Background
Many invasive breast cancers (IBC) develop through a pre-invasive stage known as ductal carcinoma in situ (DCIS) where the proliferative neoplastic cells are retained within the breast duct surrounded by an intact myoepithelial cell (MEC) layer lying in contact with basement membrane [
1]. DCIS is a non-obligatory precursor of IBC; only around half of untreated cases progress to IBC, when the tumour cells breach the MEC-basement membrane barrier [
2,
3]. However, the molecular mechanisms underlying the transition of DCIS to IBC are poorly understood, and identifying those cases that will progress and those that will not remains a major challenge [
4,
5].
Molecular profiling studies demonstrate that DCIS tumour cells and their invasive counterparts share a similar signature and DCIS is as genetically advanced as established invasive disease [
6,
7]. Therefore, in order to understand the changes that trigger invasion, attention has focused on the role of the microenvironment [
8‐
11]. MECs are a unique component of the microenvironment in the breast. In normal breast ducts, MECs adhere tightly to the basement membrane using prominent hemidesmosome (HD) formation [
12]. They show a high level of expression of anti-tumourigenic factors, such as proteinase inhibitors and anti-angiogenic mediators, and in vitro studies have demonstrated broad tumour-suppressor activities of normal primary MEC and MEC lines [
13‐
15]. In DCIS, the MECs become altered, showing changes in gene expression, epigenetics and phenotype [
13,
14,
16,
17], though the functional significance of these changes has not been established [
9,
18‐
20].
A previous study has indicated that in normal ducts MECs appear to be the main source of matrix metalloproteinase-8 (MMP-8), which is a major collagenase that cleaves Collagen type- I (Col -I) [
21]. Expression of MMP-8 appears to be lost in DCIS-associated MECs [
22]. MMPs are a large family of endopeptidases, which have the ability to remodel extracellular matrix (ECM) and are upregulated in many cancers, such that the MMP family is conventionally regarded as consisting of key enzymes that contribute to the process of cancer cell invasion and metastasis [
3,
23‐
25]. However, the failure of broad range MMP inhibitors in anticancer trials suggests an incomplete understanding of the complex function of this family; specifically when and how they act in cancer [
3,
23‐
25]. Moreover; regulatory interactions (compensation or inhibition) between MMP family members construct a complex web through which the expression and activity of MMPs are constantly controlled [
25,
26]. One particular example is MMP-9, which is implicated in the acquisition of DCIS-associated phenotype in MECs [
27] at the same stage when MECs lose MMP-8 expression. Since it was first reported that MMP-8
-/- mice exhibit increased incidence of skin tumours when challenged with chemical carcinogenesis, MMP-8 has been demonstrated to exert a clear tumour-suppressor function [
22,
26,
28,
29]. There is a growing body of evidence accumulating that MMP-8 has an anticancer role in breast cancer, malignant melanoma and tongue squamous cell carcinoma, [
22,
26,
28,
30‐
32]. Interestingly, identification of non-structural substrates of MMP-8 [
28,
33‐
37], suggests that the biological function of MMP-8 is much more complicated than just Col-I degradation.
The biological significance of loss of MEC-derived MMP-8 on MEC phenotype and MEC-breast cancer cell crosstalk remains elusive, especially whether MMP-8 may contribute to the tumour-suppressor function of MECs. In this study we employed 2D and 3D in vitro models to recapitulate the DCIS tumour microenvironment in order to investigate how MMP-8 is involved in MEC-breast cancer cell communication, and whether loss of MMP-8 contributes to loss of tumour-suppressor activity and promotes progression to invasive disease. Improving our understanding of the factors that contribute to transition of DCIS to invasion will aid in the future development of predictive signatures to help stratify patient management more appropriately, and avoid the ‘overtreatment’ that has caused such controversy in the breast screening programme.
Methods
Immunohistochemistry
Breast tissue samples were obtained from surgical specimens from patients undergoing breast surgery at Barts Health NHS Trust London. The study was performed following patient consent and approval from the local research ethics committee (reference: 05/Q0403/199 and 09/H075/39). Seven were normal breast cases from reduction mammoplasty, nine were cases of DCIS (high, intermediate and low grade) and nine were cases of DCIS with concomitant invasion.
Sections were dewaxed in xylene and antigen retrieved in citrate buffer pH6; followed by incubation with MMP-8 rabbit polyclonal antibody (Atlas, Cambridge, MA, USA, HPA02122, 1:750). Sections subsequently were incubated with goat anti-rabbit biotinylated F(ab’)2 for 30 min, developed using ABC reagent and superDAB (Dako, Glostrup, Denmark) then counterstained with hematoxylin [
7].
Cell lines and cell culture
All breast cancer cell lines were obtained from American Type Culture Collection (ATCC) and verified with STR profiling (LGC Standards, Teddington, UK, tracking number 710081047). MCF-7 and MDA-MB-231 cells were cultured in 10% fetal bovine serum (FBS, PAA Laboratories, Pasching, Austria, A15-14) containing (complete) DMEM (PAA Laboratories, E15-843). SUM159 cells were grown in complete Ham’s F12 (PAA Laboratories, E15-817) with hydrocortisone (Sigma-Aldrich, St. Louis, MO, USA, HO 888, 1 μg/ml) and insulin (Sigma-Aldrich, I9278, 1 μg/ml).
Generation of myoepithelial cell lines
The h-TERT/SV40 LgT immortalised MEC line was a gift from M. O’Hare and P. Jat, Institute of Neurology, UCL, London. αvβ6 integrin over-expressing and control (normal) MEC line was generated as previously described [
27], and termed β6-1089 and N-1089 respectively. These cells were cultured in the presence of hydrocortisone (1 μg/ml), epidermal growth factor (EGF, Sigma-Aldrich, 9644, 10 ng/ml), insulin (1 μg/ml) and puromycin (1 μg/ml, Sigma-Aldrich) in Ham’s F12.
Isolation of primary myoepithelial and luminal epithelial populations from normal breast and DCIS tissue
Primary cell isolation protocol was adapted from Gomm et al. [
38]. In brief, reduction mammoplasty tissue or DCIS tissue was digested overnight at 37 °C with 1 mg/ml collagenase (Sigma-Aldrich, C2674) and hyaluronidase (Sigma-Aldrich, H3506) on a roller-mixer. Following washing and a sedimentation step the organoids were further digested with trypsin/EDTA (PAA Laboratories, L11-004) and DNase (Roche Diagnostics, Basel, Switzerland, 10104159001) then filtered through a 40-μm filter (Becton Dickinson, Franklin Lakes, NJ, USA). Cells were incubated in a 1:1 ratio sequentially with magnetic beads (Dynal Biotech., Milwaukee, WI, USA) coated with common acute lymphoblastic leukemia antigen (CALLA) antibody (AbD Serotech, Oxford, UK, MCA1556) to isolate normal MECs or anti-β4 integrin antibody (EMD Millipore, Billerica, MA, USA, mAB 1964) for DCIS myoepithelial cells. This was followed by epithelial cell adhesion molecule (EpCAM) antibody-coated epithelial-enriched magnetic beads (Invitrogen, Carlsbad, CA, USA, 161.02) to isolate luminal epithelial cells (LECs). Primary fibroblasts were grown from the media produced by the sedimentation step. The purity of these populations has been verified by staining for CK8, CK18, EMA for the luminal epithelial cells and CK14, SMA, CD10, vimentin, p63 for the myoepithelial cells, as described in previous publications [
27,
38].
MMP-8 over-expression
The wild-type MMP-8 (WT) and inactive mutant MMP-8 (EA) coding inserts cloned into pcDNA4 (EcoRI and XhoI restriction sites, Invitrogen, K103002) have been described previously [
39]. Empty pcDNA4 vector was used as a control.
β6-1089 cells were plated at a density 7 × 104 cells per well onto 6-well plates 24 hours prior to transfection. β6-1089 cells were transfected with 1 μg DNA per well with Genejuice reagent (Novagen, Merck KGaA, Darmstadt, Germany, 70967) according to the manufacturer’s instructions for 5 hours at 37 °C. All functional experiments were performed 24 hours after transfection. For generation of conditioned media (CM), complete media was replaced with serum-free media (SFM) and harvested after 24 hours, followed by centrifugation at 1200 rpm for 3 minutes to remove cell debris and storage at -80 °C.
MMP-8 knock-down
N-1089 cells were seeded onto six-well plates at a density 7 × 10
4 cells per well and incubated for 24 hours at 37 °C, after which media was replenished and N-1089 cells were transfected with 5nM pooled MMP-8 siRNA (Thermo Fisher Scientific, Waltham, MA, USA, MU-005969-00-0020) or Luciferase GL3 Duplex control siRNA (siLUC, Thermo Fisher Scientific, D-001400-01-20) [
40] using interferin transfection reagent (Polyplus tranfection, Peqlab Biotechnologie GmbH, Erlangen, Germany, 409-01) according to the manufacturer’s instructions. All functional assays were carried out 96 hours after transfection. To generate CM, cells were serum starved for 24 hours and media was harvested as described above.
Reverse transcription and q-PCR
RNA was extracted with RNeasy kit (Qiagen, Hilden, Germany, 74104) and reverse transcribed with Superscript II (Invitrogen, 11064-014). One hundred nanograms of cDNA were used per reaction.
All primers made up at 100 mM stock concentration and used at 0.4 mM final concentration. Reactions were performed using Immomix Red Master Mix (Bioline, London, UK, Bio 25-022) in a total reaction volume of 25 μl. For nested PCR, 2 μl PCR product was used as a template for second-round PCR. The products were separated on a 1% agarose (Invitrogen, 16500-500) gel for 45 min at 100v and visualised under UV light (AutoChemi System, UVP, Cambridge, UK).
Q-PCR was carried out using SYBR Green (Applied Biosystems, Foster City, CA, USA, 4367659) chemistry. One hundred nanograms of cDNA input was used per reaction. Primers were used at 0.3 mM final concentration in 10 μl total PCR volume for one well of a 96-well plate. Reactions were carried out on a Step One Plus instrument (Applied Biosystems).
Primer sequences - MMP-8 forward: 5′-GCCGAAGAAACATGGACCAAC-3′, MMP-8 nested forward: 5′- ACTCCTCTGACCCTGGTGCC-3′, MMP-8 reverse: 5′- TGAGGATGCCTTCTCCAGAAG-3′, αvβ6 forward: 5′- GAAGGAATGATCACGTACAAG-3′, αvβ6 reverse: 5′- AGCAGGGAGTCTTCACAGGT-3′, 18 s forward: 5′-CACGGGAAACCTCACCCGGC-3′, 18 s reverse: 5′- AGCAGGGAGTCTTCACAGGT-3′.
Western blotting
Cells were lysed with radioimmunoprecipitation (RIPA) buffer. CM from N-1089 and β6-1089 cells was concentrated 20× using centrifugal units (EMD Millipore, 4FC 800324) with 3 K molecular weight cutoff (MWCO) at 4000 g for 45 minutes at 4 °C. Samples were boiled in reducing (β-mercaptoethanol containing) laemmli buffer for 5 minutes at 95 °C, then separated with SDS (National Diagnostics, Atlanta, GA, USA, EC874) acrylamide (National Diagnostics, EC890) gel for 90 minutes at 125 V at room temperature. Recombinant human MMP-8 western blot standard (R&D Systems, Minneapolis, MN, USA, WBC017) was loaded as positive control. The gel was transferred to a nitrocellulose membrane (Hybon C extra, RPN203E) for 90 minutes at 30 V at room temperature and Western blotting was performed as previously described [
27].
Primary antibodies used included mouse anti-human MMP-8 antibody (R&D Systems, mAb 908), SMAD2 (Cell Signaling Technology, Danvers, MA, USA, 3122), pSMAD2 (Cell Signaling Technology, 3101). Protein loading was confirmed with HSC70 (Santa Cruz Biotechnology, Dallas, TX, USA, sc-7298) or actin (Santa Cruz Biotechnology, sc-1615) as loading controls. For MMP-8 over-expressed samples, the membrane was probed with anti-V5 antibody (Invitrogen, R960-25). Depending on the primary antibody species, horseradish peroxidase (HRP)-conjugated mouse secondary (Dako, P0447), rabbit secondary (Dako, P0448) or goat secondary (Dako; P0160) antibody was used. The membrane was developed with ECL (GE Healthcare, Chicago, IL, USA, RPN 2106) and densitometric quantification analysis was performed with Image J software (National Institutes of Health, Bethesda, MD, USA).
Invasion assays
Boyden chamber transwells (Corning, Corning, NY, USA, 3422, pore size: 8 μm) coated with 70 μl diluted Matrigel (1 volume Matrigel: 2 volumes ice-cold serum-free media) were used to measure in vitro invasive capacity, as previously described [
27]. 2 × 10
4 modified β6-1089 or N-1089 cells were plated in serum-free media in the lower chamber. 3 × 10
4 MCF-7, MDA-MB-231or SUM159 breast cancer cells were seeded onto the Matrigel-coated insert. Invasion assays were carried out at 37 °C over 24 hours, or 48 hours for MCF-7 cells. Invaded cells were harvested with 10× trypsin/EDTA (PAA Laboratories, L11-003) from the underside of the transwell and counted with a CASY counter (Schärfe System, Reutlingen, Germany).
Viability assay
To analyse MEC viability, 3 × 104 β6-1089 or N-1089 cells were seeded onto a 24-well plate and incubated for 24 hours at 37 °C after which the media was replenished and MTS reagent (Promega, Madison, WI, USA, cell titer 96 Aqueous, G5421) was added (5:1 media:MTS v/v ratio) then incubated for 1 hour at 37 °C. The MTS containing mixed media was then collected and placed on a 96-well plate and read at 495 nm (Tecan, Männedorf, Switzerland, infiniteF50).
To analyse breast cancer cell viability 3 × 104 MCF-7, MDA-MB-231 or SUM159 cells were incubated with CM collected from modified β6-1089 or N-1089 cells for 24 hours. Viability was quantified using MTS reagent as described above.
Adhesion assays
Non-tissue culture-treated 96-well plates were coated with 100 μl of the following matrices in triplicate at given concentrations: Fibronectin at 1 μg/ml, rat tail Collagen-I (BD Biosciences, Franklin Lakes, NJ, USA, 354236) at 0.5 μg/ml, Matrigel at 0.5 μg/ml, Tenascin-C (EMD Millipore, CC 065) at 3 μg/ml, Laminin-I at 10 μg/ml, Collagen-IV at 10 μg/ml and latency-associated peptide (LAP) (Sigma-Aldrich, L3408) at 0.5 μg/ml. Three wells of each plate were coated with 0.1% bovine serum albumin (BSA, PAA Laboratories, K45-001) in phosphate-buffered saline (PBS) (w/v) as a control. Coated plates were incubated for 1 hour at 37 °C after which wells were washed twice with 100 μl PBS per well.
Modified β6-1089 or N-1089 cells were serum starved for 24 hours prior to the experiment. 3 × 103 cells were seeded in 100 μl of serum-free Ham’s F12 per well and allowed to adhere for 1 hour at 37 °C. Then 0.25 μl of Calcein AM (Invitrogen, C1430) cell tracker was added and incubated for 15 minutes at 37 °C after which the plate was washed twice with 100 μl PBS per well, and 100 μl of SFM was added. The plate was read at 490/520 nm on a fluorescent reader (BMG Labtech, Aylesbury, UK, FLUOstar Optima). The adhesion was calculated as the fluorescence value of adherent cells and normalised to that of control cells.
Migration assays
The underside of Boyden chamber transwells were coated with 100 μl of ECM or 0.1% BSA in PBS as a control, and incubated for 1 hour at room temperature. ECM components and concentrations were as for the adhesion assays. Following incubation the underside of the transwell was washed with 100 μl PBS and placed into 500 μl of SFM containing well of a 24-well plate.
3 × 104 modified β6-1089 or N-1089 cells were added onto the inner chamber in 200 μl serum-free Ham’s F12 and incubated for 8 hours at 37 °C. After that the media in the inner and outer chambers were replaced with 200 μl and 500 μl 10× trypsin EDTA respectively and incubated for 1 hour at 37 °C. The cells were then trypsinised and the migrated cells as well as the number of cells remaining in the inner chamber were counted using a CASY counter. The percentage of migrated cells was calculated using the counts of the upper chamber versus total cell number.
Organotypic culture
Organotypic gels were constructed using rat tail Collagen-I and Matrigel as previously described [
27]. Briefly; for 10 gels, 4.9 ml rat tail Collagen-I was mixed with 2.1 ml Matrigel (7:3 v/v ratio), 1 ml 10× DMEM (Sigma-Aldrich, D2429) and 1 ml FBS. The pH of the solution was neutralised by adding 1 M NaOH (Sigma-Aldrich, S8045) drop-wise until the mixture turned an orange-pink colour. 5 × 10
6 primary normal breast fibroblasts were suspended in 1 ml complete DMEM and added to the neutralised gel mixture. One millilitre of gel mixture was added to 1 well of a 24-well plate and set for 1 hour at 37 °C. Following this, complete DMEM was added drop-wise on top of the gels and equilibrated overnight at 37 °C. The media was aspirated from the top of the gels, 2.5 × 10
5 modified β6-1089 cells suspended in 500 μl complete Ham’s F12 per gel were added on top of the gels and incubated for 4 hours at 37 °C. Then 2.5 × 10
5 MDA-MB-231 or SUM159 cells in 500 μl complete DMEM per gel were added and incubated overnight at 37 °C.
Collagen-I-coated nylon membranes (100-μm pore size) were put on top of steel grids (coated side facing up) in a well of a 6-well plate and the organotypic gels placed on top. The well was filled with complete DMEM until the liquid reached the grid-membrane interface level; media was replenished every 2 days. The gels were harvested after 10 days, fixed in 10% neutral-buffered formalin (Cell Path, Newtown, UK, BAF-0010-037) overnight and then transferred to 70% ethanol for 24 hours. Gels were mounted in paraffin and sectioned.
Invadopodia assay (in situ zymography)
To analyse MEC-derived gelatinase activity an in situ zymography assay was carried out. To prepare the gelatin solution, 178.12 mg NaCl (Thermo Fisher Scientific, BB358-1), 94.57 mg NaBH4 (Sigma-Aldrich, 21,346-2) and 100 mg porcine skin type A gelatin (Sigma-Aldrich, G2500) were dissolved in 50 ml PBS for 1 hour at 37 °C (pH 9.3), following which 1.8 mg rhodamine was added to the solution and mixed for 2 hours to fluorescently label the gelatin. The gelatin solution was dialysed against PBS for 48 hours using Slide-A-Lyzer Dialysis Cassettes with 10 K MWCO (Pierce, Rockford, IL, USA, 66830), the PBS being replenished every 24 hours. After dialysis, the gelatin was collected and centrifuged at 1200 rpm for 10 minutes to remove insoluble particles, and then 1 g sucrose (Thermo Fisher Scientific, 8060153) was added and dissolved. The rhodamine-conjugated gelatin was spun at 1200prm for 10 minutes and then heated to 37 °C. Forty microlitres of solutions (drops) were aliquoted onto a 10-cm tissue culture plate. Thirteen-millimetre coverslips were placed on each drop and incubated for 20 minutes. All incubation steps were performed in the dark. In another dish, 1% gluteraldehyde solution in PBS (v/v) was aliquoted as 40 μl drops. After 20 minutes gelatin-covered coverslips were placed on the gluteraldehyde solution to create a dual layer, and incubated for 15 minutes. The coverslips were placed in a 24-well plate (dual-layered side facing up) and washed three times with 500 μl PBS, after which 500 μl 5 μg/ml NaBH4 was added and incubated for 20 minutes then washed as described. For sterilisation, coverslips were incubated with 500 μl 70% ethanol for 5 minutes, washed with PBS, then equilibrated and free aldehydes quenched using 500 μl complete Ham’s F12 for 20 minutes at 37 °C. Finally, 4 × 104 modified β6-1089 or N-1089 cells were added onto the coverslips and incubated for 24 hours.
After incubation, the cells were washed three times with 500 μl PBS and fixed in 4% formaldehyde (PFA, Sigma-Aldrich, P6148) for 15 minutes at room temperature then washed again as described. Cells were blocked in 0.1% BSA (v/v) and 0.1% NaN3 (Thermo Fisher Scientific, S227I) in DMEM (w/v) for 15 minutes in the dark at 4 °C, then stained with FITC-labelled phalloidin (Invitrogen, A22283) for 20 minutes in the dark at room temperature. The coverslips were mounted in ProLong Gold Antifade aqueous mounting reagent (Invitrogen, P36931) with DAPI.
Immunoprecipitation of conditioned media
Two hundred microlitres of 20× concentrated CM samples was cleared with 20 μl Protein A sepharose beads (GE Healthcare, CL-4B) for 1 hour at 4 °C with rotation. The samples were spun at 6000 rpm for 2 minutes at 4 °C and supernatant were collected. Pre-cleared CM were divided into 100 μl aliquots and incubated with 5 mg MMP-8 antibody (R&D Systems, mAb 908) or mouse IgG (Sigma-Aldrich, I5381) overnight at 4 °C with rotation.
Antibody treated CM was added to fresh 20 μl beads and incubated for 4 hours at 4 ° C with rotation. Samples were centrifuged as described. Supernatant were collected and kept as unbound fraction. The remaining pellet was washed with serum-free Ham’s F12 as described. In order to remove the precipitates from the beads, 40 μl reducing Laemmli buffer was added to 20 μl cleared bead pellet and boiled at 100 °C for 10 minutes then spun once again. Supernatants were collected and kept as bound fraction. Laemmli buffer was added to unbound fraction and boiled at 100 °C for 5 minutes. Twenty microlitres of prepared sample of bound or unbound fractions were loaded and separated on an SDS acrylamide gel and transferred to nitrocellulose as previously described, and probed with anti-MMP-8 or anti-V5 antibody.
Immunofluorescence staining
Cells on coverslips were fixed with 4% PFA in PBS for 10 minutes at room temperature (RT). Cells were washed three times with 500 μl PBS then blocked and permeabilized with 2% BSA and 0.1% Triton X-100 in PBS for 15 minutes. Finally, cells were incubated with primary antibody prepared in blocking solution at 1/100 dilution for 1 hour at room temperature. Primary antibodies used were α6β4 integrin (Merck Millipore, Billerica, MA, USA, MAB1964) and plectin (Epitomics, Burlingame, CA, USA, 1399). Cells were incubated with secondary antibody conjugated with FITC or Cy3 diluted in PBS. (Invitrogen, anti-mouse 546, A11030, anti-rabbit 546, A11035, anti-mouse 488, A11029, anti-rabbit 488, A11008, phalloidin 546, A22283). The cells were mounted in ProLong Gold Antifade aqueous mounting reagent with DAPI. For organotypic cultures, Neso antibody was a kind gift from Prof M. Djamgoz, Imperial College London.
For paraffin-embedded organotypic sections: sections were dewaxed and rehydrated, then boiled in 10 mM Tri-sodium citrate (pH 6) for 20 minutes. Sections were then incubated in 0.5% Triton X-100 for 5 minutes at RT. After that the sections were washed with PBS for 1 minute (three times) and blocked with 5% BSA/PBS for 1 hour at room temperature. Primary antibodies (p63; Dako, M7247, 1/50, Neso; 1/300, Ki67; Novocastra, Newcastle upon Tyne, UK, NCL-L-Ki67-MM1, 1/100) were prepared in blocking solution and incubated for 1 hour at room temperature. After that sections were washed with PBS for 5 minutes (three times). Fluorescently tagged secondary antibodies were prepared in PBS and incubated for 45 minutes at RT. Finally, the sections were washed as described followed by an extra washing step with distilled water and mounted in ProLong Gold Antifade reagent with DAPI. The Ki67 index is calculated as the percentage of Ki67-positive invading breast cancer cells out of the total cell number.
Luciferase reporter assay
MDA-MB-231 cells transfected with firefly and Renilla luciferase reporter gene fused with PAI-promoter (MDA-MB-231-Luc) were a gift from Dr Caroline Hill (London Research Institute) and cultured in DMEM in the presence of 50 μg/ml blasticidine (Sigma-Aldrich, 15205). MDA-MB-231- Luc cells were seeded onto 96-well plates at a density of 4 × 104 cells per well and incubated overnight at 37 °C, followed by serum starvation for 4 hours. 5 × 104 modified β6-1089 or N-1089 cells were seeded on top of MDA-MB-231-Luc cells in SFM and co-cultured overnight. The media was removed and cells were washed with 100 μl PBS. Cell lysis and luciferase activity quantification was performed using the Dual-Luciferase Reporter Assay System (Promega, E-1910) according to the manufacturer’s instructions.
Zymography
CM collected from modified β6-1089 or N-1089 cells was concentrated as previously described and separated by SDS-PAGE gel containing 3 mg Collagen-I for 50 minutes at 200 V. The gel was developed with developing buffer [10 ml 1 M Tris (pH 7.5), 8 ml 5 M NaCl (Thermo Fisher Scientific, BP358), 1 ml 1 M CaCl2 (Sigma-Aldrich, C7902), 1.6 ml 2.5%Triton X-100 and 179.4 ml sterile distilled water] overnight at 37 °C and stained with Coomassie Blue R-250 solution (Thermo Fisher Scientific, 20278).
Statistical analysis
Statistical significance was determined by Student’s t test or ANOVA with Bonferroni post-test where appropriate, using Prism (Graphpad Software, San Diego, CA, USA). For immunohistochemical scoring of MMP8 on a duct-by-duct basis Fisher’s exact test was used on a 2 × 3 table. Results were considered as significant with P value less than 0.05.
Discussion
Transition of DCIS to invasive cancer is a critical step in the natural history of breast cancer, converting a potentially curable lesion to a life-threatening disease. Only around half of DCIS lesions will progress to invasive cancer during a woman’s lifetime, but there is a lack of robust predictive markers to stratify patient management [
1,
48,
49]. Molecular studies have indicated a consistent similarity between DCIS tumour cells and their invasive counterparts, showing that DCIS cells are as genetically advanced as invasive cancer. This has focused attention on the microenvironment, which shows evidence of being significantly altered in DCIS compared to normal tissue [
6,
7,
50,
51]. A unique component of the breast microenvironment is the MEC population. In the normal breast, MECs form a stable interface with the basement membrane, regulate epithelial cell polarity and exert a multifaceted tumour-suppressor role [
9,
52‐
55]. We, and others, have shown that MECs are altered in DCIS and we recently demonstrated that upregulation of the integrin αvβ6 by DCIS-associated MEC confers tumour-promoter activity on MECs through TGF-β-mediated upregulation of MMP-9 [
27].
Whilst many MMPs exert tumour-promoter effects, a greater understanding of metalloproteinase biology has indicated that some family members have tumour-suppressor effects. MMP-8 was identified as a tumour suppressor nearly a decade ago when the MMP-8 knock-out mouse was found to be more susceptible to skin tumuorigenesis as compared to a WT mouse [
28]. There is also strong evidence for a role in tumour prevention in breast cancer, melanoma and squamous cell carcinoma [
22,
26,
28,
29,
31,
32,
56‐
59]. Particularly in melanoma; MMP-8 is frequently mutated by loss of heterozygosity (LOH), which is a hallmark for tumour-suppressor proteins [
22]. In addition this mutation results in a significant decrease in enzymatic activity and a concomitant loss of its cancer-protective role [
26]. However understanding the mechanism by which MMP-8 mediates this anti-tumour role is incomplete. In the breast, MMP-8 is expressed by the MEC population, and we identified loss of MMP-8 in DCIS-associated MECs [
22]. This study aimed to address the role of MMP-8 in MECs and the impact of its loss on MEC phenotype and tumour-suppressor function.
We used purified cell populations from normal and diseased breast to confirm MECs as the primary source of MMP-8 in the normal breast, and its loss in DCIS-associated MECs. We demonstrated similar loss of expression in a cell line model of DCIS-associated MEC generated by over-expression of β6 integrin (β6-1089 cells) compared to its normal MEC counterpart (N-1089). These models were used to investigate gain-of-function effects, through over-expression of MMP-8 or an inactive mutant (EA, which carries a point mutation in catalytic site) in β6-1089, and loss-of-function effects, through MMP-8 knock-down in N-1089 cells.
In normal breast, MECs have a central role in preserving normal tissue architecture, [
60], and correct MEC-matrix interactions are crucial for MECs to maintain cell polarity in the context of tumour suppression [
52‐
54,
61]. We therefore aimed to investigate if MMP-8 is involved in ECM sensing and response.
When modified β6-1089 cells were seeded on different ECM components, WT MMP-8 but not the inactive MMP-8 EA resulted in a significant increase in MEC adhesion to ECM proteins including Fibronectin, Collagen-I, Laminin-I and Tenascin-C. Furthermore, transwell migration assays indicated that MEC migration towards Fibronectin, Collagen-I, Laminin-I and Tenascin-C was significantly downregulated only in MMP-8 WT cells, suggesting that MMP-8 contributes to stable anchorage of MECs to ECM, reflecting their role in normal breast. Both adhesion to and migration towards LAP, an established ligand of αvβ6 integrin [
42], was consistently and significantly reduced, possibly reflecting a reduction in αvβ6 integrin activity. Similarly, knock-down of MMP-8 in N-1089 cells resulted in decreased adhesion and enhanced migration to ECM proteins, confirming a role for MMP-8 in matrix adhesion.
MEC attachment to basement membrane is achieved through HDs, which are stable adhesions critical for the integrity of epithelial cell monolayers. The integrin α6β4 nucleates HD formation through linkage to plectin and the intermediate filament cytoskeleton [
43,
62]. Disassociation of HDs and localisation of α6β4 to actin-rich protrusions are characteristics of the acquisition of a migratory phenotype in basal cells, and has been well studied in migrating keratinocytes during wound healing [
63‐
65]. Since loss of HD formation is recognised in DCIS-associated MEC [
43], we sought to investigate the effect of MMP-8 on the subcellular localisation of α6β4. This revealed that, in MMP-8 WT over-expressing cells, there was significantly greater co-localisation of α6β4 to HD-associated plectin, with shorter and reduced number of retraction fibres, corresponding to a reduced migratory phenotype. Moreover, phalloidin staining of modified β6-1089 cells showed that MMP-8 WT cells significantly spread more on Fibronectin, in keeping with a more adhesive phenotype. In keeping with this, knock-down of MMP-8 in N-1089 cells resulted in significantly longer and increased number of retraction fibres compared to control cells, which we speculate indicates the acquisition of a more migratory phenotype.
Compromise of the MEC-basement membrane barrier is a key event in progression of DCIS to invasive cancer. Since MMP-8 impacts on MEC adhesion and HD formation, we sought to analyse whether altered MMP-8 expression by MEC influences their tumour-suppressor activity. Transwell invasion assays were used, in which MCF-7, MDA-MB-231 or SUM159 breast cancer cells were placed onto a Matrigel-coated porous membrane and allowed to invade towards MEC populations. A significant decrease in tumour invasion was observed for MDA MD 231 and SUM159 cells co-cultured with MMP-8 WT over-expressing β6-1089 compared to empty vector and MMP-8 EA, suggesting that MMP-8 contributes to MEC invasion-suppressor effect. This was further supported by enhanced transwell invasion for all tumour cells in the presence of N-1089 MMP-8 knock-down cells. A similar effect was detected in a more physiologically relevant 3D organotypic model reflecting the microenvironmental conditions of DCIS, incorporating a fibroblast populations as well as the MEC-tumour cell bilayer [
41,
66].
Cooperation of tumour cells with the microenvironment is required to breach basement membrane [
19,
62,
67]. Thus altered proteolytic activity of MECs could contribute to the invasive phenotype of breast cancer cells. We previously have shown that β6-1089 cells promote tumour invasion through TGF-β-mediated upregulation of MMP-9 [
41]. A number of studies have indicated that MMP-8 can modulate TGF-β signalling [
68,
69], and since we show that over-expression of MMP-8 WT in β6-1089 cells modifies their invasion-promoter effect, we investigated whether MMP-8 could influence TGF-β signalling and MMP-9 expression or activity in this cell population. This showed that phosphorylation of SMAD-2 was downregulated in β6-1089 cells expressing MMP-8 WT and enhanced in N-1089 cells following MMP-8 knock-down, supporting a role for MMP-8 in dampening TGF-β signalling. However, MMP-8 WT or EA did not have a significant effect on MMP-9 expression, though when modified β6-1089 cells were incubated on fluorescently labelled gelatin, the substrate of MMP-9, the area of degradation was significantly abrogated by MMP-8 WT cells. In contrast, gelatin degradation was significantly enhanced by knock-down in N-1089 cells, and importantly this effect could be rescued by MMP-9 inhibition or by MMP-9 knock-down. This suggests some form of inhibitory interaction between MMP-8 and MMP-9: it previously has been shown that MMP-8 can establish a complex with MMP-9 but the function of this complex remains elusive [
30].
In tissues MECs of normal and benign ducts are consistently positive for MMP-8, whilst there is a significant progressive loss of this metalloproteinase through pure DCIS to DCIS with co-existing invasion (
p = 0.001, Additional file
2: Table S2). This finding should be further validated on a larger cohort ideally with long-term follow-up.