Ex vivo analysis identifies effective HIV-1 latency-reversing drug combinations

Abstract

Reversal of HIV-1 latency by small molecules is a potential cure strategy. This approach will likely require effective drug combinations to achieve high levels of latency reversal. Using resting CD4+ T cells (rCD4s) from infected individuals, we developed an experimental and theoretical framework to identify effective latency-reversing agent (LRA) combinations. Utilizing ex vivo assays for intracellular HIV-1 mRNA and virion production, we compared 2-drug combinations of leading candidate LRAs and identified multiple combinations that effectively reverse latency. We showed that protein kinase C agonists in combination with bromodomain inhibitor JQ1 or histone deacetylase inhibitors robustly induce HIV-1 transcription and virus production when directly compared with maximum reactivation by T cell activation. Using the Bliss independence model to quantitate combined drug effects, we demonstrated that these combinations synergize to induce HIV-1 transcription. This robust latency reversal occurred without release of proinflammatory cytokines by rCD4s. To extend the clinical utility of our findings, we applied a mathematical model that estimates in vivo changes in plasma HIV-1 RNA from ex vivo measurements of virus production. Our study reconciles diverse findings from previous studies, establishes a quantitative experimental approach to evaluate combinatorial LRA efficacy, and presents a model to predict in vivo responses to LRAs.

Figure 8. Mathematical model relating ex vivo virus release to predicted increases in plasma HIV-1 RNA levels in vivo.

A viral dynamic model (A, detailed in Supplemental Materials) was used to estimate changes in plasma HIV-1 RNA levels in response to the LRA treatments for which ex vivo data on virus release was available. Arrows depict routes from latently infected cells to productively infected cells after exposure to antigen or LRAs. Crosses indicate elimination/death. (B) Predicted peak plasma HIV-1 RNA levels during LRA treatment. For each LRA treatment, median fold change in supernatant HIV-1 versus the DMSO control (x axis) was used to estimate LRA-driven activation rate a′; this parameter estimate was used to predict peak plasma viral load following continuous administration of the LRA (y axis). (C) Predicted time course of viral load (y axis, log scale) following administration of single-dose LRA treatment that remains active for 1 day. (D) Predicted time course of viral load (y axis, log scale) following administration of single-dose romidepsin that remains active for 1 day (solid lines) or that continues indefinitely (dotted lines). Gray shading in C and D indicates duration of LRA activity. Parameters: d_y = 1/day; d′_y = 1 day^-1 (blue curves in B and D, all curves in C) or one-third day^-1 (red curves in B and D); a + d_z = 5.2 × 10^–4 day^–1 (reservoir half-life of 44 months), initial viral load = 2 copies/ml. For other parameters, see Supplemental Materials.

Figure 7. PKC agonists alone or in combination with another LRA do not induce substantial cytokine production.

Primary (A) rCD4s or (B) PBMCs were treated with a single LRA, or a combination of 2 LRAs were assayed for supernatant cytokine concentrations (pg/ml). Data are the mean effect of 2 or 3 independent experiments.

Figure 6. Effect of LRA treatment on T cell activation–associated surface markers and toxicity.

Primary rCD4s treated with a single LRA or a combination of 2 LRAs were assayed for (A) surface expression of CD25 and CD69 and (B) positivity for annexin V and 7-AAD staining. Data are the mean effect of 2 or 3 independent experiments. Error bars represent SEM.

Figure 5. Correlation between intracellular and extracellular HIV-1 mRNA after ex vivo LRA treatment.

Plot of intracellular HIV-1 mRNA copy number against supernatant HIV-1 mRNA copy number after exposure of rCD4s from the same infected individual to treatments containing (circles) or lacking (triangles) a PKC agonist. For PKC agonist–containing treatments, a statistically significant correlation was demonstrated by Tobit regression analysis. P = 0.008, χ² test. See Methods.

Figure 4. PKC agonists alone or in combination with other LRAs induce HIV-1 virus release by rCD4s from infected individuals on ART.

HIV-1 virion levels in the culture supernatant of rCD4s from infected individuals 24 hours after addition of a single LRA or a combination of 2 LRAs, presented as (A) HIV-1 mRNA copies/ml supernatant and (B) as a percentage of the effect of maximal reactivation with PMA/I. Dotted line indicates limit of detection (150 copies per ml). Numbers in parentheses indicate number of individuals used for each treatment. Error bars indicate mean ± SEM. Statistical significance was calculated from the HIV-1 mRNA copy number values using a ratio paired t test compared with (A) DMSO control, or (B) bryostatin-1 or prostratin alone. (C) Calculation of synergy for LRA combinations using the Bliss independence model. Data are presented as the difference between the observed and predicted fraction of supernatant HIV-1 mRNA levels in copies/ml induced by LRA combinations relative to PMA/I (f_a). See Methods for more detail. Statistical significance for the experimental f_a was calculated using a ratio paired t test compared with the predicted f_a for each combination. *P < 0.05; ***P < 0.0005; ****P < 0.00005.

Figure 3. Lower doses of bryostatin-1 synergize with romidepsin to reverse latency.

(A) Intracellular HIV-1 mRNA levels in rCD4s, obtained from infected individuals and treated ex vivo with bryostatin-1 (1 nM or 10 nM) alone or in combination with romidepsin, presented as fold induction relative to DMSO control. Statistical significance was calculated from the HIV-1 mRNA copy number values using a ratio paired t test compared with the DMSO control. **P < 0.005; ***P < 0.0005. (B) Calculation of synergy for bryostatin-1 (1 nM) and romidepsin using the Bliss independence model. Data are presented as the difference between the observed and predicted fractional response relative to PMA/I (f_a). See Methods for more details. Statistical significance for the experimental f_a was calculated using paired t test compared with the predicted f_a for each combination. *P < 0.05. rCD4s from 4 HIV-1–infected individuals were tested per condition.

Figure 2. PKC agonists synergize with JQ1 and with HDAC inhibitors to significantly increase HIV-1 mRNA expression in rCD4s from infected individuals on ART.

Calculation of synergy for LRA combinations using the Bliss independence model. Data are presented as the difference between the observed and predicted fractional response relative to PMA/I (fraction affected, f_a) presented in Figure 1. See Methods for more details. Numbers in parentheses indicate number of individuals used for each treatment. Data points represent the mean effect of 2 or 3 replicate LRA treatments of 5 million cells for each individual. Statistical significance for the experimental f_a was calculated using ratio paired t test compared with the predicted f_a for each combination. **P < 0.005; ***P < 0.0005.

Figure 1. Combination LRA treatment robustly increases HIV-1 mRNA expression in rCD4s from infected individuals on ART.

(A) Intracellular HIV-1 mRNA levels in rCD4s, obtained from infected individuals and treated ex vivo with a single LRA or a combination of 2 LRAs, presented as fold induction relative to DMSO control. Numbers in parentheses indicate number of individuals used for each treatment. (B) Induction of intracellular HIV-1 mRNA by single LRAs, PKC-agonist–containing LRA combinations, and disulfiram-containing LRA combinations presented as a percentage of the effect of maximal reactivation with PMA/I. Data points represent the mean effect of 2 or 3 replicate LRA treatments of 5 million cells for each individual. For A and B, statistical significance was calculated from the HIV-1 mRNA copy number values using a ratio paired t test compared with (A) the DMSO control, (B) bryostatin-1 or prostratin alone, or disulfiram alone. *P < 0.05; **P < 0.005; ***P < 0.0005; ****P < 0.00005. Error bars represent SEM.