BET Inhibition Overcomes Receptor Tyrosine Kinase-Mediated Cetuximab Resistance in HNSCC

Abstract

Cetuximab, the FDA-approved anti-EGFR antibody for head and neck squamous cell carcinoma (HNSCC), has displayed limited efficacy due to the emergence of intrinsic and acquired resistance. We and others have demonstrated that cetuximab resistance in HNSCC is driven by alternative receptor tyrosine kinases (RTK), including HER3, MET, and AXL. In an effort to overcome cetuximab resistance and circumvent toxicities associated with the administration of multiple RTK inhibitors, we sought to identify a common molecular target that regulates expression of multiple RTK. Bromodomain-containing protein-4 (BRD4) has been shown to regulate the transcription of various RTK in the context of resistance to PI3K and HER2 inhibition in breast cancer models. We hypothesized that, in HNSCC, targeting BRD4 could overcome cetuximab resistance by depleting alternative RTK expression. We generated independent models of cetuximab resistance in HNSCC cell lines and interrogated their RTK and BRD4 expression profiles. Cetuximab-resistant clones displayed increased expression and activation of several RTK, such as MET and AXL, as well as an increased percentage of BRD4-expressing cells. Both genetic and pharmacologic inhibition of BRD4 abrogated cell viability in models of acquired and intrinsic cetuximab resistance and was associated with a robust decrease in alternative RTK expression by cetuximab. Combined treatment with cetuximab and bromodomain inhibitor JQ1 significantly delayed acquired resistance and RTK upregulation in patient-derived xenograft models of HNSCC. These findings indicate that the combination of cetuximab and bromodomain inhibition may be a promising therapeutic strategy for patients with HNSCC.Significance: Inhibition of bromodomain protein BRD4 represents a potential therapeutic strategy to circumvent the toxicities and financial burden of targeting the multiple receptor tyrosine kinases that drive cetuximab resistance in HNSCC and NSCLC.Graphical Abstract: http://cancerres.aacrjournals.org/content/canres/78/15/4331/F1.large.jpg Cancer Res; 78(15); 4331-43. ©2018 AACR.

Figure 1. Acquired cetuximab resistant HNSCC models display heterogeneous expression of alternative RTKs

(A) PECAPJ49 and (B) FaDu parental (PAR) and cetuximab resistant (CTXR) cell lines were treated with 100nM cetuximab (CTX) for 96 hours. Cell viability was determined by a crystal violet viability assay (n = 4; *p < 0.002). (C) A phospho-RTK antibody array was used to assess phosphorylated levels of various RTKs in PECAPJ49 (PAR, PAR+CTX, CTXR#1, CTXR#4) and (D) FaDu (PAR, PAR+CTX, CTXR#2, CTXR#3) lysates (n = 1). Phosphorylated proteins that were altered in the CTXR clones compared to the parental line are indicated on the right. The densitometric ratio of duplicate spots for activated RTKs to the loading controls on the RTK array was calculated using Image J software. The ratios were compared to the PAR group and represented below. (E) Immunoblot analysis of PECAPJ49 and FaDu parental and CTXR cells for P-AXL, AXL P-MET, and MET expression was performed. Densitometry for each protein compared to actin and normalized to the parental expression is indicated below each band and is representative of three independent immunoblots. (F) AXL mRNA expression was determined in PECAPJ49 and FaDu parental and CTXRs by RT-PCR. Fold change in AXL mRNA in CTXRs were compared to their parental counterparts (n=3; *p = 0.01; **p < 0.001). Figure is representative of three independent experiments. Error bars represent SEM.

Figure 2. Acquired cetuximab-resistant models display enriched BRD4 expression and activity

Intracellular flow cytometry for BRD4 as described in the Methods was assessed in (A) PECAPJ49 and (B) FaDu parental and CTXR cell line models. Rabbit Alexa-Fluor 488 secondary antibody alone (CTL) was used to set the BRD4-positive gate. Side scatter was plotted against Alexa Fluor 488 to determine the percentage of BRD4-positive cells (n=3; * p < 0.02). Data is representative of three independent experiments. (C) PECAPJ49 parental and CTXRs were transfected with non-targeting control (NTC) and two BRD4 siRNA (#1, #2) oligonucleotides ± 100nM cetuximab for 96 h. Cells were stained with crystal violet and relative cell viability was quantified (n = 4; *p < 0.02). Figure is representative of three independent experiments (D) Lysates from PECAPJ49 parental and CTXR#4 cells treated with BRD4 siRNA ± 100nM cetuximab were resolved by SDS-PAGE for AXL, BRD4, and P21. Beta-Actin (ACTIN) was used as loading control. This figure is representative of three independent experiments. (E) Cytoplasmic (CYTO) and nuclear (NUC) extracts of PECAPJ49 PAR and CTXR#4 cells were incubated with biotinylated scrambled (SCR) or AXL promoter region oligonucleotides. Biotinylated immunoprecipitates were resolved by SDS-PAGE and probed for BRD4 and P-CDK9 Thr186 expression (ns = non-specific). Figure is representative of three independent experiments. Error bars represent SEM.

Figure 3. Bromodomain inhibition sensitizes acquired cetuximab-resistant models to cetuximab

(A) PECAPJ49, (B) FaDu and CTXR cells were treated with 100nM cetuximab ± 300nM JQ1 for 72 hours. Cell viability was assessed by crystal violet assay (n=3; *p < 0.05, **p<0.01, ns = not significant). (C) PECAPJ49 Parental and cetuximab resistant clones were treated with 100nM cetuximab±800nM I-BET-762(IBET) for 72 hours. Cells were stained with crystal violet and quantified as described in Methods (n=3; *p < 0.02). Figures are representative of three independent experiments. (D) PECAPJ49 parental and CTXR#3, #4 lines were treated with cetuximab ± JQ1 for 72 hours. Lysates were resolved and assessed for AXL (dark and light exposures), P21 and ACTIN. Densitometric analysis of AXL expression normalized to PAR CTL from three independent immunoblots is indicated below AXL bands. (E) ChIP-qPCR analysis for BRD4 occupancy on the AXL promoter of PECAPJ49 PAR and CTXR#4 cells treated with or without 300nM JQ1 was performed. Fold enrichment was determined by the ΔΔCt Method and then the BRD4 occupancy was compared to IgG occupancy and presented as a fold enrichment score (n=3; *p =0.003). Figure is representative of two independent experiments. (F) CAL33 cells were seeded at 250 cells/well in 6-well dishes and treated with control (CTL), cetuximab (CTX) or cetuximab+JQ1 every 3 days for 21 days. After 21 days, drug treatment was withdrawn for 15 days and cells were stained with crystal violet to assess viability (n=3). Figure is representative of two independent experiments. Error bars represent SEM.

Figure 4. Bromodomain inhibition abrogates RTK expression in intrinsically cetuximab-resistant models

(A) Intrinsically cetuximab-resistant cell lines were treated with 100nM cetuximab and 300nM JQ1 for 72 hours. Cell viability was assessed by crystal violet assay (n = 3; *p < 0.04, **p < 0.01). Figure is representative of three independent experiments. (B) CAL27 cells were treated with control (CTL), 100nM cetuximab, 300nM JQ1 and cetuximab+JQ1 for 72 hours. Lysates were procured and activated RTK expression was assessed using a Phospho-RTK array (n=1) (C) The densitometric ratio of duplicate spots for activated RTKs to the loading controls on the RTK array was calculated using Image J software. The ratios were compared to the CTL group and represented. (D) CAL27 lysates were also resolved by SDS-PAGE for expression of total HER3, MET, AXL and phospho-SFK. Figure is representative of three independent immunoblots. (E) SCC47 and HSC6 cells were treated with 100nM cetuximab ± 300nM JQ1 for 72 hours. Lysates were resolved and assessed for HER3, AXL and ACTIN expression. Figures are representative of two independent immunoblots. (F) Intrinsically cetuximab- resistant HNSCC cell line CAL27 was treated with Control (CTL) siRNA, CTL+cetuximab, cetuximab+BRD4si#1, and cetuximab+BRD4si#2 for 96h. Cell viability was assessed by crystal violet assay (n = 3; *p < 0.01). Figure is representative of three independent experiments .CAL27 cells treated with CTL siRNA or BRD4siRNA ± cetuximab for 96 hours were interrogated for phosphorylated HER3, BRD4 expression and Beta-tubulin which was used as the loading control. Figure is representative of three independent experiments. (G) PECAPJ49 Vector (VEC) and HER3 expressing cells were treated with vehicle, 100nM cetuximab, 300nM JQ1 or cetuximab±JQ1 for 72 hours. Cell viability was assessed by crystal violet assay. Images of representative wells stained with crystal violet are illustrated (n = 3; *p < 0.05, **p < 0.02). Figure is representative of three independent experiments. Immunoblot analysis of Vector and HER3 expressing cells for V5, HER3 and Beta –tubulin was performed and illustrated to the right. Figure is representative of two independent immunoblots. Error bars represent SEM.

Figure 5. JQ1 treatment prevents acquisition of cetuximab resistance in HNSCC patient-derived xenograft models

(A, B) Two HNSCC PDX models (6282 and 7157) were implanted into NOD/SCIDγ mice and separated into 4 treatment groups (n= 6/group): 1) Vehicle, 2) Cetuximab (20mg/kg via i.p), 3) JQ1(50mg/kg via i.p 5qd) and 4) cetuximab +JQ1. Mice were treated with two doses of cetuximab on days 1 and 5. Tumors were measured twice weekly and plotted (p < 0.01; JQ1 vs CTX+JQ1, p < 0.001; Vehicle vs CTX, p = 0.04 (PDX6282), p = 0.003 (PDX7157); CTX vs CTX+JQ1). Arrows indicate point at which tumors began to re-grow. (C) Phospho-RTK arrays using lysates harvested from PDXs 6282 and (D) 7157 at the last time point were performed. Phosphorylated RTKs that were augmented in the PDXs that acquired cetuximab resistance and abrogated with combination of JQ1 are indicated (n = 1). The densitometric ratio of duplicate spots for activated RTKs and the loading controls on the RTK arrays was calculated using Image J software. The ratios were compared to the Vehicle treated group and represented. (E) Lysates from four PDX6282 tumors in each treatment group were resolved by SDS-PAGE for HER2, HER3, ROR2, P-SFK, P-MAPK, total MAPK and GAPDH (Loading control). This figure is representative of two independent immunoblots. (F) AXL, ROR2 and (G) BRD4 gene expression data from WG-DASL (Whole-genome cDNA-mediated Annealing, Selection, extension, and Ligation) microarray data of 40 HNSCC patients treated with first line platinum and cetuximab therapy was analyzed (NCBI GEO GSE65021,(32)). The moderately and poorly differentiated tumors obtained from the oral cavity and larynx was analyzed (n=25). Y-Axis represents log2 transformed values of the processed gene expression data (AXL; p = 0.0166; ROR2: p = 0.0369; BRD4: p=0.2369). (H) BRD4 expression from The Cancer Genome Atlas (TCGA) HNSCC tumor and normal tissues was assessed using the GEPIA web server (33). The custom p-value cutoff was 0.01. Y-axis represents log-transformed relative gene expression.