Histone deacetylase inhibitors impair the elimination of HIV-infected cells by cytotoxic T-lymphocytes

Abstract

Resting memory CD4+ T-cells harboring latent HIV proviruses represent a critical barrier to viral eradication. Histone deacetylase inhibitors (HDACis), such as suberanilohydroxamic acid (SAHA), romidepsin, and panobinostat have been shown to induce HIV expression in these resting cells. Recently, it has been demonstrated that the low levels of viral gene expression induced by a candidate HDACi may be insufficient to cause the death of infected cells by viral cytopathic effects, necessitating their elimination by immune effectors, such as cytotoxic T-lymphocytes (CTL). Here, we study the impact of three HDACis in clinical development on T-cell effector functions. We report two modes of HDACi-induced functional impairment: i) the rapid suppression of cytokine production from viable T-cells induced by all three HDACis ii) the selective death of activated T-cells occurring at later time-points following transient exposures to romidepsin or, to a lesser extent, panobinostat. As a net result of these factors, HDACis impaired CTL-mediated IFN-γ production, as well as the elimination of HIV-infected or peptide-pulsed target cells, both in liquid culture and in collagen matrices. Romidepsin exerted greater inhibition of antiviral function than SAHA or panobinostat over the dose ranges tested. These data suggest that treatment with HDACis to mobilize the latent reservoir could have unintended negative impacts on the effector functions of CTL. This could influence the effectiveness of HDACi-based eradication strategies, by impairing elimination of infected cells, and is a critical consideration for trials where therapeutic interruptions are being contemplated, given the importance of CTL in containing rebound viremia.

Conflict of interest statement

The authors have declared that no competing interests exist.

Figures

Figure 1. Effects of HDACis on the viability of PBMC CD8+ and CD4+ T-cells.

A–C. PBMC were treated with HDACi drugs at the indicated concentrations for either 4 or 21 hours (drugs left in) then stained with Annexin-V-Fitc (stains phosphatidyl serine on apoptotic cells), 7-AAD (stains DNA of dead cells), CD4 pacific blue, and CD8 alexa-fluor 700 and analyzed by flow cytometry. A. Shown are representative flow cytometry data indicating gating on dead (Annexin-V+7-AAD+) and dying (Annexin-vbright7-AADdim/−) cells. B, C. Shown are summary data for treatments PBMC gated on CD8+ cells (left panel) or CD4+ cells (right panel) with romidepsin and panobinostat (B) or with SAHA (C) at the indicated doses for the indicated times. P values for comparisons between the multiple doses of romidepsin tested and the untreated condition were calculated by Kruskal-Wallis test and found to be significant for ex vivo CD4+ T-cells (p = 0.0080). Post-hoc Dunn's multiple comparison tests were performed for each of these and the indicated p values are adjusted for multiple testing. Other drug treatment conditions and cell types were not significant by Kruskal-Wallis tests. D. The effects of HDACi treatment on the viability of an HIV-Gag-SLYNTVATL-specific T-cell clone were determined in the same manner as for PBMC, including use of the same statistical tests. Error bars represent SD. * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

Figure 2. Romidepsin and panobinostat are disproportionately toxic to activated CD4+ and CD8+ T-cells.

A. HIV-Env-VL11 and HIV-Gag SL9 specific CD8+ CTL clones were isolated from subjects OM292 and OM9 respectively, and exposed to HDACis at the indicated concentrations for 4 hours. Cells were then washed to remove drugs and replated in fresh R10 medium supplemented with 50 U/ml IL-2 for an additional 17 hours. Cell viability was measured by 7-AAD staining and flow cytometry. B. PBMC from subject OM292 either directly ex vivo (left panel) or following 48 hours of stimulation with anti-CD3 and anti-CD28 (right panel) were exposed to HDACis at the indicated concentrations for 4 hours. Cells were then washed to remove drugs and replated in fresh R10 medium. Cell viability was then measured by 7-AAD staining and flow cytometry. For A and B, shown are graphs of mean values taken from triplicate wells ± SEM following subtraction of background (% 7-AAD+ in no drug controls). P values were calculated by two-way ANOVA with Dunnett's multiple comparison test (comparing to the no drug control) * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001. C. In parallel to the viability assays shown in A and B, ACH2 cells were treated with HDACis for 4 hours, then washed and replated in fresh medium. Shown are mean ± SEM values for % HIV-Gag+ cells as measured in triplicate by intracellular staining and flow cytometry.

Figure 3. HDAC inhibitors impair IFN-γ production from PMA/ionomycin stimulated CD4+ and CD8+ T-cells.

A, B. PBMC from subject OM292 were exposed to HDACis for 4 hours. Cells were then washed and cultured in fresh medium with PMA/ionomycin for an additional 5 hours in the presence of brefeldin A. A. IFN-γ production in CD4+ and CD8+ T-cells was measured by intracellular cytokine staining flow cytometry. Shown are mean ± SEM values. P values were calculated by two-way ANOVA with Dunnett's multiple comparison test (comparing to the no drug control) * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001. B. Viability in CD4+ and CD8+ T-cells was measured by Annexin-V and 7-AAD staining flow cytometry. Shown are mean ± SEM values.

Figure 4. HDAC inhibitors impair IFN-γ production from antigen-stimulated CD8+ T-cells.

A, B. PBMC or CTL clones from chronically HIV-infected ARV-treated subjects (OM265 and OM292) were cultured for 2 hours with the indicated drugs and then plated for IFN-γ ELISPOT assays with 12 hour peptide stimulation periods. A. Treated CTL clones were stimulated with overlapping 15mer peptides spanning the CMV-pp65 protein (left panel) or the MHC-I-A24 restricted HIV-Nef peptide ‘RW8’ (right panel). Each treatment condition was tested in duplicate. Shown are mean SFC/104 CTL clone cells for each condition with error bars representing SEM. B. Experiments analogous to those depicted in A were performed for 8 different optimal HIV CD8+ T-cell epitopes (5 for OM292 and 1 for OM265). Shown are summary data for all 8 responses depicting the mean SFC/106 PBMC (of triplicate wells) under different treatment conditions. For SAHA, panobinostat, and romidepsin conditions statistical significance for each drug was evaluated using two-way ANOVA tests, and p values were calculated using Dunnett's multiple comparison test to account for the use of three different drug concentrations. For IL-15SA p values were calculated using the Wilcoxon matched pair test. C. PBMC from subjects OM265 and OM292 were treated with SAHA or romidepsin at the indicated concentrations for 2 hours and then either: stimulated with peptides for 6 a hour ELISPOT assay, or washed, cultured for 24 hours in the absence of drugs, and stimulated with peptides for a 12 hours ELISPOT assay. P values were calculated by two-way ANOVA with Dunnett's multiple comparison test (comparing to the no drug control) * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

Figure 5. HDAC inhibitors impair proliferation of virus-specific T-cells.

PBMC from 11 chronically HIV-infected subjects on suppressive ARV therapy were labeled with CFSE and treated with the indicated drugs for 4 hours. HIV-Gag or CMV-pp65 peptide pools were then added to final concentrations of 1 µg/ml/peptide. 48 hours later, cells were washed thoroughly, re-suspended in medium with 20 U/ml IL-2, and cultured for an additional 5 days. Cells were then stained with a viability dye and antibodies to CD3, CD8, and analyzed by flow cytometry. A. Shown are representative flow cytometry data of a CMV-pp65 response gated on viable, CD3+CD8+ lymphocytes depicting CFSE (diminution indicates proliferation) – x-axis by CD8 – y-axis. B–E. Summary data depicting frequencies of CFSEdim cells within viable, CD3+CD8+ or CD3+CD8− (CD4− T-cell) lymphocyte populations. F. Shown are data analogous to B–E, but from a separate experiment where peptides were added back with fresh medium following HDACi wash-out at 48 hours. P values were calculated by the Wilcoxon matched-pairs signed rank test.

Figure 6. HDAC inhibitors impair CTL killing of HIV-infected primary CD4+ cells.

CD4+ T-cells were enriched from HIV-uninfected A02+ donors and either infected with HIV JR-CSF or maintained as mock infected controls. 24 hours post-infection, these target cells were co-cultured with an HIV-Gag- or CMV-pp65-specific CTL clones at the indicated effector (clone)∶target (CD4+ cell) ratios for 16 hours. Cells were then stained with fluorochrome conjugated antibodies to CD8, CD4, and HIV-Gag (intracellular staining) and analyzed by flow cytometry. A. Shown is the gating strategy utilized for killing assays, with the placement of Gag+ and Gag+CD4dim gates determined based on a mock-infected control (right panel). B. Data from a representative experiment with indicated effector∶target ratios of an HIV-Gag-SLYNTVATL-specific T-cell clone. Panels C-F utilize total Gag+ cells and calculate a % Infected Cells Killed as (%Gag+ No Effectors - %Gag+ at Given E∶T ratio)/%Gag+ No Effectors*100, ex. in panel B %Infected Cells Killed at 1∶36 ratio = (22.1−6.7)/22.1*100 = 69.7%. C–F. Shown are the results from 4 independent experiments. In C CTL were treated with romidepsin or panobinostat at 50 nM or with SAHA at 500 nM for 2 hours, and then co-cultured with target cells without a washing step. In D–F CTL were treated with romidepsin or panobinostat at 25 nM or with SAHA at 500 nM for 6 hours, then washed thoroughly before initiating a 16 hour co-culture with target cells. D and E both utilize an HIV-Gag-SLYNTVATL-specific CTL clone at high and low E∶T ratios respectively. E utilizes an HIV-Gag-FLGKIWPSHK-specific CTL clone. C–F. Shown are means ± SEM of data from triplicate wells. P values were calculated by two-way ANOVA with Tukey's multiple comparison test.

Figure 7. Impairment of CTL killing is sustained for at least 14 hours after removal of romidepsin or panobinostat.

An HIV-Gag-specific CTL clone was treated with panobinostat, romidepsin, or SAHA at the indicated doses for 5 hours, or maintained as an untreated control. Cells were then washed two times, resuspended in R10–50, and cultured for 4 or 14 hours. These CTL were then co-cultured with autologous HIV-infected CD4+ T-cells at a clone∶target ratio of 1∶10 for 8 hours, and levels of infection were measured as described above (see Fig. 4 legend). A. Shown are flow cytometry plots gated on the viable (FSC/SSC) CD8− population, and depicting HIV-Gag staining (x-axis) by CD4 staining (y-axis). For the plots show, CTL that had been treated with romidepsin were given a 14 hour wash-out period prior to co-culture with infected cells. B. Shown are summary flow cytometry for 4 hour wash-outs (left panels) and 14 hour wash-outs (right panels). Each condition was tested in duplicate. Error bars represent SEM. C. Portions of the CTL used in the above elimination assays were maintained in culture. At the conclusions of the 14 hour wash-out elimination assay, 29 hours in total since addition of drugs, the frequencies of dead (7-AAD+) and apoptotic (Annexin-V+) CTL were measured by flow cytometry. Shown are % Annexin-V+ following treatment with the indicated doses of drugs. Error bars represent SEM. P values were calculated by two-way ANOVA with Dunnett's multiple comparison test (comparing to the no drug control) * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001.

Figure 8. Time-lapse microscopy of CTL killing of peptide pulsed BLCL target cells in 3D collagen matrices.

An HIV-Gag-SLYNTVATL-specific CTL clone was labeled with Alexa-Fluor555 conjugated cholera toxin subunit B either cultured with 500 nM SAHA or 25 nM romidepsin for 20 hours, or maintained as an untreated control. These effector cells were combined with SLYNTVATL peptide pulsed target cells, matched on the restricting allele, in a collagen matrix medium containing sytox green viability dye. These mixtures were then plated in three separate wells of an 8-well cover slip and imaged by time-lapse brightfield and fluorescent microscopy. A. Shown are representative fields of view from the no treatment (upper panel), 500 nM SAHA (middle panel), and 25 nM romidepsin (lower panel) conditions advancing in time from left to right. Time stamps are given in hh∶mm format. Clones described in the results are indicated with yellow arrows and killed target cells are indicated with white arrows in the upper right panel. B. The number of killed (sytox green positive) BLCL at T = 20 minutes were counted in each of the four fields of view acquired for each condition. Each field of view is plotted as a single point on the graph along with means and SD. P values were calculated by the Kruskal-Wallis test. The reported values are corrected for multiple testing using Dunn's multiple comparison test.