Controlling HIV-1: Non-Coding RNA Gene Therapy Approaches to a Functional Cure

The current treatment strategy for HIV-1 involves prolonged and intensive combined antiretroviral therapy (cART), which successfully suppresses plasma viremia. It has transformed HIV-1 infection into a chronic disease. However, despite the success of cART, a latent form of HIV-1 infection persists as integrated provirus in resting memory CD4+ T cells. Virus can reactivate from this reservoir upon cessation of treatment, and hence HIV requires lifelong therapy. The reservoir represents a major barrier to eradication. Understanding molecular mechanisms regulating HIV-1 transcription and latency are crucial to develop alternate treatment strategies, which impact upon the reservoir and provide a path toward a “functional cure” in which there is no detectable viremia in the absence of cART. Numerous reports have suggested ncRNAs are involved in regulating viral transcription and latency. This review will discuss the latest developments in ncRNAs, specifically short interfering (si)RNA and short hairpin (sh)RNA, targeting molecular mechanisms of HIV-1 transcription, which may represent potential future therapeutics. It will also briefly address animal models for testing potential therapeutics and current gene therapy clinical trials.

The current treatment strategy for HIV-1 involves prolonged and intensive combined antiretroviral therapy (cART), which successfully suppresses plasma viremia. It has transformed HIV-1 infection into a chronic disease. However, despite the success of cART, a latent form of HIV-1 infection persists as integrated provirus in resting memory CD4 + T cells. Virus can reactivate from this reservoir upon cessation of treatment, and hence HIV requires lifelong therapy. The reservoir represents a major barrier to eradication. Understanding molecular mechanisms regulating HIV-1 transcription and latency are crucial to develop alternate treatment strategies, which impact upon the reservoir and provide a path toward a "functional cure" in which there is no detectable viremia in the absence of cART. Numerous reports have suggested ncRNAs are involved in regulating viral transcription and latency. This review will discuss the latest developments in ncRNAs, specifically short interfering (si)RNA and short hairpin (sh)RNA, targeting molecular mechanisms of HIV-1 transcription, which may represent potential future therapeutics. It will also briefly address animal models for testing potential therapeutics and current gene therapy clinical trials.
(ribosomal and transfer RNAs). The regulatory class can be divided into small ncRNAs (<200 nucleotides) and long ncR-NAs (>200 nucleotides) (7), with small ncRNAs being further categorized into microRNA (miRNA), short interfering RNAs (siRNAs), and antisense RNAs (asRNAs). This review will primarily focus on siRNAs and short hairpin RNAs (shRNAs) as there have been several recent reviews of the roles of long ncRNAs and other small ncRNAs (8,9).
The HIV-1 5′ long terminal repeat (LTR), which acts as the promoter for integrated virus, consists of ~450 base pairs. It includes multiple binding sites for host transcription factors, including NF-κB, NFAT, SP1, and AP-1, which act to enhance viral transcription. HIV-1 transcription from integrated provirus is a highly controlled process, regulated by a number of epigenetic modifications, with recent studies identifying the involvement of ncRNAs. The major HIV-1 transactivation protein, Tat, dramatically enhances viral transcription by binding to a dynamic stem loop structure in viral RNA, the transactivation response element (TAR), which is also coded within the 5′LTR. In contrast, in cells harboring latent HIV-1, transcription is severely restricted and the 5′LTR promoter region carries a specific epigenetic profile, which includes increased histone methylation, decreased histone acetylation, which follows the recruitment of histone deacetylases (HDACs). These biochemical changes involve the histones of two nucleosomes (nuc0 and nuc1) that associate with specific regions of the 5′LTR. In particular, these changes are associated with repositioning of nuc1 (10)(11)(12)(13)(14)(15). In latent virus, nuc1 overlies the transcription start site, whereas in actively transcribing virus, the demethylated/acetylated nucleosome slides upstream of the transcription start site (13).
This review aims to highlight a highly innovative approach to current HIV-1 therapies that utilizes small RNAs, which may represent an alternative strategy to eradication through a functional cure. We will discuss the latest developments in ncRNAs, specifically siRNA and shRNA, for targeting the molecular mechanisms of HIV-1 transcription as well as briefly address humanized murine models, which provide a vehicle to assess potential therapeutics, current gene therapy clinical trials, and future directions of promising therapeutics.

Limitations of Hiv-1 Therapy
Combined antiretroviral therapy (cART) has revolutionized the treatment of HIV-1 infection in the developed world, changing a fatally acute disease into a manageable chronic condition. However, there are significant caveats accompanying lifelong cART therapy, which is currently necessary to control HIV-1 replication. These include the ongoing burden of compliance, drug toxicities, and residual excess morbidity and mortality, mostly due to serious non-AIDS events (16). Whilst effective cART rapidly suppresses the plasma viral load (pVL) to near undetectable levels (17) and permits reconstitution of immune cells (18), the integrated HIV-1 provirus DNA remains essentially unaffected, and therefore provides a viral reservoir, from which recrudescence can occur upon interruption of cART (19,20). The current barrier to HIV eradication is this reservoir that persists in resting memory CD4 + T cells and cells of myeloid lineage (17,(21)(22)(23)(24). While multiple approaches have been explored to surmount this obstacle, including early intervention with cART (25), cART intensification (18,(26)(27)(28)(29)(30), and purging of the reservoir using reactivation strategies (31)(32)(33)(34)(35), the fundamental limitation remains that in the overwhelming majority of patients, the virus rebounds with cessation of cART (36).

Current Therapeutic Approaches to Achieve a Functional Cure
While the ultimate goal is often stated to be the eradication of HIV, that is achieving a sterilizing cure, recent studies have suggested a more realistic approach might be to achieve a functional cure, whereby therapeutic interventions mediate a clinical state of undetectable pVL in the absence of cART. So far, the "Kick and Kill Approach" to eradication is the most studied to date, with the objective of reactivating the latent reservoir while the patient continues cART. The expected outcome is that infected cells would be killed either directly by virus reactivation or through the cytotoxic T-lymphocyte immune response as viral proteins are expressed on the cell surface following reactivation. The presence of cART should prevent new cells becoming infected. The desired result is reduction in or elimination of the latent reservoir. However, despite various interventions including pan T cell activation through OKT3 (37), recombinant IL-2 (38), or IL-7 (34,39), and activation of the protein kinase C (PKC) or protein kinase B pathways using prostatin, bryostatin, or disulfram, respectively (40), these approaches have had only limited or no success in achieving a reduction in the size of the latent reservoir.
Most "kick and kill" approaches have focused on driving virus reactivation by modifying the epigenetic profile of the virus in latent reservoir by using HDAC inhibitors (HDACi). These have included valporic acid (VPA), suberoylanilide hydroxamic acid (SAHA, vorinostat), romidepsin, and panobinostat (41)(42)(43)(44)(45)(46). In vitro studies have shown substantial reactivation of integrated virus in cells from certain patients. However, in vivo the extent of viral reactivation has been limited and these drugs induce substantial off-target effects with significant non-specific host gene activation (47,48). This has been most clearly demonstrated for SAHA, which was observed to inhibit CTL function (49), and would therefore impede CTL mediated "kill" of previously latently infected cells following the HDACi-induced "kick" (49). These findings all point to the need to dissect more precisely the molecular mechanisms involved in HIV-1 latency, particularly reactivation, to develop a more specific and targeted approach in manipulating the viral reservoir.

Post-Transcriptional Gene Silencing of Hiv-1 by si/shRnAs
A major limitation of the post-transcriptional gene silencing (PTGS) approach is the opportunity for viral escape due to targeting at the mRNA level, which allows the transcription process to potentially incorporate resistance mutations in the targeted sequence. To address this limitation, a similar strategy to combat HIV-1 drug resistance has been adopted, using dual or triple combination therapy of anti-HIV-1 shRNAs and/or other anti-HIV-1 gene therapeutics (50)(51)(52).
The first Phase 2 cell-modified gene therapy clinical trial using a combination approach involved a tat-vpr-specific anti-HIV ribozyme, termed OZ1, delivered in autologous CD34 + HPSCs (53). Although there were no significant viral load differences reported between the OZ1 and placebo groups, this study demonstrated cell-modified gene therapy was safe and biologically active, with no adverse events and higher CD4 + cell counts in the OZ1 group (53). A substantially decrease in the blood therapeutic gene level throughout the 100-week trial (53) may provide an explanation for the lack of effect.
A combinatorial approach uses three anti-HIV-1 shRNAs, each specifically targeting highly conserved regions of the HIV Integrase, Protease and tat-rev genes, delivered in a single LV construct termed R3A (52). This study first demonstrated multiple shRNAs being efficiently expressed in a single LV construct, if expression is driven by different promoters for each shRNA, e.g., the human H1, 7SK, or U6 polymerase III promoters and the human U1 polymerase II promoter each driving one of four shRNAs (52). Potent inhibition of HIV-1 was reported in vitro, with combined LV shRNAs constructs showing greater inhibition compared to single LV shRNA constructs (52). More recently, a preclinical in vivo study demonstrated safety of the R3A, triple shRNA expressing construct, using a Balb/c Rag2(−/−) IL-2Rγc(−/−) (BRG) humanized mouse model (54). Future clinical trials are planned to develop this potential therapeutic.
Another combinatorial HIV-1 approach involves shRNA targeting the ccr5 gene and the C46 peptide fusion inhibitor in a LV construct termed LV sh5/C46 (or Cal-1), which has been developed by Calimmune Inc. It will be discussed further in the humanized murine model section.
Transcriptional Gene Silencing of Hiv-1 by si/shRnAs The transcriptional gene silencing (TGS) approach has several distinct advantages over a PTGS approach; first, it directly targets the integrated virus, locking transcription, thus the opportunity for virus escape is minimized; second, epigenetic changes induced during TGS are heritable resulting in daughter cells maintaining the suppressive phenotype, providing a prolonged therapeutic response; and third, the exquisite sequence specificity of promoter-targeted si/shRNAs reduces the chance of non-specific off-target effects and toxicities.
Our extensive investigations have demonstrated TGS of HIV-1 using the promoter-targeted si/shRNA PromA sequence complementary to the 5′LTR tandem NF-κB motifs initially in vitro using cell lines (64,67,(70)(71)(72) and more recently in an in vivo humanized murine model (73). Characterization of the effects of si/shPromA revealed prolonged and profound TGS with up to a 1000-fold decrease in viral replication after a single siRNA transfection or retroviral-delivered shRNA construct. We also reported si/shPromA suppressed viral mRNA expression and de novo virion formation, while proviral DNA was still detected, indicating that virus suppression occurs post-integration. In further studies, we confirmed si/shPromA-mediated suppression of HIV-1 occurred via the TGS pathway using nuclear run-on assays to confirm transcription initiation rates, which clearly distinguished transcriptional suppression in si/ shPromA treated HIV-1 infected nuclei from potential posttranscriptional effects (67,72). Similarly, our investigations using a 3′LTR HIV-1-driven luciferase reporter construct also confirmed a limited contribution of PTGS to the observed virus suppression (72). Finally, chromatin immunoprecipitation (ChIP) analyses provided insight into the mechanism underlying the TGS, demonstrating that silencing was associated with increased histone methylation (H3K9me2 and H3K27me3), decreased histone acetylation, and recruitment of HDAC1 in the 5′LTR promoter region (64,67,72,74). These structural histone changes are consistent with epigenetic-induced latent HIV infection (10)(11)(12)(13)(14)(15)72).
A recent study by Singh et al. using S4-siRNA targeted to the unique triple repeat of NF-κB binding motifs found in subtype C virus demonstrated long-term suppression of HIV-1 (69). The mechanism of silencing was shown in vitro to act through TGS, as determined by ChIP analysis of histone methylation, which revealed enrichment of H3K9me2 and H3K27me3 in S4-siRNA-transfected reporter cell lines (containing the subtype C LTR, luciferase reporter, and subtype C Tat protein expression cassette), but no enrichment in mock-or control siRNA-transfected cells (69). The suppressive effect was confirmed ex vivo in human PBMCs using four different subtype C isolates, which were shown to produce significantly lower p24 levels when transfected with S4-siRNA compared to mock-or control siRNAs (69). Interestingly, while a single mismatch at position 15 of the S4-siRNA sequence and a double mismatch at positions 1 and 15 were sufficient to disrupt the suppressive effect, one particular single mismatch at position 1 of the target sequence showed no disruption of the suppressive effect (69). This is intriguing considering the highly sequence specific nature of siRNAs and our experience of single mismatches disrupting any prolonged suppressive effect with loss of responses by ~6 days (75). Nevertheless, the unique S4-siRNA presents an interesting potential therapeutic approach in the specific targeting of subtype C, which is present in ~50% of the world's HIV-positive population and demonstrates that much is still to be learnt regarding the properties of a targeted sequence that allows effective induction of viral silencing by promotertargeted siRNAs.
We have also recently discovered another novel highly conserved sequence in the 5′LTR promoter region (including subtype C), upstream of siPromA, which when targeted by an si/sh RNA dubbed 143, induces potent transcriptional suppression of HIV associated with epigenetic modifications similar to those induced by siPromA (75). ChIP analyses revealed that the epigenetic changes induced by siRNA143 consisted of increased levels of both H3K27me3 and H3K9me3, reduction in H3K9Ac, and recruitment of Argonaute-1 (75), which are all characteristic heterochromatin marks observed during shRNAinduced TGS. Further, these changes in histone methylation and acetylation are consistent with epigenetic modifications found in latent HIV-1 infection. Following both siRNA PromA and 143 transfections, we observed that silencing can be partially reversed by highly potent, but also highly toxic, HDACi, such as trichostatin A (TSA), but not by those undergoing clinical trial evaluation, such as vironostat (SAHA) (75). Further, TNF at supra-physiological concentrations partially reduced the silencing. Additionally, using the J-Lat 9.2 cell latency model transduced with shPromA and/or sh143, we also observed robust resistance to viral reactivation by various stimuli, including SAHA and/or TNF, used at pharmacological or physiological concentrations (75). These observations are important in the context of the proposed gene-therapy applications as successful therapy will require sustained viral silencing despite activation of CD4 + T cells by inflammatory, homeostatic, or immune response proteins.
Identification of a second effective TGS target provides the opportunity to combine the shRNA targets (shPromA and sh143) in a single therapeutic. This combinatorial approach addresses concerns of any sequence-based target by covering HIV sequence variability. The PromA and si143 sequence targets are both highly conserved, however where variability does occur, the two constructs are complementary; e.g., subtype C viruses have a single nucleotide deletion at position 14 of the PromA target, while the sh143 sequence target has minimal subtype C variability. Thus, a multiplexed approach, delivered by a single LV construct expressing multiple shRNAs simultaneously, would likely provide increased viral coverage and more entrenched enforcement of epigenetic changes, which more robustly resist viral reactivation from alterations in the host's inflammatory or immunological status. We are currently testing these constructs in humanized mouse models, described below.

Humanized Mouse Models for Assessing Potential Hiv-1 Therapeutics
Animal models for assessing HIV-1 therapeutics include various humanized murine models and non-human primates. Although the latter species contain host-restriction factors that impede HIV-1 replication and experiments performed using this model must instead use the closely related Simian immunodeficiency virus or chimeric Simian/HIV (SHIV) (76). We recently utilized a (NOD)/SCID/Janus kinase 3 (NOJ) knockout humanized mouse model to demonstrate in vivo TGS activity of shPromA, delivered via a LV (Figure 1) (73). NOJ knockout mice were reconstituted with human PBMCs transduced with the shPromA carrying lentiviral construct, which was processed into mature siPromA by cellular ribonucleases (77). HIV-1JRFL challenge of mice reconstituted with the PromA-M2 inactive control transduced PBMCs showed acute HIV-1 infection (Figure 1A) as determined by high pVL, CD4 + T cell depletion and extensive immunodeficiency (78,79). In stark contrast, mice reconstituted with shPromAtransduced PBMCs demonstrated significantly lower pVL and normal human CD4 + to CD8 + T cell ratios in mononuclear cells recovered from the peritoneal cavity and spleen at sacrifice 14 days post HIV-1 challenge (Figure 1B) (73). This corresponds to a protective effect in the form of an induced HIV-1 "latent-like" state, which locks down active virus transcription even in this acute model of HIV-1 infection.
Our aim is to generate sufficient data using the humanized mouse model to merit further therapeutic development of these TGS-inducing constructs. An alternative humanized murine model, which circumvents the highly acute infection reported in NOJ knockout mice, is the humanized BLT murine model. This model uses the non-obese diabetic (NOD)/SCID, common gamma chain −/− (NSG) mouse, humanized with implanted bone marrow, fetal liver, and thymus tissue (BLT) and results in systemic repopulation with human T cells, B cells, monocytes/ macrophages, and dendritic cells (76,80). This system has been used to perform in vivo studies of autologous CD34 + HSPCs transduced with the H1-CCR5 shRNA 1005 vector and showed effective down-regulation of the HIV-1 co-receptor, CCR5, which protected mouse-derived human splenocytes ex vivo (80) and CD4 + T cells in vivo (81) from CCR5-tropic HIV-1 infection. This CCR5-shRNA vector has also been analyzed in non-human primates by delivery through HSPC transplantation (82). Although this required a single nucleotide mutation in the human CCR5 shRNA 1005 sequence to match the rhesus macaque CCR5 target sequence, this analysis successfully showed specific inhibition of rhesus macaque CCR5 expression, with no modulation of human CCR5 expression observed (82).
The H1-CCR5 shRNA 1005 vector has been further developed to include an additional anti-HIV target in the form of a sequence that encodes for the C46 fusion inhibitor. This HIV-1 entry inhibitor is a mimetic derived from the peptide sequence of the C-terminal heptad repeat of HIV-1 gp41, which interacts with the N-terminal coiled-coil domain of the intermediate HIV-1 gp41 to block the six-helix bundle formation and subsequent fusion between HIV-1 envelope and the host cell membrane. C46 has been tested in Phase 1 clinical trials without any adverse effects in HIV-1 positive patients infused with autologous T cells transduced with C46 expressing retrovirus vector (83). Combination of H1-CCR5 shRNA 1005 and C46 into a single LV termed LVsh5/C46 (or Cal-1) has been tested in a preclinical trial discussed below and is currently being assessed in a Phase 1 trial in patients with chronic HIV-1 infection (NCT01734850). We are planning to assess the new sh143 target in combination with shPromA contained within the Cal-1 lentivirus construct backbone in the BLT model Non-coding RNA gene therapy approaches Frontiers in Immunology | www.frontiersin.org  Gene Therapy Approaches to Hiv-1 Treatment In 2009, the extraordinary success of utilizing bone marrow transplant to deliver HPSC from a CCR5-Δ32 homozygous donor opened the door for gene therapy approaches to HIV-1 treatment. The "Berlin patient" effectively received a functional cure for HIV-1, and currently 6 years post-transplantation, does not require ART and has no detectable pVL or proviral DNA (84,85). Although this truly represents a modern day medical success, the circumstances of aggressive malignant disease and the rarity of matching HLA/CCR5-Δ32 homozygous donors present a formidable challenge in repeating this feat [reviewed in Ref. (50)] Further, this approach has been tried in six other patients without success (86). There are, however, several gene therapy clinical trials in various stages of completion involving HIV-1 infected patients with non-malignancies using constructs aimed at knocking down CCR5 to protect cells from HIV-1 infection. These current gene therapy clinical trials are summarized in Table 1.
One of the preclinical studies assessing safety and efficacy of an anti-HIV-1 lentiviral vector containing CCR5 shRNA and the C46 fusion inhibitor, termed Cal-1, has recently been reported (87). Effective delivery of LVsh5/C46 (Cal-1) was demonstrated in human T cell lines, PBMCs, CD4 + T cells, and CD34 + HSPCs, with both the CCR5-shRNA and C46 peptide being stably expressed in the target cells (87). Importantly, the study showed the Cal-1 construct was able to effectively protect gene-modified cells from both CXCR4-and CCR5-tropic HIV-1 strains (87). Further, the Cal-1 construct treatment was shown to be nontoxic, non-inflammatory, and had no adverse effect on HSPC differentiation (87). These encouraging data have led to the Cal-1 construct currently being tested in a Phase 1/2 clinical trial for the treatment of HIV-1 ( Table 1).
Our envisaged future application of the TGS-inducing PromA and 143 sequences as a proposed therapeutic strategy is outlined in Figure 1C. The approach would be to use shPromA/sh143 TGS-inducing constructs to enforce latency in HIV-1 positive individuals who have cART-suppressed pVL and latent viral reservoirs. This strategy may provide a functional cure, by inducing and enforcing HIV-1 latency, thereby maintaining transcriptionally inactive virus and effectively render patients free from cART.

Conclusion
In terms of approaches to a functional cure for HIV-1, ncRNAmediated transcriptional regulation of HIV-1, particularly in the context of the viral reservoir, is starkly juxtaposed to current activation approaches, which rely on pan T-cell activation or extensive histone acetylation modification and are associated with substantial off-target responses. Instead, the ncRNA approach relies on sequence specificity to provide a highly focused approach in manipulation of the latent reservoir. Developing ncRNAtherapeutic approaches to control HIV-1 may have the potential to enforce HIV-1 latency and block initial infection, allowing control of the viral reservoir, free from traditional antiretroviral therapies.

Author Contributions
CA, AK, KS, and GS wrote the manuscript; and KM prepared the figures.