Rheb1-independent Activation of mTORC1 in Mammary Tumors Occurs Through Activating Mutations in mTOR

Summary

Mechanistic Target of Rapamycin Complex 1 (mTORC1) is a master modulator of cellular growth, and its aberrant regulation is recurrently documented within breast cancer. While the small GTPase Rheb1 is the canonical activator of mTORC1, Rheb1-independent mechanisms of mTORC1 activation have also been reported but have not been fully understood. Employing multiple transgenic mouse models of breast cancer, we report that ablation of Rheb1 significantly impedes mammary tumorigenesis. In the absence of Rheb1, a block in tumor initiation can be overcome by multiple independent mutations in Mtor to allow Rheb1-independent re-activation of mTORC1. We further demonstrate that the mTOR kinase is indispensable for tumor initiation as the genetic ablation of mTOR abolishes mammary tumorigenesis. Collectively, our findings demonstrate that mTORC1 activation is indispensable for mammary tumor initiation, and that tumors acquire non-canonical mechanisms of mTORC1 activation.

Introduction

Mammary tumorigenesis is a multistep evolutionary process involving the selection for genetic or epigenetic alternations that allow the preneoplastic epithelial cell population to subvert barriers to uncontrolled growth and survival, correlating within progression through a series of pathologic stages (Liggett and Sidransky, 1998, Sherr and McCormick, 2002, Wright and Shay, 2000, Visvader, 2011). The multistep evolutionary nature of mammary tumor initiation is closely recapitulated by genetically engineered mouse models (GEMMs) of breast cancer (Andrechek et al., 2003). For example, mammary-specific expression of the oncogenic receptor tyrosine ErbB2 or the Polyomavirus Middle T antigen (PyV mT) is sufficient to drive tumorigenesis through a series of pre-malignant stages that culminate in the formation of metastatic mammary tumors (Guy et al., 1992a, Guy et al., 1992b, Guy et al., 1996, Schade et al., 2013). However, despite intensive efforts to elucidate the molecular events crucial for mammary tumor initiation (Huck et al., 2010, Ursini-Siegel et al., 2010, Pontier et al., 2010), many of the underlying mechanisms remain undefined.

The mechanistic Target of Rapamycin (mTOR) is a serine/threonine protein kinase that frequently undergoes aberrant activation in cancer (Zoncu et al., 2011). mTOR can form two multi-protein complexes, mTORC1 and mTORC2, that regulate distinct molecular processes (Laplante and Sabatini, 2012). Specific functions of mTORC1 include stimulation of protein synthesis through phosphorylation of p70 S6 Kinases 1 and 2, and the eukaryotic initiation factor 4E-binding protein (4E-BP1–3) (Nojima et al., 2003, Ma and Blenis, 2009, Schalm et al., 2003), whereas mTORC2 is best known for its role in regulation of the Akt family of kinases (Sarbassov et al., 2005). Although regulation of mTORC2 is not well understood, mTORC1 is activated in response to diverse extracellular and intracellular stimuli including growth factors and amino acids (Long et al., 2005c). A critical mechanism by which these factors control mTORC1 involves the activation of the small GTPase Rheb1 (Dibble and Cantley, 2015). The Tuberous Sclerosis complex (TSC1/2), composed of TSC1 (hamartin) and TSC2 (tuberin) (Garami et al., 2003), functions as the GTPase-activating protein (GAP) that governs Rheb1 function by converting active GTP-bound Rheb1 into its inactive GDP-bound form (Inoki et al., 2003, Zhang et al., 2003). The GAP activity of TSC1/2 is disabled by phosphorylation of TSC2 through PI3K-Akt (Inoki et al., 2002), and ERK1/2-MAPK signaling pathways (Ma et al., 2005), allowing for de-repression of Rheb1 and activation of mTORC1 (Garami et al., 2003, Sato et al., 2009). Although mTORC1 plays a prominent role in growth of established tumors cells (Mosley et al., 2007), its role in mammary tumor initiation has not been explored.

Herein, we report that Rheb1-mediated mTORC1 activation plays a crucial role in the initiation of mammary tumorigenesis in both an ErbB2 and Luminal B GEMM of breast cancer. Our data further indicate that in a subset of Rheb1-deficient mammary tumors, oncogenic mTORC1 activation occurs through mutations within the mTOR kinase. In contrast to ablation of Rheb1, abrogation of mTOR resulted in a complete block in mammary tumorigenesis. Collectively, our data suggest that the mTOR kinase is a critical signaling node required for mammary tumor initiation

Results

Mammary Ablation of Rheb1 Delays Mammary Tumorigenesis.

To evaluate the involvement of mTORC1 signaling in mammary tumor initiation, we used mammary epithelial-specific conditional gene targeting to delete the upstream activator Rheb1 in two GEMMs representative of the ErbB2-positive and Luminal B breast cancer subtype (Herschkowitz et al., 2007). To explore the role of Rheb1 in ErbB2-positive breast cancer, mice carrying the loxP-flanked Rheb allele (Rheb1^fl/fl) were crossed with a strain expressing bicistronic transgene containing activated ErbB2 and Cre recombinase linked by an internal ribosome entry sequence (IRES) under the transcriptional control of the Mouse Mammary Tumor Virus long terminal repeat (MMTV-LTR) (referred to as NIC) (Schade et al., 2013). This strategy couples the overexpression of ErbB2 with mammary-specific excision of the conditional knockout Rheb allele within the Rheb1^fl/fl NIC mice (Fig 1A). Mammary deletion of Rheb1 significantly delayed ErbB2-driven mammary tumorigenesis in the Rheb1^fl/fl NIC strain (TD₅₀ = 375 days) compared to wildtype controls (TD₅₀ = 125 days, p<0.0001) (Fig 1B). Although tumor initiation is fully penetrant in the NIC strain (Utermark et al., 2012, Huck et al., 2010), only 60.7% of Rheb1^fl/fl NIC mice (Fig 1B) developed tumors expressing both ErbB2 and Cre. Albeit with a reduced penetrance, ErbB2-positive tumors that do arise are lacking detectable Rheb1 expression (Fig 1D & 1F). PCR designed to specifically detect the excision of Rheb1 allele (Suppl Fig 1A) revealed that the Rheb1 allele is excised within the tumor epithelium and retained with the tumor stroma (Suppl Fig 1B & 1C). Given that defects in mammary gland development can greatly impair tumorigenesis (Marcotte et al., 2012), we further examined whether mammary deletion of Rheb produced adverse effects on mammary ductal outgrowth. Analysis of whole-mounted mammary glands of 7 and 11 week old Rheb1^fl/fl MMTV-Cre mice confirmed that Rheb1 was dispensable for early mammary ductal development (Suppl Fig. 2A–2C). To validate our observations in another GEMM of breast cancer, we crossed the Rheb1^fl/fl mice with a doxycycline (Dox)-inducible PyV mT-IRES-Cre transgenic system (MIC strain) (Rao et al., 2014) which also closely recapitulate human breast cancer progression through defined pre-malignant stages, culminating in aggressive, metastatic disease (Guy et al., 1992a) (Fig 1A). To standardize the initial mammary epithelial content prior to oncogene expression, Rheb1^fl/fl and Rheb1^wt/wt MIC mice were placed under Dox induction at 8 weeks of age. As observed with the ErbB2-driven tumor model system, mammary deletion of Rheb1 also delayed mammary tumorigenesis in Rheb1^fl/fl MIC mice (TD₅₀ = 220 days) compared to their wildtype counterparts (TD₅₀ = 60 days, p<0.05) (Fig 1C). Despite this initial defect in tumor induction, mammary tumors ultimately do develop within the Rheb1^fl/fl MIC strain albeit with reduced overall tumor penetrance (55%) compared to the wild-type control (75%) (Fig 1C). Like the ErbB2 model system, mammary tumors arising within the Rheb1^fl/fl MIC strain lacked detectable Rheb1 expression (Fig 1G) while maintaining expression of the PyV mT oncogene and Cre (Fig 1E). Collectively, these results argue that ablation of Rheb1 elicits a significant delay in tumor onset in both ErbB2- and PyV mT-driven mammary tumorigenesis. However, a proportion of the Rheb1-deficient mammary epithelium evolves mechanisms that permit tumor formation in the absence of Rheb1 in GEMMS representative of both ErbB2-amplified and luminal B breast cancer.

Figure 1. Mammary Deletion of Rheb1 Delays Mammary Tumorigenesis.

(A) Left: Schematic representation of the MMTV-driven activated ErbB2 transgenic mice (NIC strain) and the loxP-site flanked Rheb allele. Right: Schematic representation of the Doxycycline-inducible MMTV-rtTA PyV mT transgenic mice (MIC strain), and the loxP-site flanked Rheb allele. MMTV, Murine mammary tumor virus. IRES, Internal ribosomal entry site. TE, Tetracycline-responsive operator. rtTA, reverse tet-responsive transactivator. (B) Kaplan-Meier plot illustrating percentage of tumor-free Rheb1^wt/wt NIC (n=20), and Rheb1^fl/fl NIC mice (n=17). The p-values were calculated using a Mantel-Cox Test. (C) Kaplan-Meier plot illustrating percentage of tumor-free Rheb1^wt/wt MIC (n=22), and Rheb1^fl/fl MIC mice (n=26). The p-values were calculated using a Mantel-Cox Test. (D) Immunofluorescence staining of Rheb1^wt/wt NIC and Rheb1^fl/fl NIC tumor sections for ErbB2 and Cre. Scale bar represents 50μm. (E) Immunofluorescence staining of Rheb1^wt/wt MIC and Rheb1^fl/fl MIC tumor sections for PyV mT and Cre. Scale bar represents 50μm. (F) Western blot analysis of Rheb1^wt/wt NIC and Rheb1^fl/fl NIC tumor protein extract with indicated antibodies. (G) Western blot analysis of Rheb1^wt/wt MIC, and Rheb1^fl/fl MIC tumors protein extract with indicated antibodies.

Rheb1 function is required at the early stages of mammary tumor progression.

One important advantage of the Dox-inducible MIC model system (Fig 1A) is that it grants temporal control of PyV mT oncogene expression thereby allowing for interrogation of early pre-malignant stages of mammary tumorigenesis. As such, we evaluated the immediate effects of Rheb1 ablation on the early stages of mammary tumorigenesis by examining the mammary glands of Rheb1^fl/fl MIC mice following 4 and 14 days of Dox induction. In contrast to rapid expansion of epithelial content, and induction of mammary epithelial hyperplasias and adenomas observed within Rheb1^wt/wt MIC mice, mammary ablation of Rheb1 within Rheb1^fl/fl MIC mice resulted in normal mammary ducts with scarce mammary hyperplasias (Fig 2A & 2B). The block in mammary tumor progression as a result of Rheb1 ablation was also observed as early as four days post-induction, where early mammary hyperplasias were observed in the Rheb1^wt/wt MIC mice (Fig 2A). Consistent with the significant delay in tumor onset observed in Rheb1^fl/fl MIC mice, these results indicate that Rheb1 is required at an early stage of mammary tumor induction.

Figure 2. Ablation of Rheb1 Stalls Mammary Tumorigenesis at Early Stages of Progression.

(A) Top: Representative H&E-stained histological sections of mammary glands collected from Rheb1^wt/wt MIC and Rheb1^fl/fl MIC mice following 4 days, or 14 days of Doxycycline induction. Scale bar represents 2mm. Bottom: panel is a 35X representation of area boxed in red in top panel. Scale bar represents 200 μm. (B) Top: Representative Immunohistofluorescence staining for E-Cadherin. Lymph node encircled by dash lines. Scale bars represent 100 μm. Bottom: Quantification of E-Cadherin IHC staining of mammary glands of Rheb1^wt/wt MIC and Rheb1^fl/fl MIC mice following 4-day or 14-day doxycycline administration. Values represent mean (+/− SEM) of 5 mice per genotype at indicated timepoints. Statistical significance was assessed by two-tailed unpaired Student’s t-test. (C) Left: Representative hematoxylin-stained whole mounts of mammary glands collected from Rheb1^wt/wt NIC and Rheb1^fl/fl NIC mice at 130 days of age. Scale bar represents 500 μm. Right: Panel is 35X magnification of left panel. Scale bar represents 100 μm. (D) Top: Representative H&E-stained mammary gland sections of Rheb1^wt/wt NIC and Rheb1^fl/fl NIC mice collected at tumor endpoint. Right panel is magnification of left panel. Scale bars represent 500 μm. Bottom: Quantification of pathology types found in the mammary glands collected from Rheb1^wt/wt NIC (n=17) and Rheb1^fl/fl NIC (n=20) mice at tumor endpoint. Frequency of normal ducts, Hyperplasia, Adenoma, and Adenocarcinomas is relative to total number of ductal structures per histology section, and values represents mean from at least 5 biological replicates per genotype.

To confirm that this early block in mammary tumor progression was also occurring within the ErbB2 model system, we performed histological analysis of mammary glands of at 120–125 day old Rheb1^wt/wt NIC and Rheb1^fl/fl NIC mice and examined for early hyperplasia and adenomas. While mammary glands of control mice at early time points exhibited signs of hyperplasia, and small pre-malignant lesions (Fig 2C; black arrows); we failed to detect any histological evidence of transformation in the absence of Rheb1. To determine which particular pathological stage tumor progression was stalled at following loss of Rheb1, we examined the frequencies of different types of lesions within adjacent mammary tissue from Rheb1^fl/fl NIC and Rheb1^wt/wt NIC mice at tumor endpoint. Under these conditions, we noted a wide range of pathology types including hyperplasias (4.4%), adenomas (5.8%), and even early carcinomas (2.5%) in wild-type control mice (Fig. 2D), whereas in Rheb1^fl/fl NIC mammary glands exhibited a normal histological appearance, with the rare instance of early hyperplasia (0.8%).Taken together, these observations along with the tumor onset data argues that Rheb1 plays a critical role in the initiation phase of tumor progression in both ErbB2- and PyV mT-driven GEMMS.

Rheb1-dependent activation of mTORC1 is critical for the initiation phase of mammary tumor progression.

While it is well established that Rheb1 is an obligate activator of mTORC1 in various in vitro systems (Sato et al., 2009, Long et al., 2005a, Parmar and Tamanoi, 2010, Buerger et al., 2006), the relative contribution of Rheb1 in regulating mTORC1 activity within mammary epithelium in vivo is remains to be established. Since the downstream phosphorylation targets of mTORC1, p70 S6K1 and 4E-BP1 exhibit differential sensitivity to loss of mTORC1 activity (Kang et al., 2013, Yang et al., 2013) we sought to establish the degree of mTORC1 inhibition upon Rheb1 ablation by monitoring the phosphorylation status of these targets. Given that the MMTV promoter of our transgenic models exhibits mosaic activity within the mammary epithelia, it is crucial to identify the epithelium within the mammary ducts that exhibit Rheb1 ablation to accurately discern the effects of gene loss. Using RNA probes to visualize intact and excised Rheb1 transcript (Suppl Fig 3A–3C), we were able to detect an increased amount of excised Rheb1 transcript (Suppl Fig 3D), and decreasing amount of intact Rheb1 transcript (Suppl Fig 3E) within the Cre-expressing pre-neoplastic epithelium of Rheb1^fl/fl MIC mice thus confirming that Cre expression can serve as an appropriate surrogate for Rheb1 excision. To this end, histological sections of mammary glands from Rheb1^fl/fl NIC and Rheb1^wt/wt NIC mice were subjected to immunohistochemistry analysis for Cre, phospho-4E-BP1 (S65) and phospho-S6 ribosomal protein (pS6 S240/244), which is a downstream target of S6Ks (Fig 3C; Suppl Fig 4C). The Cre-positive, and hence ErbB2-expressing, mammary epithelium from Rheb1^fl/fl NIC mice exhibited decreased levels of both p4E-BP1 and p-rpS6 compared to the Cre-positive mammary epithelium of their wildtype counterparts (Fig 3A and 3D; Suppl Fig 4A & 4D). Collectively, these results indicate that Rheb1 ablation in vivo resulted in loss of both p70 S6 kinase function and 4E-BP1 phosphorylation in the ErbB2-overexpressing mammary epithelium thus indicating an acute loss of mTORC1 activity. Similar observations were also made within the Rheb1^fl/fl MIC mammary glands (Fig 3B & 3D; Suppl Fig 4B & 4D), and raptor^fl/fl MIC mammary glands (Suppl Fig 4E & 4F).

Figure 3. Loss of Rheb1 inhibits ErbB2-mediated 4EBP1 phosphorylation in Mammary Epithelium.

(A) Representative images of immunofluorescence and immunohistochemistry staining of mammary ductal structures from Rheb1^wt/wt NIC and Rheb1^fl/fl NIC mice at tumor endpoint for indicated antibodies. Scale bars represent 20 μm. (B) Representative images of immunohistofluorescence and immunohistochemistry staining of mammary ductal structures from Rheb1^wt/wt MIC and Rheb1^fl/fl MIC mice after 14 days of Dox induction for indicated antibodies. Scale bars represent 20 μm. (C) Grading scheme of phospho-4EBP1 S65 IHC staining intensity in Cre-positive mammary epithelium. Scale bar represents 50 μm. (D) Graph depicting frequency of grade 1, 2 and 3 p4EBP1 S65 staining in Cre-positive mammary epithelium in adjacent mammary glands of Rheb1^wt/wt NIC and Rheb1^fl/fl NIC mice at tumor endpoint (top), and Rheb1^wt/wt MIC and Rheb1^fl/fl MIC mice after 14 days of Doxycycline induction (bottom). Values represent mean (+/− SEM) frequency of each grade within minimally 150 Cre-positive cells from at least 3 biological replicates from each genotype. Statistical significance was assessed by two-tailed unpaired Student’s t-test.

Subsequently, we sought to examine the proliferative capacity of the neo-plastic mammary epithelium following ablation of Rheb1 by assessing the levels of the proliferative marker PCNA (Suppl Fig 5A). We observed a decrease in proliferation within the pre-neoplastic mammary epithelium following ablation of Rheb1, which was closely recapitulated with genetic ablation of Raptor (Suppl Fig 5B–5C). Taken together, these observations argues that mammary ablation of Rheb1-dependent mTORC1 activity elicits an early block in the initiation of the tumorigenic process.

While this correlative data implicates mTORC1 as the prominent output of Rheb1 function, we next sought to functionally evaluate contribution of mTORC1 function in the initiation of tumor progression. As such, we next examined if disruption of the mTORC1 scaffold raptor within the mammary epithelium elicits a comparable defect in mammary tumor induction. To address this, a conditional raptor knockout strain (Polak et al., 2008) was crossed with both the MIC and NIC model systems (Fig. 4A). Consistent with our observations with Rheb1 ablation, mammary ablation of raptor within the raptor^fl/fl NIC and raptor^fl/fl MIC stains exhibited a dramatically delayed tumor onset (Fig. 4B and 4C) which was correlated with a block in the initiation of tumor progression (Fig. 4D–4G). Collectively, these observations also argue that disruption of mTORC1 through Raptor deletion phenocopies the defect in mammary tumor initiation observed in our Rheb1^fl/fl MIC and NIC model systems.

Figure 4. Loss of Raptor Delays Mammary Tumor Initiation.

(A) Left: Schematic representation of the MMTV-driven activated ErbB2 transgenic mice (NIC strain) and the loxP-site flanked Rheb allele. Right: Schematic representation of the Doxycycline-inducible MMTV-rtTA PyV mT transgenic mice (MIC strain), and the loxP-site flanked Rptor allele. (B) Kaplan-Meier plot illustrating percentage of tumor-free Raptor^wt/wt NIC (n=11), and Raptor^fl/fl NIC mice (n=11). The p-values were calculated using a Mantel-Cox Test. (C) Kaplan-Meier plot illustrating percentage of tumor-free Raptor^wt/wt MIC (n=16), and Raptor^fl/fl MIC mice (n=13). The p-values were calculated using a Mantel-Cox Test. (D) Representative H&E-stained mammary gland sections of Raptor^wt/wt NIC and Raptor^fl/fl NIC mice collected at tumor endpoint. Right panel is magnification of left panel. Scale bars represent 500 μm. (E) Quantification of pathology types found in the mammary glands collected from Raptor^wt/wt NIC (n=17) and Raptor^fl/fl NIC (n=20) mice at tumor endpoint. Frequency of normal ducts, Hyperplasia, Adenoma, and Adenocarcinomas is relative to total number of ductal structures per histology section, and values represents mean from at least 5 biological replicates per genotype. (F) Top: Representative H&E-stained histological sections of mammary glands collected from Raptor^wt/wt MIC and Raptor^fl/fl MIC mice following 4 days, or 14 days of Doxycycline induction. Scale bars represent 2mm. Bottom: 35X representations of area in top panel. Scale bars represent 200 μm. (G) Top: Representative Immunohistochemistry staining for E-Cadherin. Lymph node encircled by dash lines. Scale bars represent 100μm. Bottom: Quantification of E-Cadherin staining of mammary glands of Raptor^wt/wt MIC and Raptor^fl/fl MIC mice following 4-day or 14-day doxycycline administration. Values represent mean (+/− SEM) of 5 mice per genotype at indicated timepoints. Statistical significance was assessed by two-tailed unpaired Student’s t-test.

Mammary Tumors Evolve Different Rheb1-independent mechanisms of mTORC1 activation.

Although initiation of mammary tumorigenesis is hindered in the absence of Rheb1(Fig 1F & 1G), Rheb1-deficient tumors do ultimately develop (Fig. 1F and 1G) which exhibit similar range of pathology and keratin expression profile as wildtype tumors (Suppl Fig 6A). In vivo growth, as measured by changes in tumor volume, of Rheb1-deficient and Rheb1-proficent NIC tumors was comparable up to 6 weeks after initial tumor palpation (Suppl Fig 6E). Similar trends were observed in immunohistochemical analyses of proliferative (Ki67 and PCNA) and apoptotic markers (Cleaved Caspase 3) (Suppl Fig 5D–5E & Suppl Fig 5B–5D). Given that we observed loss of p70 S6K1 activity and 4E-BP1 phosphorylation observed in the pre-malignant mammary epithelium of Rheb1^fl/fl NIC and Rheb1^fl/fl MIC mice (Fig 3A and 3B), we next examined the phosphorylation of rpS6 and 4E-BP1 in endpoint Rheb1-deficient NIC and MIC tumors. In contrast to the lack of mTORC1 activity in early lesions, the Rheb1-deficient NIC and NIC end-stage tumors exhibited robust phosphorylation of 4E-BP1 that was comparable to Rheb1-proficient controls (Fig 5A and 5B; Suppl Fig. 7B & 7C). However, examination of rpS6 phosphorylation in the Rheb1-deficient NIC and MIC tumors reveals two distinct subsets exhibiting either low rpS6 phosphorylation (Fig 5A; lanes 6–11 & Suppl Fig 7B; lanes 7–9 & 12), or levels of S6 phosphorylation comparable to wildtype NIC or MIC tumors (Fig 5A; lanes 12–14, and Suppl Fig 7B; lanes 10–11 and 13–14).

Figure 5. Rheb1-deficient Mammary Tumors Maintain mTORC1 Activation and Function.

(A) Immunoblot of protein extracts from end-stage Rheb1^wt/wt NIC and Rheb1^fl/fl NIC mammary tumors, with antibodies against indicated proteins. Bottom: Quantification of immunoblots for indicated proteins in (A). Error bars represents SEM (B) Immunoblots of protein extracts from end-stage Rheb1^wt/wt MIC and Rheb1^fl/fl MIC mammary tumors, with antibodies against indicated proteins. Bottom: Quantification of immunoblots for indicated proteins in (B). (C) Immunoblots of protein extracts from Rheb1^wt/wt NIC and Rheb1^fl/fl NIC tumors cell lines, with antibodies against indicated proteins. Bottom: Quantification of immunoblots for indicated proteins in (A). (D) Immunoblot showing puromycin incorporation within Rheb1^wt/wt NIC and Rheb1^fl/fl NIC tumor cell lines. Actin was detected as loading control. Bottom: Graph represents mean level of puromycin incorporation within three Rheb1^wt/wt NIC and Rheb1^fl/fl NIC tumor cell lines from three independent experiments. (E) Immunoblot showing puromycin labelled Cyclin D1 and Bcl-XL within three Rheb1^wt/wt NIC and Rheb1^fl/fl NIC tumor cell lines after Cyclin D1 or Bcl-XL immunoprecipitation. Bottom: Graph represents mean level of puromycin-labelled cyclin D1 and Bcl-XL within three Rheb1^wt/wt NIC and Rheb1^fl/fl NIC tumor cell lines from three independent experiments. * p-value <0.05. Error bars represent SEM, and statistical significance was assessed by two-tailed unpaired Student’s t-test for blot quantifications (A-E). (F) Immunoblot of Rheb1^wt/wt NIC and Rheb1^fl/fl NIC tumor cell lysate treated with Everolimus and Torin 1 probed using the antibodies indicated. Data are representative of three independent experiments. Rheb^fl/fl NIC cell line carrying mTOR mutation indicated by #. (G) Graph showing cell viability of Rheb1^wt/wt NIC and Rheb1^fl/fl NIC tumor cells treated with Everolimus, or Torin 1. Values represent mean value (+/− SEM) of three biological replicates per genotype, with eight replicates per each treatment. The growth assay was repeated independently three times. Statistical significance was assessed by two-tailed unpaired Student’s t-test where * represents p-value <0.05.

Since mTORC1 activity can be influenced by insulin stimulation (Garami et al., 2003), and fluctuations in systemic insulin levels contribute to varying translational rates within tissue (Baillie and Garlick, 1991), we also examined levels of p-rpS6 and p4E-BP1 in four Rheb1-deficient NIC tumor cell lines under constant growth factor conditions. Consistent with our in vivo analyses, we observed a similar bifurcation in phospho-rpS6 S240/244 levels among the Rheb1-deficient NIC cell lines thus indicating the cell intrinsic nature of this phenotype (Fig 5C). In addition, Rheb1-deficient NIC cells also exhibited levels of p4E-BP1 comparable to Rheb1-proficient NIC cells. Consistent with these observations, global translation monitored by puromycin incorporation and production of puromycin-labelled cyclin D1 and Bcl_XL, known to be translated by mTORC1-sensitive mRNAs (Averous et al., 2008, Wu et al., 2005), were also comparable between Rheb1-deficient and proficient NIC cells (Fig 5D & 5E). Given that other kinases such as CDK1 (Shuda et al., 2015), were reported to phosphorylate 4E-BP1 independently of mTORC1, we further examined the mTORC1-dependence of 4E-BP1 phosphorylation in these Rheb1-deficient NIC tumors cells. We observed comparable inhibition of 4E-BP1 and rpS6 phosphorylation in both Rheb1-proficient and Rheb1-deficient NIC cells following 48hr treatment with either the rapalog everolimus, or the ATP-competitive mTOR inhibitor torin 1 (Fig 5F), indicating that 4E-BP1 regulation in Rheb1-deficient NIC cells remained mTORC1 dependent. Similarly, Rheb1-deficient NIC cells demonstrated comparable impairment in cell viability following mTORC1 inhibition as wildtype controls (Fig 5G). Taken together, these results suggest that mTORC1 activity, initially lost in the Rheb1-deficient normal mammary epithelium of Rheb1^fl/fl NIC mice, is re-activated in the arising Rheb1-deficient NIC tumors.

Activating MTOR mutations Occur in Rheb1-Deficient Mammary Tumors.

Given that mTORC1 activity is fully restored within a subset of Rheb1-deficient tumors (Fig 5A; Suppl Fig. 7B), we next sought to determine the underlying Rheb1-independent molecular mechanism of mTORC1 re-activation. Notably Akt S473 phosphorylation, a marker of mTORC2 function (Sarbassov et al., 2006), was elevated in end-stage Rheb1-deficient NIC tumors compared to their wildtype counterparts. This trend was conserved within the primary Rheb1-deficient NIC tumor cells, and demonstrated mTORC2-dependency as long-term treatment with either everolimus or torin 1 ablated Akt phosphorylation (Fig 5F). The elevated mTORC2 function in Rheb1-deficient NIC tumor cells is likely attributable to alleviation of a negative feedback loop involving p70 S6K (Liu et al., 2013) as most Rheb1-deficient NIC tumors cells have lower p70 S6K activity compared to their wildtype counterpart (Fig 5C). Noticeably, two of the Rheb1-deficient NIC cell lines displayed concurrent elevation in mTORC2 activity and restoration of p70 S6K activity as measured by p-rpS6 (S240/244) levels (Fig 5C). This phenotype was reminiscent of the effects of activating mutations in mTOR, which drive increased basal activity of both mTORC1 and mTORC2 (Grabiner et al., 2014). Although it has been shown that mutant TOR supports the growth of Rheb1-null yeast cells (Urano et al., 2007), this has not previously been examined as a mechanism of compensation for loss of Rheb1 function in mammalian cells. This prompted us to examine the Mtor mutational status within a cohort of Rheb1-deficient NIC tumors. The sequencing analysis revealed four different Mtor mutations were uncovered in the subset of Rheb1-deficient NIC tumors with comparable rpS6 phosphorylation to wild-type controls while no mutations in Mtor were detected in NIC tumors (Suppl Fig 8A). Interestingly, the mTOR F1888C mutation was discovered in two independent Rheb1-deficient NIC tumors, and their corresponding cell lines (Suppl Fig 8A). Given that the concurrent increase in pAkt (S473), and p-rpS6 (S240/244) was also observed in Rheb1-deficient MIC tumors (Fig 5B), the mutational status of Mtor was also examined in the PyV mT-driven tumors. Three additional Mtor mutations were discovered in the Rheb1-deficient MIC tumors exhibiting high p-rpS6 (S240/244), while no mutations were detected in either wildtype MIC tumors or Rheb1-deficient MIC tumors exhibiting low p-rpS6 phosphorylation (Suppl Fig 8B). Of clinical relevance, many of these mTOR mutations uncovered have been previously reported within human disease (Fig 6B). All mTOR mutations uncovered from both transgenic tumor models (Fig 6A) were shown to confer elevated mTORC1 activity following ectopic expression in 293T cells (Fig 6C). These results are consistent with other studies as the mTOR T1977K mutation has been previously reported to confer elevated mTORC1 activity (Xu et al., 2016, Grabiner et al., 2014, Ghosh et al., 2015). In addition to exhibiting comparable levels of mTORC1 activity as the wildtype counterparts (Fig 6D), Rheb1-deficient NIC tumors carrying mTOR mutations also exhibited comparable expression of the mTORC1-sensitive targets cyclin D1 (Averous et al., 2008), cyclin D3 (Larsson et al., 2012) and Bcl-X_L (Wu et al., 2005). These observations argue that one of the compensatory mechanisms that restore mTORC1 activity in the absence of the Rheb1 GTPase involves the occurrence of activating mutations within the mTOR kinase.

Figure 6. Gain-of-Function Mtor Mutations Develop in Subset of Rheb1-deficient Mammary Tumors.

(A) Schematic illustrating MTOR mutations discovered in Rheb1-deficient NIC (Black) and Rheb1-deficient MIC (Red) mammary tumors along with surrounding amino acids in murine and human mTOR. (B)Table depicting human mTOR mutations found in cancer patients similar to mTOR mutations uncovered within Rheb1^fl/fl MIC and Rheb1^fl/fl NIC tumors. (C) Top: Representative Immunoblots of 293T cells expressing either wildtype or mutant mTOR probed with indicated antibodies. Bottom: Graphs of quantification of S6 and 4EBP1 phosphorylation within 293T cells expressing either wildtype or mutant mTOR. Values represent mean (+/−STEM) of three independent experiments. Statistical significance was assessed by two-tailed unpaired student’s t-test where * represents p<0.05. (D) Immunoblot analysis of Rheb1^fl/fl NIC tumors carrying mTOR mutations, and Rheb1^wt/wt NIC controls with indicated antibodies. Bottom: Quantification of p4EBP1 Thr37, p4EBP1 S65, Cyclin D1, and Bcl-XL from (D). Values represent mean (+/− SEM) where statistical significance was assessed by two-tailed unpaired Student’s t-test.

Given that transgenic mammary tumors accumulate multiple molecular alterations along tumor progression, we wondered if mTOR mutations alone were sufficient to elicit Rheb1-independence within our Rheb1-deficient tumor models, or these genetic alterations required additional factors present in our tumor system. While technical challenges limited us from directly re-introducing wildtype mTOR into primary Rheb1-deficient tumor cells carrying mutant mTOR to evaluate the relative contribution of mTOR mutation to mTORC1 reactivation, we examined if expression of mutant mTOR were sufficient to elicit Rheb1-independent mTORC1 activation in 293T cells. To address this, 293T cells were depleted of Rheb1 via siRNA after ectopic expression of either wildtype mTOR or a panel of mutant mTOR uncovered within our Rheb1-deficient mammary tumors. While 293T cells ectopically expressing wildtype mTOR exhibited decreased rpS6 and 4EBP1 phosphorylation following Rheb1 depletion via siRNA, we observed that mTORC1 activity within 293T cells expressing the mTOR I2501F and I1417T mutations exhibited insensitivity to Rheb1 depletion by siRNA (Suppl Fig 9A). Conversely, we observed that 293T cells expressing the mTOR T1977K, F1888C, A2416D, A1518D and D2424A mutations displaced variable sensitivity to Rheb1 depletion (Suppl Fig 9A). These results suggest that certain mTOR mutations may confer Rheb1-independent mTORC1 activation which is consistent with previous reports that different mTOR mutations elicit distinct mechanistic functions (Xu et al., 2016). Furthermore, it is feasible that other mTOR mutations may require the presence of additional factors within the mammary tumor setting to elicit Rheb1-independence, which requires further elucidation in future work. Additionally, given that mTOR mutations were only uncovered within a subset of Rheb1-deficient tumors exhibiting comparable p-rpS6 levels as wildtype controls, the underlying molecular mechanism leading to mTORC1 re-activation within other subsets of Rheb1-deficient tumors may be distinct.

mTOR function is indispensable for ErbB2 mammary tumor induction.

Our studies with the Rheb1^fl/fl MIC and Rheb1^fl/fl NIC strains have demonstrated that mammary tumors ultimately evolve molecular mechanisms to allow mTORC1 re-activation. Despite our observations with the conditional Rheb1 knockout model systems, whether the mTOR kinase itself is dispensable for mammary tumor progression is unclear. To directly address the contribution of the mTOR kinase in mammary tumor induction, we crossed the conditional mTOR knockout mice (mTOR^fl/fl strain) (Risson et al., 2009) with our NIC strain (Fig 7A). In contrast to the conditional Rheb1 knockout GEMMs (Fig 1A), mammary ablation of mTOR resulted in the abrogation of tumor formation within the mTOR^fl/fl NIC strain within the 1-year observation period (Fig 7B). Consistent with the indispensable role of mTOR in mammary tumor formation, histological examination of mammary glands from end-point mTOR^fl/fl NIC failed to reveal any evidence of abnormal histology (Fig 7C and 7D). Altogether, these observations argue that in contrast to Rheb1, the mTOR kinase is an essential signaling node within ErbB2-driven mammary tumor progression.

Figure 7. mTOR is Indispensable for Mammary Tumorigenesis.

(A) Schematic representation of the MMTV-driven activated ErbB2 transgenic mice (NIC strain) and the loxP-site flanked Mtor allele. (B) Kaplan-Meier plot illustrating percentage of tumor-free mTOR^wt/wt NIC (n=11), and mTOR^fl/fl NIC mice (n=12). (C) Left: Representative H&E-stained mammary gland sections of mTOR^wt/wt NIC and mTOR^fl/fl NIC mice collected at tumor endpoint. Scale bar represents 500μm Right: Panel is magnification of left panel. Scale bar represents 200μm. (D) Quantification of pathology types found in the mammary glands collected from mTOR^wt/wt NIC (n=17) and mTOR^fl/fl NIC (n=11) mice at tumor endpoint. Frequency of normal ducts, Hyperplasia, Adenoma, and Adenocarcinomas is relative to total number of ductal structures per histology section, and values represents mean from at least 5 biological replicates per genotype. (E) Cartoon representation of working model for mTORC1 re-activation in Rheb1-deficient tumors. A high threshold for mTORC1 activity is required for mammary tumorigenesis than normal mammary gland development. In the absence of Rheb1-mediated mTORC1 activation, alternative mechanisms of mTORC1 activation arise within the pre-neoplastic tissue to allow for tumor initiation.

DISCUSSION

The adaptive selection of molecular features crucial for tumor cell survival and growth associated with mammary tumor progression, is closely recapitulated within GEMMs of breast cancer (Ursini-Siegel et al., 2007). Following loss of genes or disruption of processes typically required for tumorigenesis, molecular compensation frequently occurs within the pre-neoplastic mammary epithelium, allowing the adoption of alternative mechanisms of transformation (Tung et al., 2017, Simond et al., 2017, Zhang et al., 2011). We demonstrate using two GEMMs of breast cancer that loss of Rheb1 within the preneoplastic mammary epithelium stalls tumor progression at early stages. Although recent in vitro evidence indicate that other alternative small GTPases exhibit the capacity to activate mTORC1 (Martin et al., 2014, Tee et al., 2005), mTORC1 function within the pre-neoplastic mammary epithelium of the in vivo models examined here is diminished upon Rheb1 ablation (Fig. 3A & 3B). Consistent with the importance of Rheb1-dependent mTORC1 activation in the early phase of tumor induction, targeted deletion of Raptor in both MIC and NIC models resulted in a similar delay in the initiation of mammary tumors (Fig. 4A & 4C). Interestingly, despite the critical role of mTORC1 function during the initial stages of tumor progression with these GEMMS, normal mammary gland development was refractory to Rheb1 ablation. These observations suggest that mTORC1 function may be required at different thresholds within normal and malignant cells. Consistent with this concept, haploid insufficiency in the downstream of mTORC1 signaling eIF4E, had little impact on normal development but resulted in severe impairment of lung tumor progression (Truitt et al., 2015). Given the key role of 4E-BPs in regulating translation of specific mRNAs crucial to oncogenesis such as cyclin D1 and D3 (Averous et al., 2008, Zhang et al., 2011), these observations stress the importance of maintaining an oncogenic threshold of translation within tumor cells. While mTORC1-mediated regulation of the 4EBP1-eIF4E axis is disrupted with Rheb1 ablation within these two selected GEMMs, the resultant delay in tumorigenesis may derive from the disruption of additional mTORC1-dependent processes including regulation of autophagy (Kim et al., 2011), and cell metabolism (Li et al., 2010, Land and Tee, 2007). While not the focus of this study, the relative contribution of these mTORC1-dependent processes to tumor initiation requires further investigation.

Despite the initial delay in tumor initiation observed in the conditional Rheb1 knockout MIC and NIC strains, tumors eventually developed with variable penetrance in both model systems. Biochemical analyses of Rheb1-deficient tumors revealed that they had recovered mTORC1’s capacity to phosphorylate downstream target such as S6 kinase and 4E-BP1 (Fig. 5A & 5B). The Rheb1-deficient tumors originating from both GEMMS exhibited variable levels of rpS6 phosphorylation, while 4E-BP1 phosphorylation was comparable to the wildtype controls. An important facet of the Rheb1-deficient NIC cell lines is that phosphorylation of rpS6 and 4E-BP1 is abrogated by either rapamycin or torin 1 (Fig 5C), which indicates a requirement of mTORC1 kinase. While the mTOR mutations L2185A/C, A2034V and F2108L have been reported to confer resistance to mTORC1 and mTOR inhibitors (Wagle et al., 2014a, Wu et al., 2015, Rodrik-Outmezguine et al., 2016), the Rheb-deficient NIC cell line 4927 which carried mutant mTOR still retains sensitivity to mTORC1 inhibition (Fig. 5F).

Although mTORC1 function is restored to varying degrees within arising Rheb1-deficient tumors, we did observe restoration of mTORC1 activity to comparable level as wildtype controls within a subset of mammary tumors (Fig 5A & 5B). The use of GEMMs of breast cancer provided an opportunity to interrogate these processes in a biologically relevant setting. In a proportion of these tumors, restoration of mTORC1 activity was associated with activating mutations within mTOR kinase (Fig. 6A). Recent clinical studies have identified mTOR-activating mutations to be associated with therapeutic benefit in both urothelial/bladder cancer and renal cell carcinoma (Wagle et al., 2014b, Kwiatkowski et al., 2016). Despite the identification of these mTORC1-hyperactivating mTOR mutations in numerous cancers in the clinical setting (Grabiner et al., 2014, Wagle et al., 2014b), their exact mode of action is still unclear (Grabiner et al., 2014, Xu et al., 2016). Although, mTOR mutations that arise in cells while under the selective pressure of mTORC1 or mTOR inhibitors elicit specialized modes of action in disrupting inhibitor function. In the presence of mTOR ATP-competitive inhibitor AZD8055, mTOR mutations arise in the kinase domain leading to altered kinetic properties of the kinase (Rodrik-Outmezguine et al., 2016), while mTOR mutations that disrupt binding of the FKBP12 complex arise in cells under Rapalog treatment (Wagle et al., 2014a, Rodrik-Outmezguine et al., 2016). Additionally, the L2185A mTOR mutation in the ATP-binding pocket has also been demonstrated to confer resistance to mTOR inhibitor AZD8055 (Wu et al., 2015). Taken together, the mTOR mutations that developed in the absence of Rheb1 in our transgenic system may elicit a specialized mode of action to confer activation. Of clinical relevance, the mTOR mutations discovered from both Rheb1-deficient tumor systems have been previously reported within human malignant disease (Fig 6B). Considering the evolutionary nature of tumor progression within GEMMs, these mTOR mutations likely play a compensatory role in the reactivation of mTORC1 in the absence of Rheb1 (Fig 7E). While molecular mechanism of Rheb1-mediated mTORC1 activation is lacking, it is well established that activation requires direct interaction between Rheb1 and mTOR (Long et al., 2005b). Recent structures of Rheb-bound mTORC1 reveal that mTORC1 undergoes a conformational change to adopt a conformation primed for ATP hydrolysis upon binding to Rheb1 (Yang et al., 2017). Yang et al., also demonstrated that a lower dose of Rheb1 is required for activation of mutant mTORC1 which suggests that the Rheb1-mTOR complex forms more efficiently in the presence of mutant mTOR. Conceivably, the mTOR mutations discovered within our Rheb1-deficient tumor systems may support a similar conformational change. While Rheb1 is absent within our tumor systems, alternative small GTPases such as RalB GTPases have demonstrated a modest binding capacity to mTOR and increasing RalB-mTOR interaction does elicit more robust mTORC1 activation (Martin et al., 2014). It is also plausible that acute activation of this or other closely related GTPases may provide an additional compensatory role that allows restoration of mTORC1 function (Tee et al., 2005). Future studies will focus on establishing the contribution of alternative small GTPases to mTORC1 activation within the reported Rheb1-deficient NIC cell lines.

Although mammary tumors evolved in the absence of Rheb1, mammary ablation of mTOR within the NIC model resulted in complete abrogation of tumors. These observations indicate that the kinase component of mTORC1, mTOR is indispensable for mammary tumor formation. Using our model systems, we highlight that compensatory mechanisms can occur to allow re-activation of mTORC1 upon deletion of Rheb1. Given the essential role of mTOR in mammary tumorigenesis, future development of mTOR targeted therapeutics may have important clinical implications in treatment and management of breast cancer.

STAR Method

Resource Availability

Lead Contact

Further information and requests for resources and reagents should directed towards, and will be fulfilled by the Lead Contact, William Muller (william.muller@mcgill.ca).

Material Availability

All unique/stable reagents generated in this study are available from the Lead Contact with a complete Materials Transfer Agreement.

Data and Code Availability

No unpublished custom code, software, or algorithm were generated in this paper.

Experimental Model and Subject Details

Experimental Animal Models.

Generation of MMTC-Cre, MMTV-NIC and TetO-MIC mice was described previously (Rao et al., 2014, Ursini-Siegel et al., 2008, Andrechek et al., 2003). The MMTV-rtTA strain was obtained from Dr. Lewis Chodosh as previously described (Gunther et al., 2002). The ROSA26 Cre-activated b-galactosidase reporter strain was obtained from Dr. Phillipe Soriano as previously described (Soriano, 1999). The conditional Rheb1 knockout mice were obtained from Dr. Richard Lamb, University of Liverpool. The conditional Raptor knockout and conditional mTOR knockout mice were obtained from Dr. Nahum Sonnenberg, McGill University. The study of mice was approved by the Animal Resources Centre at McGill University and follows with the guidelines set by the Canadian Council of Animal Care. All mice were backcrossed to a pure FVB/N. Female mice were monitored for tumor formation by physical palpation 8 week after birth, and were necropsied 6 weeks post tumor onset with exception when tumor burden exceeded maximal volume allowed by our animal protocol.

Primary Cell Culture and Cell Lines.

Primary tumor cells were generated from transgenic tumors as previously described (Huck et al., 2010), and maintained in culture in DMEM media supplemented with 5% FBS, 5ng/mL EGF, 35 ug/mL bovine pituitary extract, 5ug/mL Insulin, 1ug/mL Hydrocortisone. 293T cells were cultured HEK-293T cells were obtained from ATCC® (ATCC® CRL-3216), and cultured in DMEM supplemented with 10% FBS.

Method Details

Histological and Immunohistochemical Analysis.

Tumors were fixed in 10% neutral buffered formalin for 24hrs, and transferred to cold 70% ethanol. Samples were paraffin-embedded and sectioned at 4um. Sections were de-paraffinized with xylene, and antigen retrieval was performed in 10mM Citrate buffer (pH6) using a pressure cooker. Endogenous peroxidase was quenched with 3% Hydrogen Peroxide. Sections were blocked with 10% Power Block agent (BioGenex) in PBS for 10 mins, and incubated with primary antibody overnight at 4°C. Subsequently three PBS washes, sections were incubated with secondary antibody (Vector Elite) for 1hr at room temperature. For immunohistochemistry, staining was visualized with Vectastain ABC kit (Vector Laboratories) per manufacturer’s instructions follow by Hematoxylin counterstaining. For immunofluorescence, staining was visualized with either Tyraminde Signal Amplification reagent (Thermo Fisher Scientific) for 10 minutes per manufacturer’s instructions, followed by DAPI stain for 5 minutes. Immunohistochemistry images were acquired using Aperio-XT Slide scanner (Aperio Technologies). Immunofluorescent images were acquired using an LSM510 Confocal Microscope (Carl Zeiss), and analyzed using ZEN software.

Whole-mount Mammary Glands and X-Gal Staining.

No 4. Mammary glands were dissected, spread on glass slide and placed within acetone overnight. Mammary glands were incubated in Harris Modified Hematoxylin stain (Fischer Chemical) overnight, and destained in 1% HCl in 70% ethanol. Mammary glands were washed with ethanol, and cleared in xylene for 1 day before mounting with ClearMount (American Mastertech Scientific). For X-Gal staining, mammary glands were dissected, fixed in Fixing buffer (2% paraformaldehyde, 0.25% glutaraldehyde, 0.01% NP-40 in PBS) for 2hr, incubated within PreStain buffer (2mM MgCl₂, 0.01% Na-deoxycholate, 0.02% NP-40 in PBS) for 2hrs. Subsequently, glands were added to Staining buffer (500mM K₄Fe(CN)₆, and 40mg/ml in Prestaining buffer) for 24hrs at room temperature. Mammary glands were dehydrated in 70% EtOH followed by 100% EtOH, and cleared in xylene before being mounted with Clear Mount (American Mastertech Scientific).

Immunoblotting.

Tumor lysates were prepared from flash-frozen tumor pieces in RIPA buffer (150mM sodium chloride, 50mM Tris-HCl pH7.4, 2mM EDTA,10m NaF, 10mM Sodium pyrophosphate, 1% sodium deoxycholate, 1% NP-40) supplemented with 1mM sodium orthovanadate, 1mM PMSF, 10mM NaF, 10ug/ul leupeptin and 10ug/ul apotinin. Whole cell lysate was also prepared in RIPA buffer. Protein concentration was quantified by Bradford Assay (Bio-Rad) and 25ug of total protein was resolved by SDS-PAGE, and transferred to Immobilion FL PVDF membranes (Millipore), and block with Odyssey Blocking buffer (Li-COR Biosciences). Primary antibodies were diluted in Odyssey Blocking buffer overnight at 4°C. Secondary fluorophore-conjugated antibodies, IRDye 800 anti-rabbit, IRDye 680 anti-goat and IRDye 680 anti-mouse were incubated for 1hr at room temperature. Immunoblots were scanned with Odyssey CL-X imaging system, and images were processed with Image Studio Lite V4.0.

Antibodies.

Antibodies used for immunoblots include Rheb (Cell signalling, Cat# 1387, 1:1000), Cre (Cell signalling, Cat# 12830, IHC/IHF - 1:50), Cleaved Caspase 3 (Cell signaling, Cat# 9661, IHC - 1:500), pAkt S473 (Cell signaling, Cat# 4060, 1:1000), pAkt Thr308 (Cell signaling, Cat# 13038, 1:1000), pS6 S240/244 (Cell signaling, Cat# 5364, WB - 1:1000, IHC/IHF – 1:500), Total S6 (Cell signaling, Cat# 2317), p4EBP1 S65 (Cell signaling, Cat# 13443, WB-1:1000, IHC/IHF-1:500), p4EBP1 Thr37/46 (Cell signaling, Cat# 2855, WB-1:1000, IHC/IHF-1:500), 4EBP1 (Cell signaling, Cat# 9644, WB-1:1000), eIF4E (Cell signaling, Cat# 9742, 1:1000), Cyclin D1 (Cell signaling, Cat# 2978, 1:1000), Cyclin D1 (Cell signaling, Cat# 2922, IP-1:50), Cyclin D3 (Cell signaling, Cat# 2936, 1:1000), Actin (Sigma, A5316, 1:5000), PyV mT (clone 762, a gift from Dr. Steven Dilworth, IHC/IHF-1:100), Akt1/2 (Santa Cruz, SC-1619, 1:200), E-Cadherin (BD Bioscience, Cat# 610182, 1:1000), Neu (Dako, A0485, 1:500), Ki67 (Abcam, ab15580, 1:500), Cytokeratin 8/18 (Fitzgerald, 20R-CP004, 1:500), Cytokeratin 14 (BioLegend, PRB-155P, 1:500), and Anti-puromycin (Millipore, Clone 12010 – MABE343, 1:1000). Goat anti-rabbit IRDye 800, Goat anti-mouse IRDye 680, Donkey anti-goat IRDye 800 (Li-COR, 925–32211, 925–68070, 925–68074, 1:10 000)

DNA extraction and Sanger Sequencing.

Genomic DNA was collected from mammary tumors using the DNA/RNA All Prep Mini kit (Qiagen. Cat No. 80204). All Exons of mTOR gene was sequenced by Sanger sequencing as conducted by the Genome Quebec Innovation Centre.

Cloning and Cell Transfection.

pcDNA3.1 Puro+ V2 was generated by subcloning small DNA fragment carrying XbaI-NotI-XhoI-KpnI restriction sites into pcDNA3.1Puro+ using NheI and KpnI sites. MTOR was subcloned pcDNA3-Flag mTOR WT (Vilella-Bach et al., 1999) (Addgene; #26603) into pBluescript (Agilent Technologies) and the pcDNA3.1Puro+ V2. Part I of MTOR was subcloned into pBluescript using NotI and KpnI. Part II of MTOR was subcloned into pcDNA3.1Puro+ V2 using KpnI and XbaI. Mutant fragments of mTOR were generated by two part PCR strategy using primer sequences provided, and subcloned into pBluescript or pcDNA3.1Puro+ V2. Complete MTOR was reassembled within pcDNA3.1Puro+ V2 using NotI and KpnI. Point mutations were verified using Sanger sequencing as conducted by the Genome Quebec Innovation Centre using primer sequences provided. Wildtype and mutant pcDNA3.1Puro+ V2 mTOR was transfected into 293T cells (ATCC) using Lipofectamine 3000 per manufacturer’s instructions, and subsequently lysed in RIPA buffer 48hrs post-transfection.

Drug Treatment and Cell Viability Assay.

Tumors cells were seeded in 96-well plates with 8 replicates at 5,000 cells per well. 24hrs later, tumor cells were treated with either DMSO, 50uM Everolimus (LC Laboratories, E-4040), or 250nM Torin 1 (Selleckem, S2827) in growth medium for 72hrs. Tumor cells were fixed with 2% PFA and stained with 0.05% Crystal Violet (Sigma, HT90132) subsequent to treatment. Crystal violet staining was visualized in 10% acetic acid at 590nm.

SUnSET Assay and Immunoprecipitation.

Tumor cells were starved overnight and stimulated with full growth medium for 9hrs before being pulsed with 10ug/mL of Puromycin for 3hr. Whole cell lysate was collected in modified RIPA buffer supplemented with protease inhibitors. For Cyclin D1 immunoprecipitation (IP), 500μg of whole cell lysate was incubated with primary antibody overnight at 4°C under continuous agitation. Immune complexes were recovered with magnetic beads (Millipore, LSKMAGAG10), washed three times in lysis buffer, and resuspended in 50 μl of Laemmli buffer before separation by SDS PAGE. Analysis of Puromycin incorporation was detected with anti-puromycin antibody. IP experiments were repeated three times with three independent tumor lines per genotype.

siRNA transfection

Wildtype and mutant pcDNA3.1Puro+ V2 mTOR plasmids were transfected into 293T cells (ATCC) using Lipofectamine 3000 per manufacturer’s instructions. Subsequently, 100μM of siRNA Rheb1 was transfected into 293T cells carrying either wildtype or mutant pcDNA3.1Puro+ V2 mTOR plasmids using HiPerfect transfection reagent per manufacturer’s instructions. 293T cells were lysed in RIPA buffer 48hrs post-transfection of siRNA.

Quantification and Statistical Analysis

All statistical analysis was completed using GraphPad software. Two-tailed Student t-tests, and log-rank Mantel-Cox tests were applied accordingly. The results of the statistical tests can be found within the figures, and the figure legends.

Data and Code Availability

This study did not generate any unique datasets or code.