Targeting VEGFR and FGFR in head and neck squamous cell carcinoma in vitro

Head and neck squamous cell carcinoma (HNSCC) is a heterogeneous disease characterized by a tumor microenvironment (TME) that overexpresses vascular endothelial growth factor receptor (VEGFR) and fibroblast growth factor receptor (FGFR), which can lead to neovascularization, tumor growth and metastasis. Therapeutic strategies inhibiting these signaling pathways might lead to innovative HNSCC treatments. Five HNSCC cell lines were characterized based on VEGFR1-3 and FGFR1-4 expression by sqRT-PCR and treated with three different tyrosine kinase inhibitors (TKIs) (nintedanib, dovitinib and pazopanib), all of which are effective against VEGFR and FGFR family members. Crystal violet assays were performed to analyze the effect of the treatments on cell growth (viability). Additionally, VEGFR1-3 and FGFR1-4 expression data were retrieved from The Cancer Genome Atlas (TCGA), and statistical analyses were performed to investigate the receptor expression level in the different cell lines and the efficacy of the single-agent treatments. A correlation analysis was performed to quantify the degree of relationship between receptor expression and drug efficacy. With the exception of VEGFR2, the targeted receptors were expressed at different levels in all of the cell lines. The cell lines exhibited concentration-dependent responses with cell line-specific differences toward two of the three TKIs (nintedanib and dovitinib). Notably, all of the cell lines were resistant to pazopanib. TKIs have potential as therapeutic agents for HNSCC. Cell line-specific differences were observed in our in vitro experiments. The observed pazopanib resistance could be explained by receptor expression. Further investigation is required to determine TKI efficacy in HNSCC.


Introduction
With approximately 13,800 new cases per year in Germany and 600,000 new cases worldwide according to data from the Robert Koch Institute, head and neck squamous cell carcinoma (HNSCC) is one of the most common tumors (1,2).Despite the new therapeutic and diagnostic procedures, a poor 5-year survival rate of 55-60% has remained nearly unchanged in recent decades (1,3).Infiltration of locoregional tissue and lymph node metastasis occur at a high frequency in 66% of HNSCC patients (4,5).The recently approved immune checkpoint inhibitors pembrolizumab and nivolumab, which have shown antitumor efficacy in non-small-cell lung cancer (6)(7)(8), represent the few innovations in the treatment of recurrent or metastatic HNSCC during the last decades (9).However, tyrosine kinase inhibitors (TKI) are being investigated in clinical trials, some of which have shown favorable survival data (10).
Complex relationships exist between the oncogenic stroma and its surroundings in this heterogeneous disease (11).Cells from the immune system, the tumor vasculature and lymphatic systems and cancer associated fibroblasts (CAF) can encourage tumor growth, invasion and metastasis (12,13).Molecules in this environment, such as vascular endothelial growth factor (VEGF) and fibroblast growth factor (FGF), stimulate the neovascularization needed for cancer growth (13).
The amplification and overexpression of FGFR1-3 have been observed in HNSCC in several studies (14)(15)(16).Fibroblast growth factors (FGF) -A, -B, -C and -D, which play key roles in embryonic development, are crucial for angiogenesis and nerve and cartilage regeneration in adult tissue (17,18).Schultz-Hector and Haghayegh (19) demonstrated a correlation between FGF production in tumor cells and growth rate, making FGF a potential target in HNSCC (20).
Rapid tumor growth causes hypoxia in different areas of the malignant tissue, which is a feature of most solid tumors and is associated with reduced radiotherapy and chemotherapeutic efficacy (21).Oxygen deficiency releases signaling molecules, including VEGF, to induce vasculogenesis, which subsequently correlates with faster invasion and metastasis (22).As demonstrated by Mărgăritescu and colleagues (23) VEGF is expressed in 87% of HNSCC specimens, and increased expression levels are noted in neoplastic compared with dysplastic epithelium (24).Inhibiting the expression of either the VEGFR or the ligands decreases HNSCC cell proliferation (25).
As previously mentioned, we focused on investigating multi-targeted TKIs (pazopanib, dovitinib and nintedanib), which target VEGFR and FGFR family members.Thus, we conducted the present study to investigate the efficacy of nintedanib, dovitinib and pazopanib as HNSCC treatments in vitro.To the best of our knowledge, this study constitutes the first in vitro investigation of pazopanib and nintedanib in HNSCC cell lines.

Materials and methods
The Cancer Genome Atlas (TCGA) analysis.Data of VEGFR1-3 and FGFR1-4 mRNA expression in HNSCC were retrieved from The Cancer Genome Atlas (TCGA) via cBioPortal (26,27), and data from 530 cancer samples were analyzed to assess VEGFR1-3 and FGFR1-4 mutations, amplifications and gains.Cases with and without alterations were compared in terms of overall and median (months) survival.
Cell lines.The following cell lines were used: PCI-1, laryngeal carcinoma of the glottis from a male patient (pT2N0M0G2); PCI-9, primary carcinoma at the base from the tongue of a male patient (pT4N3M0G2); PCI-13, male patient who suffered from oral squamous cell carcinoma of the retromolar triangle (pT4pN1M0G3); PCI-52, primary carcinoma of the aryepiglottic fold from a male patient (pT2N0M0G2); and PCI-68, Primary tongue carcinoma from a male patient (pT4N0M0G1).
Crystal violet assay.After detachment with 0.25% trypsin and 0.53 mM ethylenediaminetetraacetic acid (EDTA) from the culture flask, the cell numbers were measured using a Casy cell counter (Roche Diagnostics GmbH, Penzberg, Germany).A total of 10,000 cells/well were seeded on a 96-well plate, and after 24 h of incubation, the cells were exposed to various concentrations (log2 dilutions) of nintedanib (starting concentration of 100 µM), dovitinib (starting concentration of 200 µM) and pazopanib (starting concentration of 800 µM).After 72 h of incubation, medium was removed, and the cells were stained with crystal violet (0.1% crystal violet/20% MetOH) for 12 min.Afterwards, the cells were washed with distilled water four times, and the 96-well plates were dried for 24 h in air.For the photometric determination, 100 µl of MetOH was added to each well.After 10 min, the absorbance was measured with a plate reader at 595 nm (Rainbow Spectra; Tecan, Maennedorf, Switzerland).Each value was measured twice and every experiment was performed in triplicate.Representative extracts from the data were obtained for this publication.

Target
The primers used in the present study are listed in Table II.Each value was measured twice and every experiment was performed in duplicate.
The comparative ΔCT method was used for normalization of the PCR data (29).using β-actin as a standard gene (assuming an expression level of 100%), we quantified the expression of each gene relative to that of β-actin.The relative expression levels were classified into four different groups: very strong expression (≥0.1%); strong expression (0.01-0.09%); intermediate expression (0.001-0.009%); and low expression (≤0.0009%).

Statistical analysis. Statistical analysis was performed using
Microsoft Excel 2010 (Microsoft Corp., Redmond, WA, USA) and Prism 6.05 (GraphPad, Inc., La Jolla, CA, USA).Different experimental aspects were investigated [data derived from TCGA database (26,27)].First, we examined the potency of nintedanib, dovitinib and pazopanib in single agent therapy against HNSCC in vitro.Afterwards, we compared single drug efficacy between the drugs.Following at least three experimental replicates, data were evaluated with a non-parametric Mann-Whitney test, and the significance level was set at P≤0.05.A correlation analysis was performed for the variables 'receptor expression' and 'fraction of viable cells' at a specific drug concentration (6 µM) (Tables I-III).The statistical analysis was based on several repeated and representative experiments.

Results
TCGA analysis.Data on VEGFR and FGFR mRNA expression in HNSCC were retrieved from TCGA.In 10% of the cases, a mutation, gain or amplification was detected for VEGFR1.VEGFR2 alterations were noted in 16% of the cases, whereas anomalies in VEGFR3 were detected in 9% of the cases.Thus, based on the collection of mutations, 30% overall among the cases showed alterations.VEGFR1+3 mutations were significantly (P=0.0166)decreased with increased overall survival (median survival, 57.88 vs. 30.45months) (Fig. 1).Changes to FGFR family members were evident in 46% of the cases, whereas FGFR1 amplifications and gains were noted in 31% of patients.In addition, FGFR3 amplification, gain or mutation was noted in 12% of the cases.FGFR2 and FGFR4 mutations were detected in 9 and 8% of the cases, respectively (26,27).No significant (P=0.63)degree of correlation could be identified between FGFR alterations and overall survival (Fig. 2).
Although the viable fraction of 99.39% (SD±7.17%)at 6 µM nintedanib did not exhibit significant differences compared with the control group, we found highly significant differences with the concentration increases (P=0.0022) and detected the lowest viable fraction of 18.37% (SD±1.61%)with the highest nintedanib concentration (100 µM).In the PCI-52 cell line, we detected the first significant (P=0.0022)decrease in the viable fraction of 79.01%(SD±4.7%)with 13 µM nintedanib.A viable fraction of 12.77% (SD±0.69%)was obtained with the highest nintedanib concentration of 100 µM.PCI-68 cells showed varied responsiveness to single-agent therapy with log2 dilutions of nintedanib.All concentrations >13 µM exhibited significant (P<0.05/P<0.01)cell count decreases compared with the control samples, and a maximum cell reduction of 39.79% (SD±3.39%)was achieved with the highest nintedanib concentration of 100 µM (Fig. 4 and Table III).
Impact of dovitinib treatment on HNSCC cell lines.Each cell line exhibited varying responsiveness toward dovitinib monotherapy.The untreated control was set to 100%.In the PCI-1 cell line, a high significant (P=0.0022)cell reduction was achieved with 6 µM dovitinib.The highest cell reduction of (19.45% (SD±3.37)was achieved with a concentration of 25 µM dovitinib.No additional cell reduction was achieved with subsequent concentration increases to 50 and 100 µM.In the PCI-9 cell line, we observed a similar effect to the applied drug.A viable fraction of 75.74% (SD±11.77%)was detected with 6 µM dovitinib, and a further significant (P=0.0022)cell reduction to 58.09% (SD±7.05%)was observed with the next concentration increase to 13 µM.In addition, 25 µM dovitinib decreased the viable fraction to 38.31% (SD±11.58%).
As previously demonstrated in PCI-1, no significant further cell reductions were observed with higher drug concentrations.Similar results were observed for the PCI-13 cell line.

Concentration (µM) -
Impact of pazopanib treatment on HNSCC cell lines.The cell lines exhibited a minimal response to varying concentrations of pazopanib monotherapy.In summary, we demonstrated significant differences in the lower and upper concentration ranges for each cell line.The control was set to 100%.
In PCI-1, the viable fraction at 6 µM pazopanib was 77.05% (SD±10.63%),which was significantly different (P=0.0043)from the control.Similar results were noted in PCI-9, in which we detected viable fractions ranging from 90.63 (SD±5.74%) to 76.86% (SD±7.57%).A significant difference (P=0.0411) was noted incubating the cells with a concentration of 6 µM pazopanib.Comparing the concentrations 25 and 50 µM with the control no significant differences were noted.PCI-13 exhibited varied reactions toward different pazopanib concentrations.A viable fraction of 71.73% (SD±4.22%)was detected with 6 µM pazopanib, and this effect was highly significant (P=0.0022)compared with the control.Incubating PCI-52 cells with a concentration of 13 µM pazopanib we detected the first significant (P=0.0043)decrease of the cell number.
No other significant differences could be detected in this cell line.In PCI-68, we detected viable fractions ranging from 81.2 (SD±3.58%) to 62.46% (SD±4.33%).Highly significant differences (P=0.0022) were noted for all concentrations applied by comparing with the control (Fig. 4 and Table V).
Correlation analysis.A correlation analysis was performed for the variables 'receptor expression' and 'fraction of viable cells' at a specific drug concentration (6 µM) to quantify the degree of relationship between receptor expression and TKI efficacy.No significant degree of association was observed between receptor expression and TKI efficacy (data not shown).

Discussion
Squamous cell carcinoma of the head and neck is one of the most common tumor presentations worldwide (1).Therapy, which consists of surgery, radiation and systemic chemotherapy, has changed minimally in recent years and provides a devastating 5-year survival rate of 55-60%.
Locoregional recurrence and lymph nodal metastasis are the most common mortality patterns for HNSCC and impair function and quality of life as well as overall survival (1,3).The literature suggests that neoangiogenesis is one of the main reasons for this poor progress (30).Pentheroudakis et al (31) demonstrated high VEGFR1+3 transcriptional activity driven by pro-angiogenic factors at the time of relapse.Similarly, these mutations occur in relevant quantities in HNSCC and result in a significant (P=0.0166)decrease of overall survival (57.88 vs. 30.45median months survival) (26,27).The role of the FGFR family remains unclear.Ipenburg et al (32) demonstrated the prognostic value of FGFR1 expression in CAFs and an association with poor survival, although the literature examining the role of other family members is both limited and conflicting.Although alterations in FGFR1-4 were detected in 46% of the cases, a TCGA analysis did not reveal a significant (P=0.630)correlation between FGFR1-4 alterations and overall survival (26,27).Given that angiogenesis plays a critical role in tumor growth in solid tumors (33) and that VEGFR-targeted therapies, such as bevacizumab, do not generate the desired effect, multi-targeted TKI might enhance therapeutic strategies for HNSCC by inhibiting several downstream targets (34).
In view of the above facts, our cell line model exhibited receptor expression patterns comparable to those detailed in the literature, and TCGA appears to be a useful method for investigating the efficacies of nintedanib, dovitinib and pazopanib for HNSCC treatment in vitro.These agents are potent VEGFR and FGFR inhibitors that exhibited different effects on different HNSCC cell lines.
Nintedanib showed a concentration-dependent efficacy in all cell lines, and a significant reduction in the viable cell fraction was observed even at the lower concentration ranges.In contrast, Kutluk Cenik et al (35) did not observe any antiproliferative effects in their in vitro investigation of lung and pancreatic cancers.However, in their in vivo models, tumor growth was inhibited in all cases (35), indicating that the tumor microenvironment might have a large impact on drug efficacy.Kudo et al (36) suggested that VEGFR2 is a feasible pharmacodynamic biomarker for hepatocellular carcinoma (HCC) in vivo, whereas the cell lines used in the present study did not show high VEGFR2 expression levels, even though significant cell count reductions were observed.A different pharmacodynamic effect might occur when treating different tumor types and cannot be excluded as a possibility, although the tumor microenvironment likely plays a crucial role during multi-targeted inhibitor therapy due to interactions with CAF.In contrast to VEGFR1+3, our TCGA data analysis did not reveal a significant correlation between VEGFR2 alteration and overall survival.
A significant decrease in the viable cell fraction was observed in response to treatment with dovitinib, even at lower concentration ranges.Because other receptors targeted by the drug are not expressed at appreciable levels, the effect is assumed to be mainly mediated through VEGFR and FGFR.Sweeny et al (11) demonstrated that dovitinib yielded significant cell count reductions in their HNSCC in vitro models.Konecny et al (37) demonstrated significant cell count reductions in endometrial cancer cell lines harboring activating FGFR2 mutations, and FGFR2 was also highly expressed in all of our cell lines.An in vivo investigation revealed significant tumor regression and growth inhibition after dovitinib treatment (38).Although our correlation analysis did not obtain a significant relationship between receptor expression and drug efficacy, the expression strength appears to play a decisive role.Furthermore, Ku (39) reported a strong interaction between dovitinib and VEGFR3, which is supported by our data.

Concentration (µM) -
In conclusion, the present study demonstrated a significant decrease in cell proliferation after the in vitro treatment of HNSCC with nintedanib and dovitinib.In contrast, the cell lines appeared to be resistant to pazopanib, which could be explained by their lack of VEGFR2 expression.The results should be critically considered due to the fact that the cell lines were treated outside their normal environment resulting in a lack of interaction with surrounding tissue, blood flow or a missing supply of nutrients.Further in vivo investigation is required to determine a specific role for TKIs in HNSCC treatment.

Figure 1 .
Figure1.The Kaplan-Meier plot illustrates overall survival curves with regard to VEGFR1+3 expression in head and neck squamous cell carcinoma (HNSCC).The grey line shows cases with alterations in VEGFR1+3 expression and the black line shows the cases without alterations.A significant (P=0.0166)decrease in overall survival was detected in the cases with alterations [data derived from TCGA database(26,27)].

Figure 2 .
Figure 2. The Kaplan-Meier plot illustrates overall survival curves with regard to FGFR1-4 expression in head and neck squamous cell carcinoma (HNSCC).The grey line shows cases with alterations in FGFR1-4 expression, and the black line shows the cases without alterations.No significant (P=0.630)difference in overall survival was detected.

Figure 3 .
Figure 3. Representation of receptor (VEGFR1-3 and FGFR1-4) expression.The expression of the receptors is plotted on the y-axis as a percentage of the level of the housekeeping gene β-actin.The expression levels were determined as a function of PCR cycles: very strong expression ≥0.1; strong expression, 0.01-0.09;intermediate expression, 0.001-0.009;and weak expression ≤0.0009.

Figure 4 .
Figure 4. Treatment efficacy of nintedanib, dovitinib and pazopanib in different cell lines (PCI-1, PCI-9, PCI-13, PCI-52 and PCI-68) as measured by crystal violet assay.In all cell lines, a concentration-dependent effect of the different TKI used was detected.The horizontal bracket indicates the first concentration that yielded a significant decrease in the number of viable cells compared with the control.The data presented are shown as the means standard deviations.* P≤0.05; ** P≤0.01.