Introduction
Total protein tyrosine kinase (PTK) activity is elevated in breast cancer [
1] and this condition is associated with poor prognosis [
2]. PTKs and their downstream signaling pathways contribute to critical biological functions relevant to the cancerous phenotype, such as increased cellular proliferation, pro-survival, invasion and migration/metastasis. One such cancer-associated PTK is breast tumor kinase/protein tyrosine kinase 6 (Brk/PTK6). Brk was cloned in a screen for tyrosine kinases expressed in a metastatic breast tumor [
3]. The murine Brk-ortholog, Src-like intestinal kinase (Sik), was independently cloned from the small intestine and skin and found to share 80% identity with Brk [
4]. Although considered to be only distantly related to c-Src, Brk shares a similar domain structure, consisting of an N-terminal SH2 domain, an SH3 domain, and a C-terminal kinase domain that is subject to autophosphorylation and autoinhibition [
5]. However, the Brk C-terminus lacks a motif required for myristoylation (that is, as found in c-Src), rendering it truly "soluble" or mobile within and between cellular compartments [
3,
4].
Brk is overexpressed in up to 86% of invasive ductal breast carcinomas [
6,
7], prostate and colon carcinomas [
8,
9], 70% of serous ovarian carcinomas [
10], 37.5% of a limited sampling of head and neck squamous cell carcinomas [
11], and a small percentage of metastatic melanomas [
12]. Brk expression levels increase in association with the carcinoma content of breast tumors [
7], tumor grade [
13], and invasiveness of breast cancer cell lines [
14]. Normal tissues that express Brk include the intestinal epithelium, melanocytes, keratinocytes [
4,
15], prostate luminal epithelium [
16], and lymphocytes [
17]. However, Brk appears to be absent from normal mammary tissue [
8]. The list of Brk substrates and interacting proteins is limited, but consists largely of signaling or signal transduction-related adaptor molecules, and RNA- or DNA-binding proteins, including signal transducers and activators of transcription (STATs). Notably, both STAT3 and STAT5b have been shown to be direct substrates of Brk
in vitro [
18,
19]. These molecules are also required regulators of mammary gland lactogenic differentiation (STAT5, [
20]) and regression (STAT3, [
21]).
Mammary gland development is a highly dynamic and hormonally-driven process; functional glands are not fully mature until early adulthood or pregnancy. Beginning as an invagination of dermal epithelium (that is, in the embryo), the mammary anlage migrates into the mesenchyme, eventually elongating into a rudimentary branched ductal tree [
22]. The gland remains at this primitive state until puberty, when the terminal end buds (TEBs) respond to hormonal cues and lead the advancement and secondary branching of the ductal network further into the mammary fat pad resulting in a network of hollow ducts. Outside of fluctuations in secondary branching due to cycling hormonal cues during the estrous cycle [
22], further functional differentiation is temporarily halted until pregnancy. Upon pregnancy, a marked increase in ductal branching and alveolar proliferation and differentiation occurs, preparing the gland for lactogenesis.
Once the suckling stimulus of the offspring is removed, involution is initiated. In mouse models, completion of this regressive 10-day process returns the gland to a near virgin state. Post-lactational involution is characterized by events that can be classified into two distinct stages. First, milk stasis and the resulting mechanical stresses initiate a tightly regulated wave of apoptosis in alveolar epithelial cells and their concomitant removal [
23], followed by the second stage in which remodeling of the ECM and the expansion of the stromal adipocyte compartment occurs [
24]. Mouse models have been extensively used to understand genetic mechanisms of breast cancer biology [
25]. Indeed, classical models of human breast oncogene overexpression in the mouse mammary gland demonstrate altered biological processes responsible for proper ductal and alveolar development, as well as modified initiation and execution of glandular involution [
26].
In this study, we describe the first transgenic model of mammary gland specific (that is, WAP-promoter-driven) Brk expression. Using newly created Brk-WAP transgenic mice, we studied the physiological process of mammary gland involution to investigate the impact of Brk expression on the survival of mammary luminal epithelium, and altered regulation of pro-survival signaling pathways that may be permissive for mammary tumorigenesis.
Materials and methods
Mice and tissues
Transgenic mice expressing the human Brk/PTK6 gene under the control of the whey acidic protein (WAP) promoter were generated by microinjection of a WAP-Brk insert containing the wild-type Brk cDNA under the control of the WAP gene promoter into FVB/n embryos (University of MN Mouse Genetics Laboratory). The Brk cDNA was subcloned into the WKbpAII vector (a kind gift of Dr. Jeff Rosen, Baylor College of Medicine; [
27]) using EcoR1 sites within the multiple cloning sequence. Two founders (one female, one male) were identified by PCR screening of tail biopsy DNA, and confirmed by Southern blotting (data not shown). Primer sequences for genotyping transgenic animals (by collection of DNA harvested from tail biopsies) span the Brk coding sequence (sense, 5'-agcgtgcacaagctgatgct-3') and the bovine growth hormone poly-A region of the transgene (antisense, 5'-tctctggctgtctgtctgca-3'). Experiments were conducted under University of Minnesota IACUC approved protocols and NIH guidelines.
Involution timecourse
Virgin FVB/n or WAP-Brk mice were bred; litters were carried to term and normalized to eight pups upon parturition. Pups were nursed for 10 days at which time the litter was force weaned. Mammary glands were harvested at one day post-weaning; involution Day 1 (INV1), INV4, INV6, INV9, and INV14.
Whole mounts
Inguinal mammary glands were harvested and fixed, washed with PBS and stained with Carmine Alum. Glands were then dehydrated in graded ethanols, cleared with xylenes, and affixed to slides.
IHC and differential stains
A Leica Microsystems 1020 automated processor was used to process tissues after fixation. After paraffinization, three- to five-micron thick sections were cut and mounted on slides.
Imaging and analysis
Digital images were taken of three fields per gland from three glands at 200 × or 400 × total magnification. For epithelial content determination, a grid of 360 boxes was overlaid on 200 × images and boxes containing epithelial cells were counted. For IHC quantification, NIH ImageJ [
28] was used with a cell counter plug-in to manually count positively stained mammary epithelial cells vs. total epithelial cells in multiple fields. Annotated regions were drawn on each digital H&E image using a pen tablet (Intuos3, Wacom, Kazo-shi, Saitama, Japan) for area calculations by determining epithelial pixel count relative to the entire gland, and selecting regions of interest for digital IHC analysis. For digital IHC quantification, slides were scanned at 40 × magnification (0.25 microns/pixel) using a whole slide scanner (ScanScope CS, Aperio Technologies, Vista, CA, USA) fitted with a 20x/0.75 Plan Apo objective lens (Olympus, Center Valley, PA, USA). Images were saved in SVS format (Aperio) compressed with JPG2000 at 70% quality and retrieved from a secure server using whole slide image management software (Spectrum, Aperio).
For automated quantification of molecules visualized by IHC, five annotated regions were drawn on each slide using a pen tablet screen (Cintiq 21UX, Wacom, Kazo-shi, Saitama, Japan) on whole slide images viewed at high resolution using the Aperio system's annotation software (ImageScope 10, Aperio).
To detect individual cells in tissue sections, a nuclear cell quantification image analysis algorithm (IHC Nuclear Quantification, Aperio) was trained on control slides by defining the color vectors for the hematoxylin nuclear counterstain and primary positive chromagen DAB, minimum and maximum size for nuclei, and threshold ranges for intensity of nuclear staining. The analysis algorithm was trained to detect nuclei in four intensity ranges for cells with no positive staining, weak positive staining, medium positive staining, and strong positive staining. Analyses were performed on each annotated region using defined settings and nuclear count results were collected from each slide. Data were represented as an H-score [
29], which accounts for staining intensity and percentage of positively stained cells. The H-score = (% of 0 intensity staining nuclei*0) + (% of 1 intensity staining nuclei)*1 + (% of 2 intensity staining nuclei)*2 + (% of 3 intensity staining nuclei) *3. Each H-score represents five fields each from three mice per time point.
Mammary epithelial cell enrichment
Mammary glands were harvested and weighed. Following disruption with scalpels, tissue homogenates were incubated at 37°C in digestion buffer (Ham's F12/DMEM, 2 mg/mL collagenase A (Roche, Indianapolis, IN, USA), 100 U/ml hyaluronidase (Sigma-Aldrich, St. Louis, MO USA)). Digested mammary tissue was pelleted and washed with Ham's F12/DMEM+1% serum three times at 1, 500 rpm, then twice at 800 rpm. Cell pellets were lysed as in [
30] with the addition of Roche PhosStop and Complete tablets.
Immunohistochemistry (IHC)
Formalin fixed paraffin embedded (FFPE) sections of mammary glands were deparaffinized with xylenes, and rehydrated through graded alcohols (70% to 100%). Rehydrated sections were equilibrated in PBS and microwaved in antigen retrieval buffer (10 mM sodium citrate, pH 6.0 for 20 minutes or 1 mM EDTA, pH 8.0 for 10 minutes). Slides were washed with ddH2O then PBS and placed in 3% H2O2 for 10 minutes to block endogenous peroxidases.
Sections blocked with serum-free protein block (Dako X0909) were incubated overnight at 4°C with primary antibodies diluted in Dako Antibody Diluent (S0809), washed with PBST and incubated in biotinylated secondary antibody (Vector Laboratories Vectastain Elite Kit, PK-101) for 30 minutes at room temperature. Slides were washed, then incubated with Vectastain Elite RTU ABC reagent (PK-7100) and subjected to colorimetric detection with ImmPACT DAB substrate (Vector Laboratories, SK-4105). Antibodies used for IHC are as follows: Brk was purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA); phospho-p38 mitrogen-activated protein kinase (MAPK), phospho-STAT3, phospho-STAT5, and cleaved caspase 3 were purchased from Cell Signaling (Danvers, MA, USA).
Growth factors and cell lines
HC11 murine mammary epithelial cells were plated and transfected with pCMV-3X-FL-Brk constructs using FuGene HD (Roche), serum starved post-transfection, and treated with 500 ng/mL prolactin (Sigma). HMEC-Brk and T47D shRNA stable cell lines described previously [
7,
31] were treated with 25 ng/mL epidermal growth factor (Sigma). Cells were lysed as previously described [
30]. Immunoblotting was performed with Brk (Santa Cruz, in-house antibody), total and phospho STAT5 and total and phospho-p38 MAPK (Cell Signaling), and E-cadherin antibodies.
Anchorage independence
Six-well dishes were coated with PolyHEMA (20 mg/mL in 98% EtOH) and dried in an incubator overnight. Previously described HMEC+Brk cells [
31] or HC11 cells transiently transfected with Brk (as above) were plated at a density of 300 K cells per well, and maintained in culture for 48 hr. Cells in suspension were collected, trypsinized and stained with 0.4% Trypan Blue. Viable cells from each sample were counted in triplicate.
Tissue microarray
A series of human breast tissue samples surgically obtained from healthy women undergoing reduction mammoplasty (n = 23), or with pathological conditions including fibroadenoma (n = 22), infiltrating ductal carcinoma (n = 23) and infiltrating lobular carcinoma (n = 23) were made available as FFPE archival material from the Third Medical Faculty (Charles University, Prague, Czech Republic). The original slides were re-evaluated by a pathologist (DH) to confirm the initial pathology diagnosis, and representative tissue blocks were selected for further processing. Informed consent was obtained, and the use of biopsy material for research was approved by the Ethics Committee of the Third Medical Faculty.
Tissue microarrays were constructed from routinely prepared FFPE tissue blocks in parallel, using a manual tissue arrayer TA1 (A Fintajsl, Czech Republic). The representative area of interest was selected on the original glass slide and corresponding area on donor tissue block was inked. Tissue cylinders, 1.6 mm in diameter, were punched from the marked regions of each donor tissue block, and transferred to a recipient block for the array. One hematoxylin and eosin section was made from each block to ensure the presence of tumor regions.
Scoring of positively stained regions was performed with arbitrary establishment of a threshold for positive IHC staining intensity. The scores consist of 0 = no staining relative to no primary controls, 1 = weak diffuse staining, 2 = moderate diffuse staining, 3 = strong diffuse staining. Any focal staining present increased the score by 1 (that is, a weakly diffuse stain with regions of strong focal staining was scored a 2). Only the adenoma or carcinoma compartment was scored, except in the reduction mammoplasty group, where the entire section was analyzed.
Statistics
Unless otherwise noted, all results are presented as means +/- SEM. Paired t-tests were conducted on IHC quantification and in vitro assays. Tumor incidence was compared between WAP-Brk and wt mice using Fisher's exact test. Tumor latency was estimated using Kaplan-Meier methodology and curves compared between WAP-Brk and wt mice using the Wilcoxon test. A chi-squared test was used to compare the association of tumors staining positive for phospho-p38 MAPK with tumors staining positive for Brk. All statistical tests were conducted at a significance level of 0.05.
Discussion
Herein, we report the first transgenic mouse model of mammary specific (inducible) Brk expression (Figure
1). We illustrate a delay in mammary gland involution following forced weaning (Figure
2). We detected evidence of Brk-mediated signaling through increased phospho-p38 MAPK (Days 4 and 6; Figure
5). Brk expression also partially prevented anoikis in non-transformed HMEC and HC11 cell lines
in vitro (Figure
7a). Aged, multiparous WAP-Brk mice exhibited a trend towards higher tumor incidence and significantly decreased tumor latency relative to wild-type mice (Table
1); these tumors were Brk-positive (Figure
8b). Finally, we detected significant association of phospho-p38 MAPK in biopsies of Brk-positive human breast cancer (Figure
9 and Table
2). Our studies suggest that Brk confers p38-associated pro-survival signals to non-transformed (luminal) mammary epithelium. Given time, these events may conspire to induce or permit the formation of latent mammary tumors (Figures
8 and
9).
Involution
Mouse mammary gland involution represents a highly sensitive read-out of human oncogenic action; numerous breast oncogenes induce delayed involution in mouse models [
26]. Similar to other models of mammary oncogene expression [
34,
35], our model undergoes delayed, but ultimately complete mammary regression, highlighting a distinct window of mammary signaling events (Days 4 to 6 of involution) that are perturbed without completely halting the involution process. We observed fewer apoptotic figures, decreased caspase3 cleavage, and reduced TUNEL staining in glands from WAP-Brk transgenic mice, while clearance of apoptotic mammary epithelial cells did not appear to be affected. Notably, WAP-Brk transgenic mice eventually undergo complete mammary regression, consistent with the decline of WAP-driven Brk expression over the time course of involution in this model. Upon multiple rounds of parity induced mammary expansion and contraction, amplified survival signaling may increase the chances for mammary epithelial cells to encounter and fix potentially oncogenic combinatorial events.
Tumor biology (in vivo model and human tumors)
Inducible Brk expression in our WAP-driven transgenic model results in a tumorigenesis rate of 30% in aged multiparous mice (Table
1 and Figure
8). Two wild-type FVB mice from the same litter also developed tumors. Indeed, this strain has a weak propensity to develop adenosquamous mammary tumors at an advanced age [
43]. Because of the sibling wild-type FVB tumors, the comparison of the number of tumors between wild-type and Brk transgenic animals did not reach statistical significance (
P = 0.13). However, the age at tumor onset decreased (Table
1) and this reduced tumor latency was significantly different from wild-type controls (
P = 0.03), indicating an effect of Brk expression on the promotion of tumorigenesis relative to wild-type FVB mice. Brk strongly promotes breast cancer cell proliferation [
7,
44], survival [
13] and migration
in vitro [
7,
14,
44,
45]. We did not observe pulmonary metastatic lesions in tumor-bearing WAP-Brk mice, suggesting that other cooperating factors are necessary for invasion and migration
in vivo. We are currently crossing WAP-Brk mice with other mouse models of breast cancer in order to identify additional oncogenic events that may cooperate with Brk overexpression.
Brk protein is readily detectable in hyperplastic regions of WAP-Brk mammary tumors (Figure
8b). The loss of Brk protein in regions of squamous metaplasia of WAP-Brk tumors is likely due to the loss of mammary epithelial differentiation, an event(s) that may ultimately lead to silencing the WAP promoter. Note that Brk expression may drive the appearance of the squamous metaplasia phenotype directly, as Brk expression in the skin increases during the maturation of keratinocytes, promoting squamous differentiation of the epidermis [
46].
Brk appears to predominantly mediate cellular survival/resistance to involution-associated apoptosis in this model. This phenotype is consistent with Brk-dependent activation of p38 MAPK [
7], as measured by its increased phosphorylation (Figures
5 and
6). Elevated phospho-p38 (Days 4 and 6) was detected in our involution time course experiments,
in vitro experiments with Brk-expressing HC11 cells, and in Brk-positive tumors derived from both WAP-Brk mice (not shown) and humans (Table
2 and Figure
9b). As expected, IHC analysis of the human breast tumor tissue array revealed Brk expression in only one (a fibroadenoma) of the 84 non-transformed tissue samples (Figure
9a), and in Brk-positive tumors, Brk expression was significantly associated with increased phospho-p38 (Table
2); the samples in this group were mostly derived from premenopausal women.
The separation of normal physiological cues from transgene-mediated signaling is critical to understanding events that may contribute to mammary oncogenesis. Initial characterization of WAP-Brk mammary glands focused on STAT3 signaling as a marker of mammary gland involution. STAT3 is a required mediator of involution-related cell death [
21], and has been reported to be a Brk substrate in studies using cell lines [
18]. Our IHC analysis illustrated that glands of WAP-Brk mice contain less p-STAT3 during involution (Days 4 and 6) relative to wild-type animals (Figure
4). These results suggest that while Brk mediated STAT3 phosphorylation may be relevant to the growth of established mammary tumors (that is, breast cancer cells [
47]), forced expression of Brk does not appear to drive phospho-STAT3 during the initiation of involution. Similar to p-STAT3, Brk expression
in vivo also suppressed early (Day 4) p-STAT5 levels; rapid loss of STAT5 phosphorylation, characteristic of involution, occurred in both wt and transgenic animals (data not shown). Thus, Brk does not appear to be a positive regulator of STAT3 or STAT5
in vivo; STAT3 phosphorylation may serve primarily as an indicator of the progress of mammary involution herein. Other factors (not addressed herein) that may contribute to Brk-dependent pro-survival include amplification of signaling pathways downstream of erbB family members [
33,
48], including activation of ERK5 signaling [
7,
45]. These pathways are frequently associated with breast cancer progression and invasive tumor behavior [
49,
50].
We have previously described Brk mediated p38 MAPK signaling as primarily promoting cell migration in EGF or heregulin-treated breast cancer cell lines [
7]. However, there are limited studies investigating the role of p38 MAPK activity in mouse models of breast cancer. Demidov
et al. [
51] expressed an MMTV-driven active MKK6 (an upstream kinase in the p38 MAPK module), and showed resistance to development of ErbB2
and Wip1 induced mammary tumors; however, when overexpressed, MKK6 may regulate other MAPKs [
52]. Similar to our studies, Leung
et al. [
35] expressed MMTV-V12Rac3 and described incomplete involution associated with elevated phospho-p38 MAPK. Additionally, Wang
et al. [
53] overexpressed activated Pak1 under the β-lactoglobin promoter and reported a 20% tumorigenesis rate and elevated phospho-p38 MAPK. In both of these studies, as well as ours, there were detectable levels of phospho-p38 in wild-type cohorts, strongly suggesting an as of yet under appreciated physiological role for p38 MAPK in mammary gland biology. Mammary glands from WAP-Brk transgenic mice exhibited higher phospho-p38 levels relative to wild-type glands during Days 4 and 6 of the involution time course (Figure
5), again consistent with an increased survival stimulus in WAP-Brk mice. Interestingly, expression of Brk in HC11 or HMEC cells increased basal phospho-p38 in serum-starved cells (Figure
6), indicating that the presence of Brk is sufficient to promote p38 MAPK activation and survival of mammary epithelium. These data suggest that p38 phosphorylation induced by Brk expression in non-transformed mammary epithelium could contribute to breast disease as either an early event (allowing pro-survival and/or luminal filling) or late event (migration/dissemination, therapy resistance) in tumorigenesis, thereby leading to a poor prognosis. Recent literature [
13] and our observations in HMEC and HC11 cell lines (Figure
7) illustrate that Brk promotes anchorage independent survival. Importantly, this has been shown to be a p38 MAPK-dependent phenotype in Brk-positive MDA-MB-468 cells [
54]. Taken together, these data suggest that Brk-mediated p38 activation is likely a critical node for mammary epithelial cell pro-survival and relevant to early oncogenic signaling; p38 inhibitors may present an opportunity for therapeutic intervention aimed at long term breast cancer prevention and/or increased sensitivity to chemotherapeutic agents.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
KAL, KLS and CAL designed the experiments. KAL and GKH purified mammary epithelium, while KAL, DH and GKH performed IHC and histology. KAL, JHO and AL performed in vitro experiments. DH prepared breast tumor tissue arrays, and KAL and DH analyzed pathology data. RB performed statistical analysis on animals and TMA studies. KAL and CAL wrote the manuscript.