Combined HDAC8 and checkpoint kinase inhibition induces tumor-selective synthetic lethality in preclinical models

Abstract

The elevated level of replication stress is an intrinsic characteristic of cancer cells. Targeting the mechanisms that maintain genome stability to further increase replication stress and thus induce severe genome instability has become a promising approach for cancer treatment. Here, we identify histone deacetylase 8 (HDAC8) as a drug target whose inactivation synergized with the inhibition of checkpoint kinases to elicit substantial replication stress and compromise genome integrity selectively in cancer cells. We showed that simultaneous inhibition of HDAC8 and checkpoint kinases led to extensive replication fork collapse, irreversible cell-cycle arrest, and synergistic vulnerability in various cancer cells. The efficacy of the combination treatment was further validated in patient tumor-derived organoid (PDO) and xenograft mouse (PDX) models, providing important insights into patient-specific drug responses. Our data revealed that HDAC8 activity was essential for reducing the acetylation level of structural maintenance of chromosomes protein 3 (SMC3) ahead of replication forks and preventing R loop formation. HDAC8 inactivation resulted in slowed fork progression and checkpoint kinase activation. Our findings indicate that HDAC8 guards the integrity of the replicating genome, and the cancer-specific synthetic lethality between HDAC8 and checkpoint kinases provides a promising replication stress-targeting strategy for treating a broad range of cancers.

Keywords: Cancer; Cell biology; Cell cycle; Genetic instability; Therapeutics.

Figure 1. Coinhibition of HDAC8 and replication checkpoints elicits severe replication stress, culminating in replication-dependent DNA double-stranded breaks.

(A) Western blot analysis of the DNA damage response in U-2 OS cells treated with the indicated compounds for 4 hours. Representative results from 1 of 2 biological replicates are shown. The lanes were run on the same gel but were noncontiguous. (B) PFGE analysis of DNA breaks in U-2 OS cells treated with the indicated compounds for 15 hours. Relative intensities of broken DNAs were obtained by normalizing individual values to the corresponding untreated control group values. Representative results (upper) and quantification of broken DNAs from 3 biological replicates (lower; n = 3) are shown. Triangles represent the relative intensities of each biological replicate; lines indicate the mean ± SDs of the biological replicates. **P < 0.01, by 1-way ANOVA. (C) Western blot analysis of the DNA damage response in U-2 OS cells treated with the indicated compounds in the presence or absence of roscovitine for 4 hours. Representative results from 1 of 2 biological replicates are shown. HDAC8 inhibitor: 40 μM; AZD-7762: 50 nM; MK-1775: 300 nM; roscovitine: 50 μM. (D) Heatmap showing the co-occurrence of either gain or loss of copy numbers of the indicated genes when the copy number of the HDAC8 gene was gained in individual TCGA cohorts.

Figure 2. Combined treatment of HDAC8 and checkpoint kinase inhibitors leads to synthetic vulnerability.

(A) Cytotoxicity analysis of the indicated treatments in U-2 OS cells. Experimental design and percentages of surviving attached cells at R48 from 1 of 3 biological replicates are shown with means and SDs (n =3). (B) Trypan blue exclusion assay of proliferation efficiency of U-2 OS cells treated with the indicated compounds. Experimental design and numbers of viable cells and percentages of dead cells from 2 biological replicates are shown with means and SDs (n = 6). (C) Cell-cycle analysis of U-2 OS cells treated as in A. Representative profiles and percentages of the sub-G₁ population of the R48 samples from 4 biological replicates are shown (n = 4). Triangles represent the percentages of each biological replicate; lines indicate the mean ± SDs of the biological replicates. (D) Western blot analysis of apoptotic proteins in U-2 OS cells at R48. Data were collected from different sets of gel electrophoresis assays with equal loading of the same samples. Representative results from 1 of 4 biological replicates are shown. Cpd, compound; p-Casp, pro-caspase; a-Casp, active caspase. PCI-34051: 40 μM (B–D); AZD-7762: 50 nM (B–D).

Figure 3. Cytotoxicity of combination treatments is the consequence of genome replication defects.

(A) Cell-cycle analysis of thymidine-synchronized U-2 OS cells treated with the indicated compounds. Experimental design and representative results from 1 of 2 biological replicates are shown. (B) Cytotoxicity analysis of the indicated compounds in thymidine-synchronized U-2 OS cells. Experimental design and representative results from 1 of 2 biological replicates are shown. PCI-34051: 40 μM; AZD-7762: 50 nM. Thy, thymidine.

Figure 4. Synergistic cell killing by HDAC8 and checkpoint kinase inhibition is selective in cancer cells.

(A) Cytotoxicity analysis of the indicated treatments in MDA-MB-231 and M10 cells. Experimental design and percentages of surviving attached cells at R48 from 2 biological replicates are displayed with means and SDs (n ≥5). (B) Trypan blue exclusion assay of proliferation efficiency of MDA-MB-231 and M10 cells treated with the indicated compounds. Numbers of viable cells and percentages of dead cells from 2 biological replicates are shown with means and SDs (n ≥3). (C and D) Western blot analysis of the DNA damage response at R0 (C) and apoptotic proteins at R48 (D) in MDA-MB-231 and M10 cells treated with the indicated compounds. Data were collected from different sets of gel electrophoresis assays with equal loading of the same samples (D). Representative results from 1 of 3 biological replicates are shown. PCI-34051: 80 μM (A, C, and D) or 40 μM (B); AZD-7762: 100 nM (C and D) or 80 nM (B); prexasertib: 3 nM.

Figure 5. HDAC8 inactivation impairs replication elongation.

(A) Immunofluorescence analysis of DNA replication efficiency in U-2 OS cells treated with the indicated HDAC8 inhibitors for 6 hours. Relative EdU intensities in PCNA⁺ replicating cells were obtained by normalizing individual values to the median of the corresponding untreated control group. Dots indicate normalized values of individual cells from each biological replicate labeled with the corresponding colors; triangles represent the median of each biological replicate; lines indicate the mean ± SDs of the medians from biological replicates. (B) Western blot analysis of the DNA damage response in MDA-MB-231 and M10 cells treated with the indicated compounds for 4 hours. Representative results from 1 of 2 biological replicates are shown. (C) DNA fiber analysis of the replication dynamics of MDA-MB-231 cells treated with the indicated compounds for 6 hours. Experimental design and quantitation results of total length of fibers and means ± SDs of the percentages of origin firing from 3 biological replicates are shown. PCI-34051: 40 μM (A and B) or 20 μM (C); HDAC8i-1: 40 μM; AZD-7762: 50 nM. HU: 1 mM. *P < 0.05, **P < 0.01, and ***P < 0.005, by 2-tailed, paired t test (A and C, left panel) and 1-way ANOVA (C, right panel).

Figure 6. HDAC8 inhibition abrogates SMC3 deacetylation throughout the cell cycle.

(A) Western blot analysis of SMC3 acetylation and replication factors loading on chromatin in U-2 OS cells that were thymidine and nocodazole (Noco) synchronized and treated with the indicated compounds. Experimental design, representative cell-cycle profiles, and Western blot results are shown. (B) Western blot analysis of SMC3 acetylation and replication factors loading on chromatin in thymidine- or palbociclib-synchronized U-2 OS cells treated with the indicated compounds. Experimental design and representative results from 1 of 2 biological replicates are shown. HU: 1 mM; HDAC8 inhibitors: 40 μM; palbociclib (Pal): 4 μM; AZD-7762: 50 nM. Asyn, asynchronized.

Figure 7. HDAC8 inhibition leads to R-loop accumulation.

(A and B) Dot blot analysis of DNA-RNA hybrids in normal (A) or ESCO1-depleted (B) U-2 OS cells treated with the indicated compounds for 24 hours. DNA-RNA hybrids were probed with the S9.6 antibody. (C and D) Western blot analysis of the DNA damage response in RNase H1-expressing U-2 OS cells (C) or SMC3-mutant–expressing MDA-MB-231 cells (D) treated with the indicated compounds for 4 hours. (E) Cytotoxicity analysis of the indicated treatments in SMC3-mutant–expressing MDA-MB-231 cells. Percentages of surviving attached cells are displayed with means and SDs (n = 3). (F) Dot blot analysis of DNA-RNA hybrids in SMC3- mutant–expressing MDA-MB-231 cells treated with the indicated compounds for 24 hours. Representative results from 1 of 2 biological replicates are shown (A–F). HDAC8 inhibitors: 40 μM; AZD-7762: 50 nM.

Figure 8. Coinactivation of HDAC8 and checkpoint kinase suppresses tumor growth in human cancer organoids and mouse xenograft models.

(A) Cytotoxicity analysis of CRC PDOs. Experimental design and percentages of surviving PDOs at the endpoint from 2 biological replicates are shown. Relative luminescence signals were obtained by normalizing individual values to the mean of corresponding untreated control group. Triangles represent the technical repeats of each biological replicate; lines indicate the mean ± SDs of all replicates (n = 6). (B) Tumor growth analysis of athymic mice bearing MDA-MB-231 (n ≥9), DLD-1 (n ≥6), or patient pancreatic (n ≥9) established tumors. Mice were intraperitoneally injected with vehicle, 10 mg/kg AZD-7762, and/or 50 mg/kg HDAC8i-3 (once a day for MDA-MB-231; twice a day for DLD-1 and pancreatic PDX), 5 times per week. Tumor volumes and mouse body weight changes were monitored throughout the treatment schedules. (C and D) IHC analysis of the DNA damage response by γH2AX staining (DAB, brown) of tumors (n = 3) excised from MDA-MB-231, DLD-1, and pancreatic PDX xenografts 1 hour after the last injection. Representative images (C) are shown, and quantification results (D) are expressed as the mean ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001, by 1-way ANOVA.

Figure 9. Proposed model of HDAC8 functions in chromatin replication.

In the unperturbed condition, cohesin complexes are loaded on chromatin before DNA replication. The SMC3 acetyltransferase ESCO1 and the deacetylase HDAC8 coordinately regulate the turnover of SMC3 acetylation to control cohesin mobility and genome organization. SMC3 deacetylation by HDAC8 increases cohesin flexibility on chromatin and thus facilitates the fork passing through the cohesin complex. After chromatin replication, ESCO2 acetylates SMC3 to promote cohesion establishment that tethers 2 sister chromatids together. HDAC8 inhibition causes hyperacetylation of chromatin-bound SMC3, resulting in reduced cohesin mobility and accumulation of R-loops that block replisomes traveling on chromatin. This generates replication stress and activates checkpoint kinases to secure replication fork integrity. Inactivation of checkpoint activity in cancer cells further exacerbates replication stress to an intolerable level, leading to fork collapse and, thus, cancer-specific cytotoxicity.