Cells and reagents

Well characterized HNSCC cells were used throughout this study [28]. HNSCC lines UMSCC1 (a gift from Dr. Tom Carey, University of Michigan, Ann Arbor, MI) , HN5 (a gift from Dr. Jeff Myers, The University of Texas MD Anderson Cancer Center, Houston, TX), OSC19 (a gift from Theresa Whiteside, University of Pittsburgh, Pittsburgh, PA), Cal33 (cat# ACC 447 DSMZ; Braunschweig, Germany), and FaDu (cat# HTB-43, ATCC; Gathersburg, MD, USA) were maintained in high glucose Dulbecco’s Modified Eagle Medium (DMEM; Corning, NY, USA) containing 10% heat-inactivated fetal-bovine serum (Neuromics, MN, USA). Non-cancerous, immortalized esophageal Het-1A (cat# CRL-2692, ATCC; Gathersburg, MD, USA) line was maintained in bronchial epithelial cell growth basal medium supplemented with bronchial epithelial cell growth medium (BEGM) kit (cat# CC-4175, Lonza, Basel, Switzerland). Cisplatin (cat# 4015472, Fresenius Kabi, Lake Zurich, IL, USA), CC-90001 (cat# HY-138304, Med Chem Express, Monmouth Junction, NJ, USA), SP600125 (cat# 420119, EMD Millipore, Burlington, MA, USA).

Chemistry

In general, melting points (m.p., uncorrected) were determined using Electrothermal 9100 (cat# 03010-51, model: Electrothermal IA9100, TEquipment, Long Branch, NJ, USA). Infrared (IR) spectra was determined using Perkin-Elmer Spectrum One FT-IR spectrophotometer (cat# FTIR-01-0240, model: Spotlight 300 FT-IR MCT, Shelton, CT, USA) with attenuated total reflectance (ATR) sampling unit. For the nuclear magnetic resonance (NMR) spectra we used Bruker Avance 300/500 spectrometer (cat# DIS-46950, model: Avance 500, Billerica, MA, USA). Chemical shifts (δ) in parts per million (ppm) downfield from tetramethylsilane (internal standard) and mass spectra we calculated using Thermo Finnigan MAT 8500 (EI; cat# 8500, model: 2000FRM, Waltham, MA, USA).

Cytotoxicity assay

Cells (4 × 10³ cells/well, 96-well plates; cat# 229196, Cell Treat, Pepperel, MA, USA) were treated in quadruplicate with A-DiFiD or indicated compounds at various concentrations, and cell proliferation assessed using CyQUANT assay kit (Life Technologies, Waltham, MA, USA) per manufacturer’s instructions at 72 h. Half-maximal drug inhibitory concentration (IC₅₀) was calculated using non-linear curve fit with GraphPad Prism software version 9.0 (GraphPad Software Inc., San Diego, CA, USA).

Colony formation assay

A total of 300 HNSCC cells pretreated with predetermined doses of A-DiFiD (IC₅₀) for 30 min then seeded in a 6-well plate (cat# 229105, Pepperel, MA, USA). Drug-free cell culture medium was replaced every 2–3 day. When colony visibility appeared in one of the 6-well plates (10–14 days), the culture was terminated by fixing with 10% formalin (cat# HT501128-4L, Sigma Aldrich, St. Louis, USA) for 30 min. Plates were stained with crystal violet (cat# C6158 Sigma Aldrich, St. Louis, USA) for 10 min at room temperature then washed with water 3 times. The number of cell colonies was counted under the microscope (< 75 clones validated). Formed colonies are determined by visualizing at least 50 cell per cluster.

Migration and invasion assay

Cells (2 × 10⁴ cells per insert) were transferred to trans-well inserts with 8 μm pores (Becton Dickinson, Franklin Lakes, NJ, USA). For invasion assay, a layer of diluted Matrigel (2 mg/mL at 100 µL volume) in DMEM (BD Biosciences, San Jose, CA, USA) was placed in the insert. Cells in serum-free media were seeded onto Matrigel layer for invasion or directly onto insert for migration assay. The inserts were placed in duplicate holding-wells containing A-DiFiD at half IC₅₀ in complete media for 24 h. Treated and untreated cells were plated in parallel to assess viability using CyQuant assay kit. The number of cells that moved to other side of membrane were counted following fixation and staining using Hema3 kit (Fisher Scientific, Hampton, NH, USA). The numbers of migrating or invading cells were normalized to cell viability.

Cell adhesion and spread assay

A total of 6 × 10⁵ HNSCC cells were pretreated with A-DiFiD (IC₅₀) for 30 min then plated on 60 mm dishes coated with collagen (cat# A1048301, Thermo, Waltham, MA, USA), fibronectin (cat# 33016015, Thermo), vitronectin (cat# A14700, Thermo, Waltham, MA, USA), or laminin (cat# L4544, Sigma Aldrich, St. Louis, USA). Cells were allowed to adhere for 15, 30, 60, or 120 min before the plates were washed with 10 mL phosphate-buffered saline (PBS), then stained with crystal violet (cat# C6158 Sigma Aldrich, St. Louis, USA). Images were taken under the microscope and analyzed for number of adherent cells per plate as well as cell area (cell spread). Spread and adhesion were quantified using ImageJ (version 13.0.6 National Institutes of Health, Bethesda, MD, USA).

Human phospho-kinase array

Proteome Profiler Human Phospho-Kinase Array Kit (ARY003C, R&D Systems, Minneapolis, MN, USA) was used to identify signaling targets of A-DiFiD per the manufacturer’s instructions. Membranes were image by autoradiography, and ImageJ was used to quantify the signal intensity.

Immunoblotting

Cells (3 × 10⁵ cells/60 mm dish) were treated with A-DiFiD (IC₅₀) and lysed in radioimmunoprecipitation assay (RIPA) buffer containing protease/phosphatase inhibitors (complete, Mini, Roche, Indianapolis, IN, USA). For inhibition of phospho-c-Jun (p-c-Jun) stimulation, cells were serum starved for 48 h, and then treated with A-DiFiD (IC₅₀) for 0, 3, 5, 10, 20, 40, or 80 min. Cell lysates were harvested on ice and proteins separated by sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE), transferred onto nitrocellulose membranes, and probed with p-c-Jun (R&D Systems, Minneapolis, MN, USA) or pan-actin (Cell Signaling, Danvers, MA, USA). Protein bands were detected using anti-mouse or anti-rabbit immunoglobulin G (IgG) conjugated to Dylight-680 or Dylight-800 (Fisher, Waltham, MA, USA) and quantified using Li-Cor Odyssey protein imaging system (Li-Cor Biotechnology, Lincoln, NE, USA).

Analysis of The Cancer Genome Atlas

Using the cBioPortal [29, 30] interface, DNA amplification data was generated by probing two The Cancer Genome Atlas (TCGA) datasets: HNSCC Firehose Legacy (512 samples total, 497 samples primary tumor only), and recurrent and metastatic (151 samples total, 56 HNSCC samples selected) datasets which include 151 samples from patients with advanced, treatment-resistant head and neck tumors that are characterized as recurrent and/or metastatic [31]. JUN DNA copy number amplification, deletion, or mutation were assessed, and percent amplification is expressed as percent of whole. JUN mRNA (RNA Seq V2) was utilized from TCGA Firehose Legacy (512 samples) to generate comparative analysis between JUN levels in either normal or tumor samples. To generate survivorship comparison, TCGA Firehose Legacy databank was assessed by log-rank (Mantel-Cox) text assessing differences between curves. Patients were stratified into two groups and correlation between reads per kilobase [reads per kilobase million (RPKM)] and survival were evaluated. RPKM values were stratified by into high/low cutoffs of 0.5.

Clinical Proteomic Tumor Analysis Consortium analysis

Clinical Proteomic Tumor Analysis Consortium (CPTAC) database was accessed using the interface of University of Alabama at Birmingham Cancer Data Analysis Portal (UALCAN) [32, 33]. To evaluate tumor vs. normal, normal vs. tumor staging, normal vs. tumor patient weight, and normal vs. gender plots were generated using Z-values that represent standard deviations from the median across the tumor. Log2 spectral count ratio values from CPTAC were initially normalized within each sample profile, then normalized across samples.

Statistical analysis

Data are reported as mean ±  standard error of the mean (SEM). Nonparametric Mann-Whitney test utilized to evaluate significance in all experiments unless otherwise stated. For TCGA survivorship comparison, log-rank (Mantel-Cox) test evaluated differences between curves. Expression cut off was generated by stratifying patients into two groups and association between survival and RPKM was assessed. RPKM values in the highest vs. lowest quartiles was used and P-values determined. All statistical calculations were performed on GraphPad Prism software (version 9.0), with significance determined by P < 0.05.

A-DiFiD demonstrates potent HNSCC cytotoxicity

Small molecule inhibitor DiFiD, has antitumor effects in several cancers [3, 34]. To enhance the antitumor efficacy of DiFiD, we developed four analog small molecule compounds. We tested the cytotoxicity of all four compounds using serial dilutions (1:1, 1:10, 1:100, 1:1 × 10³, 1:1 × 10⁴, 1:1 × 10⁵, 1:1 × 10⁶) and DMSO control. A-DiFiD demonstrated the greatest cytotoxicity (HN5 IC₅₀: 245.4 nmol/L) compared to Cl-DiFiD (HN5 IC₅₀: 3.393 µmol/L), MeO-DiFiD (HN5 IC₅₀: 4.690 µmol/L), and DF-DiFiD (HN5 IC₅₀: 7.480 µmol/L) (Figure 1A and B). A-DiFiD demonstrated nanomolar toxicity in FaDu, Cal33, OSC19, and UMSCC1 HNSCC cells lines (Figure 1C and D).

Display full size

Figure 1. 

A-DiFiD demonstrates cytotoxic activity against HNSCC. (A) Chemical structure of four DiFiD analogs Cl-DiFiD, A-DiFiD, MeO-DiFiD, and DF-DiFiD; (B) HN5 (4 × 10³ cells/well in triplicate) were treated with various concentrations of Cl-DiFiD, A-DiFiD, MeO-DiFiD, and DF-DiFiD; (C) A-DiFiD IC₅₀ determination against HNSCC cell lines FaDu, Cal33, OSC19, and UMSCC1; (D) A-DiFiD IC₅₀ on HN5 HNSCC cell line expressed by percent survival; (E) Het-1A treated with various doses of A-DiFiD. IC₅₀ was calculated using non-linear curve fit from three separate experiments on GraphPad Prism software

A-DiFiD (HN5 IC₅₀ nmol/L) has comparable toxicity to CC-90001 (HN5 IC₅₀ 200.4 nmol/L) in HNSCC cells (Figure S1A). Additionally, A-DiFiD had a greater cytotoxicity than SP600125 (HN5 IC₅₀ 523.7 nmol/L) (Figure S1B). These results demonstrate A-DiFiD has comparable if not more potent cytotoxic effects comparted to other c-Jun inhibitors available commercially. A major barrier in the development and clinical application of small molecules is the exposure of non-cancerous cells and tissue to cytotoxic effects. To address this concern, we treated Het-1A, an immortalized esophageal squamous cell line from a cancer free patient, with A-DiFiD and observed a many fold increase in IC₅₀ (8.202 µmol/L) compared to HNSCC lines (Figure 1E). This indicates that A-DiFiD is well tolerated in normal cells.

A-DiFiD abrogates HNSCC proliferation, migration, invasion, and adhesion in vitro

For cancer cells to successfully form a tumor, they must first acquire the ability for a single cancer cell to adhere to a substrate, replicate, and form a colony. To test the ability of a single HNSCC cell to replicate when challenged with A-DiFiD, we performed colony formation assay. HNSCC cells were collected by trypsinization and washed several times with PBS and then held in a suspension of DMEM with various concentrations of A-DiFiD for 30 min. Cells were then washed again and only 300 cells were plated per well in a 6-well dish containing drug-free culture medium. HNSCC cell lines pretreated A-DiFiD demonstrated significantly reduced ability to proliferate (Figure 2A and B).

Display full size

Figure 2. 

A-DiFiD (IC₅₀, 4h) treatment abrogates HNSCC colony formation, migration, and invasion. (A) Representative image of colony formation of UMSCC1 and FaDu; (B) quantification of colony formation. Data represents cumulative results from two independent experiments with mean ± SEM; (C) invasion; and (D) migration are normalized for differences in proliferation rates over the duration of the experiment. Data represents cumulative results from three independent experiments with mean ± SEM. ^** P < 0.01, ^*** P < 0.001, ^**** P < 0.0001; n.s.: not significant; VC: vehicle control

No treatment options for the metastatic spread of HNSCC are available. Further, patients with metastatic spread have an abysmal survival rate of less than 20% [35]. For cancer to metastasize, a local tumor must first break down the extracellular matrix (ECM) that confines it to a local space. To test this, we performed a trans-well invasion assay wherein HNSCC cells must move through a layer of Matrigel before they can be stained and quantified. HNSCC cells treated with IC₅₀ A-DiFiD demonstrated a significant reduction in invasion (Figure 2C). Another important feature of successful tumor spread is the ability for cells to migrate. To test this, we utilized the same transwell assay as before, but no Matrigel was added. We observed that HNSCC cell migration was significantly reduced when treated with A-DiFiD (Figure 2D).

Cell adhesion is important for squamous cell carcinoma to participate in cell-ECM interactions, and plays a crucial role in the progression, recurrence, invasion, and distant metastasis [36]. To test the ability of HNSCC challenged with A-DiFiD treatment to adhere to ECM, we employed a cell adhesion assay. Cells were plated on collagen, fibronectin, vitronectin, or laminin and allowed to adhere for various time points before being washed, stained, and quantified. We found that A-DiFiD reduced the HNSCC cellular adhesion on collagen, fibronectin, vitronectin, and laminin ECM in a time dependent manner up to 120 min (Figure 3A). Cell adhesion is typically followed by activation of intracellular signaling pathways that regulated cell spreading over the ECM. We also measured the surface area occupied by individual cells to determine if A-DiFiD impacts cell spreading. We found that A-DiFiD treatment reduced the surface area occupied by cells on all ECM substrates we tested (Figure 3B).

Display full size

Figure 3. 

HNSCC adherence and spread is reduced by A-DiFiD (IC₅₀, 30 min) pretreatment. Collagen, fibronectin, vitronectin, and laminin. (A) Adherent cells per field; and (B) cell pixel area were counted over various time points. Data represent cumulative results from three independent experiments with mean ± SEM. ^* P < 0.05, ^** P < 0.01, ^*** P < 0.001, ^**** P < 0.0001

A-DiFiD targets c-Jun

HNSCC cells exploit a multitude of molecular pathways that regulate growth and metastasis. To assess the mechanism of action of A-DiFiD, we used a phospho-protein array to identify the molecules perturbed by A-DiFiD treatment. Phosphorylation of c-Jun was the most significantly reduced in HNSCC treated with A-DiFiD treatment (Figure 4A and B). Further, we validated a reduction in p-c-Jun levels as early as at 3 min post-treatment (Figure 4C and D, Figure S2A and B). These data demonstrate that A-DiFiD targets c-Jun, an important molecular mediator in the progression of HNSCC.

Display full size

Figure 4. 

c-Jun is the molecular target of A-DiFiD. (A) HNSCC cells (FaDu; 2 × 10⁵ cells) were treated with vehicle control or IC₅₀ concentrations of A-DiFiD for 24 h. Representative dot-blot images from phospho-kinase array; (B) densitometric analysis of array signal from A-DiFiD treated lysate normalized to those treated with vehicle control and represented as fold change in protein levels; (C) HNSCC cells (FaDu; 2 × 10⁵ cells) were treated with IC₅₀ concentrations of A-DiFiD and protein lysate collected over various time points. Image is representative of three independent experimental repeats; (D) densitometric analysis of signals from immunoblot normalized to loading control. Error bars represent ± SEM. ^* P < 0.05, ^** P < 0.01

c-Jun drives HNSCC progression

Patients with recurrent or metastatic HNSCC have significantly worse prognosis. Compared to primary HNSCC, metastatic and recurrent HNSCC patient samples had a significant increase in JUN DNA copy number amplification (Figure 5A). Interestingly, mRNA levels of JUN were not significantly different between normal and tumor samples (Figure 5B).

Display full size

Figure 5. 

HNSCC patient tumor samples have increased levels of c-Jun. (A) Alteration frequency of two independent datasets, TCGA Firehose Legacy and MSK recurrent and metastatic head and neck cancer. DNA copy number alteration and amplification shown as percent of whole data set copy number alteration frequency; (B) JUN mRNA levels from TCGA (Firehose Legacy) as shown in transcripts per million; (C) c-Jun protein levels of normal vs. tumor from CPTAC head and neck cancer dataset shown using Z-value score; (D) c-Jun protein levels comparing normal to HNSCC stages 1–4; (E) normal vs. tumor at various patient weight, and (F) normal vs. tumor patient gender. ^** P < 0.01, ^*** P < 0.001, ^**** P < 0.0001

CPTAC provides proteomic data across a variety of cancer types. CPTAC dataset for HNSCC contains 108 tumor samples and 71 normal samples. Using UALCAN to access HNSCC CPTAC dataset, we found that protein levels were significantly elevated in patient tumor samples compared to normal samples (Figure 5C). Additionally, all tumor stages demonstrated significantly elevated protein compared to normal (Figure 5D). We also noted that all patients regardless of weight or gender had elevated c-Jun protein levels compared to normal tissue (Figure 5E and F).

The treatment outcomes and future direction of HNSCC are significantly impacted by the presence of oncogenic human papilloma virus (HPV) infection. Our analysis revealed no significant variations in the levels of JUN mRNA between patients who tested positive or negative for the p16 protein (Figure S3). These findings suggest that c-Jun expression is elevated in tumors compared to normal tissue, irrespective of clinical parameters or the underlying cause of the tumor.

A-DiFiD in combination with cisplatin potentiates HNSCC cytotoxicity

Cisplatin, a chemotherapy medication that utilizes platinum, is commonly used as the standard treatment for HNSCC [37]. Nevertheless, despite advancements in therapeutic approaches, overcoming tumor resistance continues to be a major obstacle in effectively treating HNSCC. In our study, we evaluated the combined impact of A-DiFiD (IC₅₀) and cisplatin (4.0 µmol/L) on HNSCC cell lines. The combination of A-DiFiD and cisplatin exhibited a pronounced decrease in HNSCC cell viability compared to individual treatments with cisplatin or A-DiFiD alone in Cal33 (P < 0.034) and FaDu (P < 0.004) cell lines (Figure 6A and B). However, there was only modest combinatorial effect in UMSCC1 cells (P = 0.4669) (Figure 6C).

Display full size

Figure 6. 

A-DiFiD in combination with cisplatin potentiates HNSCC cytotoxicity. Survival analysis of cisplatin (4 µmol/L), A-DiFiD (IC₅₀), or combination treatment on (A) Cal33; (B) FaDu; (C) UMSCC1. Error bars represent ± SEM

c-Jun expression decreases HNSCC survival

Finally, we compared patients who had high (the top 50% of the median) JUN mRNA expression to low JUN mRNA expression. High JUN expressing patients had a significantly decreased 3-year survival compared to the low JUN expressing patients (Figure 7). This implicates c-Jun as an important clinical correlate to poor outcomes in HNSCC patients.

Display full size

Figure 7. 

High JUN expression is associated with poor patient survival. Kaplan-Meier survival curve generated from TCGA HNSCC mRNA demonstrate that high JUN levels expressing patients have decreased survival compared to low JUN expressing patients (high, n = 141; low, n = 138; P = 0.0387)