Dual targeting of EGFR and ERBB2 pathways produces a synergistic effect on cancer cell proliferation and migration in vitro

Members of the epidermal growth factor receptor (EGFR/ERBB) gene family are frequently dysregulated in a range of human cancers, and therapeutics targeting these proteins are in clinical use. We hypothesized that similar pathways are involved in feline and canine tumours and that the same drugs may be of clinical use in veterinary patients. We investigated EGFR and ERBB2 targeting using a panel of feline and canine cell lines. EGFR and ERBB2 were targeted with siRNAs or tyrosine kinase inhibitors (TKIs) and their effect on cellular proliferation, colony formation and migration was investigated in vitro. Here we report that EGFR and ERBB2 combined siRNA targeting produced synergistic effects in feline and canine cell lines similar to that reported in human cell lines. We conclude that dual EGFR and ERBB2 targeting using TKIs should be further evaluated as a potential new therapeutic strategy in feline head and neck and mammary tumours and canine mammary tumours.


Introduction
In humans, amplification or overexpression of the oncogenes epidermal growth factor receptors 1 and 2 (EGFR and ERBB2/HER2) is associated with the development and progression of certain types of aggressive breast cancer, head and neck cancer (HNSCC), ovarian, stomach and uterine cancer. They are receptor tyrosine kinases and members of the erythroblastic leukaemia viral oncogene homolog (ERBB/EGFR) gene family. This family consists of four members: EGFR/ErbB1, ERBB2, ERBB3 and ERBB4, 1 and are involved in a range of normal cellular processes including migration, survival, proliferation, and cell cycle progression. 2 The receptors are expressed in a wide range of epithelial and neuronal tissues. 3 Gene knockout studies have demonstrated that the gene family is crucial during development, as homozygous null mice die during early to mid-gestation due to multiple defects in epithelial organ development. 3,4 EGFR and ERBB2 have been reported to be amplified in a number of human, canine and feline cancers. 5 EGFR has been reported to be overexpressed in feline mammary carcinomas (FMC) 6,7 and oral squamous cell carcinomas (FOSCC). 8,9 We have previously shown that knockdown of EGFR in FOSCC has an additive effect when used in combination with radiation. 10 EGFR has been reported to be both a positive 9 and negative 8,11 prognostic factor in FOSCC and increased expression was associated with decreased overall survival in FMC. 12 Feline mammary carcinomas have also been reported to have an increase in tumourigenicity with increased EGFR expression. 6 In canine cancers EGFR overexpression has been reported in canine mammary tumours (CMT), [13][14][15] transitional cell carcinomas, 16,17 medullary thyroid cancer, 18 and gastric epithelial tumours. 19 Gama and colleagues 20 reported EGFR expression to be significantly associated with malignancy but could only report a trend towards reduced disease-free and overall survival. A more recent paper using ELISA to detect EGFR levels in canine mammary tumours reported a statistically significant association between high EGFR levels and decreased disease-free and overall survival times. 21 Epidermal growth factor (EGF, the principal ligand of EGFR) was reported to stimulate proliferation, migration, angiogenesis, and survival in canine mammary carcinoma cell lines, 22 while a correlation was reported between EGFR and microvessel density in samples from malignant CMT 23 suggesting the EGFR pathways might be involved in stimulation of angiogenesis. Kennedy and colleagues 22 demonstrated that Vascular Endothelial Growth Factor (VEGF) production by a CMT cell line was stimulated by the addition of EGF and blocked by vandetanib (ZD6474), a TKI that targets VEGF receptor 2 (VEGFR-2), EGFR and rearranged during transfection (RET) tyrosine kinases.
Investigations into ERBB2 expression levels has mainly been focused around FMC 24-26 and CMT. 27,28 A wide range of expression percentages have been reported following the use of multiple techniques and different interpretations. ERBB2 plays a role in normal development of mammary tissue, 29 with its overexpression reported to increase the tumour metastatic potential in human breast cancer (HBC), 30 as well as predicts response to the HER2 targeting drug trastuzumab. In man, targeting of EGFR in HNSCC and HER2 in HBC is well established. Drugs like small molecule tyrosine kinase inhibitors (TKIs) that block the ATP binding pocket of the receptor 31 or monoclonal antibodies (mAbs) that bind directly to the receptors on the cancer cell surface blocking ligand binding and also potentially mediate an antibody-dependent cellular cytotoxicity (ADCC) or antibody-dependent cell mediated phagocytosis (ADCP) effects 32 have been of value in a subgroup of patients where the tumours are dependent on the EGFR pathway. The use of these drug classes are still in their infancy in veterinary medicine.
The overall aim of this study was to investigate the potential for EGFR and ERBB2 targeting in veterinary medicine using a panel of feline and canine cell lines. EGFR and ERBB2 was targeted with siRNAs or TKIs, and their effect on cellular proliferation, colony formation and migration was investigated. Potential synergistic effects of dual targeting of the receptors was investigated in accordance to the Bliss Additivism model. 33 Here we report that EGFR and ERBB2 combined targeting by siRNAs produced a synergistic effect. We conclude that dual targeting of EGFR and ERBB2 should be further evaluated as a potential new therapeutic strategy in feline and canine head and neck and mammary tumours.

Cell culture and reagents
Cell culture reagents were obtained from Gibco Thermo Fischer Scientific, (Paisley, UK) unless otherwise specified. The SCCF1 cell line is a previously characterised feline cell line derived from a laryngeal squamous cell carcinoma and was a gift from Professor T.J. Rosol, Ohio State University, USA. 34 The feline SCCF-SMG cell line was isolated from a boneinvasive FOSCC, and has been characterised in our laboratory (unpublished data). Both were grown in William's E Medium with GlutaMAX supplemented with 10% FBS, 0.05 mg/mL gentamicin and 10 ng/mL EGF.
The CatMC 35 (feline mammary adenocarcinoma) and REM134 36 (REM, canine mammary adenocarcinoma) cell lines were previously characterised and generously gifted by Professor R. Else, R(D)SVS, The University of Edinburgh, UK. The LILLY cell line (also from a canine mammary tumour) was generously provided by Dr R. De Maria, Department of Animal Pathology, University of Torino, Italy and was derived from a grade three simple carcinoma from a 13 year old mixed breed entire female. These cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM) with added GlutaMAX, supplemented with 10% fetal bovine serum (FBS) and 5 ml of 100 μl/ml penicillin and streptomycin. The media was changed and the cells washed with phosphate buffered saline (PBS) every 48 hours until confluent. Once confluent, cells were re-seeded by washing with PBS and adding 3 ml of Trypsin-EDTA 0.25% to T75 flasks and incubated until detachment. The human breast cancer cell lines SK-BR-3 and BT474 cell lines were both purchased from ATCC, USA and cultured as above and according to manufacturer's recommendations.

RNA interference
A previously published and validated siRNA against feline EGFR (GenBank accession number HQ185236.1) 10 was validated for the use in dog cells together with new siRNAs designed against a feline ERBB2 sequence we previously obtained and published on GenBank (accession number: KC710349.1). Multiple siRNA sequences against ERBB2 were designed using a combination of online design tools including the siDESIGN Centre (Thermo Scientific, 2013) and i-Score designer (i-score web service program, 2009).
Potential siRNAs were evaluated for GC content and Reynolds scores were obtained. 37 BLAST searches were then performed for selected siRNAs against the feline whole genome sequence (wgs) and nucleotide collection. 38 The selected siRNAs were constructed using

Transfections
Cells were seeded into 96 or 6 well plates at 1.

Immunocytochemistry
Cells were seeded into chamber slides at a density of 3-5x10 4 cells per well in a total of 200 µl media and incubated at 37°C/5% CO2 for 24 hours. Cells were fixed in cold acetone and incubated with an appropriate primary antibody (anti-EGFR and anti-ERBB2 antibodies as listed in Table 1) for subsequent detection. Secondary antibodies goat anti-mouse AlexaFluro© 488 IgG, IgA, IgM (A11011) and goat anti-rabbit AlexaFluro© 568 IgG (A10667) both from Life Technologies™ (Paisley, UK) were incubated at room temperature for 1 hour in a dark, humid chamber. Cover slips were removed, slides were washed and mounted using anti-fade mounting media containing DAPI (Vectorshield®, Vector Laboratories Inc., Burlingame, CA) and visualised with a Leica DMLB2 microscope fitted with a digital ORCA-ER digital camera and Lumencor Spectraz LED light engine®.

Real Time PCR (qPCR)
All qPCR reactions were performed on the Roche LightCycler® 480 machine following manufacturer's instructions. The most stably expressed reference genes hypoxanthineguanine phosphoribosyltransferase (HPRT) and GAPDH were selected using geNorm™  Table 1. Analysis of relative gene expression levels were performed using crossing point (Cp) values obtained for each target and reference gene using qbase plus (Biogazelle). The delta Cp value for each sample was determined and the relative expression level of the target gene was calculated according to previously described methods. 39

Drug treatment of cells
Three commercially available TKIs were used: gefitinib (EGFR specific), AG825 (ERBB2/HER2 specific) and GW583340 (dual EGFR and ERBB2) all supplied by Tocris Biosciences, UK. TKIs were supplied in powdered form and each drug was separately dissolved in dimethyl sulphoxide (DMSO) to create stock solutions of 10 mM. Aliquots were made and stored at -20°C. The cell lines were treated with each TKI at a range of concentrations (0.1-150 µM) as indicated.

Cell proliferation assays
Cells were grown in the media as stipulated previously during all experiments, except for when their basal proliferation rate in response to EGF was determined. The cells were grown in their respective media with no serum added, and either a range of EGF (5, 10 or 50 ng/ml) was added to the 96 well plates for 24 hours, or 50 ng/ml was added and the cells were cultured for up to 48 hours. Cells were seeded at 5x10 3 cells per well and incubated at 37°C/5% CO2. Cell proliferation was assessed at 24 and 48 hours as measured by proxy by reading absorbance at 490 nm using Wallac 1420 Manager program (Perkin Elmer) following treatment with CellTitre 96® AQueous One Solution (Promega) according to manufacturer's protocol.
For all other experiments, cells were seeded into 96 well plates at a density of 1.5x10 3 cells per well and incubated at 37°C/5% CO2 for 24 hours in their respective media with serum.
Cells were then treated with TKIs at the indicated concentrations or were transfected with the indicated concentrations of siRNA. Cell proliferation was assessed at specified time points up to 96 hours following drug exposure or transfection as described above.

Colony formation assays
Twenty-four hours following transfection or drug treatment cells were seeded into 10 cm plates at a density of 300-500 cells per plate and incubated at 37°C/5% CO2. Plates were checked for colony formation every 2 days. When visible colonies had formed (approximately 5-10 days) the cells were fixed with methanol and stained with 10% Giemsa stain for counting. The colonies were electronically counted and photographed using the CCD digital camera (Bio-Rad).

Cellular migration assay
Cells were seeded into 6 well plates at a density of 2.0x10 6 cells per well and incubated at 37°C/5% CO2 to achieve 100% confluence the following day. A scratch assay was performed as previously described. 40 Briefly, cells were treated with indicated concentrations of TKIs or 24-48 hours following siRNA transfections, as previously described, and a scratch was made using a pipette tip. At regular time intervals phase contrast images of the cell monolayer were captured with a monochrome digital Axiocam camera fitted to a Zeiss Axiovert40 CFL (Carl Zeiss Ltd, UK) microscope until the gap had been filled. The gap width was measured at ten different points for each image and the mean calculated for each well at each time point and expressed as the relative migratory distance.

Statistical Analysis
One-way ANOVA or the nonparametric equivalent Kruskal-Wallis were used to compare differences between more than two samples. Two sample t-tests or the nonparametric equivalent Mann-Whitney U test were used to compare differences between two samples.
Results were considered significant when P < 0.05. All statistical analysis was performed using Minitab® 17 Statistical software (Minitab Ltd.) and all graphs were generated using Microsoft® Office Excel 2013 software.

All tumour cell lines expressed EGFR and ERBB2
Both the dog and cat cell lines showed an increase in cell proliferation in response to increasing EGF concentrations when grown in serum free media with EGF only. They also showed sustained proliferation over a 48 hour period when maintained in serum free media containing 50 ng/ml EGF (data not show). All five cell lines expressed the two proteins at The siRNAs successfully reduced mRNA and protein levels of their targets A range of siRNAs against both targets were produced and tested in the SCCF1 cell line.
They produced variable levels of mRNA knockdown when validated by qPCR at different siRNA concentrations (20-100 nM). The siRNAs produced dose dependent effects in the cell line (from 20% to up to 75% reduction). The optimum concentration for individual siRNA transfections was determined to be 60 nM (data not shown) at 72 hours following transfections. The siRNAs that produced the greatest reduction in mRNA levels of their specific targets are shown in Figure 1c and Table 3. Reduction of EGFR and ERBB2 protein levels were confirmed by western blot analysis in both cat (Figure 1d

Human TKIs blocked EGFR and ERBB2 phosphorylation in vitro
To investigate if the same effects could be achieved by using readily available drugs we selected three TKIs developed to block the ATP binding pockets of the equivalent human proteins: gefitinib (specific EGFR inhibitor), AG825 (specific HER2/ERBB2 inhibitor) & GW583340 (dual inhibitor of EGFR and HER2/ERBB2). These were validated for use in feline cells by performing phosphorylation assays following serum starvation. The cells were treated with EGF/serum with or without the TKI drugs, and the levels of phosphorylation of Human TKIs blocked proliferation and caused an reduction in colony formation and migratory ability in vitro in a panel of cell lines Drug assays were performed in the three cell lines to determine the relative effect of the drugs on cellular proliferation ( Figure 5). Similar to what was observed when using the siRNAs, the cell lines were relatively more sensitive to EGFR inhibition than ERBB2 inhibition when targeting the receptors individually. The dual inhibitor however reduced cellular proliferation most effectively, with the lowest IC50 estimated for all three cell lines ( Table 4). As these are unrelated compounds, the difference in IC50 could also be due to difference in target affinity between the compounds. The results do however exactly mirror what was observed following siRNA transfections targeting the receptors independently and combined.
For comparison, two human breast cancer cell lines with known EGFR/ERBB2 status were also treated with the same drugs, and a similar pattern was observed with minimal response to ERBB2 inhibition on its own at the concentrations 0.01-10 μM, moderate reduction in cellular proliferation with EGFR inhibition while dual receptor inhibition produced the most profound effect on cellular proliferation of all ( Figure 6).
Colony formation assays following drug treatment of the cell lines showed the same pattern as observed following siRNA transfections (Figure 7). EGFR blockage was effective in reducing the colony formation ability of the cell lines while ERBB2 had no effect.
The migratory ability of the cell lines were highly variable. The two mammary carcinoma cell lines exhibited a relatively low propensity towards migration, even when untreated (data not shown). After 35-48 hours both untreated and DMSO control treated cells (as shown in approximately a third of the distance. A slight response to ERBB2 and dual inhibition was observed (P < 0.001), and although this was statistically significant this is unlikely to be biologically significant as the overall migration of this cell line was minimal. Overall, dual targeting produced similar effects as EGFR targeting alone, and did not appear to confer any further benefits on cellular migration in the cell line. We then tested the effects of the TKIs on the full panel of cell lines. Consistently, the cell lines were more sensitive to EGFR targeting (singly or in combination) and produced response curves to the left of the ERBB2 targeting TKI. Although the different target affinity of these compound cannot be dismissed as a potential contributing factor in the effect observed, this pattern exactly mirrored what was observed when targeting the receptors individually and combined using the siRNAs.

Discussion
Both the feline and the canine cell lines had the ability to proliferate for up to 48 hours when grown in serum free media containing EGF, and an increase in proliferation rate was also Peña and colleagues 42 (2014) goes some way in trying to alleviate this problem, and one of their suggestions with respect to the Hercept Test™ is to use commercially available control slides containing HBC cells of each HER2 reactivity for internal validation. As targeted therapies are starting to become available to the veterinary patient, 45 it is of paramount importance that a robust, reliable, and repeatable system is available for the evaluation of expression levels. If this can be achieved, it would prove a valuable tool for the veterinary oncologist when choosing potential targeting therapies.
With respect to EGFR expression in mammary tumours, less is known. It has been suggested that FMC is a good model of hormone negative HBC, 7, 24 as a significant proportion of hormone negative (oestrogen receptor (ER) and progesterone receptor (PR)) and triple negative (ER, PR and HER2 negative) tumours express EGFR. These triple negative HBC are associated with a very poor prognosis. In CMT EGFR overexpression has been reported in around 40% of malignant tumours. 20,27 A functional study 46 treating CMT and HBC cell lines with humanized mAbs against EGFR and HER2 revealed that both antibodies blocked tumour cell proliferation by inducing growth arrest in G0/G1 phase, but at a lower efficiency than what the same mAbs produced in the HBC cell lines. The authors suggested that this might be due to the CMT cell lines expressing the molecules at a 2-log lower expression levels compared to the human cell lines. EGFR overexpression has been reported in FOSCC, 8 and its prognostic potential has been investigated. A small study of 22 FOSCC samples showed an inverse relationship between EGFR expression levels and survival, 11 but a larger study of 67 tumour samples could not verify this. 9 We have previously confirmed ERBB2 expression in the feline cell lines by PCR and sequencing (GenBank accession number KC710349.1), ICC, and western blot analysis.
To our knowledge, no ERBB2 IHC study of FOSCC biopsy samples has been reported.
The link between protein expression detected by western blotting from a cell line and the levels observed by IHC in associated tumour biopsy samples are unknown for the cell lines used in this study, but the fact that they all seemed more sensitive to EGFR rather than ERBB2 inhibition is interesting. The use of receptor specific siRNAs first confirmed that the effect observed in the cell lines were due to the specific knockdown of their respective receptors. Despite observing a clear reduction in ERBB2 mRNA and protein levels following transfections, no measureable effect was observed in any of the cell lines. ERBB2 is different to the remaining receptors in its gene family in that it is incapable of ligand binding, and hence relies on its heterodimer partner for ligand activation. 47 It is in fact the preferred dimerisation partner for all family members, 47 a possible explanation for why dual siRNA targeting caused a synergistic effect despite minimal effect of ERBB2 monotherapy.
EGFR silencing on the other hand consistently produced reduced cellular proliferation, colony formation and migratory ability. The effect of dual siRNA targeting when compared to individual targeting suggested that a synergistic action might be achieved when targeting the receptors simultaneously. A synergistic action is highly beneficial as drugs can be given at lower doses and achieve the same efficiency. In effect a drug that previously was only effective at toxic doses separately may yield the same results at subtoxic doses. 33 The use of small molecule inhibitors in veterinary medicine is an attractive option. The use of toceranib 48 and masitinib 49 in the treatment of mast cell tumours is well established. The feline, canine and human mRNA and amino acid sequences show high homology for the EGFR and ERBB2 genes, and modelling have shown great structural homology between the canine and human proteins, 46 suggesting that human TKIs should be effective in the feline and canine cells. Here we confirm by western blot analysis that treatment of cells with TKIs does reduce phosphorylated levels of the proteins following treatment, supporting the possible use of human TKIs in veterinary medicine. When used in man tumours that initially responded to a specific TKI have been reported to develop acquired resistance over time, and this is a great disadvantage of TKIs. First generation TKIs gefitinib and erlotinib produced overall response rates of up to 75% in patients with non-small-cell lung cancer (NSCLC) who carried activating EGFR mutations. Unfortunately, the median progression free survival was less than one year due to a secondary mutation being selected for over time. The mutation responsible for 50% of these cases is the T790M mutation in exon 20 of the human EGFR gene, but the mechanisms for the remaining 50% of cases are largely unproven. 50 ERBB2 amplification have been implicated as one of the mechanisms that confer acquired resistance in NSCLC, 51 which provides a rationale to target ERBB2 in combination with EGFR.
In human medicine, monoclonal antibodies rather than TKIs have been used in the treatment of HBC and HNSCC (trastuzumab against HER2 and cetuximab against EGFR respectively). The use of humanized mAbs in veterinary patients would not be possible as the humanization would render them useless due to patient immune responses and at worst could cause adverse reactions and anaphylaxis. 45 Singer and colleagues reported the caninization of a mAb against canine EGFR, and demonstrated its efficiency in vitro in two canine mammary cell lines. 45 This demonstrates how personalised veterinary oncology is becoming a distinct possibility.
The doses required in dogs and cats in vivo is hard to extrapolate from in vitro studies, but some comparisons to human drug doses can be made. An in vitro dose of 1 µM gefitinib is equivalent to a clinical dose of 250 mg per day used in NSCLC. 52 In NSCLC in vitro studies doses of gefitinib above 2 µM would class the cell line as insensitive to the drug. 52 Early studies on HNSCC cell lines used much higher doses than this 53 while more recent studies all have used doses below 2 µM gefitinib. 54,55 The cell lines assessed in this study would therefore all be classed as relatively insensitive to gefitinib requiring relatively high doses (IC50 doses 5-28 µM). The dual inhibitor GW583340 required a lower dose (SCCF1 and CatMC IC50 doses of 0.6 and 1.1 µM respectively), suggesting a dual inhibitor of some form (monoclonal antibodies, RNAi or TKIs) might be a better therapeutic option.
In summary, we have assessed the in vitro effect of targeting the EGFR family in feline and canine tumour cell lines from two tumour types (oral squamous cell carcinomas and mammary tumours) known to benefit from EGFR family targeting in man using RNA interference and currently available TKIs. The benefit of TKIs over mAbs is that they are cheaper, can be given orally, and the human form of the drug can be utilised without the problems associated with humanized mAbs. It is also currently easier to achieve dual targeting with TKIs, with several dual targeting TKIs readily available. A caninized version of cetuximab however has now been produced 45 and shown to stimulate ADCP as well as blocking the extracellular region of the receptor. TKIs cannot stimulate the immune response, as they act solely by blocking the ATP-binding pocket of the receptors inhibiting their phosphorylation. In addition, acquired resistance to therapy is a commonly reported problem encountered in human medicine in the clinical use of TKIs.
Veterinary medicine is moving towards the possibility of offering more personalised cancer treatments, and both mAbs and TKIs are likely to play a role in this in the future. The next step requires drugs to be identified that can be taken into controlled clinical trials for further assessment of their potential. Two things we can learn from the story so far of targeted therapies in the human field is that firstly, without effective patient selection the benefits of targeted therapies are difficult to achieve and secondly, that targeted therapies are rarely effective as sole agents and should be incorporated into multimodal treatment options, for example in combination with surgery, radiotherapy and established chemotherapy protocols.
Therefore it will be crucial to establish good protocols for expression profiling that can be used repeatedly and reliably in the veterinary species.

Figure 8
The relative migration distance of the three cell lines were reduced following TKI treatments.
a-c) The SCCF1 cell line showed a dose dependent reduction in their relative migration distance in response to gefitinib and dual inhibition (GW) (*P < 0.001). d-f) The CatMC cell line showed a reduction in relative migratory distance when treated with the gefitinib (*P < 0.001) and GW583340 (*P = 0.003). g-i) The REM cell line was minimally migratory but showed some reduction in relative migratory distance after AG825 and GW583340 treatments (*P < 0.001).

Figure 9
Dose response curves for the full panel of cell lines. All cell lines tested were more sensitive to EGFR compared to ERBB2 targeting. The feline cell lines were overall more sensitive to the TKIs than the canine cell lines were. Graphs shown are representative results from one of triplicate experiments.