Cancer-associated fibroblasts (CAFs) play a role in tumour initiation and progression, possibly by inducing epithelial-to-mesenchymal transition (EMT), a series of cellular changes that is known to underlie the process of metastasis. The aim of this study was to determine whether CAFs and surrounding normal breast fibroblasts (NBFs) are able to induce EMT markers and functional changes in breast epithelial cancer cells. Matched pairs of CAFs and NBFs were established from fresh human breast cancer specimens and characterised by assessment of CXCL12 levels, α-smooth muscle actin (α-SMA) levels and response to doxorubicin. The fibroblasts were then co-cultured with MCF7 cells. Vimentin and E-cadherin expressions were determined in co-cultured MCF7 cells by immunofluorescence and confocal microscopy as well as by western blotting and quantitative PCR. Co-cultured MCF7 cells were also assessed functionally by invasion assay. CAFs secreted higher levels of CXCL12 and expressed higher levels of α-SMA compared with NBFs. CAFs were also less sensitive to doxorubicin as evidenced by less H2AX phosphorylation and reduced apoptosis on flow cytometric analysis of Annexin V compared with NBFs. When co-cultured with MCF7 cells, there was greater vimentin and less E-cadherin expression as well as greater invasiveness in MCF7 cells co-cultured with CAFs compared with those co-cultured with NBFs. CAFs have the ability to induce a greater degree of EMT in MCF7 cell lines, indicating that CAFs contribute to a more malignant breast cancer phenotype and their role in influencing therapy resistance should therefore be considered when treating breast cancer.
Increasing evidence suggests that fibroblasts play a role in cancer initiation and progression (Elenbaas & Weinberg 2001, Tlsty & Hein 2001, Bhowmick et al. 2004, Franco et al. 2010). In particular, a subpopulation of fibroblasts, known as cancer-associated fibroblasts (CAFs), is important in tumour growth and progression (Mueller & Fusenig 2004). These can be distinguished functionally from ‘normal’ fibroblasts in tissue distal from the tumour site. CAFs are believed to originate from several sources including the resident fibroblast population, tumour epithelial cells (through epithelial-to-mesenchymal transition or EMT) and mesenchymal stem cells (Anderberg & Pietras 2009). Olumi et al. (1999) found that co-culture of non-tumorigenic human prostate epithelial cells with human prostate CAFs, and not normal prostate fibroblasts, dramatically stimulated growth and altered histology of the epithelial cells. Likewise, Hayward et al. (2001) showed that co-culture of the non-tumorigenic prostate cell line BPH-1 with CAFs resulted in permanent malignant transformation of these epithelial cells. The importance of the tumour microenvironment in cancers has also been demonstrated by the ability of a stromal derived signature to accurately predict disease outcome in breast cancer (Wald et al. 2011).
In patients with breast cancer, as in patients with other solid tumours, metastatic disease rather than the primary tumour is the main cause of death (Sporn 1996). EMT, a process in which a normal polarised epithelial cell undergoes multiple biochemical changes to assume a mesenchymal cell phenotype, including acquisition of migratory and invasive properties and the loss of cell–cell adhesion and cell polarity (Kalluri & Weinberg 2009), is thought to underlie tumour progression and metastasis. Epithelial cells that have undergone EMT to adopt a mesenchymal phenotype are distinguished by a loss of expression of E-cadherin and expression of mesenchymal markers, which include N-cadherin, ZEB1, SIP1 and vimentin. There is also a change in morphology of the cells to a spindle appearance with loss of cohesion (Gregory et al. 2008). In breast cancer, expression of EMT inducers such as Snail and Slug has been found to correlate with higher tumour grade and basal-like subtype, both of which are associated with worse prognosis (Blanco et al. 2002, Storci et al. 2008).
Several studies have shown that tumour cells that have undergone EMT, as evidenced by the expression of mesenchymal markers, tend to be more resistant to chemotherapy (Yang et al. 2006, Kajiyama et al. 2007). Epithelial cell lines from colon cancer that develop resistance to oxaliplatin have been demonstrated to acquire a mesenchymal phenotype that includes spindle appearance, increased separation between the cells and formation of pseudopodia (Yang et al. 2006). Similarly, epithelial cell lines from ovarian cancer that are resistant to paclitaxel also exhibit a mesenchymal morphology (Kajiyama et al. 2007).
Lebret et al. (2007) found that CAFs and fibroblasts from breast reduction specimens were able to induce EMT in PMC42-LA cells, a breast cancer cell line that has stem cell-like properties. Another study found that fibroblasts isolated from the interface zone located within 5 mm of the cancer as well as CAFs and normal breast fibroblasts (NBFs) were able to induce EMT in MCF7 cells (Gao et al. 2010). In addition, fibroblasts from benign hypertrophic prostates that have been activated by co-culture with aggressive prostate cancer cell lines demonstrate the ability to induce EMT in a prostate cancer cell line (Giannoni et al. 2010).
The aim of this study was to determine whether CAFs and NBFs respond differently to doxorubicin, a chemotherapeutic agent used frequently in the treatment of breast cancer patients, and to determine whether CAFs and NBFs are able to induce EMT markers in a breast epithelial cancer cell line and a phenotypically normal breast cell line as well as functional changes in the breast epithelial cancer cells.
Materials and methods
Patients and tumours
Ethics approval for the study was obtained from the Northern Sydney and Central Coast Area Health Service Human Research Ethics Committee, Sydney, Australia. Before sample collection, informed consent was obtained from patients after full explanation of the purpose and nature of all procedures.
Isolation of fibroblasts
Matched pairs of CAFs and NBFs were established from fresh human breast cancer specimens. For CAFs, tissue from within the macroscopic tumour was cultured, while for NBFs, tissue from at least 2 cm away from the macroscopic tumour margin was used. Tissue samples were collected in F-12 medium (Invitrogen) supplemented with 10% fetal bovine serum (FBS; Invitrogen), penicillin-streptomycin (Invitrogen), 50 μg/ml gentamicin (Invitrogen) and 2.5 μg/ml fungizone (Invitrogen). The tissue was minced and digested overnight at 37 °C with rotation in medium containing 0.5 mg/ml collagenase (Sigma–Aldrich, Inc.). Cells were pelleted by centrifugation, resuspended in buffer containing 0.15 M NH4Cl, 1 M KHCO3 and 0.1 M EDTA, pH 7.23, and incubated for 5 min to lyse red blood cells. After centrifugation, cells were passed through a 100 μm filter, followed by a 40 μm filter and cultured in DMEM (Invitrogen) supplemented with 10% FBS, penicillin–streptomycin and 50 μg/ml gentamicin for 1–2 weeks. Fibroblasts were separated from epithelial cells by differential trypsinisation. The cells were trypsinised for 5 min at 37 °C – fibroblasts that are more sensitive to trypsin were detached from the flask while epithelial cells remained attached.
Representative haematoxylin- and eosin (H&E)-stained slides of each cancer and of benign breast tissue from more than 2 cm away from the macroscopic lesion (corresponding as closely as possible to the sampled areas) were independently reviewed by a single pathologist (A J G) who was blinded to all other data. The pathologist classified and graded all tumours using the WHO 2012 system (Lakhani et al. 2012) and staged the tumours using the AJCC 7th edition 2009 staging system (Edge et al. 2009). Based on this morphologic review, the pathologist estimated the percentage of each cell type (invasive carcinoma, in situ carcinoma, non-adipose stromal cells (fibroblasts and myofibroblasts), adipose cells and lymphocytes) in the two sampled areas in 5% increments.
Primary fibroblasts isolated from patient tissue were maintained in DMEM containing 10% FBS at 37 °C in a humidified incubator containing 5% CO2. The human breast cancer cell line, MCF7 (American Type Culture Collection (ATCC), Manassas, VA, USA), was maintained in Roswell Park Memorial Institute (RPMI) 1640 medium containing 10% FBS. The MCF7 cells were authenticated using short tandem repeat profiling at Cellbank Australia (Sydney, NSW, Australia). Out of nine loci, 14/14 alleles were identical to the original MCF7 profile. MCF10A breast epithelial cells (ATCC) were cultured in DMEM/F12 medium (Invitrogen) containing 5% horse serum (Invitrogen), 10 μg/ml insulin (Sigma–Aldrich, Inc.), 10 ng/ml EGF (Sigma–Aldrich, Inc.) and 0.5 μg/ml hydrocortisone (Sigma–Aldrich, Inc.). Primary cells used for experiments were passaged to a maximum of six passages, with experiments performed on passage 5 and 6 cells.
Conditioned media experiments
Medium conditioned for 72 h by primary fibroblasts grown in DMEM containing 10% FBS was collected. MCF7 or MCF10A cells were seeded in six-well plates (Nunc, New York City, NY, USA) at 1×105 cells/well. After 24 h, cells were treated with a 1:1 mixture of medium conditioned by fibroblasts and fresh RPMI medium with 10% FBS for 6 or 10 days.
Indirect co-culture of fibroblasts with cancer cells
MCF7 cells (1.5×104) were seeded on a 12-well plate in 1 ml RPMI with 10% FBS and 1.5×104 fibroblasts were seeded on the 0.4 μm polyester membrane of a 12 mm transwell insert (Corning, Lowell, MA, USA) in 1 ml DMEM with 10% FBS and placed in a separate culture plate. The transwell insert was subsequently transferred to the culture plate with MCF7 cells the next day and cells were cultured in 1 ml RPMI/DMEM with 10% FBS for 6 days. Medium was replenished after 3 days of co-culture.
MCF7 cells were seeded on chamber slides (1×105 cells) and allowed to adhere for 48 h. Cells were then treated with a 1:1 mixture of medium conditioned for 72 h by fibroblasts and RPMI medium for 3 and 6 days. Control cells were treated with 1:1 mixture of RPMI and unconditioned DMEM. The cells were washed with PBS, fixed in 4% paraformaldehyde for 20 min at room temperature (RT), washed with cold PBS and incubated with 0.1% Triton X-100 at RT for 10 min. Cells were incubated with 10% normal goat serum at RT for 10 min, primary antibody (E-cadherin #AB1416 and vimentin #AB8069, Abcam, Cambridge, UK, at 1:250 dilution) at RT for 1 h and secondary antibody (anti-mouse Alexa fluor 488, Invitrogen, at 1:200 dilution) at RT for 1 h. Excess antibody was removed by washing the cells in PBS three times. Cells were stained with DAPI (1 μg/ml) and then Vectashield (Vector, Burlingame, CA, USA) was used as a mounting medium. Staining was assessed by fluorescence microscopy or confocal microscopy. For fluorescence microscopy, images were captured on an Olympus BX51 microscope (Tokyo, Japan) using the Olympus Micro DP Controller program and DP Manager was used to overlay the images. For confocal microscopy, images were captured on a Leica SP5 confocal microscope (Leica Microsystems, Wetzlar, Germany) using the Las AF software (Leica Microsystems). Quantitation of staining was performed using Adobe Photoshop CS5 (Adobe Systems, Inc.).
RNA was extracted using TRI reagent (Sigma–Aldrich, Inc.). Purified RNA (1 μg) was reverse-transcribed using Superscript III First-Strand Synthesis Supermix kit (Invitrogen) in a total volume of 20 μl according to manufacturer's instructions. The TaqMan probes used in this study were CDH1 (Assay ID #HS01023894_M1) and vimentin (Assay ID #HS00185584_M1). 18S (Assay ID #HS99999901_S1) was used as a housekeeping gene. Quantitative RT-PCRs (qPCRs) for each sample were performed in triplicate in two independent experiments. Reactions for qRT-PCR were set up using an Eppendorf epMotion5070 robotic system (Eppendorf, Hamburg, Germany) and performed on the Rotor-Gene RG-3000 real-time rotary analyser (Corbett Research, Mortlake, NSW, Australia). Cycle thresholds were chosen manually on the Rotor-Gene 6 Analysis Software. Differences between groups were assessed statistically using REST-XL©-version 2 (Relative Expression Software Tool; Qiagen; Brabletz et al. 2005) where relative expression ratios are computed based on the PCR efficiency and crossing point differences.
Cells were lysed in NuPAGE sample buffer, sonicated and equal volumes were separated by SDS–PAGE on 4–12% NuPAGE Bis–Tris gels (Invitrogen) under reducing conditions. Proteins were transferred to nitrocellulose membrane (GE Healthcare, Little Chalfont, Buckinghamshire, UK) using an XCell Blot Module (Invitrogen). Membranes were blocked in 5% skim milk powder (Diploma, Mount Waverley, VIC, Australia) dissolved in TBST (TBS/0.1% Tween 20) for 2 h at RT and then incubated with the relevant primary antibody according to the manufacturer's instructions. Immunoblots were washed three times with TBST for 5–10 min and incubated with the relevant secondary antibody conjugated to HRP. Blots were then washed three times again in TBST for 5–10 min and incubated with the ECL Plus Western Blotting Detection Reagent (GE Healthcare). The LAS-3000 system was used to capture the chemiluminescent signal (Fujifilm, Brookvale, NSW, Australia). Quantitation was performed using Multi-Gauge 3.11 software (Fujifilm).
Fibroblasts were seeded at a density of 1×105 cells/well in six-well culture plates (Nunc) and grown in serum-free DMEM medium for 72 h. Conditioned medium was collected and concentrated 50-fold using 5K MWCO concentrators (Agilent Technologies, Inc., Santa Clara, CA, USA). One hundred microlitres of concentrated conditioned medium were assayed for CXCL12 using a CXCL12 ELISA kit (R&D Systems, Minneapolis, MN, USA) following manufacturer's instructions. Any value of the concentrated conditioned medium that was less than half of the lowest standard (16 pg/ml) was assigned a value of zero before data analysis.
Fibroblasts were seeded at 1×105 cells/well in 24-well tissue culture plates and allowed to adhere overnight. Cells were then treated with vehicle, 0.5 or 1 μM doxorubicin for 6 h in serum-free medium. Medium was removed and cells were washed with PBS. Cells were then harvested using citric saline, washed once in PBS and washed in Annexin V binding buffer (BioLegend, San Diego, CA, USA). The cell pellet was resuspended in 100 μl Annexin V binding buffer, 5 μl Alexa Fluor 647 Annexin V and 10 μl 1 mg/ml propidium iodide (Sigma–Aldrich, Inc.) was added to each sample. Cells were subjected to gentle vortexing and were incubated with the dyes for 15 min at RT in the dark.
Five hundred microlitres of Annexin V binding buffer were added to each sample and analysed by flow cytometry using the BD FACSCalibur (BD Biosciences, CA, USA) and CellQuest Pro software (BD Biosciences).
Cell proliferation assay
MCF7 cells were seeded into 24-well plates at 1×105 cells/well and allowed to adhere overnight. The cells were then incubated with a 1:1 mixture of medium conditioned by fibroblasts and fresh RPMI medium with 10% FBS for 3 days. Cells were harvested by trypsinisation and viable cells were counted using Trypan blue exclusion.
MCF7 cells were cultured in control or conditioned medium from fibroblasts for 6 days. Cells were then harvested, washed twice in PBS and seeded at a density of 2×105 cells/well in the transwell of BD biocoat growth factor-reduced matrigel invasion chamber plates (BD Biosciences, Franklyn Lakes, NJ, USA) in their respective treatment medium containing 5% FBS and 5 μM of cytosine β-d-arabinofuranoside (Sigma–Aldrich, Inc.). Treatment medium containing 15% FBS and 5 μM cytosine β-d-arabinofuranoside (Sigma–Aldrich, Inc.) was used in the lower chamber. Cells were given 48 h to invade through matrigel. Medium and cells within the transwells were removed with cotton swabs. Transwells were placed in methanol for 20 min at −20 °C to fix cells to the membrane, washed with PBS and dried. Membranes were excised from the transwells and mounted onto slides using Prolong antifade gold reagent with DAPI (Invitrogen). Staining was assessed by fluorescence microscopy where images were captured on the Olympus BX51 microscope. Cells were counted using Image J 1.45S (NIH, Bethesda, MD, USA).
Experiments were generally performed in triplicate at least three times unless otherwise stated. Statistical analysis of the data was performed using ANOVA or repeated measures ANOVA where appropriate. Differences between groups were evaluated by the Fisher's protected least significant difference (PLSD) test and results were considered significant if P<0.05. The statistical package StatView for Windows Version 5.0 (SAS Institute, Inc., Cary, NC, USA) was used. Results are expressed as mean±s.e.m. unless otherwise stated.
Five matched pairs of CAFs and NBFs were established from fresh human breast cancer specimens. The patient demographics and pathology details of the breast cancer samples are shown in Table 1.
Patient demographics and pathology results of breast cancer samples used for culturing of CAFs and NBFs
|CAF/NBF cell line||Age||Side||Operation||Type of cancer||Grade||Size (mm)||Stage||ER||PR||HER2||HER2 mean copy number|
WLE, wide local excision; ANC, axillary node clearance; SLNB, sentinel lymph node biopsy; ER, oestrogen receptor; PR, progesterone receptor; HER2, human epidermal growth factor receptor type 2.
All the carcinomas were of pathologic stage p2A or p2B. Four of the carcinomas were of invasive ductal type, whereas one of the carcinomas was of invasive lobular type. All the blocks sampled as cancer contained invasive carcinoma cells accounting for from 50 to 75% of cells as well as non-adipose stromal cells accounting for between 10 and 15% of cells with few adipose cells (10% or less). None of the areas sampled as benign tissue contained invasive malignancy, although one case did contain lobular carcinoma in situ (5%). The epithelial cells in these areas accounted for between 35 and 45% of all cells, with the percentage of non-adipose stromal cells (35–45%) and adipose cells (10–25%) being significantly higher than seen in the areas sampled as cancer. The benign areas did include occasional foci of simple hyperplasia without atypia but were otherwise normal. In particular, there were no areas of atypical ductal hyperplasia or morphologically discernible stromal abnormalities.
CAFs and NBFs differ in CXCL12 secretion and α-SMA expression
For initial characterisation of CAF and NBF cultures, we measured secretion of the chemokine CXCL12, also known as stromal derived factor-1 (SDF1), and cellular expression of α-smooth muscle actin (α-SMA). These factors may be regarded as two of the hallmarks of CAFs (Mishra et al. 2008). As shown in Fig. 1A, CAFs secrete a mean of 30±14-fold (range: 5.3- to 82.5-fold) higher amounts of CXCL12, measured by ELISA, compared with corresponding NBFs (P=0.013). By qPCR (Fig. 1B), CAFs express on average 7.7±3.9-fold (range: 1.1- to 22.2-fold) more CXCL12 mRNA compared with matched NBFs (P<0.0001). While both CAFs and NBFs stain for α-SMA by immunocytochemistry, CAFs express 3.0±0.8 times more α-SMA than corresponding NBFs (P=0.003) on western blotting (Fig. 1C and D). We also found that the level of CXCL12 secretion (Fig. 2A) and α-SMA expression (Fig. 2B) in NBFs were similar to fibroblasts cultured from breast reduction specimens.
CAFs are more resistant to chemotherapy
To test the chemoresistance of the CAF cultures, the response of CAFs and NBFs to doxorubicin, a chemotherapeutic agent that is commonly used to treat breast cancer patients, was also assessed. Figure 3A shows morphological changes in one of the matched CAF/NBF cell line pairs in response to 0.5 and 1 μM doxorubicin after 72 h. In response to the higher doxorubicin dose, the NBFs showed a greater loss of cells compared with the CAFs, which appeared morphologically unchanged from control cultures. The response of these cells to doxorubicin was quantitated using a marker of the induction of DNA double-stranded breaks, the phosphorylation of histone H2AX (γH2AX; Fig. 3B and C) as well as Annexin V quantitation on flow cytometry (Fig. 3D). Averaged over five pairs of cultures, CAFs had 51±47% lower γH2AX in response to doxorubicin compared with NBFs (P=0.007; Fig. 3C). On Annexin V analysis by flow cytometry on two pairs of CAF/NBFs (CAF/NBF 6 and 7; Fig. 3D), there were 3.3 times more dead NBFs compared with CAFs after treatment with 1 μM doxorubicin (P=0.0019). We also noted that the expression of γH2AX after treatment with doxorubicin was higher in fibroblasts cultured from breast reduction specimens compared with CAFs but comparable to NBFs (Fig. 2C).
CAFs induce proliferation in MCF7 cells to a greater degree than NBFs
MCF7 cells cultured in medium conditioned by CAF cell lines, compared with corresponding NBF-conditioned media, resulted in 34±6% higher MCF7 cell counts (P<0.0001) after 3 days (Fig. 4).
CAFs induce EMT in MCF7 cells to a greater degree than NBFs
MCF7 cells exposed to culture medium conditioned by CAFs showed stronger staining for vimentin (Fig. 5A) and less for E-cadherin as demonstrated by immunofluorescence and confocal microscopy (Fig. 5B) compared with those exposed to culture medium conditioned by NBFs. Quantitation of staining of vimentin and E-cadherin on confocal microscopy confirmed visual results, showing that MCF7 cells cultured in CAF-conditioned media expressed higher levels of vimentin and lower levels of E-cadherin compared with MCF7 cells cultured in NBF-conditioned media (Fig. 5C and D).
There was also more vimentin detectable by western blotting in MCF7 cells cultured in CAF- compared to NBF-conditioned medium (P<0.0001; Fig. 5E). While the expression of E-cadherin in MCF7 cells exposed to CAF- and NBF-conditioned medium for 6 days was variable among the different pairs, E-cadherin detectable by western blotting in MCF7 cells cultured in CAF-conditioned medium was lower compared with cells cultured in NBF-conditioned medium for 10 days (Fig. 5F). We found that there was a consistent significant loss of E-cadherin as measured by western blotting after 10 days (P=0.0003) but not after 6 days (P=0.67, data not shown).
CAFs also induce EMT to a greater degree in MCF10A compared with NBFs
Exposure of CAF- and NBF- conditioned medium to phenotypically normal MCF10A mammary epithelial cells also resulted in an increase in vimentin mRNA measured by qPCR (P=0.028; Fig. 6A) and expression of vimentin as measured by western blotting after 6 days (P<0.007; Fig. 6B and C).
As EMT is known to facilitate cell invasiveness, we examined the effect of conditioned medium studies on MCF7 cells with two pairs of CAFs and corresponding NBFs on their ability to invade. MCF7 cells exposed to CAF-conditioned medium demonstrated greater invasiveness through a matrigel layer compared with cells exposed to NBF-conditioned medium with P value <0.0021 (Fig. 7A and B).
In this study, we established five pairs of matched CAFs and NBFs from fresh breast cancer tissue specimens. These two groups of fibroblasts differed in terms of CXCL12 secretion and α-SMA expression. While some studies have found that α-SMA by immunohistochemistry is only observed in CAFs, not NBFs (Giannoni et al. 2010, Huang et al. 2010), other studies, like ours, have found that α-SMA is expressed by both CAFs and NBFs (Lebret et al. 2007, Hawsawi et al. 2008, Holliday et al. 2009), but at higher levels in the CAFs by western blotting (Lebret et al. 2007).
We also found CAFs to be more resistant to doxorubicin compared with matched NBFs. Sonnenberg et al. (2008) assessed sensitivity of CAFs cultured from breast and lung cancers and found that they were less sensitive to paclitaxel and cisplatin compared with 22 cell lines from lung, breast and ovarian cancers as well as acute and chronic myeloid leukaemias. In this study, however, comparison was not made to normal fibroblasts. In response to doxorubicin, we found less DNA double-stranded breaks, as measured by γH2AX, and less apoptosis, as measured by Annexin V by flow cytometry in CAFs when compared with matched NBFs. The resistance of CAFs to chemotherapy may contribute to the resistance of some cancers to chemotherapy with resultant development of recurrence or metastasis after chemotherapy.
In addition, we have demonstrated that CAFs are able to induce EMT to a greater degree in MCF7 breast cancer cells and phenotypically normal MCF10A breast epithelial cells compared with matched NBFs, as demonstrated by increased expression of vimentin on western blotting. For measurement of E-cadherin by western blotting, we found that the expression was variable after 6 days of co-culture of fibroblasts with MCF7 cells, with no consistent changes observed. After 10 days of co-culture of fibroblasts with MCF7 cells, however, there was a significant decrease in expression of E-cadherin by western blotting in MCF7 cells co-cultured with CAFs compared with cells co-cultured with NBFs. Interestingly, Janda et al. found that with TGFβ treatment of RafER-EpH4 cells, E-cadherin protein levels were reduced 48 h after treatment while CDH1 mRNA levels were only reduced after 5 days of treatment. They concluded that loss of E-cadherin protein preceded transcriptional down-regulation of CDH1 mRNA by over 2–3 days (Janda et al. 2006). We did not find consistent mRNA changes of the CDH1 gene by qPCR even after 10 days of co-culture.
Lebret et al. (2007) studied the effect of co-culture of CAFs and NBFs from breast reduction specimens on PMC42-LA, a human breast cancer cell line with stem cell-like properties, which is able to exhibit either luminal epithelial or myoepithelial markers depending on the culture microenvironment. When PMC42-LA cells were co-cultured with CAFs and NBFs, the cells showed significantly increased expression of α-SMA and cytokeratin 14 (myoepithelial markers) compared with control cells. There was also an increase in expression of vimentin in cells co-cultured with CAFs or NBFs compared with control. There was, however, no difference in expression of α-SMA, cytokeratin 14 or vimentin in PMC42-LA cells co-cultured with CAFs or NBFs (Lebret et al. 2007). E-cadherin appeared to localise to cell–cell junctions in control PMC42-LA cells. When co-cultured with NBFs, E-cadherin was found in both junctional and cytoplasmic location while E-cadherin location in cells co-cultured with CAFs appeared to be more cytoplasmic. While we found less overall expression of E-cadherin in MCF7 cells co-cultured with CAFs compared with NBFs by immunofluorescence and confocal microscopy, we did not find a difference in intracellular localisation of E-cadherin in MCF7 cells co-cultured with CAFs or NBFs. Giannoni et al. studied the effects of conditioned medium of human prostate CAFs on PC3 cells, a cell line originating from bone metastasis of human prostate cancer cells. They found up-regulation of EMT markers Snail, Twist and vimentin and down-regulation of E-cadherin. There was also a change in morphology of the cells towards a spindle-shaped appearance, again consistent with EMT (Giannoni et al. 2010).
We found an increase in invasion of MCF7 cells grown in CAF-conditioned compared with NBF-conditioned medium. This is in keeping with a greater degree of EMT in MCF7 cells cultured in CAF- compared with NBF-conditioned medium.
Fibroblasts cultured from within breast cancer tissue are different to those cultured from macroscopically normal tissue located at least 2 cm away from the tumour margin. These CAFs have the ability to induce a greater degree of EMT in MCF7 breast cancer and MCF10A breast epithelial cell lines, indicating that CAFs are able to contribute to a more malignant breast cancer phenotype and their role in resistance to therapy should therefore be considered when treating breast cancer.
Declaration of interest
The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.
This work was funded by the University of Sydney Cancer Research Fund. P S H Soon was also funded by National Health and Medical Research Council (Health Professional Research Training Fellowship #511996) and New South Wales Cancer Institute (Clinical Research Fellowship #09/CRF/2-05).
Author contribution statement
P S H Soon conceived, designed and coordinated the study as well as drafted the manuscript. E Kim carried out cell culture, characterisation and EMT studies as well as data analysis. C K Pon carried out cell culture, EMT and flow cytometry studies as well as data analysis. A J Gill performed pathological review of breast tissue used in this study. K Moore and A J Spillane collected tissue and revised the manuscript critically. D E Benn participated in the design of the study, provided intellectual input and critically revised the manuscript. R C Baxter conceived and designed the study as well as provided intellectual input and critically revised the manuscript. All authors read and approved the final manuscript.
The authors would like to acknowledge Lily Huschtscha for providing the protocol for primary culture of fibroblasts as well as fibroblasts from breast reduction specimens and Mark Sywak for collecting some of the tissue samples.
Edge SB Byrd DR & Compton CC 2009 AJCC Cancer Staging Manual. New York NY USA: Springer
GiannoniEBianchiniFMasieriLSerniSTorreECaloriniLChiarugiP2010Reciprocal activation of prostate cancer cells and cancer-associated fibroblasts stimulates epithelial–mesenchymal transition and cancer stemness. Cancer Research706945–6956. (doi:10.1158/0008-5472.CAN-10-0785).
Lakhani SR Ellis IO Schnitt SJ Tan PH & van de Vijver MJ 2012 In WHO Classification of Tumours of the Breast. Lyon France: IARC
WaldOIzharUAmirGKirshbergSShlomaiZZamirGPeledAShapiraOM2011Interaction between neoplastic cells and cancer-associated fibroblasts through the CXCL12/CXCR4 axis: role in non-small cell lung cancer tumor proliferation. Journal of Thoracic and Cardiovascular Surgery1411503–1512. (doi:10.1016/j.jtcvs.2010.11.056).