SMAC Mimetic Birinapant plus Radiation Eradicates Human Head and Neck Cancers with Genomic Amplifications of Cell Death Genes FADD and BIRC2

Head and neck squamous cell carcinoma (HNSCC) is the sixth most common cancer worldwide, with more than 600,000 new cases diagnosed each year and a 5-year survival rate of about 50% (1). The Cancer Genome Atlas (TCGA) analysis of 279 HNSCC recently uncovered genomic alterations in cell death pathways affecting approximately 40% of HNSCC (2–4). These include about 30% that harbor chromosome 11q13/22 amplifications and overexpression of Fas-associated death domain (FADD), with or without Baculovirus inhibitor of apoptosis repeat containing (BIRC2/3) genes that encode cellular inhibitor of apoptosis proteins 1/2 (cIAP1/2). A mutually exclusive subtype (∼10%) bears mutations in caspase-8 (CASP8) together with oncogenes HRAS or PIK3CA (2, 3). Significantly, amplifications of FADD without or with BIRC2 are more often linked to the human papilloma virus–negative [HPV(−)] HNSCCs with worse prognosis (2).

Functionally, FADD, cIAP1/2, and CASP8 are critical components of the TNF receptor family signaling pathways that determine cell death or survival (5–10). Binding of TNFα, TRAIL, or other death agonists to their receptors typically leads to cell death via FADD through the activation of caspase-8 and induction of apoptosis, or alternatively, activation of RIP kinase 1 (RIPK1), MLKL, and induction of necroptosis (5–10). However, FADD overexpression has also been implicated in cell growth in lung cancers and in primary tumors associated with lymph node metastases in HNSCC (11–13), suggesting that its function in cancer is context-dependent. TNFα-FADD signaling and promotion of cell death via apoptosis or necrosis may be inhibited by cIAPs (14).

IAPs are degraded following binding by the second mitochondria-derived activator of caspases (SMAC), an endogenous protein released by mitochondria at the onset of apoptosis, (15). Targeting IAPs with small-molecule mimetics of SMAC can modulate cell death in cancer cells, giving rise to growing interest in their potential as therapeutics (16). Birinapant is a novel bivalent peptidomimetic of SMAC, shown to preferentially target cIAP1, relative to cIAP2 and XIAP (17, 18). Birinapant demonstrated antitumor activity in preclinical models of hematologic cancers and solid tumors (19, 20). In early-phase clinical trials, birinapant has demonstrated tolerability and safety at active doses, with a prolonged plasma half-life of 31 hours and tumor half-life of 52 hours (21). In a 5-arm phase I/II dose escalation study, safety, tolerability, and phase II dosing of birinapant in combination with different chemotherapies in patients with solid tumors were established (22). Interestingly, birinapant demonstrated prolonged progression-free survival in previously relapsed or refractory patients when combined with chemotherapies known to induce TNFα, such as irinotecan (21, 22). Notably, radiotherapy induces TNFα and cytotoxicity (23, 24) and is a critical modality in treatment of HNSCC.

Here we explore the potential and mechanisms by which HNSCC harboring amplifications and overexpression of FADD, without or with BIRC2, may be sensitized to SMAC mimetics alone or in combination with TNFα, TRAIL, or radiation.

TCGA data were accessed through the cBioPortal for Cancer Genomics (http://www.cbioportal.org/) and queried for FADD, BIRC2, BIRC3, and CASP8 mutation and copy number alterations for all cancer studies (25, 26). Comparison for copy number variation between HPV⁺ and HPV⁻ used one-sided Wilcoxon signed rank test.

The 11 HPV⁻and 1 HPV⁺HNSCC cell lines were authenticated by genotyping using 9 allelic markers (27) and obtained from Dr. T.E. Carey at the University of Michigan (Ann Arbor, MI) in 2010. Stocks were preserved in −80°C, cultured as described (27), and used within 3 months.

gDNA was obtained with a DNeasy Blood & Tissue kit (Qiagen) from frozen cell pellets of UM-SCC lines. Whole-exome sequencing with about 87× coverage was performed using SOLiD4 platform (Applied Biosystems) and copy number alterations for FADD and BIRC2/3 genes were analyzed using CONTRA and IGA software (28, 29). A CASP8 mutation was identified using ANNOVAR (http://annovar.openbioinformatics.org/) and Mutation Assessor (http://mutationassessor.org/r3/) using settings specified (Supplementary Methods).

UM-SCC cells were plated overnight before treatment with birinapant ± TNFα or TRAIL or 0.01% DMSO diluent control. At 12, 24, and 48 hours after treatment, whole-cell lysates were collected, and electrophoresis and Western blotting were performed, using antibodies, reagents, and conditions specified in Supplementary Methods.

RNA isolation was performed using the RNeasy Mini Kit from Qiagen and on snap-frozen tumor samples collected from mice after 9 days of treatment using the mirVana miRNA Isolation Kit from Ambion by Life Technologies (AM1561). The High-Capacity cDNA Reverse Transcription Kit (Applied Biosciences) was used per manufacturer's instructions. Thirty nanograms of cDNA was amplified with primers and TaqMan Universal Master Mix in an ABI Prism 7700 Sequence Detection System (Applied Biosystems). Relative gene expression was normalized to 18S control. Samples were assayed in triplicate and presented as the mean ± SD. Significance was determined using the Student t test and P < 0.05 was considered statistically significant.

Cells were transfected as described (30) with 50 nmol/L of nontargeting control, BIRC2, BIRC3, and/or FADD siRNA, or 100 ng/well empty control pcDNA 3.1 or FADD vector plasmids (gifts from Dr. Gabriel Nuñez, University of Michigan; ref. 31), using 0.25 μL/well of Dharmafect Duo (GE Dharmacon) in 100 μL MEM media/well. After 24 hours, cells were treated with birinapant (10 μmol/L) and/or TNFα (20 ng/mL), and viability was assessed via XTT assay 48 and 72 hours following treatment.

Cells were plated in 96-well plates and treated with 0.01% DMSO control or birinapant concentrations indicated ± 20 ng/mL TNFα (R&D systems). Cell density was measured by XTT Kits (Roche) and inhibitory concentration 50% (IC₅₀) was determined at 48 to 72 hours, as described (Supplementary Methods). To differentiate between apoptotic and necroptotic cell death, cells were treated with birinapant ± TNFα, TRAIL, ZVAD, ZIETD, or necrostatin (BD Pharmingen). Cells were collected, lysed, and assayed for activity of caspases-3 and -8 using assay kits (Biovision) per manufacturer's protocol on a microplate reader.

Cells were plated and treated over 24 to 72 hours with 0.01% DMSO control, 20 ng/mL TNFα, or 1 μmol/L birinapant ± TNFα, and prepared for flow cytometry using the Cycle TEST PLUS DNA Reagent Kit (BD Biosciences), for analysis on the FACS Canto flow cytometer (BD Biosciences). Data from 10,000 cells per treatment group and time point were analyzed using BD FACSDiva (BD Biosciences) software.

All animal experiments were carried out under protocol 1322-13 approved by the NIDCD Animal Care and Use Committee. Four- to 6-week-old female SCID/NCr-Balb/c mice and athymic nu/nu were obtained from Frederick Cancer Research and Development Center (National Cancer Institute) and housed in a specific pathogen-free animal facility. Mice were injected with UM-SCC cells, randomized, and treated with birinapant and radiation as indicated and specified in Supplementary Methods.

Mice harboring UM-SCC-46 tumors were sacrificed after 11 days of control and birinapant treatment indicated. Tissue samples were embedded in OCT, and frozen tissue sections were prepared by Histoserv, Inc. Tissue sections were fixed using 4% formalin, permeabilized with methanol, and quenched with 3% H₂O₂. Samples were blocked using rabbit serum and stained with primary and secondary antibodies specified in Supplementary Methods using the VECTASTAIN Elite ABC Kit (Vector Laboratories). Slides were imaged using the Aperio ScanScope (Leica Biosystems). Histoscores were calculated by ImageScope software (Leica Biosystems).

Data from XTT assays and colorimetric assays were presented as mean ± SD. For tumor growth analysis, significance was determined using the Student t test and P < 0.05 was considered statistically significant. For survival analysis, the Gehan–Breslow–Wilcoxon test was used and significance set to 0.05 using the Bonferroni method.

We surveyed the frequency of gene copy number, mutations, and expression alterations for FADD, BIRC2, BIRC3 on 11q13/22, and CASP8 in TCGA datasets, for 279 HNSCC tumor specimens, and other cancers (Fig. 1A). Compared with other cancer types, alterations of FADD are highest in HNSCC, with about 30% exhibiting amplifications. HNSCC is the second most common tumor with alterations in the BIRC2/3 locus (8.6%). Alterations of CASP8, predominantly mutations, are most frequent in HNSCC (11.2%). Including mRNA up- or downregulation, about 60% of HPV⁻and HPV⁺HNSCC tumors harbor alterations for these genes (Fig. 1B). FADD amplification and increased RNA expression is predominantly seen in HPV⁻tumors and is associated with early recurrence and worse prognosis (Fig. 1C and D).

Because of the prevalence and association of FADD/BIRC2/3 and CASP8 genomic alterations with HPV⁻HNSCC tumors, we used exome sequencing to examine the genomic alterations of these key cell death pathway components in a panel of 11 human HPV⁻UM-SCC cell lines from patients with poor outcomes (Supplementary Table S1). Consistent with TCGA, focal amplifications of 11q13/22 loci for FADD and BIRC2/3 were found in a subset of UM-SCC lines (Fig. 2A). Real-time PCR confirmed an increase in FADD mRNA expression of more than 3-fold in UM-SCC-11A, -11B, -22B, and -46, as compared with primary human oral keratinocytes (HOK; Fig. 2B). BIRC2 expression was increased more than 3-fold in UM-SCC-6, -11A, -11B. BIRC3 was more weakly expressed despite coamplification. UM-SCC-1 was found to harbor a heterozygous CASP8 mutation (Q107 nonsense mutation, NS), confirmed recently (3). UM-SCC-11A/B, -22A/B, -38, and -74A/B harbor a common single-nucleotide variant resulting in a K14R amino acid change in CASP8 of unknown functional significance (data not shown). Corresponding FADD, cIAP1, cIAP2, and CASP8 proteins analyzed by Western blot in HOK and UM-SCC cell lines demonstrated differences in expression largely consistent with the RT-PCR findings (Fig. 2C). Thus, differences in genomic alterations and expression of key cell death pathway components in human tumor specimens in TCGA are represented among these cell lines.

We examined the functional significance of FADD, BIRC2 and 3 by siRNA knockdown alone or with TNFα in 2 cell lines differing in amplification and expression of these genes. In UM-SCC-11A, with relatively higher FADD and BIRC2 amplification and expression, FADD and BIRC2 knockdown alone, and in combination with TNFα, decreased cell density as measured by XTT assay (Fig. 3A, left; P < 0.001). In UM-SCC-1 harboring lower amplification and expression of FADD and BIRC2 with the heterozygous CASP8 Q107 nonsense mutation, weaker inhibition of cell density was observed with knockdown in the absence or presence of TNFα. (Fig. 3A, right). The combined effect of FADD and BIRC2 knockdown resulted in similar reduction in cell density as knockdown with either individual gene (P < 0.001). BIRC3 knockdown alone had minimal effect, consistent with the low BIRC3 expression. Knockdown of the targets by more than 80% was confirmed in both lines (Supplementary Fig. S1). Together, these data suggest that FADD and BIRC2 overexpression may contribute to growth and/or TNFα resistance of HNSCC.

As antagonists of BIRCs/IAPs may inhibit cancer cells, we screened the panel of UM-SCC cell lines for sensitivity to SMAC mimetic and IAP inhibitor birinapant by XTT assay over a range of concentrations (1 nmol/L–10 μmol/L) used in previous studies (20). Figure 3B summarizes the inhibition of cell density of the panel following treatment with birinapant or cell death agonists TNFα or TRAIL alone and in combination, as assessed during exponential growth on day 3, with the IC₅₀ values provided in Supplementary Table S2. Multiple cell lines showed enhanced sensitivity to birinapant + TNFα or TRAIL, including many that showed increased expression of FADD and/or other pathway components above. Birinapant alone was most active in UM-SCC-46, with increased expression of FADD, similar BIRC2, but low CASP8 relative to normal HOK cells. An HPV⁺ cell line, UM-SCC-47, showed modest sensitivity to birinapant and death ligands. The least TNFα-sensitive cell lines included UM-SCC-9 and 38, displaying weaker expression of FADD/BIRC2.

We established that gene transfer and overexpression of FADD, but not empty vector, enhanced sensitivity of the FADD-deficient line UM-SCC-38 to birinapant and TNFα, supporting a distinct function of FADD overexpression in sensitization to death ligand when combined with IAP inhibition (Fig. 4A). To further evaluate whether decreased cell density observed was related to cell cycle and/or cytotoxic effects, DNA cytofluorometric analyses of sensitive UM-SCC-46 and less sensitive UM-SCC-1 cell lines were compared after treatment with birinapant plus TNFα for 48 hours (Fig. 4B; Supplementary Fig. S2A and S2B). UM-SCC-46 cells overexpressing FADD treated with birinapant alone, or birinapant and TNFα, exhibited strong increases in sub-G₀ DNA indicative of cell death as well as increased G₂–M accumulation. In contrast, treatment of UM-SCC-1 cells with weak FADD amplification/expression and heterozygous CASP8 mutation showed a weaker increase in sub-G₀ DNA and no accumulation in G₂–M following birinapant or combination treatment. Birinapant plus TNFα induced increased sub-G₀ and G₂–M DNA in UM-SCC-46 cells in independent experiments (Supplementary Fig. S2A and S2C). UM-SCC-11B expressing both increased FADD and BIRC2 displayed increased sub-G₀ DNA but no G₂–M DNA accumulation with the combination (Supplementary Fig. S2D).

To further elucidate the mechanism(s) leading to cell death, we analyzed the effects of pan-caspase (ZVAD), caspase-8 (ZIETD), and RIPK1 (necrostatin) inhibitors on cell density in cell lines with varying cell death pathway alterations (Fig. 5A). In UM-SCC-46, RIPK1 inhibition (necrostatin), but not caspase inhibition (ZVAD or ZIETD), reversed the effects of birinapant treatment alone or with death agonists. These data support predominantly RIPK1-mediated, caspase-independent, necroptotic cell death, consistent with the expression of high FADD, normal BIRC2, and low CASP8 expression observed in UM-SCC-46. Conversely, in UM-SCC-1, the weaker inhibitory effect of birinapant plus death receptor agonists was blocked by both caspase inhibitors ZVAD or ZIETD, but not by necrostatin, supporting predominantly caspase-dependent activity. This cell line exhibits weaker amplification and expression of FADD and reduced protein expression of BIRC2 and heterozygous wt and mtCASP8. In UM-SCC-74B with normal FADD and BIRC2, but increased expression of a CASP8 sequence variant, pan-caspase inhibitor ZVAD, but not CASP8 inhibitor ZIETD or necrostatin, restored cell density in birinapant plus TNFα or TRAIL-treated cells. UM-SCC-11B, which strongly expresses FADD, BIRC2, and CASP8, demonstrated effects by both caspase inhibitors and necrostatin, supporting mixed mechanisms of cell death in this cell line. In UM-SCC-11A, a cell line with high levels of FADD and BIRC2 derived from a biopsy prior to chemotherapy in the same patient, we observed greater sensitivity with a pattern of predominantly caspase-dependent apoptotic cell death. To directly confirm the activation of caspases in UM-SCC-11A, caspase-3 and -8 enzyme assays were used to assess activity with birinapant, TRAIL, or the combination. Birinapant plus TRAIL induced more than a 2-fold increase in both caspase-3 and caspase-8 activity in UM-SCC-11A, consistent with the effects observed using the caspase inhibitors.

The expression of key cell death pathway proteins under conditions where treatment with birinapant plus TNFα induced cell death were further examined at different time points in 3 of the lines displaying different patterns of cell death (Fig. 5B). Notably, expression of cIAP1, the most sensitive IAP targeted by birinapant, was similarly decreased in all 3 cell lines at 12 hours and remained undetectable up to 48 hours after treatment with birinapant plus TNFα. In the 3 lines, both RIPK1 and RIPK3 showed decreased expression at later time points following drug treatment. UM-SCC-46 demonstrated a greater reduction in FADD and enhancement of necroptosis mediator MLKL expression from 12 to 48 hours after treatment with birinapant (Fig. 5B), consistent with the predominant rescue by necrostatin observed. Interestingly, all 3 lines showed detectable cleavage of caspase-3, although both UM-SCC-1 and -11B showed a relatively greater and sustained increase in CASP3 and CASP8 cleavage, consistent with predominant reversal by both pan- and caspase-8 inhibitors. Although birinapant has been reported to have a lower activity for cIAP2 and XIAP than cIAP1, we performed additional experiments to examine the effects of birinapant on these IAPs, without and with cIAP1 siRNA knockdown in UM-SCC-46 in vitro. Birinapant with control siRNA effectively inhibited cIAP1, partially inhibited XIAP, and weakly enhanced cIAP2 protein expression, whereas BIRC2 siRNA effectively knocked down IAP1 but induced a compensatory increase in cIAP2 that was partially suppressed by combination with birinapant (Supplementary Fig. S3A). Consistent with this evidence that birinapant also partially inhibits compensatory induction of IAP2 and XIAP, the drug had significantly greater inhibitory effect than that observed with BIRC2/IAP1 knockdown alone, which was weakly attenuated by cIAP1 knockdown alone (Supplementary Fig. S3B). Together, these data provide evidence that birinapant inhibition of IAPs and death agonists promote cell death via apoptosis and/or necroptosis in HNSCC cell lines with varying alterations affecting cell death components.

To evaluate the activity of birinapant in vivo, studies were done using UM-SCC-46 (increased FADD expression, normal BIRC2 expression) and UM-SCC-11B (increased FADD and BIRC2 expression) cells in murine xenograft models. SCID mice were implanted with tumor cells, and treated with birinapant, 15 or 30 mg/kg intraperitoneally as reported (20), or control vehicle, when tumors reached 200 mm³. Birinapant alone versus control had a significant inhibitory effect on tumor growth in both models (Fig. 6). For the UM-SCC-46 xenograft model, a significant difference was observed in the mean tumor volumes of the control versus treated animals through day 45 post-tumor cell injection (Fig. 6A). Improvement in median survival was observed, from 39 days for control to 58 days in the treated group (Fig. 6B). For UM-SCC-11B, prolonged tumor reduction was observed during and after treatment using either 30 or 15 mg/kg/d birinapant (Fig. 6C). Prolonged host survival was observed in both treatment arms (Fig. 5D), with a median of 82 days for control, 137 days for the 30 mg/kg group (P = 0.08), and >180 days for the 15 mg/kg group (P = 0.05). The 15 and 30 mg/kg doses of birinapant were well tolerated, and all mice maintained normal body conditioning scores, and weight on treatment (Supplementary Fig. S4A and S4B).

To examine the molecular and cellular effects associated with birinapant treatment in vivo, the effects on the IAP cell death pathway components demonstrated above were examined by immunostaining of UM-SCC-46 tumor xenografts during treatment (day 25; Fig. 6E). In addition, markers of cell death (MLKL, TUNEL) and proliferation (Ki67) were also examined. Tumor immunostaining was quantified by histoscores as described in Materials and Methods. Birinapant treatment significantly reduced cIAP1, cIAP2, and XIAP during sustained treatment, supporting contribution of inhibition of these IAPs to the activity in vivo. Necroptosis marker MLKL was significantly increased in tumors during the treatment period. There were no significant differences in proliferation (Ki67) or apoptosis (TUNEL) marker staining in tumor specimens at either time point, consistent with the predominantly MLKL/necroptosis-dependent inhibitory effects observed in this model.

Radiation has previously been shown to induce TNFα (23). We examined the combinatorial effects of birinapant plus radiation in vivo using the UM-SCC-46 xenograft model, which displayed sensitivity but were not cured by birinapant treatment alone. Tumor cells were injected subcutaneously into the upper leg of athymic nu/nu mice and treated with birinapant (15 mg/kg intraperitoneally) every 3 days, radiation 2 Gy daily Monday to Friday for 2 weeks, or the combination when the tumors reached 200 mm³. Treatment with birinapant or radiation alone significantly delayed tumor growth (Fig. 7A) and improved survival (Fig. 7B). Remarkably, birinapant combined with radiation cured the animals bearing UM-SCC-46–derived xenografts, with no evidence of disease relapse to 130 days (Fig. 7A and B). Representative differences in gross tumor volume among treatment arms were observed by day 21, without visual evidence of significant radiation dermatitis, contracture (Fig. 7C) or weight loss (Supplementary Fig. S3C). As quantified by RT-PCR, TNFα expression in tumors was enhanced by the combination of birinapant plus radiation versus either alone (Fig. 7D). To examine whether radiosensitization by birinapant observed is due to intrinsic or TNFα-enhanced radiosensitivity of UM-SCC-46, cells were treated with 30 nmol/L birinapant immediately after radiation, and replated after 24 hours for clonogenic assay. Interestingly, in vitro combination of birinapant alone with radiation over a range of 0 to 9 Gy had no significant direct radiosensitizing effect over that due to radiation alone (Supplementary Fig. S5A), suggesting that the activity observed in vivo is dependent on additional factor(s). As we hypothesized and observed that birinapant plus radiation potently induces endogenous TNFα in tumors in vivo (Fig. 7D), we examined and confirmed that combination of TNFα (2 ng/mL) with birinapant significantly enhances radiation sensitivity at 3 Gy compared to the single agents (Supplementary Fig. S5B). Combination birinapant and radiation or radiation alone also significantly suppressed tumor growth in FADD- and BIRC2-overexpressing UM-SCC-11B xenografts, but not UM-SCC-1 xenografts displaying weaker FADD/BIRC2 copy gain and expression, and a heterozygous CASP8 mutation (Supplementary Fig. S6). Together, these findings support birinapant plus radiation as a therapeutic combination for HNSCC harboring genomic amplification or increased expression of FADD without or with BIRC2.

TCGA linked amplification of 11q13 containing FADD with a major subset of HPV⁻HNSCC (2). 11q22 harboring BIRC2/3 encoding IAP1/2 was often coamplified. CASP8 was mutated in a mutually exclusive subset. These molecules comprise core components of the TNF receptor family signal pathways that regulate the balance between cell death via apoptosis and necroptosis or cell survival (Fig. 7E). SMAC mimetics promote degradation of IAP proteins that inhibit the intrinsic and extrinsic cell death signaling pathways, potentially overcoming resistance to death ligands and radiation (16, 32). Here, we present evidence that HNSCC with genomic alterations in these cell death pathway components are sensitive to combinations of a SMAC mimetic plus death ligands TNFα, TRAIL, or radiation.

Our TCGA analysis indicates that FADD amplification is more frequent in HNSCC than other cancers and is associated with early recurrence and worse prognosis. We observed increase in gene copy number and expression of FADD with or without BIRC2, whereas BIRC3 was minimally expressed, among a subset of human HPV⁻ HNSCC cell lines, consistent with HNSCC tumors (2). A heterozygous mutation of CASP8 was detected in only 1 of 11 cell lines, approximating the frequency in tumors. Knockdown of FADD or BIRC2, but not BIRC3, inhibited and enhanced TNFα sensitivity of UM-SCC-11A cells harboring amplification and increased expression of FADD and BIRC2, which have been implicated in promoting cell proliferation, and resistance to TNFα-induced cell death (11–14).

Multiple HNSCC lines overexpressing one or more of these cell death pathway components exhibited pharmacologic inhibition by the SMAC mimetic birinapant in combination with TNFα or TRAIL. This inhibition accompanied a corresponding increase in cell death, supported by increased sub-G₀ DNA fragmentation, and reversibility by caspase and/or RIPK1/3 inhibitor(s). Birinapant plus TNFα promoted degradation of cIAP1, the product of BIRC2, and variably activated caspase and/or MLKL mediators of apoptosis or necroptosis in different HNSCC. These findings implicate IAPs as critical mediators of the widespread resistance of HNSCC to TNFα previously observed by our laboratory (33). Birinapant activity was also associated with reproducible effects on G₂–M cell-cycle progression in UM-SCC-46 cells, similar to that reported with natural IAP inhibitor embelin in PC3 prostate cancer cells (32), and recently linked to the role of FADD in proliferation in lung carcinoma (34). In the latter study, the induction of mitosis and lung carcinogenesis depended upon activation of KRAS and CK1α phosphorylation of FADD during G₂–M, where FADD interacted with G₂–M cell-cycle regulatory components PLK1, AURKA, and BUB1. Interestingly, UM-SCC-46 showed a greater reduction in FADD by birinapant and TNFα during 48 hours posttreatment when G₂–M arrest was observed, compared with other lines lacking this effect. However, after sustained monotherapy in vivo, resumption of tumor growth was observed, and we detected no difference in Ki67 staining. Whether there is sustained G₂–M block, or compensatory mechanism(s) of proliferation with prolonged treatment in vivo, merits further investigation. Of potentially broader clinical relevance, birinapant combined with cell death ligands was capable of inducing potent cytotoxicity in HNSCC lines that were derived from aggressive primary, metastatic, and recurrent tumors, that are often mutant or deficient for pro-death tumor suppressor TP53 (2).

We observed differences in sensitivity to death ligands and predominant cell death pathways and mechanisms upon comparison of HNSCC harboring different alterations in FADD, BIRC2, and CASP8. Evidence for caspase-dependent apoptosis, RIPK/MLKL-dependent necroptosis, or mixed mechanisms of cell death were observed by comparing the effects of reversible inhibitors and established markers of caspase and/or RIPK-dependent pathways. Birinapant and TNFα-mediated cytotoxicity of UM-SCC-11B with FADD/BIRC2 amplification and normal CASP8 expression was reversed by pan-caspase and CASP8 inhibitors and showed increased caspase-3 or -8 cleavage, supporting predominantly caspase-dependent mechanisms of apoptosis. In contrast, inhibition of UM-SCC-46 was predominantly reversed by the RIPK1 inhibitor necrostatin and associated with induction of MLKL in vitro and in vivo, supporting necroptosome-dependent inhibition in this HNSCC overexpressing FADD but low caspase-8 protein. Since caspase-8 has been shown to inhibit necrosis while promoting apoptosis (35), low CASP8 expression could explain the predominantly necroptotic cell death seen in UM-SCC-46. Interestingly, UM-SCC-74B, expressing a common K14R amino acid variant of CASP8, and lower FADD and BIRC2, demonstrated sensitivity to birinapant and death ligands reversible by pan-caspase but not caspase-8 inhibitors, consistent with caspase-8–independent cell death mechanisms. Further experiments provided evidence that birinapant can also partially inhibit XIAP expression, which could potentially enhance other caspase-dependent mechanisms. Compared with these lines, UM-SCC-1 with lower amplification and expression of FADD/BIRC2 and a heterozygous CASP8 mutation was relatively less sensitive in vitro and in vivo. Inhibition of UM-SCC-1 and 11B was reversed by both caspase and necroptosis inhibitors and demonstrated cleaved caspase and basal MLKL expression, supporting mixed mechanisms. Raulf and colleagues found evidence for mixed apoptotic and necroptotic cell death in response to a SMAC mimetic in HNSCC cell line HSC3, and not others, but the genomic alterations were undefined (36). Together, these observations suggest that the status of FADD, BIRC2, CASP8 and other components of the cell death machinery may affect sensitivity to these agents and merit further investigation as potential predictive biomarkers for response in future clinical studies of SMAC mimetics in HNSCC.

Our in vivo preclinical studies demonstrate that two independent models with genomic alterations in FADD−/+BIRC2 amplification exhibited significant sensitivity and prolonged survival following birinapant treatment alone, despite differences in intrinsic sensitivity in vitro. Birinapant alone induced endogenous TNFα mRNA in vivo, but the incomplete suppression of tumor growth with birinapant monotherapy suggested that this level alone may be insufficient to potentiate tumor regression. Birinapant plus radiation led to more durable responses in UM-SCC-46 and 11B xenograft models. Furthermore, the combination of birinapant plus radiation more potently enhanced the expression of endogenous murine TNFα compared with either therapy alone. While birinapant alone did not significantly enhance the effects of radiation in vitro, addition of TNFα enhanced radiosensitization in clonogenic assay. Because TNFα is induced in tumors following radiation treatment (23), studying the addition of birinapant plus radiation holds considerable clinical relevance to this standard treatment for HNSCC.

Recently, combinatorial activity of different SMAC mimetics with radiation and induction of TNFα has been demonstrated in lung cancer and HNSCC cell lines (37, 38). Partial response in UM-SCC-1 xenografts in the latter study was observed, similar to our results. The relative contribution of weaker amplification of FADD/BIRC2 and/or heterozygous mutation of CASP8 to the resistance to birinapant plus TNFα or radiation in this model may merit future investigation. Together, these observations underscore the importance of assessing genomic alterations, expression, and the molecular and antitumor effects of targeting cell death pathways with SMAC mimetics in preclinical studies and future clinical trials in HNSCC and other cancers. FADD/BIRC2 alterations may represent an Achilles' heel for this subset of HNSCC and for clinical investigation with SMAC mimetics.

No potential conflicts of interest were disclosed.

Conception and design: D.F. Eytan, G.E. Snow, A. Saleh, J.B. Mitchell, C.M. Annunziata, Z. Chen, C. Van Waes

Development of methodology: D.F. Eytan, S. Carlson, A. Saleh, L. Hernandez, J.B. Mitchell

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): D.F. Eytan, G.E. Snow, S. Carlson, A. Derakhshan, A. Saleh, S. Schiltz, S. Mohan, S. Cornelius, A.L. Sowers, Z. Chen

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): D.F. Eytan, G.E. Snow, S. Carlson, A. Derakhshan, A. Saleh, S. Schiltz, H. Cheng, Z. Chen, C. Van Waes

Writing, review, and/or revision of the manuscript: D.F. Eytan, G.E. Snow, S. Carlson, A. Saleh, S. Mohan, J.B. Mitchell, C.M. Annunziata, Z. Chen, C. Van Waes

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): D.F. Eytan, S. Carlson, J.F. Coupar, Z. Chen, C. Van Waes

Study supervision: A. Saleh, Z. Chen, C. Van Waes

Birinapant was provided to NIDCD for research under a Materials Transfer Agreement with the National Cancer Institute and TetraLogic Pharmaceuticals. This study utilized the high-performance computational capabilities of the Biowulf Linux cluster at the NIH, Bethesda, MD (http://biowulf.nih.gov). Thanks to Barbara Conley and James F. Battey for reading and helpful comments on this article.

The study was supported by NIDCD intramural projects ZIA-DC-000016, 73, 74 (C. Van Waes, Z. Chen, S. Carlson, A. Saleh, S. Schiltz, H. Cheng, S. Cornelius, J.F. Coupar) and the NIH-Medical Research Scholars Program (D.F. Eytan, G.E. Snow, S. Mohan, A. Derakhshan).

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.