Replication Stress Leading to Apoptosis within the S-phase Contributes to Synergism between Vorinostat and AZD1775 in HNSCC Harboring High-Risk TP53 Mutation

Head and neck squamous cell carcinoma (HNSCC) is the seventh most common cancer worldwide (1). The cure rate for patients with advanced HNSCC remains poor due to resistance to standard therapy primarily consisting of cisplatin and radiation. Mutation of TP53 in HNSCC occurs in 60% to 80% of human papillomavirus (HPV)-negative cases (2, 3) and is associated with resistance to these treatments. Recently, we developed a novel computational approach termed evolutionary action (EAp53), which can stratify patients with tumors harboring TP53 mutations as high or low risk. Patients with high-risk TP53 mutations had the poorest survival outcomes, shortest time to the development of distant metastases, and increased resistance to cisplatin-based therapy (4, 5). As the group of tumors with this high-risk mutation shows resistance to cisplatin or radiation, we are interested in the development of novel therapeutic approaches to overcome this resistance and improve treatment outcomes. WEE1 is a kinase that has been linked to DNA damage–induced S-phase replication stress and G₂–M arrest, owing to its ability to inactivate cyclin-dependent kinase 1 (CDK1) through phosphorylation of the Tyr15 residue (6–10). Recently, we demonstrated that the WEE1 kinase inhibitor AZD1775 sensitizes HNSCC cells harboring high-risk TP53 mutations to cisplatin both in vitro and in vivo through induction of persistent DNA damage response associated with mitotic delay and subsequent senescence (11).

Modulation of the acetylation status of histones and transcription factors is an essential mechanism for regulating gene expression (12, 13). Histone acetylation is generally associated with elevated transcription, whereas deacetylated histones are often linked to repressed transcription (14). Histone deacetylases (HDAC) act enzymatically to remove the acetyl group from histones and silence gene expression (14). Elevated activities of HDACs have been observed in several human malignancies, including HNSCC, and their overexpression is associated with poorer prognosis in oral cancer patients (2, 15, 16). Collectively, these findings indicate that histone deacetylation may represent a potential therapeutic target in HNSCC.

Recent reports have shown that HDAC inhibitors (HDACI) induce growth arrest, differentiation, and apoptosis in various cancer cell lines in vitro and suppress tumor growth in animal xenograft models, including HNSCC (12, 17, 18). In addition, several studies have demonstrated that vorinostat, a small-molecule inhibitor of HDAC, displays preferential cytotoxicity in vitro and in vivo in cancer cells harboring TP53 mutations (19–21). Although recent evidence suggests that defects in DNA damage repair processes contribute to the selective cytotoxic effects of HDAC inhibitors in tumor cells, the detailed molecular mechanism is not well understood (22, 23). The HDAC and WEE1 inhibitors are now emerging as attractive classes of antitumor agents being tested clinically either as single agents or in combination with conventional chemotherapeutics or targeted agents (24, 25). Taken together, these preclinical results and the ongoing clinical trials have prompted us to evaluate the combination of WEE1 and HDAC inhibitors in HNSCC with mutant TP53.

In this study, we demonstrate that vorinostat interacts synergistically with AZD1775 in vitro and in vivo in HNSCC tumor cells expressing high-risk mutant p53 (mutp53). Notably, vorinostat alone or in combination with AZD1775 results in increased markers of replication stress, DNA damage response, and impaired Rad51-mediated homologous recombination (HR), leading to an early apoptotic cell death during the S-phase and subsequently in the G₂–M cell-cycle phase. Using live cell imaging, RNA sequencing (RNA-seq) analyses and reverse-phase protein array (RPPA) proteomic profiling, we further provide evidence that the mechanism of the synergistic interaction between these two drugs may be in part due to vorinostat's ability to epigenetically modulate expression of a transcript-signature containing genes involved in regulating replication stress, mitosis, and the cell-cycle checkpoints in p53-mutant HNSCC cells. Taken together, our findings support a strategy, including a combination of WEE1 and HDAC inhibition, which is a novel therapeutic regimen warranting investigation in patients with advanced HNSCC.

The HNSCC cell line PCI13 lacking endogenous expression of p53 was obtained from the laboratory of Dr. Jennifer Grandis (University of Pittsburgh, Pittsburgh, PA) in August 2008 and engineered to stably express constructs containing wild-type p53 (wtp53), high-risk EA score mutant p53 (C238F and G245D), as described previously (4). The HNSCC cell lines, HN30-expressing wtp53 and HN31-expressing mutp53, were obtained in December 2008 from the laboratory of Dr. John Ensley (Wayne State University, Detroit, MI). OSC-19 was obtained from Health Science Research Resource Bank (HSRRB) in 2010. Detroit562 was purchased from ATCC in 2009. HN5 was obtained from Dr. D.M. Easty (Ludwig Institute for Cancer Research, London, United Kingdom) in 2003. The cell lines were maintained in DMEM, supplemented with 10% FBS, l-glutamine, sodium pyruvate, nonessential amino acids, and vitamin solution, and incubated at 37°C in 5% CO₂ and 95% air. The identity of all cell lines was authenticated using short tandem repeat testing within 6 months of cell use. The WEE1 inhibitor AZD1775 was provided by AstraZeneca through a collaborative agreement arranged by NCI-CTEP. Vorinostat [formerly known as suberoylanide hydroxamic acid (SAHA)] was obtained from Merck Corp. For in vitro studies, AZD1775 and vorinostat were prepared as 10 mmol/L stock solution in DMSO and stored at −20°C. AZD1775 and vorinostat were diluted in culture medium and used at 0.25 and 4 μmol/L final concentration, respectively, before use. The staurosporine was purchased from Sigma and used at 1 μmol/L final concentration.

The HNSCC cells were seeded in 6-well plates at predetermined densities, concurrently exposed to different fixed-ratio combinations of vorinostat (dose range, 0.01–20 μmol/L) and AZD1775 (dose range, 0.01–1 μmol/L) for 48 hours, and the clonogenic cell survival was determined as described previously (11, 26).

Drug synergy between vorinostat and AZD1775 was determined by the combination index and isobologram analyses, which were generated according to the median-effect method of Chou and Talalay (27) using the CalcuSyn software (Biosoft). For details, see Supplementary Materials and Methods section.

Antibodies used for Western blotting were phospho-H2AX (Ser139; #2577), phospho-CDC2 (CDK1)-Tyr15 (#9111), CDC2 (CDK1) (#9112), cyclin B1 (#4138), PARP-1 (#9542), Survivin (71G4B7; #2808), Rad51 (#8875), RB (Ser807/811; #9308), FOXM1 (D12D5; #5436S), p-HH3 (#9701), CHK1 (#2345), pCHK1 (Ser345; #2341), acetyl-histone H3 (Lys9/Lys14; #9677), and PLK1 (208G4; #4513S; all from Cell Signaling Technology); β-actin (#A5316; Sigma-Aldrich); p53(DO-1) (#sc-126), R2 (N-M; sc-10844), and RB (C-15; sc-50) were from Santa Cruz Biotechnology; p21WAFI (Ab-1; #OP64; Calbiochem); RPA32 (clone RPA-34-20; #MABE285; EMD Millipore); EMA (2F6; #ab156947) and CD31 (#ab28364) were from Abcam.

Cells grown on 10-cm plates were treated with minimally toxic and clinically relevant concentrations of vorinostat (i.e., 0.5–4 μmol/L) and slightly toxic AZD1775 (i.e., 0.25–0.5 μmol/L) either alone or in combination for 16 or 48 hours. Whole-cell lysates were prepared, and Western blot analysis was conducted with indicated antibodies as described previously (11).

Cells were seeded in 60-mm dishes, treated the next day with vorinostat (4 μmol/L), AZD1775 (0.25 μmol/L) either alone or in combination, and then harvested at 24, 48, or 72 hours. The cell-cycle analysis and apoptosis detection were performed as described previously (11).

For live cell imaging studies, HNSCC cell lines (PCI13-G245D) were stably transfected with histone H2B-RFP lentiviral vector (Addgene), selected in 2 μg/mL blasticidin, and sorted by flow cytometry to enrich for highly expressing cells according to published protocol (26). The histone H2B-RFP–tagged cells were then plated at a predetermined density, followed by treatment with 0.01% DMSO alone, 0.25 μmol/L AZD1775 alone, 4 μmol/L vorinostat alone, or in combination. Live video imaging, EdU labeling, and DNA content analyses were carried out as described in the Supplementary Materials and Methods.

Cells were plated on glass coverslips and treated with drugs the following day as previously indicated. Cells were then fixed in 2% paraformaldehyde for 1 hour, washed, permeabilized in 0.2% Triton X-100 in PBS for 20 minutes, washed, and blocked for 1 hour at room temperature in PBS buffer containing 2% normal goat serum, and 0.3% Triton X-100. Then, glass coverslips were incubated with primary Rad51 (MA5-14419; Thermo Fisher Scientific) antibody overnight at 4°C, followed by secondary Alexa Fluor–conjugated antibody 594 as appropriate, and Rad51 foci images were acquired on a Leica confocal microscope as described previously (11). Detailed description is included in the Supplementary Materials and Methods.

All animal experimentation was approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Texas MD Anderson Cancer Center (Houston, TX). Our orthotopic nude mouse tongue model has been described previously in the literature (11). The high-risk mutp53 HNSCC cells (PCI13-G245D) were injected into the tongues of the male athymic nude mice, and they were randomized into different groups 8 to 10 days after injection. Treatment protocol and tumor growth delay measurement are described in the Supplementary Materials and Methods.

Apoptosis was assessed in mice tissue sections with terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick-end labeling (TUNEL) assay with DeadEnd Fluorometric TUNEL System (Promega) according to the manufacturer's protocol with some modifications. A detailed description is included in the Supplementary Materials and Methods.

HNSCC cell lines (PCI13) stably expressing high-risk mutp53 (G245D) were treated with vorinostat (4 μmol/L), AZD1775 (0.25 μmol/L), or the combination. Total RNA was isolated 36 hours after treatment using the RNeasy Mini Kit reagents (QIAGEN) according to the manufacturer's instructions. Cells treated with 0.1% DMSO served as untreated controls. For RNA-seq, total RNA was submitted to the Sequencing and Microarray Facility at MD Anderson Cancer Center for next-generation sequencing. Detailed description of the RNA-seq is included in the Supplementary Materials and Methods.

PCI13-G245D–mutant cells were treated with vorinostat (4 μmol/L), AZD1775 (0.25 μmol/L), or the combination. Protein lysate was collected 48 hours after treatment from each cell line under full-serum (10% FBS) conditions and subjected to RPPA analysis as described previously (28). Detailed description of the RPPA is included in the Supplementary Materials and Methods.

Sections were prepared from formalin-fixed paraffin-embedded mice tumor tissues and subjected to IHC with indicated antibodies according to the protocol as described in Supplementary Materials and Methods.

The Student t and one-way ANOVA tests were carried out to analyze in vitro data. For mouse studies, a two-way ANOVA test was used to compare tumor volumes between control and treatment groups. For IHC analyses, a one-way ANOVA and a χ² test were used to compare control and treatment groups. Survival following drug treatment was analyzed by the Kaplan–Meier method and compared with log-rank test. All data were expressed as mean ± SE, and P values of 0.05 or less were considered to indicate nominal statistical significance.

To explore the degree of sensitivity of high-risk mutp53 HNSCC cells to the HDAC inhibitor vorinostat as a single agent, we performed dose–response studies with the drug in isogenic HNSCC PCI-13 cells with wild-type p53, lacking p53 expression (pBabe null), or expressing high-risk mutp53 (C238F, G245D) using standard clonogenic survival assays. As shown in Fig. 1A and B, the PCI-13 isogenic cells displayed differential sensitivity to vorinostat as a single agent. The cell lines expressing high-risk mutp53 were more sensitive to vorinostat as a single agent (IC₅₀; 0.5–0.7 μmol/L) than the cells expressing wtp53 (IC₅₀; 1.2 μmol/L) or pBabe vector control p53-null cell lines (1.1 μmol/L), consistent with previous publications (13, 19, 20).

We next examined whether vorinostat was synergistic with AZD1775 treatment of the isogenic PCI-13 cell lines, using the combination index (CI) and fraction affected (Fa) method of Chou and Talalay (27). Treatment of these cells with the combination of vorinostat and AZD1775 revealed strong synergism manifested by the shift of vorinostat response curves and the CI values (Fa, 0.5) of 0.90, 0.30, 0.38, and 0.34 (Fig. 1C–F, top plots; Supplementary Table S1) for PCI13-pBabe (null), PCI13-C238F, PCI13-G245D, and PCI13-Wtp53, respectively. Furthermore, the CI was less than 1.0 in all cells tested representing strong synergism at the more relevant Fa values (≥50%; Fig. 1G–J, middle plots). Conservative isobologram plots of effective dose ED₅₀, ED₇₅, and ED₉₀ were also generated and further confirmed strong synergism (Fig. 1K–N, bottom plots; Supplementary Table S1). The degree of synergy between vorinostat and AZD1775 was also confirmed in other human HNSCC cell lines with different TP53 status (Supplementary Figs. S1A–S1H and S2A–S2H; Supplementary Table S1).The data clearly demonstrated that vorinostat and AZD1775 interacted synergistically to inhibit in vitro clonogenic survival in various HNSCC irrespective of the TP53 status.

To further characterize the synergistic response of HNSCC cells expressing mutp53 to vorinostat and AZD1775, we conducted live cell imaging in PCI13-G245D cells expressing a histone H2B-RFP construct. The results of the live cell imaging studies are shown in Fig. 2A as event charts, where the history of each cell is depicted as a line over time, the color of the line indicates whether the cell is in interphase (gray) or mitosis (blue), and the time of cell apoptosis is marked by a red symbol. The majority of control PCI13-G245D cells entered mitosis within 15 hours and reached the second mitosis with an average interphase time of 15.6 hours (Fig. 2A). Although AZD1775 treatment (Fig. 2A and B) slightly increased the initial rate of mitotic entry (during the first 4 hours) and mitotic duration (from an average of 37 minutes for the control to 53 minutes for AZD1775 treatment), the mitotic entry rate then decreased for the next 7 hours. In addition, AZD1775 treatment alone decreased the fraction of cells capable of reaching the second mitosis and increased the fraction of apoptotic cells, especially in cells that had undergone a first mitosis in the presence of AZD1775 (Fig. 2A and B; Supplementary Video S1). In contrast, vorinostat treatment was associated with a slower progression rate to mitosis, a decreased fraction of cells capable of reaching the first mitosis, a slightly longer mitotic duration in those cells reaching mitosis (average, 61 minutes), and resulted in a further increased apoptotic fraction both in cells that did and did not reach a first mitosis (Fig. 2A and B; Supplementary Video S2). Interestingly, when vorinostat and AZD775 were combined, the initial rate of mitotic entry again increased compared with cells treated with vorinostat alone; however, the average mitotic duration increased to 119 minutes and an increased fraction of apoptotic cells (Fig. 2A and B; Supplementary Video S3).

The prior live cell imaging studies (Fig. 2) showed that cells began undergoing apoptosis at increased rates after 7 to 9 hours of treatment with vorinostat and AZD1775. To better understand the underlying causes and timing of apoptosis, PCI13-G245D cells expressing a histone H2B-RFP construct were treated for various durations and were subjected to pulse EdU labeling and DNA content measurements. As shown in Fig. 3A (9-hour treatment), AZD1775 alone slightly increased the rate of entry into S-phase (increased fraction of replicating cells at the G₁–S boundary) and S-phase fraction (59.3%) compared with untreated control (48.7%). Vorinostat alone also caused S-phase accumulation (66.1%; especially early S-phase; Fig. 3A), slowed DNA replication (Fig. 3B), and increased the G₂ fraction compared with untreated control (Fig. 3A and B). Furthermore, the combination of vorinostat and AZD1775 resulted in a more profound S-phase accumulation (77.3%) due to enhanced entry into S-phase but profound replication slowing with fewer cells reaching G₂ phase compared with all treatments (Fig. 3A and B). Combining these findings with the live cell imaging data, we observed that the mitotic rate in the combination treatment was slowed by 2 to 3 hours of treatment due to AZD1775 effect, and cells stopped reaching mitosis by 12 hours. Moreover, we observed increasing rates of apoptosis starting at around 8 hours, mostly in the subpopulation that had not reached its first mitosis or reached mitosis later during treatment.

The above data suggest that cells treated with the combination of drugs experienced replication stress and eventually underwent apoptosis within the S-phase. If this were the case, one might expect to detect EdU-labeled cells undergoing apoptosis after treatment. The live cell imaging studies showed that the cells underwent distinct morphologic changes (stress-induced blebbing) just prior to apoptosis (Fig. 3C). We therefore imaged populations of cells that were pulsed with EdU for 1 hour prior to fixation (9-hour sample) for morphology (DIC), EdU staining, and nuclear morphology (TOPRO-3 staining) using laser scanning confocal microscopy (Fig. 3D; Supplementary Fig. S3). The arrows indicate cells showing the stressed morphology (DIC image, left) along with the associated EdU and nuclear images (right), suggestive of cells about to undergo apoptosis during an impeded S-phase.

To better understand the longer term effects of drug treatment, PCI13-G245D cells were drug treated and examined at 48 hours. Treatment with AZD1775 alone, or vorinostat alone were associated with increased G₂ fractions (25% and 38.3%, respectively; Supplementary Fig. S4A and S4B), and the combination of drugs was associated with an even higher G₂–M fraction (57.4%; Supplementary Fig. S4A and S4B). Moreover, a sub-G₁ peak indicative of apoptosis was increased in PCI13-G245D cells after 48 hours of combination treatment (Supplementary Fig. S4A and S4B). Apoptosis in these cells was further confirmed by the presence of positive APO-BrdU tunnel staining (Supplementary Fig. S4C).

The cell-cycle kinase inhibitor p21 (CDKN1A), acetylated histone H3 and H4 are known targets for induced effects of HDACIs in various transformed cells (29–31). Significant increases in p21 and acetylated H3 protein levels were observed in all the cells tested by Western blotting following treatment for 48 hours with vorinostat alone or in combination with AZD1775 (Fig. 4A and B; Supplementary Fig. S5A–S5C). These results suggest that vorinostat increases p21 level independent of its upstream activator, p53, in HNSCC cells. The increase in acetylated histone 3 further confirmed that the concentrations of vorinostat used in vitro in our study were inhibiting the activity of HDACs.

HDAC and WEE1 inhibitors have been reported to interact synergistically to induce DNA damage and replication stress through modulation of checkpoint kinase 1 (Chk1) and CDK1 phosphorylation levels in leukemia and glioblastoma cancer cells (22, 31–33). Therefore, we evaluated whether vorinostat and AZD1775 similarly affected phosphorylation and protein levels of CDK1 and Chk1 in PCI13-G245D cells and various HNSCC cell lines. The combination of vorinostat and AZD1775 markedly decreased the inhibitory phosphorylation of Tyr15 on CDK1, total CDK1, and its partner cyclin B1 protein levels in a time-dependent manner (Fig. 4C; Supplementary Fig. S5D–S5F), suggesting a shift in cells stalled during S-phase. In addition, the levels of the mitotic marker, phospho-histone 3 (S10), were dramatically reduced within 48 hours following vorinostat treatment and in combination with AZD1775, indicating that a fraction of cells were either arrested at the G₂ phase or lost during S-phase at this time point (Fig. 4C). Vorinostat sharply abrogated AZD1775-induced Chk1 phosphorylation and further decreased the total Chk1 protein levels in a time-dependent manner (Fig. 4C; Supplementary Fig. S5D–S5F). Reduced Chk1 activation was associated with induction of DNA damage, manifested by the marked increase in the double-strand break marker, phospho-γH2AX following treatment with the drugs in these cells (Fig. 4C; Supplementary Fig. S5D–S5F). These findings raise the possibility that WEE1 and HDAC inhibition induce lethal replicative stress in HNSCC cells expressing mutp53. To test this possibility, the effects of vorinostat and AZD1775 on replication stress markers were examined in PCI13-G245D and HNSCC cells with different TP53 status. Hyperphosphorylation of the replication stress marker RPA32 associated with reduced levels of the ribonucleotide reductase enzyme RRM2 was observed in a time-dependent manner consistent with the induction of replication stress in these cells following treatment with drugs (Fig. 4C; Supplementary Fig. S5D–S5F). Western blot analyses provided evidence that treatment of HNSCC cells with vorinostat, or in combination with AZD1775 for 48 hours, induced apoptosis as indicated by PARP1 cleavage (Fig. 4D; Supplementary Fig. S5D–S5F).

One of the pleiotropic actions of vorinostat includes its effect on Rad51 expression and HR in cancer cells (34, 35). On the basis of this finding, we examined alterations in the expression levels of the HR competent marker, Rad51 in PCI13-G245D cells and various HNSCC cell lines following treatment with the drugs. Vorinostat or in combination with AZD1775 is associated with decreased Rad51 protein levels within 48 hours posttreatment (Fig. 4C; Supplementary Fig. S5D–S5F). Correspondingly, Rad51 focus formation under the same treatment conditions was markedly decreased in these cells (Fig. 4E and F). Taken together, these data suggest that vorinostat and AZD1775 exert their synergistic effects in HNSCC cells expressing high-risk mutp53 through induction of replication stress associated with insufficient Rad51-mediated HR repair.

To better understand the cellular phenotype associated with the synergistic interaction between vorinostat and AZD1775 in HNSCC cells with mutp53, cells were treated for 48 hours with the drugs, and differentially expressed genes (DEG) were examined by RNA-seq as described in Materials and Methods. A linear regression model (FDR cutoff = 0.05) and filtering criteria (i.e., DEG with an adjusted P value of <0.01, genes with 50 average reads among the comparison groups, exclusion of noncoding RNAs) were used. Principal clustering component analysis (PCA) revealed that DEGs among vorinostat and vorinostat plus AZD1775 treatment groups are alike. Approximately 1,374 genes that are significantly different between vorinostat and untreated cells with at least a 4-fold difference in expression were identified (Supplementary Table S2). Because our in vitro data presented here suggested that vorinostat and AZD1775 resulted in replicative stress and specifically affected the cell cycle at multiple checkpoints and induced cell killing (Figs. 2–4), we focused our attention on the dysregulation of genes that are typically involved in these cellular processes. To examine this possibility, the 1,374 genes were evaluated by GO enrichment and IPA analysis. Not surprisingly, 5 major pathways significantly associated with the top filtered 139 DEGs regulated by vorinostat or in combination with AZD1775 were identified (Fig. 5; Supplementary Table S3), including chromatid segregation, G₁–S phase transition, mitotic anaphase DNA replication, mitotic M phase, and cell-cycle checkpoint. All of these processes were deregulated in cancer and a heatmap associated with the target GO terms representing the distribution of the 139 DEGs is shown in Fig. 5. Many of the 139 DEGs were associated with multiple cellular cell functions. Expression of top target genes involved in induction of replication stress, HR, regulation of the S–G₁ and G₂–M checkpoints, including CHK1, CDK1, PLK1, CCNB1 (cyclin B1), RRM2, RAD51, and BIRC5 (survivin), were significantly decreased following treatment with vorinostat or in combination with AZD1775 (Fig. 5). The expression level of p21 (CDKN1A), the immediate downstream target of p53, was increased by vorinostat (Fig. 5). Furthermore, decreased mRNA expression levels of 130 genes of the 139 DEGs trended toward almost significant correlation with better survival in the TCGA HPV-negative HNSCC cohort (Supplementary Fig. S6). Change in protein expression levels of some of these key genes was confirmed by Western blotting (Fig. 4; Supplementary Fig. S7B) providing partial validation for the RNA-seq data.

To identify the signaling proteins mediating the effects of vorinostat alone or in combination with AZD1775, an RPPA was performed using PCI13-G245D cells treated with the drugs as indicated. A total of 118 proteins were significantly expressed (P < 0.05) and were similar among vorinostat and vorinostat plus AZD1775 treatment groups. GO enrichment analysis identified two significant pathways involving G₁–S phase transition and cell-cycle checkpoint as annotated in the heatmap (Supplementary Fig. S7A; Supplementary Table S4). To focus our investigation, we limited our analysis to protein level changes of >1.5 fold, with an FDR of <1% and P value of <0.05 (28). RPPA identified a number of proteins involved in cell-cycle regulation that were significantly downregulated in PCI13-G245D cells, including CDK1, cyclin B1, PLK1, EMA (mucin), FoxM1 and Rb_pS807_S811 (Supplementary Fig. S7A), suggesting decreased progression to mitosis as seen with the live cell imaging and cell-cycle analyses. Downregulation of these proteins was further confirmed by Western blotting (Supplementary Fig. S7B).

On the basis of the in vitro results, the impact of vorinostat and AZD1775 either alone or in combination was evaluated in an orthotopic nude mouse model implanted with PCI13-G245D cells in the tongue. Although vorinostat (65 mg/kg) or AZD1775 (45 mg/kg) alone had little effects on tumor growth in vivo at these doses, the combination of vorinostat and AZD1775 significantly inhibited tumor growth (P < 0.001, when compared with AZD1775 alone and P < 0.0001 when compared with vehicle or vorinostat-treated group, respectively; Fig. 6A; Supplementary Fig. S8). Furthermore, combined treatment significantly prolonged animal survival compared with animals in the untreated control group (P < 0.022; Fig. 6B). To confirm that the enhancement of antitumor efficacy by the combination drug treatment was associated with engagement of downstream targets in vivo, phosphorylation of CDK1 (Y15) and expression of p21 were evaluated in tissue sections obtained from tongue tumor xenografts. Consistent with the in vitro results, treatment with vorinostat, AZD1775, or the combination was associated with decreased levels of phospho-CDK1 in vivo in orthotopic tongue tumors compared with untreated tumors (Fig. 6C and D). Increased p21 and decreased PCNA, RRM2, and survivin immunostaining levels were also evident in these tumor sections following treatment with vorinostat and in combination with AZD1775 (Supplementary Fig. S9). During the study, mice in the combination treatment arm show no more than 10% body weight loss or any signs of drug toxicity (Supplementary Fig. S10).

Angiogenesis is a critical component of tumor progression that is regarded as an important target in cancer therapy. As vorinostat can inhibit vascularization (36, 37), we evaluated the effects of these drugs administered alone or in combination on angiogenesis by measuring intratumor vasculature using CD31 as an endothelial cell marker. A significant decrease in the fraction of CD31-positive cells in vivo in mice tumor tongue sections was observed (Fig. 6E and F), suggesting that disruption of microvessel density or lack of new vessel formation may contribute to the efficacy of vorinostat and AZD1775 combination in vivo. The occurrence of apoptosis in vivo in tongue tumor xenografts bearing PCI13-G245D cells was also examined using the TUNEL assay. TUNEL-positive apoptotic cells were significantly increased in tissue sections obtained from tumors of mice treated with combination of vorinostat and AZD1775 compared with either drug alone or untreated control (Fig. 6G and H) indicative of apoptosis.

As HPV-negative HNSCC has a high burden of TP53 mutations, contributing to low cure rate and resistance to standard therapy consisting of cisplatin and radiation, novel therapeutic approaches are needed for patients with advanced HNSCC. In this study, we investigated the antitumor effects of a regimen combining the HDAC inhibitor, vorinostat and the WEE1 kinase inhibitor, AZD1775, in HNSCC tumor cells harboring high-risk TP53 mutations. Consistent with previous reports, we show that vorinostat-induced cell death in HNSCC is independent of TP53 status (22, 38). However, HNSCC cells expressing mutp53 are more sensitive to vorinostat than isogenic counterparts with wtp53 or p53 null, in agreement with data showing that vorinostat preferentially kills mutp53 cancer cells, established from various tumor types, in vitro (19–21). This difference in sensitivity toward cells with different TP53 status could be due to the pleiotropic nature of vorinostat primarily acting at the epigenetic transcriptional level (39, 40). One significant finding of our study is that vorinostat is highly synergistic with AZD1775 in vitro in HNSCC cells with high-risk TP53 mutation. In addition, our in vitro experiments guided by live cell imaging, demonstrate for the first time that, vorinostat not only arrests cells at the G₂ cell-cycle phase but also causes accumulation of cells in early S-phase, and the combination of drugs results in significant replication stress and an early apoptotic death within the S-phase, perhaps in response to stalled and collapsed replication forks. Taxanes are drugs used in the regimen for treating HNSCC, and their effects on radiosensitization have been attributed to their abilities to block tumor cells at the mitotic (M) phase. Therefore, combining taxanes with HDAC inhibitors may seem an attractive combination to improved HNSCC cell killing compared with WEE1 inhibitor. However, this is unlikely as vorinostat is expected to arrest cells at the S and G₂-phases and cause mitotic delay.

Mechanistically, combination of vorinostat and AZD1775 significantly increases the DNA damage marker, γ-H2AX (S139) levels, associated with replication stress followed subsequently by apoptosis in all TP53-mutant HNSCC cells tested. This mechanistic possibility is supported by our finding that vorinostat alone or in combination with AZD1775 resulted in activated CDK1 and increased hyperphosphorylation of the replication stress marker, RPA32 associated with reduced protein levels of RRM2. Moreover, the concomitant downregulation of Chk1 activation and Rad51 likely contribute to the HNSCC cells' inability to deal with replication stress, thereby leading to cell death.

For the first time, we show that vorinostat treatment alone or in combination with AZD1775 impacts regulation of nearly 1,400 genes in TP53-mutant HNSCC, and based on significant fold change in expression, key pathway genes implicated in cancer are modulated. Consistent with published reports, vorinostat downregulates essential genes involved in regulation of DNA synthesis, HR, and regulation of the G₂–M checkpoints, including CHK1, CDK1, PLK1, CCNB1 (cyclin B1), RRM2, RAD51, FOXM1, and BIRC5 (survivin; refs. 41–45). Based upon their well-established functions in regulating replication stress and G₂–M cell-cycle phase in cancer cells (41–45), it is likely that these genes play an important mechanistic role in synergism between vorinostat and AZD1775 in TP53-mutant HNSCC cells. In addition, HDAC inhibitors act through prevention of the deacetylation of histones, thereby increasing histone acetylation favoring a relaxed chromatin confirmation favoring transcription; our RNA-seq data indicate that transcription levels of 9 genes were increased, and 130 genes were decreased upon vorinostat addition and when combined with AZD1775 compared with untreated control or AZD1775 alone.

As a preclinical evaluation, we demonstrated that oral administration of vorinostat combined with AZD1775 markedly inhibited in vivo growth of high-risk TP53-mutant HNSCC tongue xenografts and prolonged animal survival. Furthermore, tumor growth inhibition was associated with increased apoptosis in vorinostat alone or in combination with AZD1775 compared with the control groups consistent with the in vitro results. Moreover, vorinostat, AZD1775, or the combination significantly lead to CDK1 activation in vivo in orthotopic tongue tumors compared with untreated tumors. Although few studies have shown that vorinostat alone suppresses tumor growth of myeloid leukemia and B-cell lymphoma cells in vivo (22, 46), we observed little single-agent activity of this drug in vivo in mice injected with TP53-mutant HNSCC cells and treated with similar reported pharmacologic doses. Angiogenesis plays an important role in a broad array of physiologic and pathologic processes, including tumor growth. Consistent with published results (36, 37), we show that the fraction of cells with vasculature marker CD31 is decreased following vorinostat alone or in combination with AZD1775, supporting our previous finding that disruption of the tumor microvasculature can contribute to antitumor efficacy in HNSCC (47). Although similar findings were reported previously in myeloid leukemia (22), our data revealed the underlying causes, timing of apoptosis, and identified key pathway genes that possibly regulate the synergy between vorinostat and AZD1775 in HNSCC cells. Furthermore, we provide evidence that low expression of these genes trended toward almost significant correlation with better survival in the TCGA HPV-negative HNSCC cohort, suggesting that vorinostat synergizes with AZD1775 to downregulate basal transcriptional levels of prosurvival and proliferative genes in HNSCC.

In summary, we demonstrate that combination of vorinostat and AZD1775 is highly synergistic in HNSCC tumors bearing high-risk TP53 mutation in vitro and in vivo. One hypothesis for this synergism is that WEE1 inhibition promotes progression of cells in different phases of the cycle into the next phase and that vorinostat decreases the expression of genes that the cells need to adapt in next phase of the cell cycle, and this leads to cell death. More specifically, WEE1 inhibition enhances CDK1 activity, and this leads to either cells in G₁ to be pushed prematurely into S-phase causing replication stress, or cells in G₂ phase to be pushed prematurely into mitosis, leading to mitotic dysfunction. The loss of important replication stabilizing and HR proteins Chk1, and Rad51, and likely others leaves unrepaired DNA, which in S-phase cells causes replication fork destabilization and more DNA damage that cannot be repaired, leading to apoptosis. Taken together, our findings argue that a strategy aimed at simultaneous interruption of WEE1 and HDAC function is a novel therapeutic regimen and deserves investigation in patients with advanced HNSCC.

No potential conflicts of interest were disclosed.

Conception and design: N. Tanaka, A.A. Patel, A. Lindemann, F. Johnson, S. Fu, W.N. Hittelman, J.N. Myers, A.A. Osman

Development of methodology: N. Tanaka, A.A. Patel, A. Lindemann, M.J. Frederick, W.N. Hittelman, A.A. Osman

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): N. Tanaka, A.A. Patel, L. Tang, N.L. Silver, H. Takahashi, N.N. Kalu, T.-C. Chen, J. Wang, F. Johnson, S. Fu, W.N. Hittelman, A.A. Osman

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): N. Tanaka, N.L. Silver, H. Takahashi, R. Jaksik, X. Rao, M.J. Frederick, F.O. Gleber-Netto, S. Fu, M. Kimmel, J. Wang, W.N. Hittelman, C.R. Pickering, J.N. Myers, A.A. Osman

Writing, review, and/or revision of the manuscript: N. Tanaka, A.A. Patel, N.L. Silver, H. Takahashi, M.J. Frederick, S. Fu, W.N. Hittelman, J.N. Myers, A.A. Osman

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A.A. Patel, N.L. Silver, H. Takahashi, J. Wang, J.N. Myers, A.A. Osman

Study supervision: N. Tanaka, A.A. Osman

This work was supported by the NIH/NIDCR R01DE024601 (to A.A. Osman and J.N. Myers) and The German Research Foundation (to A. Lindemann).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.