BRD9 functions as an HIV-1 latency regulatory factor

Abstract

A major challenge for HIV type 1 (HIV-1) cure is the presence of viral latent reservoirs. The "Shock & Kill" strategy involves the combined use of latency reversal agents (LRA) and antiretroviral treatment (ART) to reactivate HIV-1 latent reservoirs, followed by elimination of infected cells. However, current LRAs are insufficient in fully reactivating the latent reservoirs. Therefore, investigation on novel HIV-1 latency regulators will be crucial to the success of HIV-1 cure research. Here, we identify bromodomain-containing protein 9 (BRD9) as an HIV-1 latency regulator. BRD9 inhibition induces HIV-1 latency reactivation in T cell lines, human resting memory CD4⁺ T cells, and PBMCs derived from people living with HIV-1 (PWH) on ART. BRD9 inhibition, gene depletion, and protein degradation consistently reactivate HIV-1 latency. Moreover, BRD9 inhibition synergizes with BRD4 inhibition in inducing HIV-1 production. Mechanistically, BRD9 binds to HIV-1 LTR promoter and competes with HIV-1 Tat protein for binding to the HIV-1 genome. Additionally, our integrated CUT&RUN DNA sequencing, transcriptomics, and pharmacological analysis revealed downstream host targets of BRD9, including ATAD2 and MTHFD2, that modulate HIV-1 latency.

Keywords: AIDS; BRD9; HIV-1 latency.

Fig. 1.

BRD9 inhibition induces HIV-1 latency reactivation. (A) Schematic diagram of the epigenetic compound library screen. (B) Ranked list of epigenetic drug candidates in ACH2 T cells. HIV-1 latency reactivation was quantified with HIV-1 p24 ELISA (n = 3). (C–F) ACH2 T cells were treated with I-BRD9 or JQ-1. HIV-1 latency reactivation was quantified with qPCR (C), HIV-1 p24 ELISA (D), and flow cytometry (E and F) (n = 3). (G and H) J-Lat T cells were treated with I-BRD9 or JQ-1. HIV-1 latency reactivation was quantified with qPCR (n = 4) and flow cytometry (n = 3 for DMSO; n = 4 for I-BRD9 and JQ-1). Data represent mean ± SD. Statistical significance in (C–E, G, and H) was determined with one-way ANOVA with Dunnett multiple comparisons test. **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Fig. 2.

Identification of BRD9 as an HIV-1 latency regulatory factor. (A) ACH2 T cells were treated with BRD9 or BRD4 siRNA. Knockdown efficiency was measured with qPCR (n = 12). (B–D) HIV-1 latency reactivation was quantified with qPCR (B) (n = 9 for the BRD9 panel, n = 11 for the BRD4 panel), ELISA (C) (n = 17), and flow cytometry (D) (n = 3 for scramble siRNA; n = 4 for BRD9 and BRD4 siRNA). (E) Representative Western blot image of BRD9 in BRD9 knockout J-Lat T cells. (F) HIV-1 gene expression in BRD9 or BRD4 knockout J-Lat T cells was quantified with qPCR (n = 5). (G) Representative Western blot image of BRD9 in VZ185-treated ACH2 T cells. (H–J) ACH2 T cells were treated with different concentrations of VZ185. HIV-1 expression was quantified with ELISA (n = 4) (H), qPCR (n = 3) (I), and flow cytometry (n = 3) (J). (K) BRD9 and BRD4 knockout J-Lat T cells were treated with VZ185 and JQ-1, respectively. HIV-1 Gag gene expression in the cell lysates was quantified with qPCR (n = 3). (L) ACH2 T cells were treated with VZ185 and JQ-1 in single or combined settings. HIV-1 latency reactivation was determined with ELISA (n = 4). (M) ACH2 T cells were treated with VZ185 and SAHA in single or combined settings. HIV-1 latency reactivation was determined with ELISA (n = 4). (N) ACH2 T cells were treated with JQ-1 and SAHA in single or combined settings. HIV-1 latency reactivation was determined with ELISA (n = 4). The same set of 5 μM VZ185 values was used in Fig. 2 L and M and SI Appendix, Fig. S4A; the same set of 10 μM VZ185 values was used in Fig. 2 L and M. Data represent mean ± SD. Statistical significance in (A, B, and K) was determined with the unpaired t test. Statistical significance in (C, D, F, H–J, and L–N) was determined with one-way ANOVA with Dunnett multiple comparisons test. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Fig. 3.

BRD9 inhibition induces HIV-1 latency reactivation in HIV-1-infected human primary T cell models. (A) Schematic of the LRA treatment in human resting memory CD4⁺ T cells. (B and C) HIV-1 infection was performed by spinoculation. The infected cells were treated with LRAs for 3 d. HIV-1 latency reactivation was determined with ELISA (n = 8 for uninfected; n = 9 for DMSO; n = 11 for I-BRD9, JQ-1, and SAHA) (B) or flow cytometry (n = 10 for uninfected and DMSO; n = 11 for JQ-1 and SAHA; n = 13 for I-BRD9) (C). (D) Representative flow cytometry dot plots. (E) Schematic of the LRA treatment in PBMCs from PWH on ART. (F) PBMCs from PWH on ART were treated with VZ185 or SAHA for 3 d. HIV-1 latency reactivation was quantified with TZM-bl qVOA (n = 13 for DMSO and 10 µM I-BRD9; n = 4 for 20 µM I-BRD9; n = 7 for SAHA). (G) PBMCs from PWH on ART were treated with VZ185 or SAHA for 3 d. HIV-1 latency reactivation was quantified with flow cytometry (n = 4). (H) PBMCs from PWH on ART were treated with different concentrations of VZ185 for 3 d. HIV-1 Gag gene expression was quantified with qPCR (n = 8 for DMSO and 10 µM VZ185; n = 5 for 2 µM and 5 µM VZ185). (I) PBMCs from PWH on ART were treated with VZ185 and different concentrations of I-BRD9 for 3 d. HIV-1 Gag gene expression was quantified with qPCR (n = 8 for DMSO and 10 µM VZ185; n = 5 for 5 µM VZ185; n = 6 for 5uM I-BRD9, 10 µM I-BRD9, and I-BRD9/VZ185 combined treatments). (J) PBMCs from PWH on ART were treated with VZ185 and JQ-1 in single or combined settings for 3 d. HIV-1 Gag gene expression was quantified with qPCR (n = 13 for DMSO; n = 7 for 100 nM JQ-1; n = 9 for 200 nM JQ-1; n = 11 for 300 nM JQ-1; n = 5 for 5 µM VZ185, 5 µM VZ185/100 nM JQ-1, and 5 µM VZ185/200 nM JQ-1 combined treatments; n = 4 for 5 µM VZ185/300 nM JQ-1 combined treatment). Data represent mean ± SD. Statistical significance in (B, C, and F–J) was determined with one-way ANOVA with Dunnett multiple comparisons test. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Fig. 4.

Interaction of BRD9 with the HIV-1 genome and host genome of HIV-1-infected human resting memory CD4⁺ T cells. (A) Representative DNA electrophoresis gel of sonicated DNA samples from HIV-1-infected human resting memory CD4⁺ T cells. (B) ChIP-qPCR assay using anti-BRD9 and anti-BRD4 pulldown (n = 3). (C) ChIP-qPCR assay using anti-BRD9 pulldown under I-BRD9 treatment (n = 4). (D) Schematic of our working modeling on BRD9 binding to the HIV-1 genome. (E) Modified CUT&RUN assay that evaluates HIV-1 Tat protein binding to HIV-1 LTR promoter with or without added BRD9 protein in ACH2 T cells (n = 3). (F) CUT&RUN-seq reveals BRD9 and BRD4 binding region on the HIV-1 genome. (G) Ranked list of host genes bound by the BRD9 protein. (H) Aligned-read peak spectra of BRD9 protein on ROCK1P1 and DUX4. (I) Binding motif estimation of BRD9 and BRD4 protein on human T cell genome. Data represent mean ± SD. Statistical significance in (B, C, and E) was determined with one-way ANOVA with Dunnett multiple comparisons test. *P < 0.05, **P < 0.01, and ***P < 0.001.

Fig. 5.

Functional characterization of downstream cellular pathways of BRD9 in modulating HIV-1 latency. (A) Ranked list of downregulated BRD9-bound genes upon VZ185 treatment. (B) ACH2 T cells were treated with 10 µM VZ185, 10 µM AZ13824374, 10 µM DS18561882, 10 µM AXKO-0046, or 10 µM BAY-1816032 in single or combined settings. HIV-1 latency reactivation was quantified with ELISA (n = 6 for DMSO and VZ185; n = 8 for AZ13824374, DS18561882, AXKO-0046, and BAY-1816032; n = 4 for VZ185/AZ13824374, VZ185/DS18561882, VZ185/AXKO-0046 combined treatments; n = 6 for VZ185/BAY-1816032 combined treatment) (the same set of 10 μM VZ185 values was used in the AZ13824374, DS18561882, AXKO-0046 panels). (C) ACH2 T cells were treated with 10 µM VZ185, 10 µM AZ13824374, 10 µM DS18561882, 10 µM AXKO-0046, or 10 µM BAY-1816032 in single or combined settings. HIV-1 latency reactivation was quantified with qPCR (n = 4) (the same set of 10 μM VZ185 values was used in the AZ13824374, DS18561882, AXKO-0046 panels). (D) PBMCs from PWH on ART were treated with 10 µM VZ185, 10 µM AZ13824374, 10 µM DS18561882, 10 µM AXKO-0046, or 10 µM BAY-1816032 in single or combined settings. HIV-1 latency reactivation was quantified with qPCR (n = 5 for DMSO and DS18561882; n = 6 for VZ185; n = 5 for VZ185 (AXKO-0046 panel); n = 4 for AZ13824374, AXKO-0046, and BAY-1816032; n = 5 for VZ185/AZ13824374 and VZ185/BAY-1816032 combined treatment; n = 4 for VZ185/DS18561882 and VZ185/AXKO-0046 combined treatment) (the same set of 10 μM VZ185 values was used in the AZ13824374 and BAY-1816032 panels). (E) Ranked list of upregulated BRD9-bound genes upon VZ185 treatment. (F and G) ACH2 T cells were treated with 10 µM VZ185, 10 µM TXNIP-IN-1, or 10 µM tofacitinib. HIV-1 latency reactivation was quantified with ELISA (n = 6 for DMSO, VZ185, and TXNIP-IN-1; n = 4 for tofacitinib) (F) or qPCR (n = 3) (G). Data represent mean ± SD. Statistical significance in (B–D, F, and G) was determined with one-way ANOVA with post hoc Tukey’s multiple comparisons test. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.