licensed- visceral leishmaniasis	(Lucas et al, 2010)
		Lancemaside A Compound K Arctigenin	Eliminated cytoprotective phenotype of microglial cell lines infected with HIV upon cellular stress	N/A	(Kim et al, 2011)
Smac Mimetics	Smac mimetics bind and block anti-apoptotic IAP proteins to promote apoptosis.	Birinapant GDC-0152 Embelin	Dose-dependent increase in death of HIV latently infected CD4+ T cells	Phase I/II- Hepatitis B Phase I- Locally Advanced or metastatic malignancies Phase IV- Psoriasis	(Campbell et al, 2015)
RIG-I Inducers	Induce RIG-I that detects viral RNA and promotes apoptosis. Retinoic acid (RA) also activates p300 acetyl transferase cellular transcription factor leading to HIV transcription.	Acitretin	Induced HIV expression, reduced HIV DNA and induced preferential death in latently infected CD4+ T cells from HIV infected individuals on suppressive ART	Licensed- Psoriasis	(Li et al, 2016)
			No induction in HIV expression or death in latently infected cells		(Garcia-Vidal et al, 2017)

Bcl2: B-cell lymphoma 2, PI3K: phosphoinisitide-3-kinase, Smac: second mitochondria-derived activator of caspase, RIG-I: retinoic acid-inducible gene I

Bcl-2 antagonists

The overexpression of Bcl-2 is common in many cancers and thus is a target of cancer therapy (Reed & Pellecchia, 2005). Many of the Bcl-2 antagonists currently being investigated are BH3-mimetics (Lessene et al, 2008). Compounds such as Venetoclax and Navitoclax mimic the BH3 binding domain present in pro-apoptotic molecules and bind with higher affinity to anti-apoptotic Bcl-2 family proteins (Balakrishnan & Gandhi, 2013; Cummins et al, 2016b), blocking their anti-apoptotic function thereby favouring apoptosis. Venetoclax has recently been licensed for the treatment of chronic lymphocytic leukaemia (CLL) and is in advanced clinical development for non-Hodgkin’s lymphoma (NHL) (Souers et al, 2013).

In vitro, Bcl-2 antagonists have been shown to deplete latently infected T-cells following T-cell receptor activation with antibodies binding to CD3, a co-receptor for the T-cell receptor (TCR), and CD28, a co-stimulation signal (Cummins et al, 2016a). During active HIV replication, the HIV protease results in cleavage of pro-caspase-8 to generate the Casp8p41 fragment containing a BH3-like domain that subsequently binds to and activates Bak, triggering apoptosis (Sainski et al, 2014). However, during reactivation of HIV latency in resting CD4+ T-cells, HIV protease-dependent apoptosis is prevented as the cells contain high levels of anti-apoptotic Bcl-2, which can sequester the pro-apoptotic Casp8p41 to prevent apoptosis (Cummins et al, 2016b). Therefore, inhibiting Bcl-2 may prevent the Bcl-2-mediated sequestration of pro-apoptotic Casp8p41, liberating Casp8p41 generated by HIV protease to induce apoptosis of these activated latently infected cells.

Encouragingly, pre-treatment of latently infected cells from individuals on ART with the Bcl-2 antagonist Venetoclax and subsequent reactivation with anti-CD3 plus anti-CD28 T-cell stimulation reduced the frequency of latently-infected T-cells in cultures from 8 of 11 individuals (Cummins et al, 2016b). Whether combinations of Bcl-2 inhibitors like Venetoclax with LRAs that do not induce maximal T-cell stimulation also lead to the death of latently infected cells remains unclear. Interestingly, Venetoclax also leads to the selective killing of HIV infected primary T-cells during productive infection in vitro (Cummins et al, 2017) and therefore Venetoclax could also potentially be used to reduce the establishment of latency.

Similar effects on latently infected cells were observed with the Bcl-2 inhibitor, Navitoclax, but there was increased toxicity in uninfected cells. Dose-limiting thrombocytopenia from Navitoclax also makes this a less attractive drug for future clinical trials (Cummins et al, 2016a).

PI3K/Akt inhibitors

PI3Ks are lipid kinases that produce secondary messengers that control a wide range of intracellular signalling pathways in leukocytes. Activation of the PI3K pathway has the cumulative effect of promoting cell survival (Vanhaesebroeck et al, 2010). PI3Ks also have roles in apoptosis and the survival of virus-infected cells (Cooray, 2004). Therefore, PI3K inhibitors could enhance the apoptosis of HIV infected cells.

The major effector of the PI3K pathway is the serine/threonine kinase Akt (also referred to as Protein Kinase B or PKB) (Chandarlapaty et al, 2011). Binding of the Pleckstrin homology (PH) domain of Akt to the phosphoinositide products of PI3K (PIP₃) results in the recruitment of Akt to the plasma membrane and Akt activation (She et al, 2010). Activated Akt then interferes with pro-apoptotic molecules either through direct inhibitory phosphorylation of the Bcl-2 family member Bad, or indirectly through the phosphorylation of transcription factors such as FOXO1 that subsequently translocate out of the nucleus, thereby preventing transcription of pro-apoptotic genes (Rodrik-Outmezguine et al, 2011). This prevents apoptosis and fosters cell survival.

Two HIV proteins expressed early in the virus life cycle interact with the PI3K pathway (Figure 3). The HIV protein Nef activates the PI3K signalling pathway through binding the p85 unit of PI3K (Wolf et al, 2001), which results in inhibitory phosphorylation of the pro-apoptotic factor Bad, blocking premature apoptosis in T-cells (Chugh et al, 2008). HIV-1 Tat protein also interferes with cellular PTEN, a negative regulator of Akt (Chugh et al, 2008; Kim et al, 2010). PTEN converts PIP₃ back to PIP₂, thereby inhibiting activation of Akt. PTEN is regulated by binding to host cell p53. p53 is a pro-apoptotic molecule that enhances PTEN expression, leading to the repression of the PI3K/Akt signalling pathway. However, HIV-1 Tat also binds p53, preventing p53 binding to PTEN and causing downregulation of PTEN. This leads to activation of PI3K/Akt signalling and impaired apoptosis (Chugh et al, 2008; Wolf et al, 2001). Therefore, both HIV Nef and Tat expressed early during the virus life cycle can promote the pro-survival PI3K/Akt signalling pathway to prevent apoptosis. Given the role of PI3K/Akt signalling in preventing apoptosis and promoting survival, inhibition of key factors in the PI3K/Akt pathway may redirect the balance towards apoptosis and death of infected cells (Lucas et al, 2010).

Effects of PI3K inhibition in HIV-infected macrophages

During HIV infection of macrophages, the HIV Tat protein promotes the stress-induced activation of the PI3K/Akt survival pathway (Chugh et al, 2008). Activation of this pathway limits cytotoxicity and may be important for long-term survival of infected macrophages. Inhibition of the downstream effector of the PI3K signalling pathway, Akt, prevented activation of the PI3K/Akt signalling pathway in infected macrophages (Chugh et al, 2008). Three Akt inhibitor compounds, Edelfosine, Perifosine and Miltefosine were able to inhibit Akt activation induced by HIV, ultimately leading to the death of macrophages upon cell stress (Lucas et al, 2010). Perifosine also antagonised the cytoprotective phenotype of HIV-infected macrophages and reduced HIV production. In a microglial cell line, three compounds: Lancemaside A, Compound K and Arctigenin induced cell death following cellular stress (Kim et al, 2011). All three compounds inhibit different steps in the PI3K/Akt pathway. Arctigenin binds to the PI3K enzyme, preventing conversion of PIP₂ to PIP₃, thus decreasing levels of Akt protein. Lancemaside A blocks the localisation of Akt to the plasma membrane, thereby inhibiting Akt activation. Lastly, Compound K is thought to directly bind and inhibit Akt (Kim et al, 2011).

These studies show the potential of PI3K/Akt inhibitors to eliminate HIV-infected macrophages by sensitizing the cells to apoptosis. Whether these compounds can also sensitize latently infected T-cells to apoptosis warrants further investigation. Potentially priming and pre-sensitizing cells with PI3K/Akt inhibitors to decrease levels of phosphorylated Akt followed by LRA-induced expression of viral proteins from latent HIV may tip the balance of reactivated cells towards apoptosis, thus eliminating these cells.

Combining PI3K inhibitors with LRAs

CUDC-907 is a small molecule inhibitor currently in clinical trials for lymphoma or multiple myeloma (Ma et al, 2014). Interestingly, CUDC-907 has dual activities of potently obstructing class I PI3K (α, β and δ) isoforms as well as inhibiting HDAC class I and II enzymes (Younes et al, 2013). As HIV latency is regulated by HDACs (Sheridan et al, 1997; Verdin et al, 1993) and reactivated by HDAC inhibitors (reviewed in (Margolis, 2011; Mboyne & Karn, 2014)), CUDC-907 could potentially both reactivate latent HIV via HDAC inhibition and also trigger death of reactivated cells through blocking the PI3K/Akt pathway.

Of note, some LRAs can have pro-survival effects on the PI3K/Akt pathway and thus act in a counterproductive manner to the PI3K inhibitors. For instance, the anti-alcohol drug disulfiram acts as an LRA via depletion of the PTEN regulator of the PI3K/Akt pathway, thus promoting pro-survival Akt activation and NF-κB driven HIV transcription (de Almagro & Vucic, 2012; Xing et al, 2011). Additionally, the HDACi vorinostat and the PI3Kα isoform agonist 55704 also promote pro-survival PI3K/Akt signalling (Chen & Huerta, 2009). The activation of the PI3K/Akt pathway by these LRAs may enhance cell survival and might explain why clinical trials with vorinostat (Archin et al, 2012) and disulfiram (Elliott et al, 2015; Fulda, 2015) both failed to reduce the levels of HIV DNA in HIV infected individuals on ART. Given that some LRAs potentially enhance survival of infected cells, this is a critical consideration when devising shock and kill strategies that include both PI3Ki and LRAs. LRAs with pro-survival effects on the PI3K-Akt pathway are expected to compete with the pro-apoptotic effects of PI3K/Akt inhibitors and may prevent apoptosis. Additionally, combining PI3K/Akt inhibitors with pro-survival LRAs like disulfiram, vorinostat or 55704 may also reduce the potency of latency reversal. This is because HIV transcription depends on an increase in transcription factors like NFκB that are induced by LRAs that activate the PI3K/Akt pathway, but will be impaired by PI3K/Akt inhibitors. In summary, combining PI3K inhibitors and LRAs is a potential strategy to clear the latent HIV reservoir. However, the choice of LRAs requires careful selection to avoid opposing effects of the LRAs on the PI3K/Akt pathway. Further pre-clinical evaluation is required to determine combinations of various PI3K/Akt inhibitors and LRAs that might effectively deplete latently infected cells.

Smac Mimetics and XIAP inhibitors

Inhibitor of apoptosis proteins (IAPs) are defined by the presence of one to three baculoviral IAP repeat (BIR) domains (Bai et al, 2014), and eight human IAPs have been identified: NAIP (BIR1), cellular-IAP1 (BIR2), cIAP2 (BIR3), XIAP (BIR4), survivin (BIR5), BRUCE/Apollon (BIR6), livin (ML-IAP1/BIR7) and ILP2 (BIR8) (de Almagro & Vucic, 2012). XIAP, cIAP1 and cIAP2 have major roles in both the intrinsic and extrinsic apoptosis pathways while the other IAPs regulate cell survival through cell cycle control or inflammation (de Almagro & Vucic, 2012). IAPs regulate apoptosis either by direct or indirect inhibition. XIAPs directly bind and inhibit caspase-3, -7 and -9 activity, thus preventing the caspase cascade and apoptosis. In contrast, cIAPs do not directly bind caspases but instead regulate caspase activation through their E3 ligase activity, modulating TNF-mediated apoptosis and NF-κB pathways (Bai et al, 2014; Chen & Huerta, 2009; de Almagro & Vucic, 2012).

Upon activation of the intrinsic apoptosis pathway, mitochondria release both cytochrome c and second mitochondria-derived activator of caspase (Smac), also known as DIABLO (direct inhibitor of apoptosis protein-binding protein with low pI) (Fulda, 2015). Smac is a pro-apoptotic protein that binds to IAPs to inhibit their function. Most Smac proteins directly compete with caspases for XIAP binding, liberating caspases for apoptosis. Smac proteins can also activate the ubiquitin activity of cIAP1/2 leading to the degradation of these IAPs (Pache et al, 2015). Through their ability to bind and inhibit IAP activity, Smac mimetics can induce pro-apoptotic mechanisms and therefore are attractive compounds to induce the apoptosis of reactivated, latently infected cells.

Targeting XIAP may potentially selectively kill latently infected cells. The flavopiridol cyclin-dependent kinase 9 (CDK-9) inhibitor that also downregulates XIAP expression (Rosato et al, 2007) increased apoptosis of the latently infected ACH2 cell line compared to untreated cells (Berro et al, 2007). The XIAP antagonists GDC-0152, embelin and the Smac mimetic Birinapant also resulted in a dose-dependent increase in apoptosis of latently infected primary memory CD4+ T-cells. This was achieved without the addition of LRAs, providing intriguing evidence that HIV latently infected cells may be selectively killed without reactivation (Campbell et al, 2015). Interestingly, Birinipant has been shown to clear Hepatitis B virus-infected cells in vitro and in a mouse model, through a similar mechanism (Ebert et al, 2015a; Ebert et al, 2015b). Finally, pre-treatment of monocyte-derived macrophages (MDMs) with the AEG40730 Smac mimetic that depletes cIAP1, cIAP2 and XIAP also sensitizes MDMs to HIV Vpr-induced apoptosis (Busca et al, 2012). Whether this pathway is similar in latently infected T-cells treated with an LRA warrants further examination.

Further evaluation of these compounds, especially those in clinical development, should be considered. The Smac mimetic, Birinapant is undergoing clinical evaluation to treat refractory haematological and solid tumours, including lymphoma (Amaravadi et al, 2015) and epithelial ovarian cancer (Noonan et al, 2016). Flavopiridol decreases XIAP and is under investigation for the treatment of chronic lymphocytic leukemia (Awan et al, 2016; Blachly et al, 2016). The impact of these inhibitors on uninfected cells also requires careful investigation to minimise undesirable toxicities on uninfected cells.

Aside from apoptosis, Smac mimetics can also overcome blocks in the death receptor pathway of necroptosis. In FADD- or caspase-8-deficient leukemia cells, the Smac mimetic BV6 primed the cells for necroptosis (Laukens et al, 2011) in a RIPK1-dependent and caspase-independent manner. In apoptotic-proficient cells, the Smac mimetic primed cells for TNF-α, caspase-dependent death (Laukens et al, 2011). HIV infected cells share similarities with leukemia cells such as prolonged survival and defective cell death pathways. This finding that Smac mimetics can prime cells for necroptosis proposes an alternative cell death mechanism that can be exploited. Although, care should be taken with this approach as necroptosis also results in the release of intracellular contents and damage-associated molecular patterns (DAMPs) into the extracellular matrix, which promotes inflammation (Pasparakis & Vandenabeele, 2015). The role of necroptosis induced by Smac mimetics in the context of HIV latency reversal is unclear but warrants further investigation.

RIG-I Inducers

The induction of the innate immune response plays a critical role in the host cell defence against invading viruses. Recognition of viral pathogens through the innate immune response pattern recognition receptors (PRRs), which include toll-like receptors (TLRs) or retinoic acid-inducible gene I (RIG-I), can also drive the apoptosis of virus infected cells (Mogensen, 2009). RIG-I is a DExD/H box RNA helicase that senses viral RNA within the cytoplasm of infected cells (Kato et al, 2005; Wang et al, 2008). During HIV infection, RIG-I can sense HIV RNA and induce anti-viral signalling and apoptosis (Wang et al, 2013). However, the HIV protease can also degrade or sequester RIG-I to abrogate RIG-I mediated apoptosis (Solis et al, 2011) and circumvent this innate immune response.

Retinoic acid (RA) can induce RIG-I expression as well as activate the p300 acetyl transferase cellular transcription factor, thus leading to HIV transcription (Li et al, 2016). Therefore, treating HIV latently infected cells with drug compounds that mimic retinoic acid might similarly activate both HIV transcription and RIG-I mediated apoptosis to both reactivate latent HIV and tip the balance towards apoptosis of these infected cells despite HIV protease expression impeding RIG-I.

Acitretin, an FDA-approved retinoic acid derivative used to treat psoriasis (Ortiz et al, 2013), was recently shown to reactivate latent HIV and induce the selective apoptosis of HIV latently infected cells in vitro (Li et al, 2016). In the study by Li et al, acitretin induced HIV expression in latently infected T-cell lines and preferential death of HIV infected cells compared to uninfected samples. Moreover, this effect was enhanced when the LRA HDACi vorinostat was combined with acitretin. Cell death was driven by RIG-I-mediated increases in the pro-apoptotic protein Bax, which can trigger caspase-mediated apoptosis (Li et al, 2016). Unfortunately, these findings were not replicated in a subsequent study that also evaluated latently infected cells lines and patient derived cells (Garcia-Vidal et al, 2017).

Potential challenges

Although many of the pro-apoptotic compounds discussed in this review are now in clinical development or licensed (Table 2), several issues require consideration when being evaluated in HIV infection. First, it is critical that there is minimal depletion of non-infected cells. This could be reduced if cell death is dependent on the expression of HIV proteins and we believe that this will be a highly desirable strategy for drug development. To ensure only reactivated latently infected cells are eliminated, we propose first administering the “kill” compounds to sensitize latently infected cells towards apoptosis, followed by subsequent reactivation with LRAs to drive the expression of pro-apoptotic viral products and selectively clear the HIV infected cells. Sensitizing the cell towards an apoptotic state followed by the production of pro-apoptotic HIV proteins such as protease and HIV RNA could tip the balance towards apoptosis and specific killing of the cells that only produce these HIV proteins. This has been demonstrated with the pro-apoptotic drug Venetoclax (Cummins et al, 2016b) that when combined with LRAs led to the selective apoptosis and clearance of HIV infected cells. Additionally, the acitretin RIG-inducer that acts as an LRA as well as RIG-I inducer to drive apoptosis also leads to selective death of HIV-infected cells (Li et al, 2016). A second issue for consideration involves examining interactions with LRAs to determine potential synergy, antagonism or toxicity. Third, effects on non-T-cell reservoirs should also be evaluated as cells such as infected macrophages may be particularly resistant to apoptosis. Fourth, the effect of these compounds on dividing and non-dividing cells should be evaluated to determine how effectively various HIV infected cell types will be cleared. To this end, a recent study found that the Bcl-2 inhibitor Venetoclax induced apoptosis of the latently infected J-Lat10.6 T cell line model (Cummins et al, 2017), in which the J-Lat10.6 cells proliferate slowly in the absence of stimulation. While Venetoclax can induce apoptosis in both dividing and non-dividing HIV infected T cells, it will be important to evaluate the impact of any kill compounds on diverse cell types to ascertain selective clearance of HIV infected cells. Finally, penetration of these compounds into tissue sites such as the central nervous system and gut-associated lymphoid tissue (GALT) (Eisele & Siliciano, 2012) will be important to determine. Potency and specificity could be potentially enhanced using drug delivery systems such as nanoparticles coated with specific antibodies targeting CD4 or a latency marker, such as the recently described CD32a (Descours et al, 2017).

The timing of administration of the apoptosis inducing compounds relative to the LRAs also requires careful examination to maximize the depletion of latently infected cells. Maximal depletion may be achieved by pre-sensitizing latently infected cells with the apoptosis inducing compound before adding LRAs to induce pro-apoptotic viral products that drive apoptosis of HIV infected cells. Alternatively, greater depletion of latently infected cells may be achieved by adding the apoptosis-inducing compound at the same time or potentially after LRA reactivation, in which pro-apoptotic viral products have been generated. The sequence of drug administration will require pre-clinical and in vitro assessment.

Ethical considerations must always be a top priority for all clinical trials but are especially challenging in this context. Individuals on ART have a near normal life expectancy (Wandeler et al, 2016) and in early phase studies aimed at achieving HIV remission or cure, participants are unlikely to have a direct benefit from the intervention while there are also potential risks. These issues have been reviewed elsewhere (Lo et al, 2013; Rennie et al, 2015). In contrast to interventions for malignancy, toxicity of any intervention must be low and the pre-clinical rationale for the study well justified in order to proceed to early phase clinical trials.

Conclusion and Future Perspectives

Latently infected resting CD4+ T-cells remain a major barrier to curing HIV infected patients on long-term ART (Elliott et al, 2014; Rasmussen et al, 2014). There is a critical need for novel compounds and therapies that not only potently reactivate latently infected cells but also lead to the death of these reactivated cells. Identifying compounds that can be combined to both induce reactivation and death of latently infected cells have great appeal as this strategy is not dependent on an effective immune response or understanding immune escape in latently infected cells. Additionally, this strategy is potentially scalable and cost effective to administer to individuals living in low income countries. While the potential benefits of the administration of LRAs together with compounds that enhance HIV-induced death is yet to be investigated in HIV-infected individuals, more work is needed put the kill into “shock and kill”.