Introduction
Elevated expression of FAK has been associated with highly invasive human breast cancers [
1,
2]. In particular, several groups have reported a correlation between FAK and human epithelial growth factor receptor (ErbB2, Neu) overexpression in ErbB2-positive human breast cancer [
3‐
5]. Activation of FAK has also been observed in human breast cancer cell lines expressing elevated levels of ErbB2 [
6,
7]. Moreover, recent studies have indicated that FAK and the related kinase Pyk2 are expressed in ErbB2-positive breast cancer and contribute to the proliferative and invasive potential of breast cancer cell lines [
8,
9].
Direct evidence for the
in vivo importance of FAK in tumourigenesis derives from several recent studies in which components of the integrin signaling pathway were selectively ablated in the germline of mice. For example, mice heterozygous for a FAK null allele exhibit a dramatic delay in tumour induction in a chemically-induced skin carcinogenesis model [
10]. Because germline deletion of FAK results in embryonic lethality [
10,
11], it is difficult to assess whether complete ablation of FAK could impact on chemical skin carcinogenesis. To circumvent this limitation, the same group demonstrated that conditional ablation of FAK in the skin resulted in an absolute block in the progression of benign papilloma lesions to malignant carcinomas in this model [
12]. More recently, it has been shown that prostate-specific ablation of FAK in an SV40 T antigen mouse model resulted in the inability of prostate tumours to progress to the aggressive neuroendocrine phenotype [
13].
Although these studies have largely focused on tissues such as skin, there is compelling evidence suggesting that activation of FAK is directly involved in the induction of mammary tumours
in vivo. In the polyomavirus middle T (PyVmT) model of mammary tumour progression, mammary epithelial disruption of FAK prevented the transition of mammary hyperplastic growths into mammary adenocarcinomas [
14]. This result was subsequently confirmed by a number of independent laboratories [
15‐
17]. By contrast, another group claimed that FAK function in PyVmT tumour progression played a critical role in the initial progression of primary epithelium to the hyperplastic state [
17]. The minor difference between these groups likely reflects the timing at which FAK-deficient lesions were monitored. Another suggested explanation is that the block in PyVmT tumour progression incurred by abrogation of FAK signaling was due to a deficit in the tumour-initiating cell population [
15]. Taken together, these observations confirm that FAK plays a critical role in converting PyVmT mammary epithelial hyperplasias into the malignant phenotype.
Given the dramatic impact of FAK deletion on PyVmT tumour induction, we evaluated whether deletion of FAK in an activated ErbB2 mouse model resulted in a comparable phenotype. To accomplish this, we first intercrossed the mouse mammary tumour virus (MMTV)-activated ErbB2 strain (NDL2-5) to separate strains of mice bearing the MMTV-Cre and conditional FAK alleles and monitored virgin female cohorts for the development of mammary tumours. Although a delay in mammary tumour onset was observed, all animals developed mammary tumours that were highly metastatic. Genetic and biochemical analyses revealed that the majority of these tumours had escaped Cre-mediated recombination of the conditional FAK alleles and were proficient for FAK expression, indicating a strong selection for retention of FAK signaling. Cre-mediated deletion of FAK in established ErbB2 mammary tumour cells resulted in a profound delay in tumour growth that was further associated with a proliferative defect. Finally, using a transgenic mouse model that co-expressed ErbB2 and Cre recombinase, we demonstrated that while mammary epithelial ablation of FAK delayed tumour onset and reduced the number of neoplastic lesions, animals developed mammary tumours with 100% penetrance. Collectively these observations argue that while FAK contributes to ErbB2 tumour cell proliferation, it is ultimately dispensable for ErbB2 mammary tumour initiation and progression.
Materials and methods
Transgenic mice
MMTV-Cre and GTRosa26 mice were described previously [
18,
19]. MMTV-NDL2-5, MMTV-NIC and FAK
flox mice were generated and characterized as described [
12,
20,
21]. FAK
flox mice were interbred with MMTV-NDL2-5, MMTV-Cre and MMTV-NIC mice and routine genotyping was performed by PCR as described [
14]. Experimental and control virgin female mice were monitored for mammary tumour formation by twice-weekly palpation. All experiments involving mice were conducted in accordance with McGill University Facility Animal Care Committee guidelines which approves the use of animals (mice) for research.
Tissue preparation, histology and β-galactosidase assay
The No. 4 inguinal mammary glands were used for wholemount analyses and the No. 3 thoracic glands for histological analyses. Mammary gland wholemounts were prepared as previously described [
14]. For
in situ β-galactosidase assays, lungs were processed as described previously [
22]. For histological analyses, mammary glands, tumour samples and lung lobes were fixed in 10% neutral buffered formalin (Leica Microsystems Inc., Concord, ON, Canada) and transferred to 70% ethanol the next day. Samples were then paraffin-embedded, sectioned at 5 μm and H&E-stained. Quantitative image analysis of the mammary gland wholemounts and H&E-stained sections was performed using the Aperio ImageScope software (Vista, CA, USA). For lung lobe examinations, five step sections were performed at 50 μm intervals on lungs harvested from mice bearing tumours for 8 weeks. Pulmonary metastases were identified by microscopic analyses of H&E- or X-gal-stained sections.
Immunostaining of tissue sections
For immunohistochemistry (IHC) and immunohistofluorescence (IHF), tissue sections were deparaffinized in xylene, endogenous peroxidase activity was blocked with 3% hydrogen peroxide in methanol, and antigen retrieval was done in 10 mmol/L sodium citrate (pH 6) using a pressure cooker (Cuisinart, Woodbridge, ON, Canada). Sections were then blocked with Power Block Universal Blocking Agent (Biogenex, Fremont, CA, USA), incubated in primary antibody as described previously [
14], then incubated with biotinylated (Vector Laboratories, Burlington, ON, Canada) or Alexa Fluor 488 and 555 (Invitrogen/Life technology, Grand Island, NY, USA)-conjugated secondary antibodies for IHC or IF, respectively. Primary antibodies used for immunohistology were Cre recombinase (PRB-106, Covance, Denver, PA, USA), cleaved caspase-3 (9661, Cell Signaling, New England Biolabs, Pickering, ON, Canada), Ki67 (NCL-L-Ki67-MM1, NovoCastra, Leica Microsystems Inc.), and FAK (05-537, Millipore, Billerica, MA, USA). For IHF analysis, the slides were visualized using a Zeiss LSM 510 META confocal microscope (Carl Zeiss Canada, Toronto, ON, Canada).
Immunoblotting and immunoprecipitations of tumour tissue
Lysates were prepared from mammary tumour tissues flash frozen in nitrogen, and immunoblots were performed as described previously [
20]. Antibodies used for immunobloting analysis included Sigma (Oakville, ON, Canada) β-actin (clone AC-15, A5441) and Pyk2 (P3902), as well as Millipore phosphotyrosine (clone 4G10, 05-321) and c-Src (clone GD11, 05-184). Neu (sc-284) and CK8 (sc-101459) antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA, USA) and Paxillin (610051), FAK (610087) and p130Cas (610271) were from BD Biosciences (Mississauga, ON, Canada). Phospho-p130Cas (4011), AKT (9272), phospho-AKT (9271), ERK1/2 (9102), phospho-ERK1/2 (9106), phospho-c-Src (2101), and phospho-Paxillin (2541) antibodies were purchased from Cell Signaling. All membranes were incubated with horseradish peroxidase-conjugated secondary antibodies (Jackson immunoresearch Laboratories, West Grove, PA, USA) and visualized using enhanced chemiluminescence (GE Healthcare Life Sciences, Baie d'Urfé, QC, Canada). Immunoprecipitations of lysates from mammary tissues were performed as described previously [
20]. Pyk2 and phospho-p130Cas were immunoprecipitated with the same antibodies used for immunoblotting.
Primary cell culture, transfection and viral transduction
Tumours at 8 weeks post-palpation were processed using a McIlwain tissue chopper (Mickle Laboratory Engineering, Guildford, Surrey, UK), dissociated in collagenase B/dispase II (Roche, Mississauga, ON, Canada) for 2 hours at 37°C, washed with PBS + 1mM EDTA and plated in DMEM supplemented with 2% FBS and Mammary Epithelial Growth Supplement (MEGS - Invitrogen). pMSCV retroviral vector containing Cre recombinase or empty pMSCV vector were used for virus production in 293VSV cells and subsequent transductions were carried out as previously described [
23]. Transduced cell lines were selected and maintained in DMEM + mammary epithelial growth supplement (MEGS) with 2μg/ml puromycin (Sigma).
Proliferation, migration and invasion assays
CellTiter Aqueous MTS (Promega, Madison, WI, USA) proliferation assays were performed according to the manufacturer's protocols using 2,500 cells per well in 96-well optical bottom plates (Nalge Nunc International, Rochester, NY, USA). For migration and invasion assays, cells were seeded in serum-free medium in transwell chambers (Falcon, BD Biosciences) with complete medium as a chemoattractant, and for the invasion assay chambers were coated with 5% matrigel (BD Biosciences). Cells were incubated for 24 hours, formalin-fixed, and stained with crystal violet. Cells that passed through the membrane were visualized by microscopy and pixel counts were determined using ImageJ software.
Cell spreading assay
Cells of approximately 80% confluence were trypsinized, washed in serum-free DMEM/0.5% BSA + MEGS containing 100ug/ml soybean trypsin inhibitor (STI), and re-suspended with serum-free DMEM/0.5% BSA + MEGS. Suspension cells were seeded at 5 × 104 cells per well and allowed to adhere for either 15 or 120 minutes at 37°C in 24-well plates coated with fibronectin (BD Biosciences, #354411). At various time points, wells were washed twice with PBS and adherent cells were fixed with 10% neutral buffered formalin and then stained with crystal violet (0.1%). After extensive washing to remove the free dye, the stained cells were visualized by microscopy and cell spreading was determined by counting at least 200 cells.
Mammary fat pad injection assay
Stable cell lines (2.5 × 105 cells in PBS) were injected into the fourth inguinal mammary fat pad of athymic nude mice (NCr; Taconic, Hudson, NY, USA). Mice were monitored twice weekly for tumour formation by palpation. Tumour growth was assessed by weekly caliper measurements.
Discussion
Elevated expression of FAK has been previously implicated as an important determinant in both the initiation and the progression of ErbB2-positive breast cancers [
6,
7]. Consistent with these observations, several
in vitro studies with ErbB2-transformed mammary tumour cells have identified FAK and its closely related family member Pyk2 as critical signals in ErbB2-dependent proliferation [
8,
9,
17]. Although these studies provide compelling data supporting a role for FAK in ErbB2 tumour progression, whether FAK is required
in vivo for ErbB2 mammary tumour progression has not been addressed; this question is imperative given that the integrin/FAK signalling output from ErbB2 tumour cells is highly dependent on the microenvironment within the mammary fat pad. Here we demonstrate using two well-characterized transgenic mouse models of ErbB2 breast cancer that while the presence of functional FAK confers a proliferative advantage to ErbB2 tumour cells, metastatic ErbB2 mammary tumours can be generated in the complete absence of FAK expression.
To evaluate the role of FAK in ErbB2 mammary tumour progression, we initially interbred the MMTV-activated ErbB2 strain (NDL2-5) [
20] with separate strains of mice bearing MMTV-Cre and conditional FAK alleles [
12,
18]. Although a delay in mammary tumour onset was observed (Figure
1), the majority of tumour epithelia maintained expression of FAK due to loss of Cre expression (Figure
2). This phenomenon persisted throughout the progression to metastasis since we could not detect any FAK-deficient lesions in the lungs of FAK
flx/flx tumour-bearing mice (Figure
3). One potential explanation for the presence of these escapee populations of tumour cells that have not undergone Cre-mediated recombination is that the few FAK-ablated tumour epithelia detected were at a proliferative disadvantage relative to their FAK-proficient counterparts (Figure
2). This observation is consistent with results from similar experiments in the PyVmT tumour model system where FAK was shown to influence the proliferative status of these tumour cells [
14,
16,
17]. Likewise, mammary disruption of β1-integrin or c-Src signalling pathways (which are immediately upstream of FAK) also impairs PyVmT tumour cell proliferation [
22,
23]. Collectively these data argue that the integrin/c-Src/FAK signalling network is a key regulator of the proliferative capacity of both ErbB2 and PyVmT mammary tumour cells.
While these studies indicate the importance of FAK in the initiation phase of ErbB2 mammary tumour progression, acute deletion of FAK in established tumour cells also impacts on their ability to proliferate in both
in vitro and
in vivo settings (Figure
4) (Figure S2a in Additional file
2). In fact, from a population of tumour cells that were largely FAK-deficient (Figure
4a), the tumours that eventually arose regained FAK expression (Figure S2c in Additional file
2). The presence of FAK protein suggests that these tumour cells presumably derived from the small population of cells that had not undergone productive excision of the FAK conditional allele; alternatively, it could also represent FAK from the stromal compartment of the mammary gland which was the case in the FAK-null/MMTV-NIC tumours (Figure
6a and
6b). Again these data suggest that retention of a functional FAK gene confers a strong proliferative advantage to ErbB2 tumour cells over their FAK-deleted counterparts.
Although these data demonstrate that FAK activation is involved in the induction and maintenance of ErbB2 tumours, our studies with the MMTV-NIC model indicate that ErbB2 tumours completely lacking FAK can develop eventually. By contrast, deletion of FAK in the PyVmT model prevented tumour progression entirely [
14]. One possible explanation for this difference is that the tumour-initiating cell population in the PyVmT model is absolutely reliant on FAK, whereas this population in the ErbB2 model can function independently of FAK signalling. Consistent with this hypothesis, FAK function has been reported to be important in tumour-initiating cells in the PyVmT model [
15].
Examination of downstream signaling pathways revealed that loss of FAK resulted in a reduction in tyrosine phosphorylation of p130Cas (Figure
6). However, in contrast to complete loss of tyrosine phosphorylation of p130Cas in an ErB2 model with abrogation of either β1-integrin or integrin linked kinase (ILK) [
26,
27], FAK-deficient tumours still retained a basal level of tyrosine-phosphorylated p130Cas that was further correlated with its ability to couple with the FAK-related kinase Pyk2 (Figure S4 in Additional file
4 and Figure S5d in Additional file
5). Although it has been proposed that FAK was critical for migratory behaviour whereas Pyk2 was involved in the proliferative phase of ErbB2 tumour induction
in vitro [
8], the observation that FAK-null ErbB2 tumours can form metastatic lesions argues that at least in the
in vivo setting, Pyk2 can functionally substitute for FAK. One of the main targets for Pyk2 is likely p130Cas, as it has been reported that p130Cas is essential for ErbB2 tumour progression [
28]. While these data implicate the Pyk2/FAK/p130Cas signalling complex as an important signalling element in ErbB2 tumour progression, both ILK- and β-1 integrin-ablated ErbB2 tumours lack detectable tyrosine phosphorylated p130Cas [
26,
27] indicating that tyrosine phosphorylation of the p130Cas scaffold is not essential for ErbB2 tumour induction. However, tumours from both these models are poorly metastatic [
26,
27] suggesting that engagement of Pyk2/FAK/p130Cas signalling likely plays a more important role in the metastatic phase of ErbB2 tumour induction. Consistent with this concept, upregulation of Pyk2 has been observed in metastatic foci derived from PyVmT cancer stem cells devoid of FAK function [
25].
One of the most interesting findings of these studies was the observation that acute knockout of FAK from established ErbB2 tumour cells (Figure
4) was more deleterious than early abrogation of FAK function in the MMTV-NIC model (Figure
6). A hypothesis for this difference in response between these two models is that established ErbB2 tumour cells have become addicted to the requirement for FAK function whereas MMTV-NIC-initiated tumour cells have evolved to circumvent the FAK signalling network through compensation by Pyk2. Indeed, the phenomenon of oncogene
addiction has been observed in a number of human cancers [
29].
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
HL and VSG were involved in generating all primary data in the paper and contributed to the writing of manuscript. MF provided the FAK conditional strain. WJM was involved in experimental design and contributed to the writing of manuscript. All authors read and approved the final manuscript.