TiHo-0906: a new feline mammary cancer cell line with molecular, morphological, and immunocytological characteristics of epithelial to mesenchymal transition

Feline mammary carcinomas (FMCs) with anaplastic and malignant spindle cells histologically resemble the human metaplastic breast carcinoma (hMBC), spindle-cell subtype. hMBCs display epithelial-to-mesenchymal transition (EMT) characteristics. Herein we report the establishment and characterization of a cell line (TiHoCMglAdcar0906; TiHo-0906) exhibiting EMT-like properties derived from an FMC with anaplastic and malignant spindle cells. Copy-number variations (CNVs) by next-generation sequencing and immunohistochemical characteristics of the cell line and the tumour were compared. The absolute qPCR expression of EMT-related markers HMGA2 and CD44 was determined. The growth, migration, and sensitivity to doxorubicin were assessed. TiHo-0906 CNVs affect several genomic regions harbouring known EMT-, breast cancer-, and hMBCs-associated genes as AKT1, GATA3, CCND2, CDK4, ZEB1, KRAS, HMGA2, ESRP1, MTDH, YWHAZ, and MYC. Most of them were located in amplified regions of feline chromosomes (FCAs) B4 and F2. TiHo-0906 cells displayed an epithelial/mesenchymal phenotype, and high HMGA2 and CD44 expression. Growth and migration remained comparable during subculturing. Low-passaged cells were two-fold more resistant to doxorubicin than high-passaged cells (IC50: 99.97 nM, and 41.22 nM, respectively). The TiHo-0906 cell line was derived from a poorly differentiated cellular subpopulation of the tumour consistently displaying EMT traits. The cell line presents excellent opportunities for studying EMT on FMCs.

Cytomorphologic features of the cell line. After cell culture establishment, TiHo-0906 cells proliferated in a monolayer. Subconfluent cells were spindle-shaped with a bipolar to multipolar morphology (Fig. 2a). At confluence, cells revealed a more epithelioid morphology with polygonal to bipolar shape (Fig. 2b). Cellular morphology and proliferation rate remained stable during prolonged subculturing (over 100 passages). Microscopic examination of TiHo-0906 cells in H&E stained paraffin sections from pellets revealed round to polygonal cells and a moderate to partly abundant cytoplasm. They had large, prominent, round to oval nuclei that showed moderate to severe atypia and contained coarsely stippled to vesiculated chromatin and distinct nucleoli. TiHo-0906 cells were further characterized by marked anisokaryosis and anisocytosis, atypical mitotic figures and several large multinucleated syncytia (Fig. 2c).
Tumour immunophenotyping. Tumours displaying EMT features usually express epithelial and mesenchymal markers concurrently 7 . In general, a co-expression of epithelial markers and Vim was observed in tubular and solid anaplastic areas of the tumour. In contrast, spindle cells were negative to all epithelial markers and positive to all mesenchymal markers used. Additionally, all parts of the tumour were negative to the specific myoepithelial markers cytokeratin 5/6 (CK5/6) and p63 (Table 1).
Tubular epithelial cells were strongly immmunoreactive to the luminal epithelial marker E-cad (Fig. 3a), while polygonal and spindle cells were negative (Fig. 3b,c). Tubular epithelial cells and polygonal cells were positive for the general epithelial marker pan-cytokeratin [pan-CK] while spindle cells were negative (Fig. 3d-f). A positive reaction to Vim was observed in all areas of the tumour (Fig. 3g-i).
HMGA2 elicits EMT by activating a series of transcription factors 24,25 , while CD44 participates in the downregulation of E-cad commonly observed during EMT 12,28 . Immunohistochemically, there were no specific cross-reactivity of the HMGA2 selected antibody with feline tissue. On the other hand, tubular epithelial cells were negative for CD44 (Fig. 3j), while numerous CD44-positive polygonal and spindle tumour cells were observed (Fig. 3k,l).
Tubular epithelial cells were slightly positive to oestrogen and progesterone receptors (ER and PR) and negative to the human epidermal receptor-2 (HER-2). Polygonal cells were negative to ER and PR, but slightly positive to HER-2. Whereas, well-differentiated spindle cells were negative for all hormonal markers evaluated. The role of cyclooxigenase-2 (COX-2), p53, claudin 2 (CLDN-2), and ki67 as indicators of tumor aggressiveness and prognosis has been documented in FMCs [40][41][42][43][44] . Moreover, loss or reduced expression of claudins, in general, may contribute to the development of EMT [45][46][47] . In this case, the tumour was negative to COX-2. The expression of p53 was weak in tubular and polygonal cells, whereas well-differentiated spindle cells were negative. CLDN-2 was  slightly positive in tubular and polygonal cells; spindle cells were negative. Additionally, a high Ki-67 proliferation index was observed in all areas of the tumour 44 . Results of the immunohistochemical characteristics of the tumour are detailed in Table 2.
Cell line phenotyping. TiHo-0906 cells at low and high passage reacted similarly to the markers employed; all pellets evaluated coexpressed epithelial and mesenchymal markers ( Table 3). The immunohistochemical characteristics of the cell line resembled those of the polygonal epithelial cells present in the anaplastic areas of the tumour with exception of E-cad that was positive in most TiHo-0906 cells at low (Fig. 4a) and high passages (Fig. 4b). TiHo-0906 cells at low (Fig. 4c) and high passages (Fig. 4d) were positive for pan-CK, and Vim (Fig. 4e,f). Interestingly, some of the pellets evaluated in both low-and high-passaged cells were positive for the myoepithelial marker p63 (Fig. 4g, low-passaged cells), while the tumour was negative. An intense membrane staining for CD44 was observed on almost 100% of the cells in all pellets evaluated (Fig. 4h,i, low and high passages respectively). TiHo-0906 cells were slightly positive to the hormonal receptors (ER, PR), and HER-2. COX-2 labelling was negative in all of the pellets evaluated. p53 was moderately expressed in all of the pellets evaluated. CLDN-2 was weakly expressed in all pellets (score 1 to 3) 42 . The mean Ki-67 proliferation index of TiHo-0906 cells in pellets was 39.3% and 42.2% in early and high passages, respectively.
Copy number variation analysis. TiHo-0906 cells at low (P7) and high (P79) passages showed a high level of genomic instability when compared to the original tumour. The frequency of CNVs in TiHo-0906 cells increased during in vitro culturing. FCAs B2, B3, B4, and F2 in TiHo-0906 cells displayed most of the genomic regions with CNGs (Fig. 5).
CNGs in chromosomes B2, B4 and F2 remained similar during subculturing. Interestingly, chromosomes B4 and F2 harbour some of the previously reported EMT-implicated genes associated with more frequent CNGs in different types of human cancer 36 (Table 4).
In general, the amount of CNLs also increased during cultivation. High-passaged TiHo-0906 cells showed losses on proximal chromosome A1, distal chromosome A2 and a focal deletion on chromosome E1 that were not observed in low-passaged cells. However, low-passaged TiHo-0906 cells showed a focal deletion of proximal chromosome B1 that was not observed in high-passaged cells. Additionally, the size of CNLs observed in chromosome E3 decreased during sub-cultivation. Some of the most important breast cancer-related genes in humans are located in the analysed corresponding feline regions 7,32,33 showing amplifications as well as deletions (details in Table 5).
Real-time PCR expression analyses of HMGA2 and CD44. The levels of HMGA2 and CD44 in TiHo-0906 cells were examined by absolute RT-PCR and compared to the corresponding reference tissues (feline testis, and healthy mammary feline tissue, respectively). The level of HMGA2 expression in low and high passaged cells was higher than those of the reference tissue (p = 0.001, and p < 0.0001, respectively). Similarly, the CD44 absolute expression in TiHo-0906 cells at low and high passages was significantly higher in comparison to the reference tissue (p = 0.01, and p = 0.0015, respectively). No significant differences in HMGA2 and CD44 expression were observed between low and high passages (Fig. 6).
Growth behaviour and migration activity. No statistical differences in BrdU incorporation and growth curves were found between low and high passages of the TiHo-0906 cell line (Fig. 7a,b). The doubling time for the cell line was 28.9 h at low passage and 27.4 h at high passage. Fourteen hours after the scratching, migrating cells completely covered the wound (400 µm) in early and high passages of the TiHo-0906 cell line ( Fig. 7c-j). No significant differences regarding the time to wound closure were observed between low and high passages.  Metabolic activity of TiHo-0906 cells after doxorubicin treatment. According to the MTS assay results, the metabolic activity of doxorubicin-treated TiHo-0906 cells at low and high passage starts to significantly decrease at 100 nM of doxorubicin (p = 0.01, and p = 0.03, respectively), for details see Fig. 8a. After incubation with different concentrations of doxorubicin, the metabolic activity of TiHo-0906 cells at low and high passages was not statistically different. However, the IC50 of cells at low passage was approximately 2-fold higher than the IC50 of cells at high passage (Fig. 8b,c).  Doxorubicin induces dose-dependent apoptosis in TiHo-0906 cells at low and high passages, ranging from 5.2% (SD, 1.4%), and 4.6% (SD, 0.7%) at 1 nM doxorubicin to 45.1% (SD, 10.6%), and 65.9% (SD, 1.6%) at 2000 nM doxorubicin. The amount of apoptotic cells was significantly higher in high-passaged cells at all doxorubicin concentrations up to 50 nM (Fig. 9c). Surprisingly, incubation with doxorubicin reduced the amount   of viable cells while increasing that of apoptotic cells but was not able to significantly increase the amount of dead cells. Nonetheless, part of the cell debris excluded from viability gating could represent fragments of dead cells. Incubation with 50 nM doxorubicin induced the highest amount of dead cells observed (5.9%; SD, 1.6%). Comparing low and high passages, the amount of dead cells was only higher in cells at high passage incubated with 2000 nM doxorubicin (Fig. 9d).

Discussion
Epithelial to mesenchymal transition can be observed in vitro and has been immunohistochemically portrayed in mammary cancer in dogs and humans [48][49][50] . The importance of EMT on FMCs has not been evaluated, and to the extent of our knowledge, EMT-associated genomic aberrations have not been documented in an FMC cell line. TiHo-0906 cells displayed a combined epithelial/mesenchymal phenotype and a high expression of EMT-related markers HMGA2 (qPCR) and CD44 (qPCR and immunohistochemistry). Moreover, the cell line   The cell line was derived from a tumour composed of malignant tubular-epithelial cells, malignant spindle cells, and small anaplastic areas with polygonal cells. Immunohistochemically, tubuli-forming areas showed a higher affinity for CKs while spindle areas were characterized by a mesenchymal profile. Interestingly, polygonal cells in anaplastic areas and TiHo-0906 cells co-expressed CKs and mesenchymal markers. In human breast cancer, this is associated with poor prognosis 27,55,56 , in canine and human cellular models, with malignancy 38,55 .
TiHo-0906 cells displayed an increasing genetic instability during sub-cultivation. Genomic aberrations observed in the cell line were not present in the tumour. This may be related to the fact that DNA isolated from the tumour represents a genomic mixture of all cellular subpopulations included in the sample 19 , whereas the in TiHo-0906 cells at low (P8) and high (p80) passage versus selected reference tissues (feline testis, and healthy mammary feline tissue, respectively). Data are displayed as mean (SD); *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. cultivated cells show clonal selection favouring a genomic shift. TiHo-0906 cells protein expression profile was more similar to that of the anaplastic polygonal cells in the tumour. This sub-population was, in fact, the least frequent of the three main cellular components of the tumour. However, considering that during the process of cell line establishment, cells with the higher rate of proliferation are likely to have a better survival chance in culture, the polygonal tumour cells may represent the cellular subpopulation with the faster growth rate and the higher amount of somatic aberrations. These findings also highlight the importance of cellular models as unique tools to study distinct aspects of the oncologic process specifically.
The tumour described in this study may represent a different histological subtype "FMCs with anaplastic and malignant spindle cells" from those included in the last WHO classification 2 . Matsuda et al. 57 reported a  "tubulopapillary carcinoma" with spindle cells (CKs+, and SMA+) in a cat 57 . In contrast, the spindle-cell component in the tumour in this study was negative for CKs but positive for SMA. Paniago et al. 58 reported a "feline mammary carcinosarcoma" mainly composed of malignant pleomorphic cells, but no spindle cells were described 58 . Similar to the survival of the patient described in this study (three months), the animals reported by Matsuda et al. 57 and Paniago et al. 58 developed pulmonary metastasis and died three and two months after diagnosis, respectively 57,58 .
Similar to the tumour described in this study, spindle-hMBCs are characterized by a dominant fibromatosis-like pattern with epithelial cells and limited regions of polygonal cells arranged in clumps or cords 59,60 . Immunohistochemically, spindle-hMBCs coexpress CKs (cytokeratin 8/18 [CK8/18], pan-CK, cytokeratin 14 [CK14], and CK5/6) and mesenchymal markers (SMA and Vim) 59,60 . Spindle cells in these tumours are malignant and display a variable reaction to CKs [61][62][63] . Therefore, a wide spectrum keratin panel (including pan-CK) may be useful for differentiating spindle-hMBCs from other spindle-cell lesions 60 . In the tumour described in this study, the spindle cells expressed SMA, CALP and Vim but lacked CKs including pan-CK, which according to the aforementioned would suggest a more likely mesenchymal phenotype. Since the tumour and derived cell line were enriched with EMT traits, and both, polygonal and spindle cells in the tumour expressed CD44, it appears reasonable to consider that both components may have arisen from EMT. However, we cannot completely rule out the presence of some non-neoplastic reactive mesenchymal cells (i.e. myofibroblasts) within the spindle component of the tumour. Considering the presence of both epithelial and mesenchymal components, carcinosarcoma would represent another possible differential diagnosis 64 . In dogs, mammary carcinosarcomas (commonly carcinoma and osteosarcoma) are described to express CK8/18 and Vim, but not CALP, p63 and CK14 64 . Although this expression profile has not been demonstrated in cats so far, a mammary gland carcinosarcoma is very unlikely in the present case on the basis of histological and immunohistochemical findings. Nonetheless, whether the spindle cells were originated from EMT or represented an additional neoplastic mesenchymal population remains at this point unanswered. In summary, the tumour described in this study shows some similarities with the spindle-hMBC, and it also resembles the FMC cases described by Matsuda et al. and Paniago et al. 57,58 .
HMGA2 induces EMT 7,[22][23][24][25][26]65,66 . In this study, we compared the absolute HMGA2 qPCR-based expression of TiHo-0906 cells with a well-known reference tissue (testis) characterized by high expression of HMGA2 67 . The quantitative expression of HMGA2 was significantly higher in TiHo-0906 cells when compared to the reference tissue. These findings are concordant with the CNGs observed in chromosome B4 harbouring HMGA2, which in combination with the characteristic fibroblastoid-shape and the highly migratory ability (demonstrated by in vitro scratching assay) also reveal a profoundly altered mesenchymal gene expression program; consistent with the combined epithelial/mesenchymal phenotype observed.
hMBCs coexpress EMT-and CSCs-features 12 . Accordingly, these tumours may originate directly from epithelial cells that have undergone EMT and have therefore acquired CSCs properties, or from pre-existing CSCs expressing EMT-associated markers 8 . In FMC and human breast cancer cell lines, CSCs express high levels of CD133, CD44 and low (or none) CD24 30,68,69 . In this study, the immunohistochemical expression of CD24 and CD133 was not assessed due to the unavailability of the specific antibodies. However, the fact that only some TiHo-0906 cells expressed p63 when the original tumour was negative, may suggest the presence of some cells with stem cells features which may have acquired a myoepithelial phenotype during sub-cultivation as previously described in a human mammary gland derived cell line 70 . All cell types in the tumour expressed CD44, and there was a specific intense membrane staining for CD44 on almost all TiHo-0906 cells. The quantitative gene expression of CD44 was also significantly higher in the cell line than in the reference tissue. These findings together with the comparative expression profile and HMGA2 expression also suggest that the cell line was originated from the anaplastic polygonal cells in the tumour.
Several well-described breast cancer-and EMT-associated genes were affected by CNVs in the genome of the cell line. As reported by Zhao et al. 36 in a study exploring the relationship between EMT-implicated gene expression and CNVs in multiple cancer types in humans 36 ; we observed that EMT-associated genes were more frequently affected by CNGs than by CNLs. As was mentioned before, most of the EMT-related genes affected by CNVs were located in amplified regions of FCAs B4 and F2. Affected regions of chromosome F2 harboured four of the most commonly EMT-associated genes reported by Zhao et al. 36 ; this includes ESRP1, MTDH, YWHAZ, and MYC 36 . Among those, ESRP1 regulates CD44 alternative splicing producing a more malignant isoform (CD44v), promoting invasiveness and distant metastasis in human breast CSCs 71 . TiHo-0906 cells overexpress CD44 and displayed CNGs in chromosome F2 harbouring ESRP1 gene. Therefore, TiHo-0906 cell line might be a suitable model for the study of ESRP1 and CD44 as possible therapeutic targets for FMC.
CNGs in chromosome F2 observed in the TiHo-0906 cells resemble one of the most common DNA aberrations reported in human breast cancer cell lines-CNGs at the long arm of human chromosome 8 (HSA 8q)- 35,72 . When comparing the human and feline karyotypes using fluorescence in situ hybridization (FISH), the entire HSA 8q is hybridized by the paint from FCA F2 73,74 . CNGs in chromosome F2 observed in the cell line also included some proposed breast cancer-related genes like EIF3E, PKHD1L1, CSMD3, ZHX2, SAMD12, EXT1, MRPL13, MTBP, SNTB1, RNF139, and TATDN1 35 . These findings highlight the importance of FCA F2 in FMC development and the close evolutionary relationship with the homologous disease in humans. Additionally, TiHo-0906 cells at low and high passage also displayed CNGs in regions harbouring other known breast cancer-related genes like FOXO3 and MYB (FCA B2), AKT1 (FCA B3), and GATA-3, CCND2, CDK4 and PFDN5 (FCA B4).
TiHo-0906 proliferation and migration activity remained comparable during subculturing. Growth curves and doubling times (28.9 h at low passage and 27.4 h at high passage) revealed a high proliferation similar to those reported for previously characterized FMC cell lines 39,[75][76][77] . Migration activity was assessed by in vitro scratch assay revealing a shorter time to complete wound closure (approximately 14 h for both, early and high passages) than that previously reported in a recently characterised FMC cell line (approximately 24 h) 39 .
After doxorubicin treatment, we observed a significant dose-dependent reduction of viable cells with active metabolism only after 72 h incubation with 100 nM of doxorubicin at low and high passages. These data may suggest a higher doxorubicin resistance of TiHo-0906 cells to that reported in an FMC cell line 78 . Nonetheless, methodological differences in viability assessment and exposure time to doxorubicin make comparison difficult. Resistance to doxorubicin, in terms of metabolic activity remained comparable during subculturing. However, flow cytometric assessment of doxorubicin effects on TiHo-0906 cells revealed a significant increase of apoptotic cells in high-passaged cells when compared to low-passaged cells at all doxorubicin concentrations up to 50 nM (except for 1000 nM). These findings are consistent with the observed two-fold reduction of the IC50 during sub-cultivation, suggesting a reduction in chemoresistance during prolonged subculturing. Similar findings have been reported for different cellular models in humans 79 . However, the specific mechanisms underlying this process are still not clear. To the best of our knowledge, resistance to doxorubicin has not been previously assessed by flow cytometry in FMC-derived cell lines. In this case, we observed a dose-dependent reduction of vital cells with simultaneous increasing of apoptotic cells.
Some immunohistochemical studies in veterinary medicine support the EMT hypothesis to explain the origin of the spindle-cell component on canine mammary metaplastic tumours 48,49 . There is also evidence suggesting a direct CSCs origin in cell lines derived from canine mammary spindle-cell tumours and sarcomas 80 . Although these two theories have been considered separately, a growing body of literature in human medicine suggests a link between them 8,13,14,50 . In vitro culture models are pivotal to understanding the role of EMT in the oncogenic process of FMC, determining the presence of possible subtypes and developing of new therapeutic approaches. However, the number of established FMC cell lines is still small 30,39,[75][76][77][78][81][82][83] .
In summary, the present study demonstrated that TiHo-0906 cells co-expressed epithelial and mesenchymal features, and some EMT markers like HMGA2 and CD44. Additionally, TiHo-0906 cells are characterized by stable metabolic activity, proliferation rate, and ability to migrate during prolonged subculturing. Furthermore, our results indicate that specific CNVs observed in the cell line harbour a considerable amount of genes implicated in breast cancer development and EMT. CNGs in chromosomes B4 and F2 may be essential for EMT activation in the cat genome and closely resemble some of the typical genomic aberrations observed in human breast cancer. Nonetheless, further studies are necessary to determine the importance of these genomic aberrations in FMCs. The TiHo-0906 cell line represents a unique model consistently displaying EMT-associated traits at baseline. Considering the poor prognosis of affected animals, further studies using in vitro models that closely resemble the uniqueness of the original tumours are of major value for the development of potential therapeutic approaches.

Methods
Primary tumour tissue. The TiHo-0906 cell line was obtained from the primary lesion of a 13-year-old female intact British Shorthair cat diagnosed with spontaneous FMC at clinical stage three according to McNeill et al. 84 . The cat was admitted to the Small Animal Clinic at the University of Veterinary Medicine Hannover and underwent unilateral chain mastectomy. No chemotherapy was performed and the cat was euthanised three months after surgery due to pulmonary metastasis. Tissue samples were obtained during the medically necessary surgery, after signed informed owner consent. Consequently, an ethical approval was not required (German Animal Welfare Act, § 7).
Cell culture establishment and maintenance. Immediately after surgery, a tumour tissue sample was placed in a tube with Hank's Balanced Salt Solution (Gibco) and kept overnight at 4 °C. The remainder of the surgical specimen was fixed in 10% neutral-buffered formalin and prepared for histopathological examination. The selected tumour tissue was minced into small pieces and digested with 4 mL 0.26% collagenase (SERVA) for 3 h at 37 °C with constant stirring. The digested cells were transferred into a sterile 25 cm 2 cell culture flask containing 5 mL medium 199 (Gibco) supplemented with 20%foetall bovine serum (FBS, Biochrom) and 2% penicillin/streptomycin (Biochrom). The culture was incubated at 37 °C in a humidified atmosphere of 5% CO 2 and split at confluency once or twice a week utilizing 1 mL TrypLE TM (Gibco) as dissociation reagent. Cryogenic preservation was performed once a week at 90% confluence. Cells were dispersed in TrypLE TM and resuspended in growth medium (medium 199 supplemented with 10% FBS, and 2% penicillin-streptomycin). The cell suspension was centrifuged at 1000 g for 10 min, the supernatant was discarded and the cell pellet was resuspended in 1 mL freezing medium (medium 199 supplemented with 20% FBS, 2% penicillin-streptomycin and 10% DMSO [AppliChem]) and transferred into cryogenic storage vials. Cryogenic vials were accommodated in a freezing container (Mr. Frosty, Nalgene), and transferred to a −80 °C freezer, after 24 h the vials were stored in a −150 °C freezer or in a liquid nitrogen storage vessel. TiHo-0906 was considered established when overcoming the passage 30 th (P30).
All experiments described below were carried out in triplicate; those involving cells were performed using TiHo-0906 cells, or cell-pellets (5 × 10 6 cells) at low and high passage (P7-P10, and P76-P80, respectively). For pellets preparation, cells were detached with 1 mL TrypLE TM , rinsed with PBS, counted with a PC-based Cellometer (Auto T4, Nexcelom Bioscience), and centrifuged at 1000 g for 10 min. For DNA and RNA isolation, pellets were stored at −80 °C. For cytology and immunohistochemistry, pellets were fixated in 10% neutral-buffered formalin and stored at room temperature.
Histopathology and cytology. After formalin fixation, paraffin sections (4 µm) of the original tumour were stained with H&E for histopathologic evaluation. The samples were examined under light microscopy and the morphological diagnosis was performed following the WHO classification 2 . Histological grading of the tumour was performed according to the method described by Elston and Ellis 1991 and Castagnaro et al. 85,86 . The cytomorphologic features of the cell line during long-term cultivation were examined microscopically based on H&E stained formalin-fixed, paraffin-embedded cell pellets at low (P7) and high (P80) passages.

Immunohistochemistry.
Immunophenotyping by the avidin-biotin method was performed in the original tumour and cell pellets of TiHo-0906 (low [P7], and high [P80] passages) as described elsewhere 46 using the following markers (details see Table 6): E-cad, CK8/18, pan-CK, CK14, CK5/6, p63, SMA, CALP, Vim, HMGA2, CD44, ER, PR, HER-2, COX-2, p53, CLDN-2 and proliferation marker Ki-67. Positive and negative controls used for immunohistochemistry are listed in the Supplementary Table 1. For negative controls, the specific primary antibody was replaced by either an isotype control antibody or normal serum. For positive controls, only tissue samples were used that were sent to the department of pathology for diagnostic purposes not related to the present study. No animal was killed for the generation of positive controls.
Copy number variation analysis. Four 10-µm-thick sections were sliced from the tumour FFPE-block, using a microtome (pfm Slide 2003). Afterwards, the sections were deparaffinized and the DNA was isolated with the AllPrep DNA/RNA FFPE kit (QIAGEN) according to manufacturer's instructions. Before DNA isolation, 4 × 10 6 TiHo-0906 cells (P7 and P76) were homogenized using QIAshredder TM columns (QIAGEN). DNA isolation was performed with the AllPrep DNA/RNA Mini Kit (QIAGEN) following the manufacturer's protocol. DNA yields and purity (260/280 ratio) were quantified with the Synergy 2 plate-reader (BioTek).
Starting from the DNA a sequencing library for each sample was prepared using the NEBNext ® Ultra TM II DNA Library Preparation Kit for Illumina ® (New England Biolabs). Before library preparation, the DNA was sheared by ultrasound to an approximate size of 500 bp. An input of 500 ng DNA was used for each library. Sequencing was conducted on an Illumina NextSeq500 according to manufacturer's instructions. The program CNV-seq (http://bmcbioinformatics.biomedcentral.com/articles/10.1186/1471-2105-10-80) was used for copy-number analyses; using feline healthy mammary tissue as normal reference control and a fixed window size of 1,000,000 bp. Control-tissue was collected from a recently euthanized intact female cat with no pathologies of the mammary gland. The raw results (log2ratios) were smoothed using a circular binary segmentation algorithm implemented in the program "Copynumber" (http://bioconductor.org/packages/release/bioc/html/copynumber. html). Based on the results of the normal reference control, regions with log2-copy-number ratios of >0.25 or <−0.25 were scored as significantly aberrant. Additionally, some bins at telomeres were removed. All bins with smoothed log2-copy-number ratios of >0.25 were scored as amplified (CNGs) and all bins with <−0.25 were scored as deleted (CNLs).
Structural rearrangements (CNGs, and CNLs) observed were compared with those reported in human breast cancers 33,34 , and were complemented by important EMT-related CNVs reported in multiple human cancer types 36 , specific hMBCs-related CNVs 7,32 , breast cancer cell lines-related CNVs 35 , and additional cancer-associated genes reported on FMCs.  Real-time PCR expression analyses of HMGA2 and CD44. HMGA2 expression is nearly undetectable in adult tissues except for meiotic and post-meiotic cells, in which its expression is high 67,87 . Consequently, we used feline testis as reference tissue. CD44 is overexpressed in metastatic mammary cancer tissues and CSCs, while its expression in the normal mammary gland is very low [88][89][90] . Accordingly, we used healthy mammary tissue as reference tissue. Testicular tissue was collected from a healthy tomcat during elective orchiectomy and mammary tissue from a recently euthanized (for medical reasons) intact female cat with no pathologies of the mammary gland. Total RNA was isolated from TiHo-0906 cells (P8 and P80), fresh-frozen feline testicular tissue (HMGA2 reference) and non-neoplastic feline mammary tissue (CD44 reference) using the RNeasy Mini Kit (QIAGEN) according to manufacturer's protocol. The RNA yield and purity (260/280 ratio) were determined using a plate reader (Synergy 2, BioTek). The RNA was stored at −80 °C until use.
HMGA2 and CD44 mRNA levels were measured using absolute quantification, relative to amplicon-specific standard curves. HMGA2 standard curve sequence and amplification primers were based on a custom canine assay 91  Real-time PCR was carried out using the Mastercycler ® ep realplex Real-Time PCR System (Eppendorf). The qPCR reactions were performed using the QuantiTect SYBR ® Green RT-PCR Kit (QIAGEN) in a total reaction volume of 20 µL and a primer concentration of 1 µL of 10 µM stock, for each primer. 100 ng of isolated total RNA were used per sample to measure mRNA copy numbers. The standard curves resulted from seven dilution steps, ranging from 10 2 to 10 8 copies. PCR conditions were as follows: 50
For calculating doubling time and growth curves, TiHo-0906 cells (P10 and P79) were seeded (7.5 × 10 5 cells/well) in 6-well plates in triplicates and incubated at 37 °C in a humidified atmosphere of 5% CO 2 for four days. Every 24 h, three replicative wells were dissociated with 1 mL TrypLE TM and the cells were counted using a PC-based Cellometer Auto-T4 (Nexcelcom Bioscience). The growth curves were established, and the doubling times were calculated using the "Doubling Time" online tool (http://doubling-time.com/compute.php).
TiHo-0906 cells migration ability was evaluated using the in vitro wound-scratching assay as previously reported 39 . Cells at low (P7) and high (P76) passages were cultured in 6 well culture plates in triplicates, once the cell confluence reached 100%, the growth medium was removed, the cell monolayer was gently scratched using a 200 µL pipette tip and rinsed with PBS, and the growth medium was replaced. Afterwards, cells were photographed every 10 min during 24 h using a live cell imaging microscope (DMI 6000 B, Leica Microsystems). Time to wound closure was defined as the time at which the wound was completely filled with cells.
Sensitivity to doxorubicin. The anti-proliferative effects of doxorubicin on TiHo-0906 cells were evaluated by assessing the cells metabolic activity with the CellTiter 96 ® MTS Proliferation Assay (Promega). Cells at low (P7) and high (P79) passages were seeded in 96-well plates (7.5 × 10 3 cells/well) and incubated for 18 h. The medium was aspirated and the cells were treated with different dilutions of doxorubicin (1 nM, 10 nM, 100 nM, 200 nM, 1000 nM, and 2000 nM [200 μL/well]). These concentrations of doxorubicin were below the range of the in vivo maximum plasma concentration in cats treated with clinically relevant dosages 92 . Each concentration was added in quadruplets, every plate contained a control of cells in growth medium and a negative control of the medium. After 72 h of incubation, the medium was changed, and 20 μL of the CellTiter 96 ® solution were added to each well. The plates were incubated at 37 °C in a humidified atmosphere of 5% CO 2 for 2 h and the absorbance values were recorded (reference wavelength 492 nm) and normalized to the medium negative control using a plate reader (Synergy 2, BioTek). The inhibitory activity of doxorubicin on TiHo-0906 cells at low and high passage was assessed by dose-response curves and IC50 using GraphPad Prism 3.0 (GraphPad Software Inc., San Diego, CA, USA), as calculated from the data on metabolic activity.
Apoptosis induced by doxorubicin. TiHo-0906 cells (P7 and P77) were seeded in 6-well plates (2 × 10 5 cells/well) and incubated for 18 h at 37 °C in a humidified atmosphere of 5% CO 2 . Afterwards, the cells were exposed to different dilutions of doxorubicin (1 nM, 10 nM, 50 nM, 100 nM, 200 nM, 500 nM, 1000 nM, and 2000 nM [3 mL/well]) and then re-incubated for 72 h. After this period, media containing non-adherent and dead cells were collected. Adherent cells were dissociated with 1 mL TrypLE TM and centrifuged (1000 rpm for 6 min) together with the previously collected media. The supernatant was discarded, and cell pellets were resuspended in 500 μL of 1X Binding Buffer. Cell suspensions were incubated (10 min at room temperature in the dark) with 5 μL Annexin V-FITC and 1 μL SYTOX Green Dye (Annexin V-FITC Detection Kit plus, PromoCell). Flow cytometry was performed using a MACSQuant ® Analyzer 10 (Miltenyl Biotec). Dead cells treated with 1 mL 0.2% Triton TM X-100 (Sigma-Aldrich) and non-treated viable cells were used to set the gates. Annexin V-FITC and SYTOX Green Dye were measured in the FL-1 channel, and results were analysed with FlowJo Version 7.6.5 (FlowJo, Ashland, OR, USA). For data analysis, cell debris on the left-side of the plot were set as negative gate. Afterwards, intact cells were gated for viability on three different populations: viable, apoptotic, and dead. Statistical analysis. All statistical analyses were performed using SAS software 7.1 (SAS Institute Inc., Cary, NC, USA); the significance threshold was set at p ≤ 0.05. Data distribution was tested using Shapiro-Wilk test. Statistical analysis of BrdU cell proliferation test, growth curves, and real-time PCR expression analyses of HMGA2 and CD44 were performed using a 2-tailed Mann-Whitney-U test. Times to in vitro wound-scratch closure were analysed by two-tailed t-test. Results from MTS cell proliferation test and flow cytometry were evaluated using Wilcoxon's two-sample test.