High HDAC8 Creates Targetable Vulnerabilities in Acute Myeloid Leukemia through Dysregulation of BRCA1-Related RNA Splicing Machinery and Defective DNA Damage Repair

Inv(16) [inv(16)(p13q22) or t(16;16)(p13.1;q22)], a common recurrent chromosomal translocation in acute myeloid leukemia (AML), creates CBFb-MYH11 (CM) leukemogenic fusion gene by fusing core binding factor CBFb with smooth muscle myosin heavy chain gene MYH11. Despite the relatively favorable prognosis of inv(16) patients, only about 50-60% achieve long-term survival with standard chemotherapy, highlighting the need for novel therapies to achieve cure. We previously reported that CM fusion protein physically interacts with HDAC8 and enhances its activity, which is critical for CM-driven leukemogenesis. We also presented CM expression increases DNA damage and impairs homologous recombination-directed repair through novel HDAC8-mediated mechanisms. HDAC8 interacts with BRCA1, BCLAF1, and U2AF1 as part of the RNA splicing machinery, critical for efficient DNA damage repair. HDAC8 deacetylates U2AF1, thereby disrupting the recruitment of BRCA1/BCLAF1 to RNA splicing machinery. Accordingly, CM expression, through enhanced HDAC8 activity, leads to dysregulated RNA splicing and defective DNA repair. Our analysis of the TARGET dataset revealed higher HDAC8 expression in AML samples compared to healthy controls (HL) and high HDAC8 (top 25%) is significantly correlated with increased DDR signature score (GO:00006974) in both TARGET and Beat AML datasets. We postulated that HDAC8-high malignant cells, including CM-AML, may exhibit increased sensitivity to splicing modulators or PARP inhibitors (PARPi).

First, we evaluated the effects of H3B-8800 (H3B; an orally bioavailable small molecule inhibitor for SF3B complex) and PARPi (Olaparib or Talazoparib) in primary murine CM-AML cells. Compared to normal bone marrow (BM) cells, CM-AML cells showed significantly increased sensitivity to H3B (IC₅₀: 55.21 ± 9.585 vs. 432.0 ± 21.49 nM; p<0.0001) and Olaparib (IC₅₀: 4.394 ± 1.008 vs. 57.67 ± 2.358 µM; p<0.0001). Similar results were obtained in primary human inv(16) AML samples. Compared to HL-CD34⁺, inv(16)-CD34⁺ cells were significantly more sensitive to H3B (IC₅₀: 40.45 ± 12.54 vs. 270.7 ± 18.04 nM; p<0.0001), Olaparib (IC₅₀: 3.251 ± 1.498 vs. 28.04 ± 4.827 µM; p=0.0004) and Talazoparib (IC₅₀: 130.4 ± 27.57 vs. 2681 ± 487.9 nM; p<0.0001). To test in vivo effects, cohorts of CM-AML mice were generated and given H3B (8 mg/kg), Talazoparib (0.25 mg/kg), or vehicle (0.5% methylcellulose) by daily oral gavage for 10 days. Compared with vehicle, H3B or Talazoparib group showed significantly reduced splenic disease burden (H3B: 0.253 ± 0.0214 g vs. 0.422 ± 0.0734 g; p=0.0401; Tala: 0.143 ± 0.033 g vs. 0.422 ± 0.0734 g; p=0.0051), decreased ckit+ blasts in BM (H3B: 7.825 ± 2.08% vs. 32.22 ± 2.689%; p<0.0001; Tala: 18.5 ± 2.422% vs. 32.22 ± 2.689%; p=0.0042), decreased ckit+ blasts in spleen (H3B: 9.4 ± 1.118% vs. 52.84 ± 5.326%; p<0.0001; Tala: 24.09 ± 6.973% vs. 52.84 ± 5.326%; p=0.0115) with prolonged survival (H3B: 95 vs. 68 days; p=0.0003; Tala: 95.5 vs. 68 days; p=0.0007).

In support of our hypothesis, analysis of Beat AML dataset showed that Olaparib-sensitive patients (IC₅₀<4 μM) are enriched for HDAC8-high (top 50%) expression. In addition, HDAC8 overexpression (HDAC8-OE) in several cell lines representing different disease subtypes (MV4-11, HEL, MDS-L) were more sensitive to H3B, Olaparib, and Talazoparib compared to empty vector (EV) control. To assess in vivo effects, MV4-11-luciferase cells transduced with EV or HDAC8-OE vector were transplanted into NSGS mice (i.v. 0.5x10⁶ cells/mouse, no radiation) and given H3B (8 mg/kg), Talazoparib (0.25 mg/kg), or vehicle (0.5% methylcellulose) by daily oral gavage for 10 days. Bioluminescence imaging showed a significant reduction in tumor burden in both H3B and Talazoparib treated groups compared to vehicle. MV4-11-Luci-HDAC8 cells were selectively more sensitive to H3B (HDAC8 vs. EV; p<0.0001) or Talazoparib (HDAC8 vs. EV; p=0.001) treatment with reduced tumor burden compared to MV4-11-Luci-EV cells. Compared to MV4-11-Luci-EV mice, significantly prolonged survival was seen for MV4-11-Luci-HDAC8 mice treated with H3B (HDAC8: 39 vs. EV: 31 days; p=0.0001) or Talazoparib (HDAC8: 36 vs. EV: 29 days; p=0.0002). Overall, these results indicate that both splicing modulator and PARPi can effectively target HDAC8-high AML cells, including inv(16) AML, offering novel therapeutic avenues for inv(16) AML and AML with aberrant high-HDAC8 activity.