HIV-1 persists in a latent reservoir (LR) despite antiretroviral therapy (ART) 1–5 . This reservoir is the major barrier to HIV-1 eradication 6,7 . Current approaches to purging the LR involve pharmacologic induction of HIV-1 transcription and subsequent killing of infected cells by cytolytic T lymphocytes (CTL) or viral cytopathic effects 8–10 . Agents that reverse latency without activating T cells have been identified using in vitro models of latency. However, their effects on latently infected cells from infected individuals remain largely unknown. Using a novel ex vivo assay, we demonstrate that none of the latency reversing agents (LRAs) tested induced outgrowth of HIV-1 from the LR of patients on ART. Using a novel RT-qPCR assay specific for all HIV-1 mRNAs, we demonstrate that LRAs that do not cause T cell activation do not induce significant increases in intracellular HIV-1 mRNA in patient cells; only the PKC agonist bryostatin-1 caused substantial increases. These findings demonstrate that current in vitro models do not fully recapitulate mechanisms governing HIV-1 latency in vivo. Further, our data indicate that non-activating LRAs are unlikely to drive the elimination of the LR in vivo when administered individually. HIV-1 cure is hindered by viral persistence in a small fraction (~1/106) of resting CD4+ T cells (rCD4s) that harbor latent but replication-competent proviruses 1–3 . Upon cellular activation, latency is reversed and replication-competent virus is produced. Although T cell activation reverses latency, global T cell activation is toxic, generating interest in small molecule latency-reversing agents (LRAs) that do not activate T cells. Due to the low frequency of latently infected rCD4s in vivo, cell models have been used to identify a number of mechanistically distinct LRAs. These include: (1) histone deacetylase (HDAC) inhibitors, thought to function through epigenetic and other mechanisms 11–14 ; (2) disulfiram, postulated to involve nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) 15,16 ; and (3) the bromodomain-containing protein 4 (BRD4) inhibitor JQ1, which elicits effects through positive transcription elongation factor (P-TEFb) 17–20 . Acting through signaling pathways associated with T cell activation, protein kinase C (PKC) agonists such as phorbol esters, prostratin 21–23 and bryostatin-1 12,24–26 also reverse latency in cell models. Evidence that putative LRAs reverse latency ex vivo in primary rCD4s from HIV-1-infected individuals is limited; disulfiram and the HDAC inhibitor vorinostat have been tested in patient cells with inconsistent results 11,13,16,27,28 . Clinical trials in patients on ART are ongoing with disulfiram and the HDAC inhibitors vorinostat, romidepsin, and panobinostat 27,29 . A recent trial of disulfiram showed no consistent evidence of latency reversal 30 . In another clinical trial, a single dose of vorinostat modestly increased intracellular RNAs containing HIV-1 gag sequences in rCD4s of patients on ART 27 . Ex vivo treatment of patient cells with vorinostat induced outgrowth in some studies 11,13 but no virion production in another study 28 . Importantly, no LRA has been shown to reduce the size of the LR. A consistent ex vivo validation strategy has not been employed to compare putative LRAs. Given the costs and risks associated with clinical trials, such a strategy is important for HIV-1 eradication research. Therefore, we utilized three independent assays to evaluate the efficacy of LRAs in cells from HIV-1 infected individuals on suppressive ART (participant characteristics in Supplementary Table 1). We first tested LRAs in a modified viral outgrowth assay 1 . In the original assay, patient-derived rCD4s were activated and co-cultured with CD4+ T lymphoblasts from healthy donors to expand released virus. Induction of outgrowth provides conclusive evidence of latency reversal. In the modified assay, T cell activation was replaced with LRA treatment. The subsequent co-culture of patient rCD4s with healthy donor lymphoblasts constitutes a mixed lymphocyte reaction, which induces background reactivation of latent HIV-1 31 and complicates LRA evaluation. Therefore, we treated rCD4s with LRAs and then cultured the cells with a transformed CD4+ T cell line (MOLT-4/CCR5) (Fig. 1a) that supports robust HIV-1 replication but does not induce allogeneic stimulation of rCD4s (Supplementary Fig. 1a–c). We treated five million purified rCD4s from infected individuals on ART with single LRAs for 18 h and then co-cultured the cells with MOLT-4/CCR5 cells for 14 days to permit viral outgrowth. T cell activation with phorbol 12-myristate 13-acetate + ionomycin (PMA/I) served as a positive control. We concurrently measured the frequency of latently infected cells 32 . We evaluated vorinostat, romidepsin, panobinostat, disulfiram and bryostatin-1 at clinically relevant concentrations that effectively reversed latency in a primary cell model (see below) and that were not toxic to rCD4s. No drug treatment induced cell death as shown by the lack of 7-AAD staining (Fig. 1b). Surprisingly, none of the LRAs induced viral outgrowth from cells from any individual tested while PMA/I-treated cultures were positive for every patient with a detectable LR (Fig. 1c). We next asked whether LRA treatment induced rapid virus release. We collected culture supernatants from rCD4s from five infected individuals (S26–S30) after 18 h of LRA treatment and prior to addition of MOLT-4/CCR5 cells for measurement of viral outgrowth. PMA/I induced virus release as detected by HIV-1 mRNA in the supernatant from four out of five individuals (S26–S29) (Fig. 1D). Bryostatin-1 treatment induced detectable supernatant HIV-1 mRNA from one infected individual (S27), whereas no other LRA had a measurable effect (Fig. 1d). None of the LRAs induced subsequent viral outgrowth from these treated cells, including the cells from the single individual (S27) that released HIV-1 mRNA after bryostatin-1 treatment (Fig. 1c). The most widely used method to detect induction of HIV-1 transcription 16,27 in cells from infected individuals involves the measurement of RNAs containing HIV-1 gag sequences. Because this method lacks a stringent selection for poly-adenylated RNAs, it does not exclusively detect fully elongated and correctly processed HIV-1 mRNAs. Therefore, we devised a new assay specific for intracellular HIV-1 mRNA using a primer/probe set that detects the 3′ sequence common to all correctly terminated HIV-1 mRNAs (Fig. 2a). We detected baseline intracellular HIV-1 mRNA in rCD4s from ten out of 11 infected individuals. Stimulation with PMA/I for 18 h dramatically increased intracellular HIV-1 mRNA (mean increase = 115.5-fold, Fig. 2b). However, at clinically relevant concentrations that reverse latency in a primary cell model (Fig. 3B, C), vorinostat, romidepsin, panobinostat, disulfiram, and JQ1 failed to increase intracellular HIV-1 mRNA in rCD4s from infected individuals when used as single agents (Fig. 2b, c). Bryostatin- 1 caused significant increases in some infected individuals (Fig. 2c). We observed similar results after 6 h of LRA treatment (Supplementary Fig. 2). While no effect was seen in latently infected cells from infected individuals, LRA treatment increased intracellular HIV-1 mRNA in a B-cell lymphoma 2 (BCL-2) transduced primary rCD4 model of latency (Fig. 3a). LRA-induced increases in HIV-1 mRNA were consistent with measurements of the fraction of cells that up-regulate HIV-1 gene expression, as assessed by GFP reporter (Fig. 3b). The frequency of latent infection in this model is substantially higher than that observed in vivo 4 . To confirm that our assay effectively detects intracellular HIV-1 mRNA increases at frequencies of latent infection seen in vivo, we treated model cells with a known percentage of latent infection and then serially diluted these cells into rCD4s from uninfected individuals immediately prior to RNA isolation. We detected proportionate increases in intracellular HIV-1 mRNA in vorinostat-treated cells down to a frequency of 1/106 cells (Fig. 2d, e). Therefore, the lack of LRA efficacy in cells from HIV-1 infected individuals is not a result of assay insensitivity. Rather, our findings demonstrate that freshly isolated latently infected cells from infected individuals responded differently to LRAs than latency model cells. RT-qPCR assays that detect gag-containing sequences in total RNA are frequently used to detect latency reversal. These sequences do not necessarily represent bona fide unspliced HIV-1 mRNA. HIV-1 integrates into host genes that are actively transcribed in rCD4s 33,34 , allowing for the production of chimeric host/HIV-1 primary transcripts. Such transcripts, initiated at host promoters, could contain gag sequence and would be indistinguishable from LTR-initiated transcripts by conventional gag RT-qPCR assays (Fig. 4a). We therefore designed a primer/probe set that amplifies a region of the LTR that is not transcribed during LTR-initiated and correctly terminated HIV-1 transcription. This primer/probe set is specific for transcripts containing read-through of the 5′ LTR or 3′ LTR, independent of proviral orientation (Fig. 4a). We treated ten million rCD4s from infected individuals on ART with vorinostat or PMA/I for 6 h and compared the levels of HIV-1 mRNA, read-through transcripts, and transcripts containing gag sequence (Fig. 4a, b). We detected a small increase (~2-fold) in transcripts containing gag sequence in vorinostat-treated rCD4s from four out of five infected individuals, consistent with previous reports 27 (Fig. 4b). Vorinostat treatment also induced increases in read-through transcripts (Fig. 4b) comparable to the increases in transcripts containing gag sequence but had no effect on levels of HIV-1 mRNA (Fig. 4b). To prove that the read-through signal is amplified from a transcript that initiated upstream of the 5′ LTR and contains gag sequence, we primed cDNA synthesis with a gag primer (Fig. 4c). We detected comparable, statistically significant inductions of read-through and gag transcripts after 6 h of vorinostat treatment (Fig. 4d) (P = 0.027, P = 0.011, respectively; ratio paired t-test of transcript copies), indicative of read-through transcription. PMA/I induction of gag transcripts greatly exceeded that of read-through transcripts, indicative of LTR-initiated transcription (Supplementary Fig. 3). While not every potential LRA will induce read-through transcription by activating a host gene, our data show that chimeric host/HIV-1 transcripts can have a confounding effect on the RT-qPCR signal obtained with standard gag primers. Such an effect should be taken into consideration when evaluating LRAs using conventional gag RT-qPCR assays. The novel assays presented herein facilitated the first comparative ex vivo evaluation of candidate LRAs. Our data demonstrate that none of the leading candidate non-T cell activating LRAs tested significantly disrupted the LR ex vivo. The striking discordance between the effects of non-stimulating LRAs in in vitro models of HIV-1 latency and the ex vivo effects in rCD4s from infected individuals on ART indicates that these models do not fully capture all mechanisms governing HIV-1 latency in vivo. These compounds are unlikely to drive the elimination of the LR in vivo when administered individually. The only active single agent was the PKC agonist bryostatin-1, which is likely too toxic for clinical use. Whether other PKC agonists or other compounds that stimulate signaling pathways associated with T cell activation can be safely administered remains to be seen, and further progress may depend on finding safe and active combinations of LRAs. Methods Cell isolation and culture The Johns Hopkins Institutional Review Board approved this study and all research participants in this study gave written informed consent. Infected individuals were enrolled under the criteria of suppression of viremia to undetectable levels ( 0.5. We determined that the limit of quantification for all transcripts was 10 copies. A PCR signal of less than 10 copies (1–9 copies) was treated as 10 copies in calculations of fold change and marked as 10 copies on graphs depicting RNA copies. Undetectable PCR signal was treated as 10 copies in calculations of fold change and marked as 1 copy on graphs depicting RNA copies. Levels of RNA polymerase II (Pol2) and Glucose-6-phosphate dehydrogenase (G6PD) RNA were also measured for each sample to use as an endogenous control. Voronistat, romidepsin, panobinostat, JQ1 and PMA/I treatment consistently increased expression Pol2 and G6PD. Samples treated with the same drug had even levels of Pol2 and G6PD, indicating that the template inputs were approximately equal. Measurement of supernatant HIV-1 mRNA HIV-1 mRNA was extracted from 0.2mL of supernatant from five million cultured rCD4s after 18 h of LRA treatment using the ZR-96 Viral RNA kit (Zymo Research). cDNA synthesis was performed using qScript cDNA Supermix (Quanta Biosciences). Real-time PCR was performed using TaqMan Fast Advanced mastermix (Applied Biosystems) on an ABI Viia 7 Real-Time PCR machine. Primers and probes listed below. Manufacturer’s thermal cycling conditions were used. Molecular standard curve was generated as described above. Primer and probe sequences Nucleotide coordinates are indicated relative to HXB2 consensus sequence. HIV-1 mRNAs were detected using the following primers and probe, modified from Shan et al. 37 : Forward (5′→3′) CAGATGCTGCATATAAGCAGCTG (9501–9523) Reverse (5′→3′) TTTTTTTTTTTTTTTTTTTTTTTTGAAGCAC (9629-poly A) Probe (5′→3′) FAM-CCTGTACTGGGTCTCTCTGG-MGB (9531–9550) Transcripts containing HIV-1 gag sequence were detected using the following primers and probe, described previously 27 . Forward (5′→3′) ACATCAAGCAGCCATGCAAAT (1368–1388) Reverse (5′→3′) TCTGGCCTGGTGCAATAGG (1453–1471) Probe (5′→3′) VIC-CTATCCCATTCTGCAGCTTCCTCATTGATG-TAMRA (1401–1430) Chimeric host/HIV-1 read-through transcripts were detected using the following primers and probe: Forward (5′→3′) CAGATGCTGCATATAAGCAGCTG (416–438, 9501–9523) Reverse (5′→3′) CACAACAGACGGGCACACAC (556–575, 9641–9660) Probe (5′→3′) FAM-CCTGTACTGGGTCTCTCTGG-MGB (446–465, 9531–9550) cDNA synthesis reaction with gag primer sequence: Reverse (5′→3′) GTCACTTCCCCTTGG (1480–1494) Supplementary Material 1
