BAP1 Is a Novel Target in HPV-Negative Head and Neck Cancer

Head and neck squamous cell carcinoma (HNSCC) is the sixth most common type of cancer worldwide, with an annual incidence of approximately 600,000 new cases and approximately 320,000 deaths each year (1). Radiotherapy is the primary treatment modality for most patients with locally advanced HNSCC. Unfortunately, about half of such patients experience recurrence after treatment and ultimately succumb to the disease, suggesting the existence or emergence of radioresistant clones that survive after radiotherapy (2, 3). Finding the means of overcoming resistance to radiation is critical to improving survival for patients with HNSCC.

Recently, our group used proteomic and transcriptomic analysis to screen for potential targets associated with poor outcome following radiation in human papillomavirus (HPV)–negative HNSCC clinical samples (4). One novel candidate identified on reverse-phase protein array analysis (RPPA) screening was the BRCA1-associated protein-1 (BAP1), a nuclear deubiquitinating enzyme consisting of 729 amino acids. BAP1 regulates a number of cellular processes, including cell cycle, cell differentiation, transcription, and DNA damage response (DDR). Moreover, germline mutations in BAP1 can lead to a tumor predisposition syndrome, possibly due to altered DDR (5). BAP1 can modulate the DDR via several different interactions, including deubiquitinylation of histone H2A (6, 7). BAP1 regulates the DDR by interacting with the BRCA1/BARD1 heterodimer and with RAD51, a BRCA1/BARD1 complex with E3 ubiquitin ligase activity (8, 9).

Because of its ability to modulate DDR, BAP1 may also modulate response to radiation. Indeed, Nishikawa and colleagues (8) first found that depletion of BAP1 in HeLa cells resulted in hypersensitivity to ionizing radiation. This discovery, in combination with our own identification of BAP1 in an RPPA screen for proteins associated with treatment failure following radiation in HPV-negative HNSCC, led us to hypothesize that BAP1 participates in the development of radioresistance in HNSCC, potentially via modulation of DDR after irradiation. In testing this hypothesis, we found that BAP1-knockdown (KD) and knockout (KO) cells were more sensitive to radiation regardless of HPV or p53 status, and this sensitivity was reversed by forced reexpression of BAP1-WT in the BAP1-KD cells. We also observed a significant enhancement of radiosensitivity after BAP1-KD in a mouse xenograft model of HNSCC. Finally, in the TCGA HNSCC cohort, we found that BAP1 was associated with disease-free survival. Collectively, our findings suggest that BAP1 promotes radioresistance in HNSCC and is a possible target for therapeutic development.

The HNSCC cell lines used in this study were generously supplied by Dr. Jeffrey Myers via The University of Texas MD Anderson Cancer Center Head and Neck cell line repository. Of the six cell lines tested, HN31, HN30, and TR146 were HPV negative and UMSCC47, SCC154, and SCC152 were HPV positive. Cells were maintained in DMEM (Gibco), supplemented with 10% FBS, 1% penicillin/streptomycin, 1% glutamine, 1% sodium pyruvate, 1% nonessential amino acids, and 1% vitamins and incubated at 37°C in a humidified atmosphere containing 5% CO₂.

Cell lines were transfected with shRNAs specific for BAP1 or control scrambled shRNA via lentiviral vectors containing puromycin resistance gene (GE Dharmacon). Lentiviral-transfected control (pGIPZ; shControl) cells and BAP1 shRNA KD cells were subjected to puromycin selection. After antibiotic screening, pooled clones of control cells and BAP1-KD cells were assessed for BAP1 protein expression by immunoblotting. HN31 and UMSCC47 BAP1-KD cell lines were further transduced with lentiviral vectors (pLVX) encoding wild-type (WT) BAP1 vectors (kindly shared by Dr. Boyi Gan from The University of Texas MD Anderson Cancer Center), to rescue BAP1 expression. BAP1 expression was confirmed by immunoblotting in these cells as well. BAP1 CRISPR/Cas9 knockout plasmid and control CRISPR/Cas9 plasmids were constructed as described previously (10). The BAP1 KO gRNA sequences are shown in the Supplementary Information. Knockout of BAP1 expression was confirmed in these cell lines via immunoblotting.

Cell cultures were irradiated with an X-Rad 320 Biological Irradiator (320 KV, 12.5 mA, SSD 50 cm, Precision X-Ray), and in situ tumors were irradiated with a Shepherd Mark I ¹³⁷Cs irradiator (662 keV Model, 68-A), with a custom block. For the Shepard Mark I, absolute dosimetry was performed employing ion chambers calibrated by the Accredited Dosimetry Calibration Laboratory, in-air employing AAPM protocol TG61. For mouse irradiation, EBT3 (Gafchromic-Film) was employed for dosimetry in simulated irradiation geometry. The mouse was simulated by tissue-equivalent Gel "Superflab." Ratio of EBT3 response in simulated geometry versus ion chamber calibration in reference geometry provides the dose rate in the animal.

Radioresponse (sensitivity or resistance) was assessed via colony formation assay. Known numbers of cells were plated in 6-well plates or 6 dishes for 12 to 14 hours and then irradiated at the indicated doses of radiation. The cells were allowed to form colonies over a 10- to 14-day incubation period and were then fixed in a 1.5% crystal violet/25% methanol solution. Numbers of colonies containing more than 50 cells each were then counted to determine surviving fraction.

Protein expression levels were assessed by immunoblotting from whole-cell lysates. Total protein was extracted by using nuclear and cytoplasmic extraction reagents with protease and phosphatase inhibitors according to the manufacturer's instructions (Thermo Fisher Scientific). The protein concentrations of lysates were measured by the ABC assay (Bio-Rad). The following primary antibodies were used: BAP1 (C-4), p53 (DO-1), and BRCA1 from Santa Cruz Biotechnology; γH2AX, H2Aub (K119), 53-BP1, Chk1, Chk2, p-Chk1, p-Chk2, and β-actin from Cell Signaling Technology. Goat anti-mouse and anti-rabbit secondary antibody conjugated to horseradish peroxidase (Santa Cruz Biotechnology) were used, and signal was generated with the SuperSignal West Chemiluminescence System (Pierce Biotechnology).

Immunofluorescence was measured as described previously (11). In brief, cells were cultivated on cover slips placed in 35-mm cell culture dishes. At specified time points after exposure to radiation (2 Gy), cells were fixed in 4% paraformaldehyde for 15 minutes at room temperature, briefly washed in PBS (Mediatech), and placed in 70% ethanol overnight at 4°C. Then, fixed cells were permeabilized with 0.1% IGEPAL (octylphenoxypolyethoxyethanol) for 20 minutes at room temperature, blocked in 2% BSA (Sigma) for 60 minutes, and then incubated in anti-γH2AX (1:400) or anti-BRCA1 primary antibody (1:400) overnight at 4°C. To assess H2Aub level, cells were permeabilized with 0.4% Triton X-100 for 5 minutes at room temperature, blocked in 5% normal goat serum for 60 minutes, and then incubated in anti-H2Aub (1:1,000) overnight at 4°C. Cells were then washed four times with PBS and incubated for 1 hour in secondary anti-rabbit antibody conjugated with Cy3 (Jackson ImmunoResearch) to visualize γH2AX foci and H2Aub or in secondary anti-mouse antibody conjugated to Cy3 to visualize BRCA1. DNA was stained with DAPI (4′, 6-diamidino-2-phenylindole; Sigma) followed by mounting on labeled slides with mounting media (DAKO). Immunoreaction results were visualized with a Leica Microsystems microscope, and foci were counted with ImageJ software (https://imagej.nih.gov/ij/).

In HEK293T cells, the pDR-GFP plasmid (3 μg) was transfected using PEI transfection reagent in Opti-MEM Reduced Serum Media (5 ug/mL). After an incubation of 4 to 6 hours at 37°C, media containing 10% FBS were added to the cells. Stably expressing cells harboring the pDR-GFP plasmid were selected with (2 μg/mL) puromycin. BAP1 expression in these cells was transiently knocked down by transfecting 10 pmol ON-TARGET plus siRNA Human BAP1 (Dharmacon) by electroporation (Nucleofactor II, Amaxa). After 24 hours, cells were transfected with SceI-expressing plasmid (as described above) to induce double-stranded breaks. Two days after transfection with SceI plasmid, the dishes were washed with PBS and cells were harvested and transferred to filter-top FACS tubes. Flow cytometric analysis for GFP was performed immediately using a BD Accuri C6 Flow Cytometer (BD Biosciences) and the results analyzed using the BD Accuri C6 Software.

Forty male athymic nude mice (6–8 weeks old, ENVIGO/HARLAN) were randomly assigned to one of 4 groups of 10 mice each: untreated HN31 shControl, untreated HN31 BAP1-KD, irradiated HN31-shControl, and irradiated HN31 BAP1-KD. HN31 tumor cells (4 × 10⁶ in 0.1 mL of serum-free medium) were injected subcutaneously in the right dorsal flank of each mouse. After palpable tumors had developed, tumor diameters were measured with digital calipers, and tumor volume was calculated as A × B² × 0.5, where A represents the largest diameter and B the smallest diameter of tumor. When the tumor volumes reached approximately 150 mm³, tumors were irradiated with 10 Gy (2 Gy/day for 5 consecutive days) and tracked for 3 weeks. At that time, the experiment was terminated and tumors harvested.

Clinical data and BAP1 protein expression were abstracted from the Head and Neck Cancer cohort of The Cancer Genome Atlas (TCGA). Of the 530 patients in the cohort, information on BAP1 protein expression and disease-free survival data were available for 248, and these patients formed the basis for this analysis. Survival curves were generated with the Kaplan–Meier method, with log-rank statistics used to identify significant differences between groups. Patients were grouped by upper, middle, and lower tertiles of BAP1 protein expression for this analysis.

Data were reported as mean ± SEM, and Student t tests (unpaired, unequal variance) were used to compare two groups of independent samples for in vitro radiosensitivity, γH2A foci, and BRCA1 foci. GraphPad Prism (v6.0) and SPSS Statistics (v22) were used to analyze the data. A P value <0.05 was considered to indicate statistical significance in all analyses.

Studies of the effects of BAP1 on cell proliferation have shown conflicting results, with some reporting that BAP1 suppresses cell proliferation (12, 13), and others reporting that BAP1 enhances proliferation (14, 15). To investigate the role of BAP1 in HNSCC cell proliferation and clonogenic survival, we tested HN31 (HPV-negative) and UMSCC47 (HPV-positive) HNSCC cell lines transfected with BAP1 shRNA to knock down BAP1 (BAP1-KD). We performed standard MTT assay to access the effect of BAP1-KD on cell proliferation and colony formation assay was used to examine baseline clonogenic survival (Fig. 1). In both cell lines, knockdown of BAP1 significantly inhibited cell proliferation (P < 0.05; Fig. 1A and B) and clonogenic survival (Fig. 1C–F). To determine whether this phenomenon could be reversed, we rescued BAP1 expression via transfection of stably expressing BAP1-WT in HN31-BAP1-KD and UMSCC47-BAP1-KD cell lines. In both sets of experiments, forced reexpression of BAP1-WT reversed the effect of BAP1-KD phenotype.

We previously observed that clinical failure after radiation is associated with high BAP1 expression in tumors. We next examined the role of BAP1 on radioresponsiveness in HNSCC cell lines. To do so, we utilized stable knockdown-BAP1 versions of several cell lines. Because the two most important biomarkers of radioresistance in HNSCC are HPV and p53 mutation, we tested cell lines of different HPV and p53 status, namely HN31 (HPV⁻/p53 mutant), TR-146 (HPV⁻/p53 mutant), HN30 (HPV⁻/p53 WT), and UMSCC47 (HPV⁺/p53 WT). We also generated cells rescued from BAP1-KD by forced expression of BAP1 WT. Colony formation assays after irradiation were used to examine the relationship between BAP1 expression and response to radiation. Interestingly, we found that inhibition of BAP1 universally increased sensitivity to radiation in all cell line variants tested and that forced expression of BAP1 WT reversed this phenomenon (Fig. 2). To rule out shRNA off-target effects, BAP1 gene was knocked out using CRISPR (KO) in UMSCC47 cells and confirmed the KO of BAP1 expression by immunoblotting. Colony formation assay revealed a significant reduction in the survival fraction in BAP1 KO cells after radiation treatment as compared with control cells (P < 0.05; Fig. 2E).

BAP1 has been reported to deubquitinate H2Aub (7), which in turn inhibits transcription in chromatin regions flanking DNA double-strand breaks, facilitating DNA repair (16, 17). Thus, BAP1 is thought to modulate DDR through H2Aub. To verify this assumption in HNSCC, we examined H2Aub expression in BAP1-KD HNSCC lines. Specifically, we transfected HN31, HN30, TR146, UMSCC47, SCC154, and SCC152 cell lines with BAP1 shRNA to generate BAP1-KD cell lines and noted increased H2Aub in these cell lines (Fig. 3A). To determine whether this phenomenon could be reversed, we transfected HN31-BAP1-KD cells and UMSCC47-BAP1-KD cells with stably expressed BAP1-WT and found that reexpression of BAP1-WT successfully reversed the effect of BAP1-KD on H2Aub expression (Fig. 3B). Moreover, in a coculture experiment in which control HN31 or UMSCC47 cells were cultured with their BAP1-KD counterparts, BAP1 was inversely associated with H2Aub expression (Fig. 3C). In addition, BAP1 KO in UMSCC47 cells led to increased levels of H2Aub as compared with control cells (Fig. 3D). Thus, BAP1 seems to regulate H2Aub in HNSCC cells and may be a mechanism by which BAP1 induces radioresistance via facilitating DNA repair.

To better understand the mechanism underlying BAP1-induced radioresistance in HNSCC cells, we counted numbers of γ-H2AX foci (indicative of double-strand breaks) at different times after irradiation. We found that the number of foci in all variants had increased to their peak levels in all cells at 1 hour (Fig. 4A); however, the number of γ-H2AX foci was no different among variants in either UMSCC47 cells or in HN31 cells (Fig. 4B), findings confirmed by Western blotting (Supplementary Fig. S1). Because BRCA1 has an important role in DDR, specifically in homologous recombination (HR)–mediated repair, we also examined BRCA1 foci after irradiation at different time intervals (Fig. 4B). We found significantly fewer numbers of BRCA1 foci in BAP1-KD cells relative to BAP1 vector control and BAP1-WT cell lines at 1, 4, and 24 hours after irradiation in both UMSCC47 and HN31 cell lines, suggesting that BAP1 may affect the recruitment of BRCA1 to DDR. Because others have found that CHK1, CHK2, 53BP1, and ATM proteins have important roles in nonhomologous end joining (NHEJ) in DNA double-strand break repair DRR, we next examined the effect of BAP1-KD on proteins known to be related to NHEJ after irradiation with 4 Gy and found that BAP1-KD did not affect levels of Chk1, pCHk1, CHk2, pCHk2, 53-BP1, or ATM (Supplementary Fig. S1), indicating that NHEJ is not likely the pathway by which BAP1 regulates the DDR. We next examined HR in BAP1-KD and control HEK293T expressing the pDR-GFP plasmid. Inhibition of BAP1 led to a significant reduction in I-SceI–induced HR (P < 0.05; Fig. 4E). Thus, our results indicate that knockdown of BAP1 may lead to radiosensitization via inhibition of HR.

To further confirm the association between BAP1 and radioresistance in vivo, we implanted athymic mice with relatively radioresistant HN31 shControl and HN31 BAP1-KD tumor cells, allowed tumors to develop, and irradiated the tumors with 2 Gy per day for 5 consecutive days (total dose 10 Gy). We observed that 2 Gy × 5 days of radiation treatment had minimal effect on control tumors (Fig. 5) but led to a significant decrease in growth of BAP1-KD tumors (Fig. 5; P < 0.05), suggesting that BAP1 mediated radioresistance in this xenograft model.

We previously identified BAP1 as one of several targets associated with treatment failure following radiation in a screen of HNSCC clinical samples (4), leading us to examine potential associations between BAP1 protein expression and outcomes in the HNSCC cohort of The Cancer Genome Atlas database. We confirmed that BAP1 was associated with poorer disease-free survival in this group (Fig. 6), highlighting the importance of BAP1 expression in disease control in HNSCC.

Patients with radioresistant HNSCC have poor survival outcomes (2). Increasing the radiotherapy dose may overcome radioresistance; however, this is not feasible in HNSCC due to normal tissue toxicity. So, identifying genes that mediate radioresistance, and finding ways of manipulating these genes so as to enhance radiosensitivity, is critical as a therapeutic strategy for patients with HNSCC. Previously, we have identified BAP1 as a potential biomarker of failure following radiation in HNSCC (4). In the current study, we investigated the ability of BAP1 to modulate radioresistance in HNSCC as well as predict outcome in this disease. We found that inhibition of BAP1 expression led to radiosensitization irrespective of HPV or p53 status, likely due to modulation of HR, and that this phenomenon can be rescued by forced expression of WT BAP1. Inhibition of BAP1 also led to in vivo radiosensitization, and BAP1 protein expression was associated with disease-free survival in the TCGA HNSCC cohort.

The two predominant DDR repair pathways are classical NHEJ and HR. Typically, radiation-induced double-strand breaks in DNA are repaired by NHEJ, but HR can have a significant role as well. In our study, we found that BAP1 depletion decreased the assembly of constitutive BRCA1 foci, which are associated with HR, but had minimal effect on γ-H2AX foci. Moreover, proteins associated with NHEJ were not altered by BAP1 modulation. Thus, in HNSCC, BAP1 seems to be important for radioresponse not by NHEJ, but rather by modifying HR.

Previously, Calypso and others have demonstrated that, a BAP1 homolog in Drosophila, was shown to deubiquitinate H2Aub, following DNA damage and was potentially required for silencing of transcription at DSBs (12, 18). In our study, we found that BAP1-KD increased expression of H2Aub protein in 6 different HNSCC cell lines of various HPV status and that this increase was reversed by the forced expression of BAP1-WT. Thus, H2Aub may be a target of BAP1 in HNSCC. Zhou and colleagues reported that histone H2A ubiquitination/deubiquitination was a critical chromatin modification involved in regulating gene expression, maintenance of heterochromatin, and DDR responses (19). Indeed, RNF8 can physically interact with H2A and catalyze its ubiquitination in response to DSBs, leading to BRCA1 and 53BP1 accumulation at flanking regions of DSBs (19, 20). Yu and colleagues (7) found that BAP1 was recruited to DNA break sites after irradiation and H2Aub levels correlated inversely with BAP1 recruitment. In contrast, no recruitment of BAP1 was detected distal to the break sites, where high levels of H2Aub were observed. Our own findings suggest that BAP1 inhibition leads to increased level of H2Aub, which interferes with chromatin and histone modifications at double-strand breaks. Moreover, increased H2Aub is inversely proportional to cell proliferation, and indirect stabilization of H2Aub leads to reduced cell proliferation (7, 18, 21).

In conclusion, we found that BAP1 can deubiquitinate H2Aub and induces radioresistance in HNSCC cells via modulation of HR. This effect seems to be independent of HPV status and to have clinical significance, as BAP1 protein expression was found to be associated with poor outcome in HNSCC. BAP1 expression may serve as a biomarker of radioresponse, and further investigation of the interaction between BAP1, HR, and radiation in this disease is warranted.

No potential conflicts of interest were disclosed.

Conception and design: L. Yang, H.D. Skinner

Development of methodology: X. Liu, L. Yang, H.D. Skinner

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): X. Liu, M. Kumar, L. Yang, D.P. Molkentine, D. Valdecanas, H.D. Skinner

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): X. Liu, M. Kumar, L. Yang, D. Valdecanas, S. Yu, H.D. Skinner

Writing, review, and/or revision of the manuscript: X. Liu, M. Kumar, L. Yang, S. Yu, R.E. Meyn, J.V. Heymach, H.D. Skinner

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): X. Liu, M. Kumar, L. Yang, J.V. Heymach, H.D. Skinner

Study supervision: L. Yang, S. Yu, H.D. Skinner

This work was supported by the following grants from the NCI (NIH): R01 CA 168484-022 (to J.V. Heymach), HNSCC SPORE 5 P50CA070907-16, CCSG 5 P30 CA01667239 (to J.V. Heymach), R01 CA 168485 (to R.E. Meyn and H.D. Skinner), and R21 CA 182964 (to H.D. Skinner). This work was also supported by the Cancer Prevention Institute of Texas (RP150293; to H.D. Skinner) and The University of Texas MD Anderson Cancer Center Multidisciplinary Research Program (to H.D. Skinner and R.E. Meyn).

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.