Development of mouse models of malignant phyllodes tumors by transplantation of syngeneic mammary gland cells expressing mutant H‐Ras

Phyllodes tumors (PTs) are rare fibroepithelial tumors of the breast with epithelial and stromal components, and surgical resection is the standard and only available treatment for malignant PTs. To provide a better understanding of these tumors, we developed mouse models that recapitulate the pathological and clinical properties of human malignant PTs. Mouse undifferentiated mammary gland cells were infected with a retrovirus encoding the human oncoprotein H‐RasG12V, and the infected cells were transplanted orthotopically into the mammary fat pads of syngeneic mice. The transplanted cells showed a high tumorigenic activity, with the resulting tumors manifesting pathological characteristics including stromal overgrowth similar to those of human malignant PTs. The tumors also showed high rates of both local recurrence and lung metastasis. Our models may prove useful for studies of the pathophysiology of malignant PTs as well as facilitate the development of new treatments.


Introduction
Phyllodes tumors (PTs) are fibroepithelial tumors of the breast with epithelial and stromal components (Zurrida et al. 1992;Kim et al. 2013;Spitaleri et al. 2013;Borhani-Khomani et al. 2016;Tan et al. 2016). They are characterized by stromal overgrowth and a leaf-like architecture resulting from an exaggerated intracanalicular pattern, and they are graded as benign, borderline or malignant on the basis of stromal cellularity, atypia and mitotic activity as well as the nature of the tumor margin (Zurrida et al. 1992;Kim et al. 2013;Spitaleri et al. 2013;Borhani-Khomani et al. 2016;Tan et al. 2016). PTs account for <1% of all breast neoplasms, with local recurrences developing in 8%-50% of patients depending on the grade of the primary tumor and distant metastases having been detected in up to 47% of malignant cases (Zurrida et al. 1992;Chen et al. 2005;Taira et al. 2007;Kim et al. 2013;Borhani-Khomani et al. 2016;Tan et al. 2016). Surgery remains the standard and only treatment for PTs. The development of new therapeutic approaches for refractory cases would be facilitated by the establishment of mouse models that recapitulate the biological and pathological characteristics of human PTs.
Comparative genomic hybridization has showed copy number changes in malignant PTs, with inactivation of the tumor suppressor p16 INK4a appearing to be important for progression to malignancy (Jones et al. 2008). Many molecules have also been found to affect the biological behavior of malignant PTs (Karim et al. 2013). Exome sequencing has identified highly recurrent MED12 and RARA somatic mutations in breast fibroepithelial tumors (Lim et al. 2014;Tan et al. 2015), with PTs also manifesting mutations in FLNA, SETD2, KMT2D, BCOR and MAP3K1 (Tan et al. 2015).
Ras is a small GTPase that functions as a molecular switch in the activation of multiple downstream effectors (Downward 2003;Eckert et al. 2004;McLaughlin et al. 2013). It thus activates the mitogen-activated protein kinase (MAPK) signaling pathway composed of Raf, MEK and ERK, which regulates gene expression as well as cell proliferation, differentiation and survival (Dhillon et al. 2007;Young et al. 2013). It also activates the PI3K-Akt-mTOR pathway, which plays a key role in the regulation of glucose metabolism as well as cell proliferation, migration and apoptosis (Eckert et al. 2004;Castellano & Downward 2011). The activation state of Ras is controlled at the level of its cycling between GTP-bound (active) and GDP-bound (inactive) forms (Milburn et al. 1990;Downward 2003;Karnoub & Weinberg 2008). Ras GTPase-activating proteins (RasGAPs), such as p120GAP and NF1, negatively regulate Ras signaling by promoting the conversion of Ras-GTP to Ras-GDP (Milburn et al. 1990;Karnoub & Weinberg 2008), whereas Ras guanine nucleotide exchange factors (RasGEFs) positively regulate such signaling by promoting the conversion of Ras-GDP to Ras-GTP (Milburn et al. 1990;Downward 2003). Mis-sense gain-of-function mutations in RAS genes (HRAS, NRAS, KRAS) have been detected in 27% of all human cancers, with 98% of the mutations occurring at three mutational hotspots: amino acids Gly 12 , Gly 13 and Gln 61 (Karnoub & Weinberg 2008;Baines et al. 2011;Hobbs et al. 2016). Stabilization of Ras proteins in their active state results in malignant transformation and contributes to tumor growth in cancers bearing such RAS mutations (Karnoub & Weinberg 2008;Baines et al. 2011;Hobbs et al. 2016).
In an attempt to develop a mouse model of human PTs, we infected normal undifferentiated mammary gland cells derived from C57BL/6 mice with a retrovirus encoding H-Ras G12V . The oncogene-expressing cells were then injected into mammary fat pads of syngeneic mice. The tumors that formed from the injected cells were found to be similar to human malignant PTs in terms of their pathological and clinical characteristics.

Establishment of mouse undifferentiated mammary gland cells that express H-Ras G12V
Although xenograft models based on injection of established cell lines into immune-deficient mice are often used in preclinical studies, such models alone are insufficient because they do not fully recapitulate the heterogeneous nature of native tumors that results from various microenvironmental stimuli and variability in the differentiation ability of cancer cells. We previously showed that H-Ras G12V transforms Ink4a/Arf knockout C57BL/6 mouse mammary gland cells and thereby established a syngeneic mouse model of triple-negative breast cancer (TNBC) (Kai et al. 2014). In the present study, we set out to establish malignant mesenchymal tumor models through the expression of H-Ras G12V in wild-type mouse mammary gland cells. Normal mammary glands of C57BL/6 mice were digested with collagenase, and the isolated cells were maintained in culture as mammospheres. The mammospheres were subsequently dissociated and cultured on Matrigel for the establishment of mammosphere-derived cells (MDCs) (Fig. 1A). In addition, we isolated CD24 med CD29 high cells (defined as basal stem cells, or BSCs) (Stingl et al. 2006;Lim et al. 2010;dos Santos et al. 2013) from normal mouse mammary gland cells by fluorescence-activated cell sorting (FACS) (Fig. 1B), and cells were subsequently subjected to culture on Matrigel. Flow cytometric analysis for these stromal markers showed that both cultured MDCs and BSCs were predominantly CD24 low CD29 high (Fig. 1C), indicating that the isolated BSCs changed from CD24 med CD29 high to CD24 low CD29 high during the course of in vitro culture.
We next infected the cultured MDCs and BSCs either with a retrovirus encoding both oncogenic H-Ras G12V and green fluorescent protein (GFP) or with a control virus encoding GFP alone. The infected cells were isolated by FACS on the basis of their expression of GFP. Expression of H-Ras G12V induced a change in the morphology of both MDCs and BSCs from a spindle shape to a more spherical shape as well as triggered the formation of focal cell aggregates ( Fig. 1D). It also induced phosphorylation of the kinases Akt and ERK (Fig. 1E), indicating that both the PI3K-Akt and MEK-ERK pathways were activated as downstream effectors of mutant H-Ras signaling. Furthermore, both H-Ras G12V /MDCs and H-Ras G12V /BSCs showed a marked growth advantage in comparison with parental MDCs and BSCs or corresponding cells infected with the control retrovirus ( Fig. 1F).

H-Ras G12V -induced mammary tumors are pathologically similar to human malignant PTs
To examine the tumorigenic potential of the mammary cells expressing H-Ras G12V in vivo, we transplanted these cells into mammary fat pads of syngeneic mice. Both H-Ras G12V /MDCs and H-Ras G12V /BSCs formed tumors in the recipient mice at rates of 97.9% (47/48) and 100% (20/20), respectively ( Fig. 2A).
Both types of tumors manifested marked stromal cellularity and atypia, a high mitotic rate, and pronounced stromal overgrowth (Fig. 2B). They were thus pathologically similar to human malignant PTs, which are characterized by stromal overgrowth and a leaf-like architecture resulting from an exaggerated intracanalicular pattern (Tan et al. 2016). Although the expression of hormone receptors in PTs has been examined in previous studies, the relevance of such expression is unclear (Tse et al. 2002;Tan et al. 2016).
Immunohistochemical analysis showed that the tumors formed by H-Ras G12V /MDCs or H-Ras G12V /BSCs in mice did not express estrogen receptor a (ER) or the progesterone receptor (PR) at detectable levels ( Fig. 2C). It has been reported that H-Ras genetically engineered mice developed lung bronchiolo-alveolar adenomas, lung bronchiolo-alveolar adenocarcinomas, splenic hemangiosarcomas, cutaneous squamous cell papillomas, Harderian gland adenoma and hepatocellular adenomas (Nambiar et al. 2012; Paranipe et al.  Triple-negative breast cancer is a type of human breast cancer that is characterized by a lack of ER, PR and human epidermal growth factor receptor 2 (HER2) expression. TNBC tumors have been classified into six subtypes-basal-like 1 (BL1), basal-like 2 (BL2), immunomodulatory (IM), mesenchymal (M), mesenchymal stem-like (MSL) and luminal androgen receptor (LAR)-on the basis of the activity of canonical signaling pathways and differential gene expression (Chen et al. 2012). We carried out cDNA microarray analysis with total RNA isolated from H-Ras G12Vinduced mouse mammary tumors as well as from similar tumors formed by MDCs infected with a retrovirus encoding the R1275Q oncogenic mutant of human anaplastic lymphoma kinase (ALK) (Chen et al. 2008;Mosse et al. 2008;Azarova et al. 2011), which were also negative for ER and PR expressions by immunohistochemical analysis (Fig. S1 in Supporting Information). To classify our tumor models, we converted the mouse cDNA data to the equivalent for human genes and then uploaded the results to 'TNBCtype' <http:// cbc.mc.vanderbilt.edu/tnbc>, a subtyping tool for TNBC. This tool showed that none of the tumors formed by H-Ras G12V /MDCs, H-Ras G12V /BSCs or ALK R1275Q /MDCs could be assigned to a TNBC subtype (Fig. 2D). We then compared the patterns of gene expression between each pair of the three mouse tumor types. The patterns for H-Ras G12V /MDC tumors and H-Ras G12V /BSC tumors were more similar to each other than were those for H-Ras G12V /BSC tumors and ALK R1275Q /MDC tumors or for H-Ras G12V /MDC tumors and ALK R1275Q /MDC tumors (Fig. 2E), suggesting that the gene expression signature is largely influenced by the type of oncogene.

H-Ras G12V -induced mammary tumors are clinically similar to human malignant PTs
Complete surgical excision is the standard treatment for PTs. However, the local recurrence rate for malignant PTs is 23%-30% even in patients who experience successful tumor resection (Zurrida et al. 1992;Kim et al. 2013;Tan et al. 2016). We investigated local recurrence and metastasis to the lungs after surgical resection of tumors in our mouse models. The pathology of local recurrences and lung metastases resembled that of the primary tumors for mice injected with H-Ras G12V /MDCs or H-Ras G12V / BSCs (Fig. 3A). H-Ras G12V /MDCs formed tumors in all 10 injected mice within nine days, with local recurrences and lung metastases being observed in seven and two mice, respectively, after surgical resection of primary tumors. H-Ras G12V /BSCs formed tumors in all 10 injected mice within 10 days, with local recurrences and lung metastases being detected in nine and six mice, respectively (Fig. 3B). The rate of lung metastasis was thus greater for H-Ras G12V / BSCs than for H-Ras G12V /MDCs. These observations suggested that the H-Ras G12V -induced mammary tumors are highly malignant and are clinically similar to human malignant PTs.
In conclusion, we have established mouse models of human malignant PTs through injection of H-Ras G12V -expressing mouse CD24 low CD29 high undifferentiated mammary gland cells (MDCs or BSCs) into syngeneic recipients. The clinicopathologic features of the mouse tumors are highly similar to those of human malignant PTs. Our models may therefore provide new insight into the pathogenesis of PTs as well as assist in the development of new treatments for these tumors.

Isolation of mammary gland cells
Mammary glands of six-week-old female C57BL/6J mice were minced and then digested for 3-4 h at 37°C with collagenase type II (1 mg/mL; Sigma, St Louis, MO, USA) in Dulbecco's modified Eagles' medium (DMEM)-F12 (Sigma) supplemented with 21.4 mM NaHCO 3 , 25 lM HEPES and bovine serum albumin (20 mg/mL; Sigma). The digested tissue was then treated with DNase I (0.1 mg/mL; Sigma) before isolation of single cells by passage through a 100-lm cell strainer (BD Biosciences, San Jose, CA, USA). Red blood cells were lysed by the addition of NH 4 Cl to a final concentration of 150 mM.

Isolation of BSCs
A single-cell suspension of primary mammary gland cells in phosphate-buffered saline was labeled with allophycocyaninconjugated rat monoclonal antibodies to mouse CD24 (BioLegend, San Diego, CA, USA) and phycoerythrin-conjugated hamster monoclonal antibodies to mouse/rat CD29 (eBioscience, San Diego, CA, USA) for FACS with a MoFlo XDP Cell Sorter (Beckman Coulter, Brea, CA, USA). The CD24 med CD29 high fraction was isolated as BSCs. Cultured MDCs and BSCs were also analyzed for CD24 and CD29 expressions by flow cytometry with a Gallios Flow Cytometer (Beckman Coulter) and FLOWJO software.

Cell viability assay
Cells (3 9 10 3 per well) were seeded in 96-well plates, cultured for 72 h and assayed for viability with the use of a Cell TiterGlo assay (Promega, Madison, WI, USA).

Injection of cells into mammary fat pads of mice
All mouse experiments were approved by the animal ethics committee of Keio University School of Medicine. Female C57BL/6J mice aged 6-8 weeks were anesthetized by intraperitoneal injection of pentobarbital (50 mg/kg). Retrovirus-infected MDCs or BSCs [1 9 10 6 in 20 lL of a 1 : 1 (v/v) mixture of DMEM-F12 and EZ cell matrix] were injected into the no. 4 mammary fat pads of the anesthetized mice.

Immunohistochemistry
Tumor tissue was fixed with 4% paraformaldehyde, embedded in paraffin and sectioned, after which the sections were depleted of paraffin and stained with rabbit polyclonal antibodies to ER, to PR or to GFP (Santa Cruz Biotechnology). Immune complexes were detected with biotinylated secondary antibodies and a Vectastain avidin and biotinylated horseradish peroxidase macromolecular complex reagent and 3,3 0 -diaminobenzidine (Vector Laboratories, Burlingame, CA, USA).

Microarray analysis
Total RNA was isolated from tumors formed by H-Ras G12V / MDCs, H-Ras G12V /BSCs or ALK R1275Q /MDCs with the use of Trizol (Thermo Fisher Scientific K.K., Yokohama, Kanagawa, Japan) and RNAeasy Kit (Qiagen, Tokyo, Japan). The RNA was then converted to cRNA for the analysis of gene expression with a cDNA array (Agilent Gene Expression DNA Microarray; Agilent Technologies, Santa Clara, CA, USA).