Antiproliferative effects of masitinib and imatinib against canine oral fibrosarcoma in vitro

Background Canine oral fibrosarcoma (COF) is one of the most common oral tumors in dogs and carries a guarded prognosis due to a lack of effective systemic therapeutic options. Mastinib and imatinib are two commonly used tyrosine kinase inhibitors (TKIs) in veterinary oncology but their potential efficacy against COF is uncharacterized. To begin investigating the rationale for use of these TKIs against COF, the present study tested for the presence TKI targets PDGFR-α, PDGFR-β, Kit, and VEGFR-2 and examined the in vitro effects on cell viability after TKI treatment alone or with doxorubicin. Immunohistochemistry for PDGFR-α, PDGFR-β, Kit, and VEGFR-2 was performed in 6 COF tumor biopsies. Presence of these same receptors within 2 COF cell lines was probed by reverse transcription-polymerase chain reaction and, for those with mRNA detected, confirmed via western blot. Effects on cell viability were assessed using an MTS assay after masitinib or imatinib treatment alone (0-100 μM), or in combination with doxorubicin (0-3000 nM doxorubicin). Anti-PDGFRB siRNA knockdown was performed and the effect on cell viability quantified. Results Expression of the TKI targets evaluated was similar between the 2 COF cell lines and the 6 COF tumor biopsies: PDGFR-α and PDGFR-β were detected in neoplastic cells from most COF tumor biopsies (5/6 and 6/6, respectively) and were present in both COF cell lines; KIT and KDR were not detected in any sample. Masitinib and imatinib IC50 values ranged from 7.9–33.4 μM, depending on the specific TKI and cell line tested. The addition of doxorubicin resulted in synergistic cytotoxicity with both TKIs. Anti-PDGFRB siRNA transfection reduced PDGFR-β protein expression by 77 % and 67 % and reduced cell viability by 24 % (p < 0.0001) and 28 % (0 = 0.0003) in the two cell lines, respectively. Conclusions These results provide rationale for further investigation into the use of TKIs, possibly in combination with doxorubicin, as treatment options for COF.


Background
Canine oral fibrosarcoma (COF) is one of the three most common oral neoplasms in dogs [1]. Compared to other anatomic locations, COF exhibits a biologically aggressive behavior with recurrence rates following resection of 24-59 %, metastasis in up to 30 % of cases, and reported median survival times of 7-24 months [2][3][4][5][6][7][8][9]. The most recent of these studies included 65 dogs and found significant predictors of median survival time to include tumor location (maxillary location better than mandibular), size (smaller tumors better), type of surgery (aggressive surgery better than conservative), histologic margin status, and grade (low grade better). This study included 14 dogs that received adjuvant systemic therapy (4 received doxorubicin and 10 received metronomic chemotherapy) but because of this low sample size and the fact that therapy was often initiated after relapse of disease, no conclusions could be drawn regarding the potential efficacy of this treatment strategy [8]. Currently, the prognosis for this disease remains guarded due to a lack of effective systemic therapeutic options to address potential metastasis as well as local recurrence [1][2][3][4][5][6][7][8][9].
The use of receptor tyrosine kinase inhibitors (TKIs) for targeted therapy in veterinary oncology is increasing as indicated by the growing number of clinical reports [10][11][12][13][14][15]. Although some reports describe use of TKIs alone, others have reported on observed clinical efficacy when combined with traditional cytotoxic chemotherapeutic agents and/or piroxicam [10,11,13,15,16]. Proposed mechanisms behind combination therapy include chemosensitization as well as immunomodulatory effects such as suppression of regulatory T cells and restoration of T cell-mediated immune responses [16]. Masitinib is conditionally approved by the United States Food and Drug Administration and the European Medicines Agency for use against canine mast cell tumors. Masitinib targets PDGFR-α and -β, Kit, Lyn, and to a lesser degree, the FGFR3 and FAK pathways [16]. Masitinib may also affect VEGFR-2 levels [14]. Imatinib is another TKI that targets some of the same kinases as masitinib, including PDGFR-α, PDGFR-β, and Kit [16,17]. Although not approved by the United States Food and Drug Administration for use in veterinary patients, off-label veterinary use of imatinib has been reported with favorable results in canine and feline cancer patients [18][19][20][21].
To our knowledge, there are no reports that have profiled expression of tyrosine kinases in COF, nor the potential for targeting by masitinib or imatinib. The purpose of this study was to (1) evaluate the expression of PDGFR-α, PDGFR-β, Kit, and VEGFR-2 in archived COF biopsies and immortalized cell lines and (2) assess the effects on cell viability of two TKIs (masitinib and imatinib), either alone or in combination with doxorubicin, against the cell lines in vitro. The results presented herein begin to shed light on this strategy as a potential future therapy for COF.

Results
Archived canine oral fibrosarcoma tumors express PDGFR-α and -β protein Immunohistochemistry (IHC) for PDGFR-α, PDGFR-β, Kit, and VEGFR-2 demonstrated differential expression of each protein amongst the six archived tumor specimens with good agreement between subjective observerderived assessments and semi-quantitative softwarederived results (Table 1). Representative photomicrographs for each of the proteins evaluated are shown in Fig. 1. A representative software threshold-processed image of tumor cell immunoreactivity for PDGFR-β is shown in Fig. 2. Higher percentages represent more immunoreactivity (i.e. pixels above the user-defined threshold for IHC stain). Mitotic counts in five of six sarcomas were low, ranging from 1 to 4 in ten 400x high power fields. Case 3 had a much higher mitotic count (n = 15). This tumor also had the largest nuclei, poorest overall organization, and most intense IHC staining for both PDGFRs.
Staining for PDGFR-α was detected in the cytoplasm and nuclei of 75-100 % of neoplastic cells in five of the six tumor samples, with a relatively uniform staining intensity among samples. In all sections, PDGFR-α staining was also present in the cytoplasm and nuclei of endothelial cells including neoplastic endothelial cells of a canine metastatic hemangiosarcoma control sample. Similarly, PDGFR-β was detected in the cytoplasm of 60-100 % of neoplastic cells in all six tumor samples. There was relatively uniform intensity and subcellular location of immunostaining of neoplastic cells in five of the biopsy samples with a similar distribution but weaker staining intensity in the remaining sample. Two samples also showed cell membrane associated PDGFRβ staining. PDGFR-β stained the cytoplasm of endothelial cells in all sections including neoplastic endothelial Subjective scoring of immunoreactivity of 6 archived canine oral fibrosarcoma cases for VEGFR-2, PDGFR-α, PDGFR-β, and Kit. The estimated percentage of tumor cells displaying immunoreactivity, the predominant location(s) of staining (C = cytoplasmic, M = membranous, or N = nuclear), and the subjective intensity of staining (+, ++, or +++) are displayed along with semi-quantitative measurement of immunoreactivity using computer image analysis software and threshold-processed photomicrographs cells in the hemangiosarcoma control sample. VEGFR-2 and Kit staining were uniformly negative in neoplastic cells of all six COF tumor samples. Endothelial cells around but not within the tumors had cytoplasmic staining for VEGFR-2, whereas neoplastic cells of a metastatic hemangiosarcoma did not stain. Most sections had interstitial mast cells that displayed a largely membraneassociated staining pattern for Kit. In the mast cell tumor control sample, most neoplastic cells had cytoplasmic, perinuclear, punctate staining for Kit (pattern 2); less than 5 % showed diffuse cytoplasmic staining (pattern 3) [22]. Fig. 1 Immunohistochemistry of archived canine oral fibrosarcoma tumor biopsies. Representative photomicrographs of an archived canine oral fibrosarcoma tumor specimen (dog #4) stained with (a) hematoxylin and eosin, b PDGFR-α immunohistochemistry, c PDGFR-β immunohistochemistry, d Kit immunohistochemistry, e VEGFR-2 immunohistochemistry, and f rabbit negative control; 400×. Immunoreactivity for PDGFR-α (both cytoplasmic and nuclear locations) and PDGFR-β (predominantly cytoplasmic location) is visible. No staining is seen for VEGFR-2, Kit, or in the rabbit negative control Fig. 2 Semi-quantitative assessment of immunoreactivity via image threshold-processing. Representative photomicrograph of PDGFR-β immunohistochemical staining of an archived canine oral fibrosarcoma tumor specimen (dog #6) before (a) and after (b) thresholdprocessing for semi-quantitative assessment of immunoreactivity; 400×. Red pixels are reported as percentage of total image pixels to provide a semi-quantified measure of immunoreactivity Canine oral fibrosarcoma cell lines express PDGFR-α and -β at both mRNA and protein levels PDGFRA and PDGFRB mRNA was reverse transcribed and amplified from exponentially growing MBSa1 and CoFSA cells by reverse transcription-polymerase chain reaction (RT-PCR; Fig. 3a). Amplicons were of the predicted size and sequencing reaction results matching the published sequence with 100 % homology. Transcripts for KIT and KDR were not detected (Fig. 3a) despite using two different canine-specific primer sets. Western blots showed strong expression of both PDGFR-α and -β in cell lysates from CoFSA, with weaker expression in MBSa1 (Fig. 3b). These data coincide with the apparent mRNA signals shown in these two cell lines (Fig. 3a).
Masitinib or imatinib alone, or in combination with doxorubicin, inhibit canine oral fibrosarcoma cell viability The calculated IC50 of imatinib for MBSa1 and CoFSA is 33.4 and 7.9 μM, respectively.
The combination of either 1.0 μM masitinib or imatinib with doxorubicin yielded synergistic reductions in cell viability for both cell lines (Fig. 5). Combination treatment demonstrated synergism in MBSa1 cells at all doxorubicin concentrations for masitinib and at 1, 3, 10, 30, and 100 nM doxorubicin concentrations for imatinib (Fig. 5a). Due to the greater reduction in cell viability seen with 1.0 μM of either TKI alone in CoFSA cells, synergism was shown only at the highest doxorubicin  Cell viability after PDGFRB siRNA transfection was significantly reduced in both MBSa1 (mean reduction of 24.0 %; p < 0.0001) and CoFSA (mean reduction of 27.6 %; p = 0.0003) cell lines compared to vehicle-treated control cells (Fig. 7). Visual comparison of siRNAtransfected cells to control cells revealed a greater negative effect on cell viability than was reflected by the MTS assay results (Fig. 7).
Effect of masitinib and imatinib, alone or combined with doxorubicin, on oral fibrosarcoma cell line caspase activity MBSa1 cells did not demonstrate significant changes in caspase-3/7 activity at any drug concentration tested ( Fig. 8a and c). In contrast, CoFSA cells did display significantly increased caspase activity at 1.0 μM masitinib alone (Fig. 8b) and at 300 nM doxorubicin combined with either 1.0 μM masitinib or imatinib (Fig. 8d); CoFSA cells showed significantly reduced caspase activity following treatment with 30 μM masitinib alone (Fig. 8b).

Discussion
This study begins to explore the potential rationale for using two commonly prescribed TKIs (masitinib and imatinib) as adjunctive treatment in COF. The stimulus Masitinib showed a synergistic interaction with doxorubicin at all concentrations for MBSa1 and at 10, 300, 1000, and 3000 nM for CoFSA. Imatinib showed a synergistic interaction with doxorubicin at 1, 3, 10, 30, and 100 nM for MBSa1 and at 300, 1000, and 3000 nM for CoFSA. Plotted values are mean ± standard error of the mean. Synergism was defined as being present when the surviving fraction of cells exposed to the combination of doxorubicin and either tyrosine kinase inhibitor was lower than the product of the surviving fraction of cells exposed to the tyrosine kinase inhibitor alone multiplied by the surviving fraction of cells exposed to doxorubicin alone. See Materials and Methods section for detailed synergism calculation methods for this investigation is based on the premise that targeted small molecule therapy may provide an adjuvant therapeutic strategy for control of COF following surgery, given the challenge of obtaining complete surgical tumor excision and the reluctance of many pet owners to pursue adjuvant radiotherapy. We began by testing for the presence of targets of these two TKIs, including PDGFR-α, PDGFR-β, VEGFR-2, and Kit, in six archived COF tumors and two immortalized COF cell lines. Our results demonstrate a similar expression profile between the immortalized cell lines and the archived tumor samples, leading into the second aim of the study: assessing the effects of the two TKIs on cell viability, either alone or in combination with doxorubicin. Our data show both cell lines were relatively resistant to single-agent TKI treatment, with substantial reductions in cell viability and an increase in apoptotic activity being seen only at relatively high concentrations in most experiments. Both TKIs met the criteria for synergistic in vitro cytotoxicity when combined with doxorubicin, although the magnitude of this effect was relatively small. Cumulatively, the present study provides insight into the potential validity of future in vivo investigations exploring the use of TKIs, possibly in combination with traditional cytotoxic chemotherapeutic agents, as adjuvant treatment options in COF.
Of the four potential TKI targets evaluated via IHC, PDGFR-β was the most consistently expressed (6/6 of archived tumors) and showed the strongest immunoreactivity across tumor samples. Furthermore, PDGFR-β was detected within both cell lines at the mRNA and protein levels. PDGFR-α was also frequently detected with 5/6 archived tumor samples showing immunoreactivity, but with lower subjective staining intensity and scoring lower on our semi-quantitative immunoreactivity measurements. Both cell lines also expressed PDGFR-α at both mRNA and protein levels. The COF tumor sample with the highest mitotic count and least degree of differentiation was among the tumors with the most intense IHC staining for PDGFR-β and PDGFR-α, consistent with the positive effect of PDGF on cell proliferation. Neither Kit nor VEGFR-2 were detected at the protein level within tumor samples or at mRNA levels within the cell lines. These data indicate that at least two targets of the TKIs used in this study are present within COF tumors and provided a rationale for proceeding with the evaluation of cell viability following TKI treatment.
The in vitro effect of the tested TKIs on CoFSA and MBSa1 cell viability was observed to be relatively similar between the two drugs. These observations are consistent with the shared targets between the two TKIs [15,16]. MBSa1 cells were consistently resistant to either TKI, with significant cytotoxic effects seen only at high concentrations and no significant changes in apoptosis elicited by any drug concentration tested. By comparison, CoFSA cells were similarly resistant to masitinib but slightly more sensitive to imatinib, with a modest but statistically significant reduction in cell viability at ≥ 1 μM concentrations. CoFSA cells also demonstrated significant increases in apoptosis at . Scrambled siRNA sequence used to account for nonspecific, off-target effects. "*" indicates statistically significant (p < 0.05) differences in cell viability compared to vehicle-treated controls. Because siRNA transfection was performed during two independent experiments, statistical analysis for these data was performed using the 6 replicates within one representative siRNA experiment. Visual assessment of cellular appearance in PDGFRB siRNA treated samples (e and h) display apoptotic bodies and marked cellular morphologic deterioration select drug concentrations, consistent with their increased sensitivity to the treatments evaluated in this study. CoFSA cells also showed a significant decrease in relative caspase activity at 30 μM masitinib, which may reflect a paucity of cells with caspase activity due to the extremely low cell viability at this TKI concentration, as supported by our MTS assay results. These relatively minor differences in TKI sensitivity between cell lines may reflect a combination of differences in drug targets between the two TKIs tested along with differences in the cell lines used [14,16]. To the authors' knowledge, no studies have characterized specific receptor tyrosine kinase pathway dependence in either MBSa1 or CoFSA cell lines. Several other in vitro veterinary studies have reported similarly high masitinib or imatinib IC50 concentrations in canine hemangiosarcoma and feline injection site sarcoma cell lines [12,14,23]. In contrast, the cell-based IC50s for masitinib against PDGFR-α, PDGFR-β, and Kit have been reported as 0.3, 0.05, and 0.15 μM, respectively, in an IL3-dependent hematopoietic cell line [24]. These values are markedly lower than the IC50 values seen in the present and prior in vitro veterinary studies, raising the possibility that the observed reductions in cell viability may be due to off-target effects. This is also compatible with the calculated IC50 values from another veterinary study evaluating in vitro masitinib effects on a variety of immortalized canine cancer cell lines [25]. The reason for this relative resistance to single-agent TKI treatment in the cell lines in the present report, as well as in the those cell lines used in the referenced studies, is not fully understood but likely reflects the cell lines' lack of dependence on the targeted pathways for survival [12,14,23,25]. However, as pointed out in previous reports, findings such as described in the current study do not preclude a potential clinical benefit of masitinib therapy either as a single agent targeting tumor-related angiogenesis, or perhaps more importantly, as a potential chemosensitizer [25].
To begin to investigate the potential role of either masitinib or imatinib as chemosensitizers in COF, we performed MTS assays using a range of doxorubicin concentrations, with or without 1.0 μM of either TKI. This concentration of TKI was chosen because pharmacokinetic studies in healthy Beagle dogs have shown that a clinically-relevant oral dose of 10 mg/kg of masitinib results in a serum maximum concentration of 1.3-1.5 μM [26]. Our data support an in vitro synergistic effect of either TKI with doxorubicin in both cell lines, Fig. 8 Cell apoptosis following treatment with imatinib and masitinib, alone or combined with doxorubicin. Graphical plot of relative caspase activity following treatment with escalating concentrations of masitinib and imatinib alone (a and b) and doxorubicin combined with 1.0 μM of either masitinib or imatinib (c and d) on MBSa1 and CoFSA cells, respectively. Caspase activity was assessed using a luminogenic caspase-3/7 substrate assay following treatment with the above drug concentrations after 72 h of incubation and expressed as a percentage of caspase activity within vehicle-treated (DMSO 0.1 %) control cells. COS cells treated with SB2224269 represent a positive control. Plotted values are mean ± standard error of the mean. "*" indicates statistically significant (p < 0.001) difference compared to the vehicle-treated controls indicated by a one-way ANOVA with Dunnett's correction although the effect was modest. The potential for myelosuppression, or other side-effects, may be increased when TKIs are combined with traditional cytotoxic chemotherapeutic agents in vivo, as shown in a study evaluating the safety of toceranib combined with vinblastine in dogs with mast cell tumors [16]. Maximal sensitization factors, potential alterations in drug pharmacokinetics, and alternative methods for determining pharmacologic synergism were not considered as they extend beyond the scope of the present study, but may form the basis for future investigations.
As PDGFR-β was found to be the most uniformly expressed receptor tyrosine kinase for TKI targeting in the present study, we chose to further investigate the contribution of PDGFR-β signaling to viability of COF cell lines. Knockdown of PDGFR-β protein expression via siRNA transfection was successful in both cell lines. The effect of this PDGFR-β protein reduction was associated with a significant reduction in cell viability in both cell lines as well as visibly apparent degenerative changes in cellular morphology. This suggests that PDGFR-β signaling plays a partial role in maintaining viability of COF cells, but the overall significance of this single signaling pathway to COF cell survival requires further investigation.
Although a thorough discussion of the mechanism of action of imatinib and masitinib is beyond the scope of this report, a few select points are worth highlighting. Both TKIs are considered small molecule inhibitors that selectively interfere with specific receptor tyrosine kinase activity (PDGFR-α, -β, and Kit; masitinib also targets Lyn, the FGFR3 and FAK pathways and possibly VEGFR-2 levels) [14,16,27]. Through occupying the receptor's active site, thereby blocking receptor tyrosine kinase phosphorylation, TKIs prevent subsequent activation of downstream pathways. Depending on the cell's dependence on the targeted pathways, this may result in cell death [16]. Some TKIs, such as masitinib, have shown an anticancer action that extends beyond inhibition of its primary targets, and may include disruption of additional signaling pathways associated with tumor progression, metastasis, and chemoresistance [25,26,28,29]. The in vivo tumor microenvironment is characterized by varying levels of hypoxia and acidity, which influence tumor cell behavior and drug sensitivity, potentially rendering them more or less sensitive to TKI treatment [30]. PDGFR is emerging as a key regulator of mesenchymal cells within the tumor microenvironment of many common human malignancies [31]. Blockade of PDGFR signaling has been shown to reduce metastasis in in vivo murine models of colorectal and prostate cancers [32][33][34]. These points serve to illustrate some of the impetus behind this study's investigations into the use of TKIs as a potential treatment strategy, although their applicability to imatinib and/or masitinib treatment of COF remain unknown at this time.
The primary limitations of the present study center on its in vitro nature and the limited experimental methods used. This study tested for receptor tyrosine kinase expression but did not evaluate receptor phosphorylation status (i.e. activation), receptor over-or underexpression, or effect of ligand stimulation on the receptors present. The presence of TKI targets does not necessarily imply their requirement for cell survival or a gain-of-function structural aberration conferring malignant behavior. This study evaluated effects of TKI treatment on COF cell line viability using an MTS assay, supplemented with an apoptosis assay for select drug concentrations, representing only a partial evaluation of potential TKI effects. Examples of additional treatment effects that could be examined in future studies include COF cell migration and/or metastasis. Finally, it is difficult to extrapolate in vitro results of TKI treatment to the far more complex in vivo scenario that includes interactions with the tumor microenvironment and the host immune system.

Conclusions
In conclusion, this study identified expression of PDGFR-α and -β in COF tumor biopsies and cell lines. Treatment with masitinib or imatinib yielded in vitro reductions in cell viability which was enhanced synergistically by the addition of doxorubicin. Furthermore, the tested COF cell lines exhibited partial PDGFR-β dependency for survival. Taken together, these data support further investigation into the potential use of TKIs, potentially in combination with doxorubicin, to augment existing treatment options for COF.

Immunohistochemistry of archived canine oral fibrosarcomas
Medical records from dogs seen at the Oregon State University Lois Bates Acheson Veterinary Teaching Hospital between 2007 and 2011 were searched to identify histologically confirmed COF tumor biopsies. All tumors were comprised of elongate to spindle cells arranged in streams or bundles and whorls that produced variable amounts of collagenous matrix. Four of the tumors had small heterochromatic nuclei, case 1 had medium-sized and case 3 had large euchromatic nuclei. The diagnosis of COF was confirmed by examination of a representative hematoxylin and eosin stained section from each biopsy by a single board-certified veterinary anatomic pathologist (CVL). Serial sections 4-5 μm thick from paraffin-embedded formalin-fixed tumor biopsies were mounted on positively charged slides for IHC analysis of PDGFR-α, PDGFR-β, Kit, and VEGFR-2 expression using anti-human receptor-specific polyclonal rabbit antibodies (detailed in Table 2) [35][36][37].
High temperature antigen retrieval was performed with a microwave pressure cooker using Dako Target Retrieval solution (pH 6, 10 mins) according to the manufacturer's recommendations a . IHC staining was performed on a Dako Autostainer (Dako North America, Carpinteria, CA) at room temperature (21°C) after blocking for 10 mins with 3 % H 2 O 2 (Sigma Laboratories, Santa Fe, NM) in TBST (Biocare Medical, Concord, CA) followed by Dako serum-free protein block (Dako North America, Carpinteria, CA) for 10 mins. The primary antibodies were diluted in Dako antibody diluent (Dako North America, Carpinteria, CA) and applied for 30 mins. Conditions and manufacturer information are detailed in Table 2. Specific antibody binding was detected using MaxPoly-One polymer HRP rabbit (Immu-noBioScienceIH-8064-custom-OrSU, Immuno-BioScience, Mukilteo, WA) for 10 mins followed by Nova Red (SK-4800, Vector Laboratories, Burlingame, CA) for 5 mins. Hematoxylin (Dako North America, Carpinteria, CA) diluted 1:3 in distilled water for 5 mins was used as a counter stain. Washes between steps were performed using TBST (Biocare Medical, Concord, CA), except no wash was performed for the protein block. Dako Universal Negative Control-Rabbit (Dako North America, Carpinteria, CA) was used as the negative control. Peritumoral non-neoplastic tissues were used as internal positive (endothelium or mast cells) and negative (epidermis) controls, a canine cutaneous mast cell tumor submitted as biopsy served as positive control for Kit staining, and a canine metastatic hemangiosarcoma (liver, kidney, testis, spleen) collected during necropsy was used as a positive control [37][38][39]. Evaluation of IHC staining for specificity was performed by a boardcertified veterinary anatomic pathologist (CVL).

Immunohistochemistry scoring
Immunoreactivity for PDGFR-α, PDGFR-β, Kit, and VEGFR-2 were scored by three of the investigators (CVL, MM, and SB) with the results representing a consensus agreement between the observers. The criteria evaluated included: percentage of tumor cells displaying immunoreactivity (assessed from a representative 40× field, after examining the slide in its entirety), the predominant location of staining (cytoplasmic, membranous, or nuclear), and relative visual intensity of staining (+, ++, or +++).
In addition, semi-quantitative measurement of immunoreactivity for the same proteins (using a photomicrograph of the same IHC field as described above) was carried out using a computer image analysis software package (ImageJ v1.47, NIH, Bethesda, MD) as previously described [40]. Output data recorded was the percentage of image pixels above the user-defined threshold to capture immunoreactivity.

Cell lines and reagents
Two immortalized COF cell lines were tested: MBSa1 (provided by Dr. Marlene Hauck, North Carolina State University, Raleigh, NC, USA) and CoFSA (provided by Dr. Melanie Wergin, University of Zurich, Zurich, Switzerland). Both cell lines were derived from biopsies acquired from clinically affected dogs presented for spontaneously arising COF [41,42]. Cells were cultured in RPMI-1640 medium supplemented with 10 % fetal bovine serum, 2 mM glutamine, 2 mM sodium pyruvate, 2 mM HEPES, and 1 % pen-strep in a humidified 5 % CO 2 atmosphere at 37°C.
Masitinib powder (provided by AB Science, Paris, France) was suspended in DMSO and stored at -80°C until use. Imatinib was purchased from a commercial supplier (LC Labs, Woburn, MA), suspended in DMSO and stored at -80°C until use. Doxorubicin HCl (2 mg/ ml) in isotonic solution was purchased from a commercial supplier (Amneal-Agila, Glasgow, KY). Dimethyl sulofoxide concentrations in all experiments never exceeded 0.3 %.

Reverse transcription-polymerase chain reaction
Expression of transcripts for PDGFRA, PDGFRB, KIT, and KDR was assessed in MBSa1 and CoFSA cells using RT-PCR. Cells were seeded into six-well plates (3 × 10 5 / well) suspended in 2.0 mL supplemented medium and allowed to adhere overnight. The cells were rinsed in PBS and RNA was isolated (RNeasy, Qiagen, Valencia, CA) and reverse transcribed to cDNA (High Capacity Reverse Transcription, Applied Biosystems, Foster City, CA) according to the manufacturers' instructions. Targets were amplified from cDNA using the specific

Western blot
Western blots were performed against proteins for which mRNA transcripts were detected via RT-PCR in MBSa1 and CoFSA cells. Cells were seeded in 6-well plates (3 × 10 6 /well) and allowed to adhere overnight as described above. The cells were detached using a cell scraper, transferred to a micro-centrifuge tube, and centrifuged in a tabletop centrifuge (3 min, 1200 × g). Cell pellets were rinsed by re-suspending twice in 3 mL ice cold PBS and extracted in 50 μL ice cold RIPA buffer with protease and phosphatase inhibitor cocktail (Sigma Laboratories, Santa Fe, NM To assess the potential for a chemosensitizing effect of the tested TKIs with doxorubicin, MTS cell viability experiments were performed as described above with doxorubicin either alone or combined with 1.0 μM of either masitinib or imatinib. This concentration of TKI was Three MTS experiments were performed independently and each condition was run in triplicate within each experiment.
The type of interaction between each TKI and doxorubicin was determined using the following equations [44]: SF y < SF tþy < SF t and SF y Antagonistic ¼ SF tþy > SF t or SF y SF t+y = surviving fraction of cells exposed to the combination of either TKI and doxorubicin, SF t = surviving fraction of cells exposed to either TKI alone, SF y = surviving fraction of cells exposed to doxorubicin alone. These equations are appropriate provided the drug effect is reduced cell viability [45], a condition that was met at all drug concentrations except at 3 and 10 nM doxorubicin in CoFSA where the mean cell viability values for doxorubicin treatment alone were slightly increased. In these exceptions, the SF of the control group (i.e. cells, but no doxorubicin or TKI) was substituted in the equation for the SF of doxorubicin, as it was the more stringent condition for evaluating drug interaction.
All MTS testing was performed during three independent experiments, with each condition run in triplicate during each experiment.
Inhibition of PDGFR-β expression by siRNA PDGFR-β knockdown was used to assess the role of PDGFR-β in sustaining viability of COF cell lines MBSa1 and CoFSA. Cells were seeded into 96-well plates (5 × 10 3 cells/well) and allowed to adhere overnight before transfection with a combination of four commercially purchased target-specific siRNAs (Life Technologies, Carlsbad, CA; 3 pmol per reaction) against canine PDGFRB (Dharmacon, Lafayette, CO) or a scrambled sequence (Mission siRNA Universal Negative Control #1, Sigma Laboratories, Santa Fe, NM) using Lipofectamine RNAiMAX (Invitrogen, Carlsbad, CA) according to the manufacturer's directions. Cell viability was measured in 96-well plate format 72 h after siRNA transfection using the MTS colorimetric assay (CellTiter 96 Aqueous One Solution Cell Proliferation Assay, Promega, Madison, WI) as described above.
The PDGFRB siRNAs used were (5′-3′): Sequence #1: CCUUCAAGGUGGUGGUGAUTT Sequence #2: CCAUGAACGAACAGUUCUATT Sequence #3: GAAAUGAGGUGGUUAACUUTT Sequence #4: GAAUGACCAUCGAGAUGAATT All siRNA data were normalized to the readouts taken from control cells treated with the transfection reagent alone, and included scrambled sequence controls to assess for nonspecific, off-target effects. Results from siRNA knockdown reported are representative of two independent experiments, with each condition run in sextuplicate during each experiment.
Western blot was repeated after siRNA treatment, as described above, using only the anti-PDGFR-β antibody(BioGenex, Fremont, CA). Quantification of siRNA PDGFR-β protein knockdown was performed using computer image analysis software with a gel analysis package (ImageJ v1.47, NIH, Bethesda, MD). PDGFR-β signal intensities in each lane were expressed as percentage of the total protein control (actin) loaded into respective lanes. Final PDGFR-β protein knockdown was reported as percentage of actin-normalized PDGFR-β band intensity in the siRNA treated lane relative to actin-normalized PDGFR-β band intensity in the control (vehicle-treated) lane.

Apoptosis
A luminogenic caspase-3/7 substrate assay (Caspase-Glo 3/7, Promega, Madison, WI) was used to determine relative caspase activity in COF cell lines after TKI treatment, either alone or with doxorubicin. The drug concentrations tested were selected based on MTS results in order to represent the range of observed effects on cell viability. Cells were seeded in 96-well culture plates at a density of 2,500 cells/well and challenged with the following drugs (concentrations): masitinib (0, 1, 3, and 30 μM), imatinib (0, 1, 3, and 30 μM), masitinib 1.0 μM + doxorubicin (3 and 300 nM), and imatinib 1.0 μM + doxorubicin (3 and 300 nM). For an apoptosis positive control, COS cells were incubated with 6.25 μM SB2224289 (Tocris, Bristol, UK) [46]. For a vehicle control, cells were incubated with 0.1 % DMSO in media. Drugs were dissolved in DMSO, with cells exposed to a final vehicle concentration of 0.1 %. Cells were challenged for 72 h, after which the caspase-3/ 7 activity was quantified. The caspase activity assay was performed according to the manufacturer's protocol. Cumulative luminescence over 1 s was measured using a luminometer (GloMax 96 Microplate Luminometer, Promega, Madison, WI). Relative caspase activity was calculated using the formula: relative caspase activity = (mean luminescence of treated cells)/(mean luminescence of vehicle control) × 100.