Background
Metastasis is the primary cause of death in breast cancer patients, yet there are few therapies directed at this process. As regulators of cell proliferation, cytoskeletal organization, and cell motility, Rho GTPases are essential to dissemination of cancer cells throughout the body. The Rho GTPase family consists of 20 genes in humans, with Cdc42, Rac1, and RhoA being the most thoroughly studied [
1,
2]. Rac1, RhoA, and RhoC are commonly overexpressed in human breast cancers and RNAi-mediated knockdown or deletion of these genes inhibits tumorigenesis and metastasis [
3‐
7]. Rac1 and RhoA/C contribute to metastasis by regulating separate types of invasive behavior in cancer cells, with Rac1 driving integrin-dependent, mesenchymal-type movement and RhoA/C driving integrin-independent, amoeboid movement. Importantly, cancer cells switch between these types of movement depending on the extracellular obstacles they are traversing, and inhibition of either of these forms of movement significantly inhibits invasive activity [
8‐
10].
Rho GTPases function as molecular switches, cycling between their active, GTP-bound and inactive, GDP-bound states. When active, Rho proteins interact with downstream proteins, known as effectors, to initiate signaling cascades that control cell motility [
11,
12]. Rho GTPase activation is controlled by two families of proteins, known as guanine nucleotide exchange factors (RhoGEFs) and GTPase activating proteins (RhoGAPs). RhoGEFs stimulate GDP release, thereby allowing Rho proteins to bind GTP and become active [
13,
14]. RhoGAPs stimulate the intrinsic GTPase activity of Rho proteins, thereby shutting them off [
15]. Because the wild-type forms of Rho proteins are overexpressed in breast cancers, it is commonly assumed that altered function of RhoGAPs and RhoGEFs drives their activation. For example, the RhoGAP DLC-1 is deleted or epigenetically silenced in many breast cancer subtypes, driving aberrant RhoA activation and metastatic spread to the bones [
16‐
18]. Alternatively, the Rac1 GEFs P-Rex1, Vav2/3, and Dock1 have been shown to contribute to metastatic behavior in particular breast cancer subtypes [
19‐
21]. Despite evidence for RhoA and RhoC contributing to breast cancer tumorigenesis and metastasis, RhoA subfamily GEFs that contribute to breast cancer in vivo have not yet been identified.
The neuroepithelial transforming gene 1 (Net1) is a RhoA subfamily GEF that is overexpressed in many human cancers, including breast cancer [
22,
23]. We have shown previously that Net1 is required for human breast cancer cell motility and invasive capacity in vitro [
24]. In these cells Net1 is dedicated to controlling actomyosin contraction, as inhibition of Net1 expression blocks actomyosin contractility but does not affect other RhoA-regulated events, such as Ezrin phosphorylation. We have also observed that Net1 controls FAK activation, which is necessary for focal adhesion maturation [
24]. Furthermore, Net1 has been shown to control cell motility in other cell types, and to regulate actin cytoskeletal rearrangements downstream of ligands such as TGFβ [
25‐
27]. Net1 also controls mitotic progression by regulating Aurora A activation and chromosome alignment during the metaphase [
28]. Thus, there is ample evidence to suggest that Net1 may contribute to tumorigenesis and metastasis in vivo; however, the role of Net1 in these processes has not been investigated. Similarly, the role of Net1 in human breast cancer is largely unknown.
In the present work we demonstrate that Net1 is critical for mammary gland tumorigenesis and metastasis in the mouse mammary tumor virus (MMTV)-PyMT mouse genetic model of breast cancer, and demonstrate obligate signaling pathways that are regulated by Net1. Moreover, we identify a gene expression signature indicative of Net1 function and use this signature to demonstrate that Net1 contributes to metastasis in human breast cancer patients. Together these data indicate that Net1 is required for breast cancer progression in the MMTV-PyMT mouse model and may also contribute to human breast tumorigenesis and metastasis.
Methods
Mouse husbandry and care
Mice were housed in the Center for Laboratory Animal Medicine and Care within the Medical School at the University of Texas Health Science Center at Houston, TX, USA. All studies were approved by the Institutional Animal Care and Use Committee (protocol AWC 14-007) and were conducted in accordance with the guidelines of the US Public Health Service Policy for Humane Care and Use of Laboratory Animals.
Mouse strains used, genotyping, and analysis of tumorigenesis
Net1−/− mice in the C57BL/6 strain were as described previously [
29]. Male
MMTV-PyMT mice (Tg(MMTV-PyVT)634Mul) in the FVB/J background were purchased from Jackson Labs. The
MMTV-PyMT allele was carried by crossing to female FVB/J mice. Mice lacking
Net1 were backcrossed to wild-type FVB/J mice for 10 generations to create a congenic line. Female
Net1+/− (FVB/J) mice were mated with male
MMTV-PyMT (FVB/J) mice to derive
Net1
+/−
,MMTV-PyMT mice. Female
Net1
+/−
mice were crossed with
male Net1
+/−
,MMTV-PyMT mice to produce a cohort of female littermates with
Net1
+/+
,MMTV-PyMT and
Net1
+/−
,MMTV-PyMT, and
Net1
−/−
,MMTV-PyMT genotypes for tumor studies. Primers for genotyping
Net1-deficient mice were as follows: forward primer 5GF3, 5′-TGCTATGCTATTGCTGCTT-3′, and reverse primer 3GR1, 5′-AGAACACCACCAAGTAACAA-3′ (amplifies wild-type
Net1); and forward primer 5GF1, 5′-TTGTTACTTGGTGGTGTTCT-3′ and reverse primer TV3-1R, 5′-AAGTGCTAACCTTCCTGC-3′ (amplifies
Net
−/−
allele) [
29]. The PyMT transgene was identified using previously published primers: forward, 5′-CGGCGGAGCGAGGAACTGAGGAGAG-3′; and reverse, 5′-TCAGAAGACTCGGCAGTCTTAGGCG-3′ [
30]. Tumor growth was monitored after weaning. Once palpable tumors had formed, tumor size was measured twice per week using electronic calipers. Tumor volume was calculated using the following equation:
$$ V=\left(\mathrm{Length}\times {\mathrm{Width}}^2\right)/2. $$
Mice were euthanized when the largest tumor reached 2.5 cm3.
Antibodies
The following antibodies were used: anti-pSer19-MLC2 (3675), anti-Ki67 (12202), anti-cleaved caspase 3 (9661), anti-CD31 (77699), anti-pT696-MYPT1 (5163), anti-MYPT1 (2634), anti-pY461-Src (6943), anti-pT202/Y204ERK1/2 (4370), anti-ERK1/2 (4695), anti-pT308-Akt1 (2965), anti-pS473-Akt1 (4060), anti-Akt (2920), anti-PP2A-C (2038), anti-PP2A-A (2041), anti-Shc (2432), anti-PI3 kinase p85 (4257), and anti-RhoC (3430) (Cell Signaling Technology); anti-β-actin (A5316) (Sigma-Aldrich); anti-PyMT (sc-53,481), anti-GST (sc-138), anti-Src (sc-8056), and normal rat IgG (sc-2026) (SantaCruz Biotechnology); and anti-RhoA (ARH03) (Cytoskeleton, Inc.).
Primary tumor cell isolation and mammary gland transplantation
Individual mammary tumors were isolated from Net1
+/+
,MMTV-PyMT and Net1
−/−
,MMTV-PyMT mice, manually minced, and incubated in DMEM/F12 (Hyclone) with 2 mg/ml collagenase A (Agilent), and 1× antibiotic–antimycotic (Life Technologies) for 2 h at 37 °C with 150 rpm rotation at a 45° angle. The minced tissues were then shaken vigorously and pipetted up and down to create a cell suspension. The cells were pelleted by centrifugation at 600 × g for 10 min at room temperature, resuspended in DMEM/F12 with 10% fetal bovine serum (FBS) and 1× antibiotic–antimycotic, and passed through a 70-μm cell strainer (ThermoFisher Scientific). The cells were cultured in DMEM/F12 with 10% FBS, 100 U/ml penicillin/streptomycin, 10 μg/ml insulin, and 1× antibiotic–antimycotic. The epithelial marker cytokeratin 8 (CK8) and PyMT were detected in 90–95% of cells by immunofluorescence microscopy.
Confluent primary tumor cells were harvested with 0.25% trypsin and rinsed twice with PBS. 8 × 106 cells were mixed with Matrigel (catalog no. 354,248; Corning) to a final volume of 200 μl, and were injected into the number four fat pad of wild-type female FVB mice, 6–7 weeks old, using a 28.5-G insulin syringe (catalog no. 329,424; BD). Once palpable, tumor sizes were measured twice per week. Mice were euthanized when the tumors reached 2.5 cm in length or diameter. Mammary tumors and lungs were collected for immunohistochemistry (IHC) and IF staining to analyze tumor cell proliferation, apoptosis, and angiogenesis, as described in the following.
Cell motility assays
For migration assays, confluent primary tumor cells were starved in DMEM/F12 plus 0.5% FBS for 16 h prior to trypsinization. 8 × 104 cells were placed in the upper chamber of a Transwell insert with 8-μm pores (BD Biosciences). The medium in the bottom well was supplemented with EGF (100 ng/ml; R&D Systems). Cells were allowed to migrate for 2 h, and then the cells in the upper well were removed using a cotton swab. Cells on the bottom of the membrane were fixed and stained with DAPI (1 μg/ml; Sigma-Aldrich). Cells that had traversed the membrane were counted in 10 random fields using a 20× objective and a Zeiss Axiophot microscope. Images were captured with an Axiocam MRm camera and Axiovision software. Cell numbers were quantified using ImageJ software.
Mammary gland whole mount analysis and tissue immunohistochemistry
After dissection, the fourth inguinal mammary glands were immediately fixed in Carnoy’s fixative (60% ethanol, 30% chloroform, 10% glacial acetic acid) for 2–4 h at room temperature. Glands were stained in Carmine alum solution (2 mg/ml carmine, 10.5 mM aluminum potassium sulfate dodecahydrate) overnight with gentle shaking followed by successive dehydration steps in 70%, 95%, and 100% ethanol for 1 h each, at room temperature. Glands were cleared in xylene overnight and mounted on glass slides with Permount (ThermoFisher Scientific). Mammary glands were imaged with an Eclipse 80i digital camera (Nikon) mounted on a SMZ-745 T stereo microscope (Nikon).
For immunohistochemistry (IHC) of hyperplastic regions, number four inguinal mammary glands were immediately fixed in 4% paraformaldehyde overnight at 4 °C and stored in 70% ethanol at 4 °C until paraffin embedding. Tumors were excised and fixed for IHC as already described. Five-micron sections were cut for all tissues, deparaffinized in xylene, and rehydrated. Sections were boiled for 20 min in 10 mM sodium citrate for antigen retrieval, rinsed in PBS, and quenched for 30 min in 3% H2O2 at room temperature. Sections were blocked in 5% BSA/0.5% Tween-20, or M.O.M. blocking buffer (BMK2202; Vector Labs), for 1 h at room temperature. Primary antibodies were diluted in blocking solution and sections were incubated with primary antibodies overnight at 4 °C. After washing five times in phosphate buffered saline (PBS), sections were incubated with secondary antibodies for 45 min at room temperature, washed in PBS, and incubated in ABC solution (PK7100; Vector Labs) for 30 min. Sections were then developed in diaminobenzidine (K3468; Dako) and counterstained with hematoxylin (Thermo Fisher Scientific). Images were visualized with Eclipse 80i microscope (Nikon) and Digital Sight DS-VI1 camera (Nikon), and acquired using NIS-Elements Basic Research software (Nikon).
Lung isolation and analysis
Whole lungs were isolated after euthanasia, rinsed with sterile PBS, and immediately fixed in 4% paraformaldehyde overnight at 4 °C. Lungs were stored in 70% ethanol at 4 °C until paraffin embedding. Metastatic foci on the dorsal and ventral surface of lung lobes were counted and imaged with a SMZ-745 T stereo microscope (Nikon) mounted with an Eclipse 80i digital camera (Nikon). Five-micron sections were cut, deparaffinized in xylene, and rehydrated. The lung sections were then stained with hematoxylin and eosin (ThermoFisher Scientific). Sections were visualized with a SMZ-745 T stereo microscope (Nikon), and images were acquired using NIS-Elements Basic Research software (Nikon).
Isolation of tissues and western blotting analysis
Mouse tissues were rinsed quickly in cold PBS, snap frozen in liquid nitrogen, and stored at − 80 °C until use. For extraction of proteins and mRNA, frozen tissues were pulverized with a mortar and pestle under liquid nitrogen and homogenized on ice in SDS lysis buffer for protein extraction (2% SDS, 20 mM Tris–HCl (pH 8.0), 100 mM NaCl, 80 mM β-glycerophosphate, 50 mM NaF, 1 mM sodium orthovanadate, 10 μg/ml pepstatin A, 10 μg/ml leupeptin, 10 μg/ml aprotinin) or TRK lysis buffer (E.Z.N.A.® Total RNA Kit I; Omega Bio-Tek) for RNA extraction, using a rotor-stator homogenizer. For protein analysis, lysed tissue was sonicated and protein concentrations were determined by bicinchoninic acid assay (Pierce). Equal amounts of protein were separated by SDS-PAGE, transferred to polyvinylidene difluoride membrane (PVDF), and analyzed by western blotting.
Immunoprecipitation and GST-RBD assays
For analysis of proteins coprecipitating with PyMT, pulverized mouse tumors were lysed in radioimmunoprecipitation assay (RIPA) buffer with 0.1% SDS (1.0% Triton X-100, 0.1% SDS, 0.5% sodium deoxycholate, 50 mM Tris–HCl (pH 8.0), 150 mM NaCl, 80 mM β-glycerophosphate, 10 μg/ml leupeptin, 10 μg/ml pepstatin A,10 μg/ml aprotinin, 1 mM phenylmethylsulfonyl fluoride), incubated on ice for 10 min, and homogenized using a rotor-stator homogenizer. Insoluble proteins were pelleted by centrifugation (16,000 ×
g, 10 min, 4 °C). Equal amounts of soluble lysate were precleared by incubation for 30 min at 4 °C with 2 μg of normal rat IgG plus Protein G-Sepharose (Rockland Immunochemicals). Clarified lysates were then incubated with 2 μg of normal rat IgG or rat anti-PyMT plus Protein G-Sepharose for 2 h at 4 °C. Immunoprecipitates were washed three times with wash buffer (20 mM Tris–HCl (pH 8.0), 125 mM NaCl, 5 mM MgCl
2, and 0.5% Triton X-100), resuspended in 2 × Laemmli sample buffer, and resolved by SDS-PAGE. Proteins were transferred to a PVDF membrane and analyzed by western blotting analysis as described previously [
31].
For GST-RBD assays, pulverized mouse tumors were lysed in 0.5% Triton lysis buffer plus 10 mM MgCl2 (0.5% Triton X-100, 10 mM MgCl2, 20 mM Tris–HCl (pH 8.0), 100 mM NaCl, 1 mM EDTA, 50 mM NaF, 80 mM β-glycerophosphate, 10 mM MgCl2, 1 mM Na2VO3, 10 μg/ml leupeptin, 10 μg/ml pepstatin A,10 μg/ml aprotinin, 1 mM phenylmethylsulfonyl fluoride) and homogenized using a rotor-stator homogenizer. Insoluble proteins were pelleted by centrifugation (16,000 × g, 10 min, 4 °C). Equal amounts of soluble lysate were incubated with 30 μg of GST-Rhotekin-RBD protein beads (Cytoskeleton, Inc.) for 1 h at 4 °C. Precipitates were washed three times with wash buffer (25 mM Tris–HCl (pH 7.5), 30 mM MgCl2, 40 mM NaCl), resuspended in 2 × Laemmli sample buffer, and resolved by SDS-PAGE. Proteins were transferred to a PVDF membrane and analyzed by western blotting.
Gene expression analysis
Gene expression was analyzed by the UTHealth Quantitative Genomics and Microarray Core Facility using an Illumina mouse WG6 Whole-Genome Gene Expression BeadChip. Gene expression values were estimated with the Illumina GenomeStudio software with background subtraction and quantile normalization. Changes in gene expression were initially identified using a Student’s
t test, and genes were accepted as differentially expressed if they exhibited at least 1.5-fold change with
P < 0.05 between the wild-type and
Net1 knockout samples. Gene Ontology category enrichment analysis was performed using GATHER [
32]. We scored the activation of gene expression signatures on gene expression profiles as described previously [
33]. Briefly, for a gene expression dataset, we centered and normalized each gene to a mean of 0 and standard deviation of 1, and then averaged the expression of each gene in the signature, after taking the additive inverse of the expression values for genes negatively correlated with the pathway. Finally, we created a Net1 gene expression signature using an empirical Bayes approach [
34] to find genes differentially expressed in
Net1 knockout mouse tumors with at least 5-fold change and
P < 0.05. To score this signature on human tumors, we found the human orthologs of the genes in the signature using the Homologene database [
35].
Statistical analysis
Unpaired, two-tailed Student t tests were employed for all other statistical tests. P < 0.05 was considered significant, as indicated in figure legends. All data are reported as means, and errors are the standard error of the mean. Animal cohort size was chosen based on the work of many other groups using this tumor model and all animals within the cohort were included in the results. No randomization of animals was necessary. No blinding of cohort identity was done.
Discussion
Although RhoA signaling is critically important for breast cancer cell motility and invasiveness in vitro, few studies have assessed its role in metastasis in vivo. In the present study we demonstrate that the RhoA subfamily GEF Net1 contributes to PyMT-driven tumorigenesis and is required for efficient lung metastasis. Moreover, we demonstrate that high Net1 signaling correlates with increased human breast cancer metastasis, indicating that our findings are relevant to human disease progression. To our knowledge, this is the first report of a RhoA GEF promoting breast tumorigenesis or metastasis.
MMTV-PyMT mice model luminal B-type breast tumors [
54]. However, the wide incidence of the Net1 gene expression signature in human breast cancers suggests that Net1 function is not limited to this breast cancer subtype. For example, we observed that a high Net1 gene expression signature correlated closely with human basal-type breast cancers (Fig.
7g). Moreover, the reduced distant metastasis-free survival in patients with a high Net1 gene expression signature was observed in a cohort of patients that was not subdivided according to cancer subtype (Fig.
7f). These findings, coupled with our previous results indicating that coexpression of Net1 with the β4 integrin predicted reduced distant metastasis-free survival and reduced overall survival in ERα-positive breast cancer patients [
55], indicate that Net1 may promote metastasis in a wide range of breast cancer subtypes. The idea that aberrant RhoA activation drives breast cancer metastasis fits with the findings of others indicating that reduced expression of the RhoA subfamily-specific GAP DLC-1 is predictive of metastatic spread to the bone in all breast cancer subtypes [
18]. These findings are distinct from studies focusing on Rac1 activators, which appear to function in a more subtype-specific manner. For example, the Rac1 GEFs P-Rex1 and Vav2/3 contribute to metastasis in luminal subtype breast cancers, while the Rac1 GEF Dock1 controls metastasis in HER2-positive breast cancers [
19‐
21]. The subtype independence of Net1 has important implications for therapeutic approaches, as it suggests that targeting Net1 would be a widely applicable therapeutic strategy for breast cancer. Our data indicating that
Net1 deletion switched tumors to a wild-type p53 gene expression signature (Fig.
6c) may also suggest that targeting Net1 would sensitize breast tumors with wild-type p53 to chemotherapies dependent on p53 function.
A potential caveat of our Net1 gene expression signature is that it may contain components that reflect PyMT signaling, which would also be expected to include PI3K activation. Unfortunately, it is not technically possible to isolate the activity of one pathway from the function of the rest of the signaling network. In our design, we compare the gene expression profile of Net1 within a PyMT background, and, in principle, the contribution of PyMT should not be detected. Nevertheless, the notion that Net1 function is associated with PI3K signaling and metastasis is logical given the reported roles of Rho GTPase signaling in controlling PI3K activation and extracellular matrix invasion in cell-based studies. Future work will be required to dissect the components of the Net1 gene expression signature that are conserved among different breast cancer models.
It is unclear why there was not a significant increase in the survival of mice with
Net1 deletion, given the observed delay in tumor initiation. On the surface this would suggest that
Net1-deleted tumors proliferated more rapidly, yet this was clearly not the case as Ki67 staining was reduced in both early and late tumors (Fig.
2), and unbiased gene expression analysis demonstrated significantly reduced expression of proliferation associated genes (Fig.
6b,
c). This apparent contradiction most likely reflects the short delay in tumorigenesis (only 20 days), and the fact that
Net1-deleted tumors tended to be fluid filled and less firm, which may have increased their apparent volume when measuring tumor size with calipers. The observation that they had larger necrotic cores supports this idea (Fig.
3a–
c).
The
Net1 mouse model we used is a whole-body deletion of the
Net1 gene, so some of the phenotypes we observed may be due to cancer cell extrinsic as well as intrinsic effects. That being said, our tumor cell transplant experiments indicate that many of the phenotypes we observed are tumor cell autonomous. For example, tumors arising from injection of
Net1 knockout cells exhibited less proliferation, less angiogenesis, and increased apoptosis (Fig.
6b–
g). Significantly, there was also less metastasis to the lungs (Fig.
6h,
i), indicating that the decrease in metastasis in the genetic
Net1
−/−
,PyMT mice was likely not the result of a delay in tumorigenesis. The decrease in proliferation in
Net1 knockout cells is likely attributable to decreased signaling by PyMT (Fig.
5). However, it is less clear how Net1 influences tumor angiogenesis. Presumably,
Net1 deletion inhibits the secretion of one or more angiogenic factors by the tumor cells. Whether this occurs through altered transcription, translation, or secretion is an open question, as RhoA signaling has been shown to impact each of these steps. Future work will be directed at understanding the mechanism by which Net1 controls tumor angiogenesis.
The PyMT oncogene is considered a general model for activated receptor tyrosine kinase (RTK) signaling in oncogenic transformation [
39]. Thus, the effects of Net1 on PyMT signaling to PI3K and ERK1/2 may have wider applicability. This idea is supported by our finding that our
Net1
−/−
tumors have reduced gene expression signatures for PI3K and proliferation signaling (Fig.
7c). Signaling by PyMT is initiated through interactions with PP2A and Src, and these interactions are greatly reduced in
Net1
−/−
tumors (Fig.
5d). The mechanism by which Net1 regulates recruitment of PP2A and Src to PyMT is at present unclear, as there is no precedence for regulation of these events by Rho GTPases.
Net1 knockout tumor cells tended to express slightly more PyMT than wild-type cells (Figs.
2 and
5), so reduced recruitment cannot be due to effects on PyMT expression. One possibility is that loss of Net1 inhibits delivery of signaling molecules to the plasma membrane. RhoB has been shown previously to regulate EGF-mediated delivery of Src to the plasma membrane [
56]. Moreover, RhoB has been reported to interact with the catalytic subunit of PP2A and to control its ability to recruit the B55 regulatory subunit [
57,
58]. Thus, it may be that Net1-dependent recruitment of PP2A
A,C to PyMT is also RhoB dependent. In the future it will be important to test whether Net1 is required for recruitment of Src or PI3K to activated RTKs.