The MSP‐RON axis stimulates cancer cell growth in models of triple negative breast cancer

Triple‐negative breast cancer (TNBC) is the most aggressive subtype of breast cancer with poor prognosis and high rates of relapse. The lack of actionable targets for TNBC has contributed to the high mortality rates of this disease, and new candidate molecules for potential manipulation are urgently required. Here, we show that macrophage‐stimulating protein (MSP) and its tyrosine kinase receptor, Recepteur d’origine nantais (RON), are potent drivers of cancer cell growth and tumor progression in a mouse model of TNBC driven by the loss of Trp53 and Brca1. After comparison of two genetically engineered mouse models of TNBC, we found that mammary tumors from K14‐Cre;Brca1 F/F ;Trp53 F/F (KB1P) mice exhibit high endogenous levels of MSP and RON expression. We show that MSP stimulates serine/threonine kinase 1 and extracellular regulated MAPK activation as well as cancer cell growth in cell lines derived from the two mouse models, while genetic and pharmacological inhibition of RON prevents these effects. Similarly, KB1P tumor progression in mice was robustly attenuated by treatment with a RON inhibitor with accompanied reduction in the proliferation marker, Ki‐67. Analysis of human gene expression data confirmed that the genes encoding MSP and RON are robustly expressed in human TNBC as well as other subsets of breast cancer. Our findings uncover a mouse model where MSP expression and RON expression are naturally increased, and they provide evidence that this receptor and its ligand are viable candidate molecules for targeted treatment of breast cancer.


Introduction
Triple negative breast cancer (TNBC) represents a subgroup of about 15-20% of breast tumors that lacks expression of three main biomarkers: estrogen receptor (ER), progesterone receptor and epidermal growth factor receptor 2 (HER2) amplification (1). When compared to other breast cancer subgroups, patients with TNBC generally have an unfavorable prognosis due to the aggressive nature of this disease and higher risk of local and distant recurrence (2). At the molecular level, TNBC is highly heterogeneous, but unlike ER-positive and HER2amplified breast cancers, the discovery of uniform actionable molecular targets has eluded researchers (1). Thus, the absence of approved targeted therapies for TNBC leaves chemotherapy as the only therapeutic option.
Recepteur d'origine nantais or RON is a receptor tyrosine kinase encoded by the Mst1r gene that shares high homology with the cMET oncogene (3). RON is translated as a singlechain cytoplasmic pro-protein and then further processed to a mature form, which is displayed as a transmembrane protein at the cell surface. Additionally, several isoforms of the RON protein have been reported that stem from two distinct promoters upstream or within the Mst1r gene (3). These isoforms can be constitutively active, oncogenic or biologically inactive (3). RON has only one known ligand called macrophage-stimulating protein (MSP). MSP is encoded by the Mst1 gene, and it is structurally similar to the cMET ligand, hepatocyte growth factor (HGF). RON is lowly expressed in normal mammary gland tissue, but can be overexpressed and phosphorylated in primary breast tumors (4). MSP and RON (over)-expression in breast cancer specimens correlates with poor prognosis and distant recurrence (5,6), suggesting that the MSP-RON axis contributes to disease progression. Indeed, experimental models, in which MSP or RON are overexpressed or the RON receptor is inhibited or genetically deleted, show that the MSP-RON axis drives mammary tumor progression and metastasis (6)(7)(8)(9)(10)(11)(12)(13)(14).
Here, we provide the first evidence of the importance of the MSP- RON

Animal Models and Cell Lines
The generation and characterization of K14-Cre;Trp53 F/F (KP) and K14-Cre;Brca1 F/F ;Trp53 F/F (KB1P) mice, which are maintained on the FVB/n background, is described here (15). Single cell suspensions from end-stage (15mm) mammary tumors were plated and propagated in hypoxic incubators for four weeks to generate cell lines. These cells were grown in DMEM supplemented with 10% fetal calf serum (FCS), 100U/mL penicillin/streptomycin and 2mM glutamine. Before recombinant MSP treatment (100ng/mL; R&D systems #6244-MS-025), cells were serum starved in 0.2% FSC overnight. The RON inhibitor BMS-777607 (ChemieTek #CT-BMS777) was added to cells at 1μM for 1h before MSP treatment where indicated.
The orthotopic transplantation of tumor fragments is described here (16). Briefly, KB1P tumor fragments were implanted into the fourth mammary fat pad of 10-12 week old female recipient FVB/n mice (Charles River, UK). When tumors reached 3x3 mm, mice were given a daily oral gavage of BMS-777607 (50mg/kg dissolved in DMSO and diluted in 70% PEG-300) or vehicle control (11% DMSO diluted in 70% PEG-300) and then monitored until

RNA sequencing analysis
Total tumor gene expression data from K14-Cre;Trp53 F/F (KP) and K14-Cre;Brca1 F/F ;Trp53 F/F (KB1P) mice were kindly provided by Lewis Cantley (NCBI BioProject #PRJNA398328). RNA isolation, DNA library preparation, sequencing information and generation of FPKM values for these data is described here (17). To visualize the gene expression FPKM values were log transformed in R version 3.5.0 and boxplots of generated with the ggplot2 package.
Differences between expression distributions were tested with the Wilcoxon rank sum test.

Quantitative real-time PCR
RNA was isolated from frozen mammary tumors or cell lines using Qiagen's RNeasy kit and on-column DNA digestion. RNA concentration was determined using a Thermoscietific NanoDrop spectrophotometer with NanoDrop 2000 software. cDNA was prepared from 1 μg

ELISA
MSP serum levels from autochthonous, mammary tumor-bearing KP and KB1P mice or tumor-free, wild-type mice were measured using a kit from R&D Systems according to the manufacturer's recommendations.

Lentiviral transduction
Target sequences specific for Mst1r were designed using the iRNAi program. shRNA oligos were generated and sub-cloned into the pLKO.1puro lentiviral backbone using Addgene's protocol (https://www.addgene.org/8453/). Viral supernatants were prepared following transient transfection of 293FT cells with pLKO.1 encoding shRNAs, pSPAX2 packaging vector and pVSVG envelope vector using Lipofectamine 2000 according to the Page 7 of 18 manufacturer's instructions. Two 24h supernatants were collected sequentially over a 48h period, pooled and filtered through a 0.45μm syringe filter and then concentrated using the Lenti-X Concentrator solution (Clontech). Freshly concentrated supernatants were added directly to drained sub-confluent recipient cells and incubated overnight. An equal volume of fresh medium was then added and antibiotic selection (puromycin) initiated 48h later.
Antibiotic resistant clones were expanded for further analysis. We tested 9 different shRNA oligos. Efficient Mst1r knockdown was determined by quantitative real-time PCR and two distinct shRNAs were selected for further experiments. Target sequences of the two oligos used for this study were as follows: shRON-1 sense -CCTGCTGTATGTGTCCAACTT, antisense -AAGTTGGACACATACAGCAGG; shRON-2 sense -CGTCCTAGACAAGGAATACTA, anti-sense -TAGTATTCCTTGTCTAGGACG.

Proliferation Assay
Cellular proliferation was measured using the fluorescence-based proliferation CyQuant NF kit (Thermo Fisher Scientific) according to the manufacturer's instructions. Cells were seeded in a 96-well plate, 6 replicates per condition at 10 4 cells/well in low serum (0.2%) for 72h with daily administration of 100ng/mL MSP and/or 1μM BMS-777607.

Immunohistochemistry
Immunohistochemical analyses were performed by the Histology facility at the CRUK Beatson Institute using standard protocols on Bond Rx or Dako autostainers. Anti-Ki67 (clone SP6; 1:100) was purchased from Abcam and anti-Caspase 3 (clone Asp-175; 1:500) was purchased from Cell Signaling. Quantitative analysis of positive staining was performed by counting cells in at least 3 high-power fields of view (x40) per tumor by two independent researchers who were blinded to the sample group. Images were captured on an Axio Imager A2 Bio upright microscope (Zeiss) using ZenPro 2012 software (Zeiss).
Page 8 of 18

Statistics
The non-parametric Mann-Whitney U test was used to compare two groups, while one-way ANOVA was used to compare groups of three or more. Two-way ANOVA with repeated measures was used to analyze tumor growth curves. The log-rank (Mantel-Cox) test was used to analyze Kaplan-Meier survival curves. Sample sizes for each experiment were based on a power calculation and/or previous experience of the mouse models. Analyses were performed using GraphPad Prism (version 8).

MSP and its receptor, RON, are expressed in autochthonous TNBC models
In previous studies, gene expression analysis of mammary tumors from K14-Cre;Trp53 F/F  (Fig. 1A). We validated these findings by real-time PCR. Mst1 mRNA levels were significantly higher in KB1P tumors, but Mst1r levels were the same between tumor models (Fig. 1B). Expression levels of MSP protein mirrored the mRNA levels and were greater in KB1P compared to KP derived tumor tissues (Fig. 1C). Similarly, MSP serum levels were also elevated in mammary tumor-bearing KB1P mice, when compared to tumor-bearing KP mice or tumor-free, wildtype (WT) mice (Fig. 1D). Western blot analysis of RON expression revealed multiple isoforms of the RON receptor (Fig. 1E). The short form of RON was most prevalent and, in  (Fig. 1e). Quantification of the combined full length RON band intensities revealed overall higher expression levels in KB1P compared to KP tumors (Fig. 1E). Together these data reveal that the MSP specific form of the RON receptor was differentially expressed between KP and KB1P tumor models, and the short form of RON, which lacks the MSP binding domain and remains unresponsive to MSP, was present at high levels in both tumor models.

MSP-RON signaling activate AKT and MAPK pathways
We derived cell lines from four independent KP and KB1P tumors, and performed qRT-PCR analysis to confirm continued expression of Mst1r mRNA in vitro. As shown in Fig. 2A, Mst1r transcripts persisted in all cell lines. Protein expression was then measured by western blot analysis in the same cell lines. In contrast to the primary tumor tissue (Fig. 1E), a single band corresponding to the full length RON receptor β chain was the predominant species detected and it was expressed to approximately equivalent levels in each cell line (Fig. 2B).
Expression of the short form of RON also persisted, but the ratio between the long and short forms had increased compared to the primary tumors. Given that MSP and full length RON protein were increased in tumors from KB1P mice (Fig. 1 (Fig. 2C). To confirm selectivity against RON (since BMS-777607 is not absolutely specific for the RON receptor), we repeated these experiments using an shRNAmediated knockdown approach to silence Mst1r expression (Fig. 2D). Mirroring pharmacological inhibition of RON, MSP treatment of KB1P cells lacking RON failed to increase phosphorylation of both AKT and ERK1/2 (Fig. 2E). These data indicate that MSP stimulates robust intracellular signaling pathways via the RON receptor on KB1P cells.

The MSP-RON axis promotes cell proliferation
We examined the functional effects of MSP-RON signaling on proliferation of two independent KB1P cell lines, since both pathways are known to regulate cell division. We found that MSP increased proliferation of both cell lines, and that proliferation could be attenuated by addition of BMS-777607 (Fig. 3A). Treatment of either KB1P cell line with BMS-777607 alone had no effect on cell proliferation compared with untreated control cells (Fig. 3A). Moreover, shRNA-mediated RON depletion, using two shRNAs separately in each KB1P cell line, significantly reduced proliferation after MSP treatment (Fig. 3B). These data indicate that MSP requires long form RON for its mitogenic effects.

Inhibition of RON delays tumor growth in vivo
Next, we transplanted tumor fragments from KB1P mice into the mammary glands of syngeneic FVB/n mice. When tumors were palpable, we treated the mice with BMS-777607 or vehicle control and monitored tumor growth until humane end-point. Inhibition of RON in KB1P mammary tumor-bearing mice delayed tumor growth. This difference was statistically significant (p < 0.01) beginning from day 13 (Fig. 4A). In addition, KB1P tumor-bearing mice treated with BMS-777607 survived significantly longer (7 days) when compared to the control cohort (Fig. 4B) (Fig. 4C), whereas Caspase 3+ cells were similar between both groups (Fig. 4D). Together these data indicate that RON inhibition slows cancer cell proliferation without affecting apoptosis and suggest that, in the KB1P model, the MSP-RON axis plays an important role in cancer progression.

Discussion
In this study, we found that MSP signaling via its receptor, RON, is critical for growth and proliferation of KB1P mammary cancer cells both in vitro and in vivo. These data accord well with previous publications in other mammary tumor models that rely on artificial, ectopic expression of MSP or the RON receptor (6,7,(9)(10)(11)(12)14). Most mechanistic work on the MSP-RON axis has used cell lines derived from the MMTV-PyMT model (6,9,12,14) that overexpress MSP, or genetically engineered mouse models in which Mst1r expression is driven by the MMTV promoter (7,11

Conclusion
Our Reading and approval of final manuscript: all authors.