The Neddylation Inhibitor Pevonedistat (MLN4924) Suppresses and Radiosensitizes Head and Neck Squamous Carcinoma Cells and Tumors Free

Head and neck squamous cell carcinoma (HNSCC) refers to SCC of the upper aerodigestive tract, including the oral cavity, pharynx, larynx, and nasal cavity that is characterized by poor survival rates in advanced stages. It is the sixth most common cancer type worldwide with approximately 600,000 new cases diagnosed annually (1, 2). The overall 5-year survival rate is 60%, only a slight improvement from 53% in the late 1970s (3). Many patients with advanced disease fail current treatments and those with locoregional recurrences suffer painful deaths. Treatment consists of surgery and/or radiotherapy, with the addition of chemotherapy for patients with advanced disease. There is significant heterogeneity in HNSCCs at the genomic and proteomic levels. For example, human papilloma virus (HPV)-associated HNSCC tumors exhibit elevated expression levels of p16 and low levels of cyclin D1, and unlike HPV-negative (HPV-ve) tumors, respond relatively well to radiotherapy (4, 5). Escalation of radiation doses is limited by treatment-related morbidity, which negatively impacts quality of life. Treatment is often disfiguring and can lead to a lifetime of breathing, speaking, and swallowing dysfunction. The economic and quality of life effects on the patients, families, and society are significant, with many patients requiring long-term medical care and nutritional support. Despite improvements in surgical and radio/chemotherapeutic techniques, there has been little improvement in overall survival in the past four decades. Cetuximab, an inhibitor of the EGFR receptor, is the only targeted chemotherapeutic agent approved for curative treatment of HNSCC in the past decade and provides only modest survival improvement compared with radiation alone. Because radiation dose escalation is restricted by adjacent normal tissue toxicity, and recent attempts at chemoradiation dose escalation with cisplatin and cetuximab have not proven beneficial (6, 7), cure rates will likely be improved by the development of effective novel radiosensitizers that selectively target tumor cells while preserving normal mucosa.

Protein polyubiquitylation is a posttranslational modification by which cellular proteins are targeted for proteolysis by the 26S proteasome, although in some cases, it can augment the protein's function or cellular location, in response to specific stimuli (8, 9). The highly coordinated process involves three enzymatic steps with enzymes E1, E2, and E3 functioning successively to promote ubiquitin conjugation, with E3 ligases and substrate recognition subunits (SRS) conferring specificity for the target substrates (9, 10).

Cullin RING E3 ubiquitin ligases (CRL) are the largest family of E3 ubiquitin ligases in mammals and have a number of substrates involved in tumorigenesis (11). The CRL4 E3 ligase with the SRS CDT2 (CRL4^CDT2) is emerging as a master regulator of genome stability and the cell cycle (12–17). This is mediated primarily through the targeted proteolysis of the replication licensing protein CDT1, the CDK inhibitor p21, and the histone methyltransferase SET8 during the S-phase of the cell cycle (12–27). CRL4^CDT2 is also activated, and its substrates are degraded, following DNA damage, such as that induced by ionizing radiation (IR), and this is important for PCNA-dependent DNA repair, and for IR-induced early G₂–M checkpoint (11, 26).

CRL4^CDT2-mediated ubiquitylation of CDT1, SET8, and p21 in the S-phase of unperturbed cells prevents reinitiation of DNA replication, a phenomenon termed DNA rereplication. Rereplication is deleterious to cells and, in some cases, induces cellular senescence and/or apoptosis due to the accumulation of toxic replication intermediates and replication fork stalling (11, 26). Rereplication can be induced experimentally through genetic inactivation of components of the CRL4^CDT2 complex (e.g., through CDT2 depletion or knockdown by siRNA), through CDT1 activation (e.g., by knockdown of geminin, a small protein inhibitor of CDT1), or through the activation of the APC/C^CDC20 E3 ligase (e.g., by knockdown of the APC inhibitor EMI1; refs. 11, 26). Finally, the discovery that treatment of cancer cells with the neddylation inhibitor pevonedistat (also known as MLN4924) inhibits cancer cell lines and tumors and is associated with significant induction of rereplication demonstrated that rereplication can be induced pharmacologically, and this may have an anticancer activity (27–31).

Here, we demonstrate that CDT2 is overexpressed in HNSCC and is required for the proliferation of HNSCC cells and tumorigenesis mainly to suppress aberrant DNA rereplication. Furthermore, we show that inactivation of CRL4^CDT2 or the induction of rereplication pharmacologically or genetically induces robust radiosensitization in HNSCC cells and tumors.

Cal27, FaDu, and SCC25 HNSCC cells were obtained from the ATCC in 2012. UNC7 cells were provided by Dr. Wendell Yarbrough (Vanderbilt University, Nashville, TN) in 2013. Cells were grown in DMEM/Ham's nutrient mixture F12 supplemented with 10% FBS and 1% penicillin/streptomycin. OKF6-TERT2 cells were purchased from Dr. James Rhienwald at Harvard Medical School (Department of Dermatology, Boston, MA) in 2013 and were cultured in GIBCO keratinocyte serum-free medium, supplemented with 25 μg/mL BPE, 0.4 mmol/L CaCl₂, 0.2 ng/mL EGF, and 1% penicillin/streptomycin. All cell lines were maintained under 37°C in 5% CO₂ and were regularly tested for mycoplasma contamination using MycoAlert (Lonza). Cells were authenticated on the basis of growth and morphologic characteristics as well as by DNA fingerprinting (University of Arizona, Tucson, AZ). MLN4924 was purchased from Active Biochem and was dissolved in 10% DMSO in sterile PBS. Propidium iodide, 7-AAD, and BrdU Kit were purchased from BD Biosciences. Antibodies against p21 (C19), p53 (DO-1), geminin (FL-209), and actin (I-19) were purchased from Santa Cruz Biotechnology. Antibodies against SET8, CHK1, CHK2, γH2AX, p-CHK1 (S375), p-CHK2 (T68), p-p53 (S15), and PARP were purchased from Cell Signaling Technology. Anti-Cul3 antibody was purchased from Bethyl Laboratories. Anti-EMI1 antibody was purchased from Life Technologies. Anti-CDT1 and anti-CDT2 antibodies were described before (21).

HNSCC cells were lysed using RIPA lysis buffer (50 mmol/L Tris, pH 8.0, 150 mmol/L NaCl, 1% NP-40, 0.5% sodium deoxycolate, 0.1% SDS, 1 mmol/L benzamidine-HCl, 0.5 μg/mL leupeptin, 0.5 μg/mL aprotinin, 1 μg/mL pepstatin, 20 mmol/L NaF, 20 mmol/L Na₃VO₄). Equal amounts of protein were electrophoretically separated in a polyacrylamide 8% to 12% gel (Bio-Rad), transblotted to a nitrocellulose membrane, and incubated with primary antibodies for 1 hour at room temperature or overnight at 4°C. The immunoblot signals were detected by enhanced chemiluminescence (EMD Millipore).

Transfections of different siRNAs (10 nmol/L) were performed using lipofectamine RNAimax according to the manufacturer's protocol (Invitrogen). The following siRNAs (sense strands) were used: si-GL2: 5′-AACGUACGCGGAAUACUUCGA-3′; si-CDT2: 5′-GAAUUAUACUGCUUAUCGA-3′; si-geminin: 5′-UGCCAACUCUGGAAUCAAA-3′. si-EMI1: 5′-CGAAGUGUCUCUGUAAUUA-3′.

HNSCC cells were transfected with siRNA (48 hours prior to first count) or treated with pevonedistat (24 hours prior to first count). Cells (8 × 10⁵) were seeded in 60-mm plates, and cell proliferation was determined by staining with trypan blue and counting by Countess Automated Cell Counter (Invitrogen). Depending on the cell growth rate, cell counts were recorded either every 24 or 48 hours, and growth curves were established. The effect of pevonedistat treatment or transient silencing of CDT2, geminin, or EMI1 on cell proliferation or on radiation sensitivity was tested using clonogenic survival assays. Cells were transfected with the appropriate siRNA 48 hours prior to seeding. Cells were then counted using Countess Automated Cell Counter and were seeded at 15,000 cells per plate in 60-mm dishes. Twenty-four hours after seeding, cells were irradiated with various doses and were cultured for 7 to 10 days. Once colonies reached the appropriate size (>50 cells each), they were washed twice with cold PBS, fixed in cold 100% methanol for 10 minutes, and stained with crystal violet (0.5%) for 10 minutes. The plates were washed, dried, and imaged using Image Lab software (Bio-Rad). QuantityOne software (Bio-Rad) was used to quantify the number of colonies, and survival curves were established on the basis of the linear quadratic model, using the formula S = e^−αD-βD2, where S represents the surviving fraction and D the dose of irradiation. Results are represented as mean ± SD of three independent experiments normalized to the corresponding nonirradiated plates for each group. When the effect of pevonedistat on cell radiosensitivity was tested, cells were seeded at 15,000 cells per 60-mm plate and allowed to adhere for 4 to 6 hours. Pevonedistat was then added upon cell adherence at varying concentrations, and cells were irradiated the following day. The duration of the experiment and the stopping procedure were as described above.

The effect of pevonedistat treatment or CDT2, geminin, or EMI1 knockdown on the cell cycle (and induction of DNA rereplication) was assessed by flow cytometry with propidium iodide (PI) staining. Cells were harvested at 72 or 96 hours posttreatment with pevonedistat or posttransfection, respectively. Cells were collected, washed with PBS, and resuspended in ethanol (75%). Cells were subsequently treated with 20 μg of DNase-free RNase and stained with PI following the manufacturer's protocol. FACScan (Becton Dickinson) was used to analyze the samples, and G₀–G₁, S, and G₂–M fractions were segmented. Subsequent analysis using FlowJo and ModFit softwares was used to determine apoptotic and rereplicating fractions. Where indicated, Cal27 and FaDu cells were treated with pevonedistat for 48 hours and pulsed with bromodeoxyuridine (BrdU; 10 nmol/L) for 1 hour in the dark prior to harvesting. Cells were washed with PBS and staining solution before fixation and permeabilization steps according to the manufacturer's protocol. Cells were subsequently stained with anti-BrdU antibody solution for 20 minutes at room temperature, washed, and stained with 7-AAD for 30 minutes at 4°C. Cells were resuspended in 1 mL of staining buffer and stored at 4°C overnight before analysis. Sampled were analyzed on a FACScan (Becton Dickinson), and different fractions of BrdU-positive cells were determined using FlowJo and ModFit softwares.

The animal studies were conducted in accordance with the guidelines established by the University of Virginia Animal Care and Use Committee. The effect of pevonedistat on tumor growth was tested in a flank HNSCC xenograft model. Four- to 5-week-old Foxn1^nu athymic female nude immunodeficient mice (20–25 g bodyweight; Harlan Laboratories) were used in this study. Pevonedistat was prepared in 10% DMSO containing PBS and filtered before use. Cal27 cells (5 × 10⁶; suspended in 200 μL sterile PBS) were inoculated subcutaneously in both flanks of nude mice (8 mice/group). When the tumor size reached 100 mm³ (10 days postinoculation), mice were randomized and were treated with pevonedistat (20 mg/kg) or with control vehicle (DMSO), administered intraperitoneally on a regimen of 5 days on/5 days off for 2 cycles (28). Tumors from a third group of mice were exposed to 1 Gy IR daily, 5 days per week for 3 weeks, and a fourth group of mice received both pevonedistat and IR treatments. Tumor irradiation was performed at the University of Virginia X-Ray facility, and only the tumors on both flanks were irradiated while the rest of animal body was shielded. For combination treatment, pevonedistat was given 2 hours prior to radiation exposure with the same schedule as for the individual treatments. Mice were weighed once a week during the entire course of the experiment, and no significant effect of either treatment was observed. Tumor growth was monitored every other day using an electronic caliper, for 3 weeks posttreatment, and average of tumor volumes were calculated using the formula [L × W²)/2]. The results are represented as the mean tumor volumes ± SEM, and P < 0.05 was considered statistically significant.

The Cancer Genome Atlas (TCGA) data, publicly available at cBioPortal (32, 33), were used to plot Kaplan–Meier plots on tumors divided into two groups based on CDT2 expression as a z-score (34–36).

All experiments were performed in triplicates and results with P < 0.05 were considered statistically significant. All quantitative differences were analyzed by Student t test. Synergy was determined using the Bliss model of independence (37, 38).

CDT2 expression is elevated in a number of human malignancies, including breast, gastric, liver, brain, and skin cancers (27, 39–41). In addition, the DTL gene encoding CDT2 is amplified in a subset of Ewing carcinoma (42). Using mRNA expression in public databases of HNSCC (43–46), we found that CDT2 mRNA expression is elevated in oropharyngeal and nasopharyngeal carcinoma (∼4.5- and 5.5-fold) compared with normal squamous mucosa of the oral cavity and nasopharynx, respectively (Fig. 1A and B). CDT2 ranks in the top 3% in oropharyngeal SCC and in the top 1% in nasopharyngeal carcinoma of overexpressed mRNAs in these arrays. CDT2 was also overexpressed in other HNSCCs (43, 44, 46), including oral cavity carcinoma, tonsillar carcinoma, and floor of mouth carcinoma (Supplementary Fig. S1). Elevated CDT2 expression in hepatocellular carcinoma, gastric cancer, and melanoma is associated with poor overall and disease-free survival (27, 40, 47). To test whether elevated CDT2 expression similarly correlates with patient survival in HNSCC, we stratified CDT2 expression (based on RNA-seq) in two large datasets of HNSCC tumors available through The Cancer Genome Atlas (TCGA) databases (32, 33) into high- and low-CDT2 expressors, but found no statistically significant correlation between CDT2 expression and overall or disease-free survival (Supplementary Fig. S2). We conclude that CDT2 overexpression in HNSCC is not predictive of patient outcome.

Next, we tested whether CDT2 is essential for the proliferation or viability of HNSCC cell lines. We silenced the expression of CDT2 in two HPV-ve HNSCC cell lines, Cal27 and FaDu, using a previously validated siRNA (18). We chose these two lines because they were extensively profiled and were found to harbor some of the most common mutations found in head and neck cancers, including inactivating mutations in genes encoding p53, p16, and NOTCH2/3 receptors (48–50). CDT2 depletion in either of these cell lines resulted in a significant increase in the levels of CRL4^CDT2 substrates p21 and SET8 (Fig. 2A). However, we did not detect significant increase in CDT1 protein in Cal27 or in FaDu cells, presumably because CDT1 ubiquitylation and degradation in the S-phase is additionally mediated via the activity of SCF^SKP2 E3 ubiquitin ligase following its phosphorylation by cyclin A/CDK2 (51–53). Consistent with the protective role of CDT2 for genome stability, transient silencing of CDT2 led to the accumulation of spontanous DNA damage as manifested by the induction of γH2AX and increased phosphorylation of the checkpoint kinases CHK1 and CHK2 (Fig. 2A). Importantly, CDT2 depletion inhibited the proliferation of Cal27 and FaDu cells (Fig. 2B–D). Inhibition of cell proliferation in CDT2-depleted cells was accompanied by significant morphologic changes commonly seen in cells undergoing DNA rereplication; flattening of cells, and increased nuclear and cytoplasmic volume (Fig. 2E). Consistently, FACS analysis of Cal27 or FaDu cells depleted of CDT2 demonstrated significant increase in cells with greater than 4N DNA content with 55.7% of Cal27 and 43.9% of FaDu cells undergoing rereplication (Fig. 2F and G). BrdU labeling confirmed that DNA rereplication occurred within the same cell cycle (unpublished observation). We did not however observe a substantial increase in cells with sub-G₁ DNA content (Fig. 2F), consistent with the observation that only a minor increase in cleaved PARP protein (an apoptotic marker) was detected in cells depleted of CDT2 (Fig. 2A). Furthermore, the depletion of CDT2 in these two cell lines was not associated with cell senescence, as we did not detect β-galactosidase staining in these cells (unpublished observation). This is likely due to mutations in CDKN2A (encoding p16), which we have shown previously to be critical for rereplication-induced senescence (27). These results suggest that neither apoptosis nor senescence contribute significantly to proliferation inhibition following CDT2 depletion in HNSCC cells. Collectively, these results demonstrate that CDT2 plays important roles in promoting the proliferation of HNSCC and in preventing DNA rereplication and the accumulation of DNA damage.

Pevonedistat was shown to induce rereplication in a variety of cancer cell lines (28). Furthermore, our recent study in melanoma demonstrated that pevonedistat-induced rereplication and growth inhibition are dependent on CRL4^CDT2 inhibition and the consequent increased stability of p21 and SET8 proteins (27). Thus, we examined the impact of pevonedistat on the proliferation of HNSCC cells by treating Cal27 or FaDu cells with increasing concentrations of pevonedistat. Treatment of either cell line with 50 nmol/L pevonedistat inhibited cullin neddylation at 24 or 48 hours and was accompanied by increased levels of the CRL4^CDT2 substrate p21 (Fig. 3A). In contrast, CDT1 and SET8 began to accumulate at early time points but returned to normal levels by 24 hours (Supplementary Fig. S3) Furthermore, pevonedistat treatment resulted in the accumulation of DNA damage in Cal27 and FaDu cells as manifested by γH2AX and was accompanied by activation of the G₂–M checkpoint, as evident by increased phophorylated CHK1 and CHK2 kinases (Fig. 3A). Similar to cells depleted of CDT2, pevonidestat treatment of Cal27 or FaDu cells did not result in significant apoptosis, as we only detected a minor increase in cleaved PARP (Fig. 3A). Exposure of Cal27 and FaDu cells to pevonedistat (80 nmol/L) inhibited the proliferation of both cell lines, as determined by cell counting (Fig. 3B). Colony formation and CyQuant cell viability assays demonstrated that pevonedistat inhibited the proliferation of HNSCC cells with an IC₅₀ of approximately 50 nmol/L (Fig. 3C and D), a concentration that is comparable with that seen in the most sensitive melanoma cell lines (27). Furthermore, transient treatment of Cal27 with pevonedistat (100 nmol/L) for 24 hours was sufficient to permanently halt cell proliferation (Supplementary Fig. S4). FACS analysis by PI staining showed that pevonedistat induces robust rereplication with the majority of Cal27 (61 %) or FaDu (60.4%) cells treated with 100 nmol/L pevonedisat exhibiting >4N DNA content (Fig. 3E). Furthermore, rereplication was observed with 80 nmol/L concentration as early as 24 hours following treatment (Fig. 3F). On the other hand, only 5% to 10% of the cells exhibited sub-G₁ DNA content (indicative of apoptosis) when examined 72 hours posttreatment (Fig. 3G). Importantly, DNA rereplication was observed in less than 3% of control hTERT-transformed keratonocytes (OKF6-TERT2; Fig. 3E and F), consistent with the notion that normal cells have additional mechanisms to suppress DNA rereplication (27).

Recent studies demonstrated that pevonedistat enhances the sensitivity of pancreatic, breast, and colorectal cancer cells to IR (30, 54, 55). As HPV–ve HNSCC cells and tumors are extremely resistant to IR (56–62), we tested whether pevonedistat radiosensitizes Cal27 or FaDu HNSCC cells by measuring cell survival following the incubation of cells with increasing concentrations of pevonedistat using standard clonogenic survival assays. First, we confirmed that both Cal27 and FaDu cells are significantly resistant to IR, with only 81% and 88.4% of the cells losing replicative capacity with 9 Gy (Fig. 4A and B). Pretreatment of Cal27 with increasing concentrations of pevonedistat for 24 hours significantly and dose dependently enhanced their sensitivity to radiation, with sensitivity enhancement ratios (SER) of 2.99 when measured at 10% survival and following 80 nmol/L pevonedistat treatment (Fig. 4A; Table 1). Pevonedistat similarly radiosensitized FaDu cells, albeit to a lower extent, with SER of 1.49 when measured at 10% survival and following 60 nmol/L pevonedistat treatment (Fig. 4B; Table 1). It is important to note that the treatment of either line with higher doses of pevonedistat was associated with complete suppression of proliferation (Fig. 3C), precluding any further assessment of radiosensitization at these higher doses. Pevonedistat also radiosensitized other HPV-ve HNSCC lines, including SCC25 and UNC7 cells, with 40 and 80 nmol/L concentration, respectively (Supplementary Fig. S5).

Pevonedistat-induced radiosensitizing activity in breast and colon cancer cells was attributed to the induction of G₂–M cell-cycle arrest (54, 55). In pancreatic cells, however, pevonedistat induced rereplication, which was stimulated by IR (30). To understand the mechanistic basis of pevonedistat-enhanced radiation sensitivity in HNSCC cells, we analyzed the cell-cycle distribution of Cal27 or FaDu cells exposed to IR with or without pevonedistat treatment (80 nmol/L). As expected, exposure of these cells to 4 Gy failed to upregulate p21 protein (Fig. 4C), presumably due to inactivating mutations in the gene coding p53 in these cells. Consistently, exposure of Cal27 or FaDu cells to 2 or 4 Gy did not result in G₁ growth arrest and instead was accompanied by G₂ cell-cycle arrest, most noticeable at 4 Gy (Supplementary Fig. S6). However, an increase in the number of Cal27 cells, and more prominently FaDu cells, with polyploid nuclei was evident (Fig. 4D and E). As expected, treatment of Cal27 or FaDu cells with 40 nmol/L pevonedistat for 48 hours resulted in 16.8% and 21.6% of the cells undergoing rereplication, respectively (Fig. 4D and E). Importantly, exposure of these cells at the 24-hour time point to 2 Gy resulted in significantly higher percentage of cells undergoing rereplication (32.9 and 42.2, respectively), with further increases at 4 Gy. Increased DNA rereplication by the combined treatment resulted in a small, but reproducible, increase in γH2AX (Fig. 4C), indicating the accumulation of more DNA damage.

The results above suggest that pevonedistat suppresses HNSCC cells through the induction of DNA rereplication, and this additionally elicits radiosensitizing activity. To test whether pevonedistat exhibits anticancer and/or radiosensitizing activities in HNSCC tumors, we inoculated nude mice with Cal27 cells and monitored tumor growth. A randomized group of mice was treated with DMSO or with 20 mg/kg pevonedistat (intraperitoneally) for 5 consecutive days for two cycles separated by 5 days of no treatment, a scheme previously reported in murine studies with this drug (28). A third group of animals received 1 Gy of radiation daily, and a fourth group received both treatments. Animals were weighted and monitored daily and the drug was well tolerated (Fig. 5A). The results demonstrate that pevonedistat significantly suppressed Cal27 xenografts (P = 0.0211; Fig. 5B). Moreover, although IR suppressed these xenografts (P = 5.5 × 10⁻³), its combined treatment with pevonedistat resulted in more suppression than either treatment alone (P = 2 × 10⁻³; Fig. 5B).

Pevonedistat inhibits all cullin-based E3 ligases and additionally exhibits cullin-independent activity (28, 29, 63–67). However, the results described above suggest that the radiosensitizing activity of pevonedistat is due to its ability to induce rereplication, which can be enhanced by IR. To test whether rereplication is sufficient to confer radiosensitivity in HNSCC cells, we silenced the expression of CDT2 in Cal27 or FaDu cells by si-CDT2 for 72 hours to inactivate the CRL4^CDT2 ligase and carried out clonogenic survival assays. CDT2 depletion in Cal27 cells enhanced radiation sensitivity with an SER of 1.34 when measured at 10% survival (Fig. 6A; Table 2). As expected, silencing of CDT2 or exposure of CDT2-proficient cells to 4 Gy IR resulted in the accumulation of DNA damage (γH2AX) and activation of S and G₂–M checkpoints (phosphorylated CHK1 and CHK2 protein), and this was augmented by the combined treatment (Fig. 6B). Furthermore, and similar to pevonedistat treatment, low doses of IR increased the percentage of CDT2-depleted Cal27 cells undergoing rereplication (Fig. 6C). Similar results were obtained in FaDu cells, although CDT2 depletion was less radiosensitizing in this cell line (Fig. 6B and C).

To test whether rereplication was sufficient to confer radiation sensitivity in HNSCC in cells with intact CRL4^CDT2 activity, we depleted Cal27 or FaDu cells of geminin, an endogenous inhibitor of CDT1 protein known to induce rereplication in some cancer cells (27, 68). Consistently, depletion of Cal27 or FaDu cells of geminin induced rereplication (Fig. 6F). Importantly, geminin silencing prior to IR exposure induced robust radiation sensitivity in both lines with an SER of 1.79 and 1.59, respectively (Fig. 6D; Table 2). Furthermore, silencing of EMI1, an inhibitor of the APC/C ubiquitin ligase whose depletion induces rereplication (69), in Cal27 or in FaDu cells similarly induced rereplication (Fig. 6I), and sensitized both lines to IR with an SER of 1.45 and 1.49, respectively (Fig. 6G; Table 2). As is the case for cells treated with pevonedistat (Fig. 4C), or depleted of CDT2 (Fig. 6B), cells depleted of geminin or EMI1 and exposed to 4 Gy IR exhibited more DNA damage and cell-cycle checkpoint activation than cells depleted of these two proteins or exposed to IR (Fig. 6E and H). In addition, and similar to pevonedistat treatment or CRL4^CDT2 inactivation, low doses of IR (2 and 4 Gy) greatly increased the percentage of rereplicating Cal27 or FaDu cells depleted of geminin or EMI1 (Fig. 6F and I). Collectively, these results demonstrate that induction of DNA rereplication is sufficient to confer radiosensitivity in HNSCC cells and that exposure of cancer cells to low or moderate doses of radiation renders them more susceptible to rereplication induction.

Previous studies demonstrated that the CRL4 substrate receptor CDT2 is overexpressed in various human cancers (27, 39–42). This study extends this observation to HNSCC, where CDT2 is found to be significantly overexpressed in HNSCC from various tissues. Unlike the case for hepatocellular carcinoma, gastric cancer, or melanoma where CDT2 expression is correlated with poor patient outcome (27, 40, 47), elevated CDT2 expression in head and neck cancers does not correlate with patient outcome. Thus, although CDT2 is essential for the proliferation of HNSCC cancer cells, it is not likely to be involved in HNSCC tumor progression. Instead, and given that CDT2 depletion in HNSCC cells is associated with DNA rereplication and the accumulation of DNA damage, CDT2 appears to be critical for coping with the replication stress in these highly proliferative cells, functioning as a cancer-associated gene, which is critical for the capacity of cells to survive the consequences of oncogenic transformation; the so called “nononcogene addiction” (70).

We have recently demonstrated that pevonedistat suppresses melanoma in vitro and in vivo through the induction of DNA rereplication downstream of CRL4^CDT2 inhibition and the stabilization of its ubiquitylation substrates SET8 and p21 (27). Here, we demonstrated that pevonedistat suppresses the proliferation of HNSCC cells and tumors, and this was also due to the induction of robust rereplication. In melanoma cells, the main cytotoxicity associated with pevonedistat-induced rereplication appeared to be the induction of senescence, which correlated with the presence of functional p16 tumor suppressor protein (27). However, we failed to detect senescence in Cal27 or in FaDu cells, presumably because they both lack functional p16 protein and additionally have inactivating mutations in the p53 tumor suppressor proteins (48–50). Furthermore, pevonedistat-induced growth inhibition in Cal27 or in FaDu cells was not associated with apoptosis. Importantly, both the induction of rereplication and lack of apoptotic response in Cal27 or in FaDu cells treated with low pevonedistat concentrations (20–100 nmol/L) reported here is in contrast to the robust apoptosis observed in a previous study in SqCC/Y1 or in Tr146 HNSCC cells (55% and 65% of cells) treated with 5- to 10-fold higher doses of pevonedistat (0.5 and 1 μmol/L, respectively; ref. 67). This significant apoptotic response was attributed to the induction of c-FLIP degradation, which the authors showed to be independent of NEDD8 inhibition (67). Furthermore, it is now abundantly clear that hundereds of cellular proteins are regulated through neddylation, and consistently, pevonedistat inhibits several signal transduction pathways, including the NFκB, AKT, and the mTOR signaling pathways, in addition to cullin-signaling (28, 29, 63–66). Thus, pevonedistat may exhibit differential toxicities in HNSCC cells and tumors depending on the genetic backgrounds of the cells or tumors as well as on the concentrations employed. Such wide inhibitory activity must be considered carefully for future clinical trials with this promising anticancer agent.

In this study, we also demonstrated that low doses of pevonedistat significantly radiosensitized HNSCC cells. Furthermore, pevonedistat synergized with IR to suppress HNSCC tumorigenesis in nude mice. In vitro studies demonstrated that low doses of IR greatly and synergistically increased the percentage of cells undergoing rereplication in response to low doses of pevonedistat. The finding that induction of rereplication through direct inactivation of CRL4^CDT2 or through CDT1 activation or stabilization in cells with intact CRL4^CDT2 activity also radiosensitized HNSCC cells strongly supports the conclusion that the main radiosensitizing activity of pevonedistat is mediated through its rereplication-inducing activity. The results also demonstrate that induction of rereplication is sufficient to confer radiosensitizing activity in HNSCC cells. It will be interesting to test the generality of these observations in other radioresistant cancers.

In vitro experimentations also demonstrate that the maximal radiosensitizing activity of pevonedistat was achieved if the drug was administered 24 and 48 hours prior to irradiation (unpublished observation). The lower radiosensitization achieved either by the simultaneous treatment of IR and pevonedistat or by treatment with pevonedistat following IR most likely results from the IR-induced cell-cycle arrest, which may impede rereplication induction. Tumors with functional p53 may also be less susceptible to the radiosensitizing activity of pevonedistat due to the induction of a G₁ cell-cycle block imposed by IR, and because of the protective effect of p53 against rereplication (71). Finally, because CDT2 overexpression confers susceptibility to pevonedistat-induced rereplication (27), we suspect that HNSCC patients whose tumor exhibit elevated CDT2 expression would benefit the most from pevonedistat treatment, both as a single agent, and as an adjuvant for radiotherapy. Because pevonedistat does not induce rereplication in nonmalignant keratinocytes, our results also suggest that pevonedistat-induced radiosensitization may be selective to cancer cells over normal tissue, which is key to reducing treatment-related toxicity in HNSCC.

In summary, our results demonstrate that the CRL4^CDT2 ubiquitin ligase represents a novel molecular target for inhibition and radiosensitization in HNSCC. Furthermore, the results demonstrate that pevonedistat, currently in several clinical trials for human malignancies, exhibits promising antitumor and radiosensitizing activity in HNSCC and that induction of rereplication represents a novel therapeutic strategy for radiosensitization.

No potential conflicts of interest were disclosed.

The authors have filed a patent application related to the radiosensitizing activity of pevonedistat in HNSCC.

Conception and design: V. Vanderdys, A. Allak, M.J. Jameson, T. Abbas

Development of methodology: V. Vanderdys, A. Allak, F. Guessous, T. Abbas

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): V. Vanderdys, A. Allak, M. Benamar, M.J. Jameson, T. Abbas

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): V. Vanderdys, A. Allak, M. Benamar, P.W. Read, T. Abbas

Writing, review, and/or revision of the manuscript: V. Vanderdys, A. Allak, P.W. Read, M.J. Jameson, T. Abbas

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): V. Vanderdys, T. Abbas

Study supervision: M.J. Jameson, T. Abbas

T. Abbas was supported by the NCI R00CA140774. A. Allak was supported by the Farrow Fellowship and by the NCI Cancer Center Support Grant P30 CA44579. M. Benamar was supported by the Farrow Fellowship and by the NCI Cancer Center Support Grant P30 CA44579.

We thank Joanne Lannigan for technical assistance with flow cytometry and members of the Abbas laboratory for helpful discussions.

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.