Background
During metastatic progression, cancer cells detach from the primary tumour into the surrounding extracellular matrix (ECM). A subpopulation of these cells will intravasate into the blood stream or lymphatic vessels, entering the circulation [
1]. Due to their size, these disseminated cancer cells will eventually extravasate via transendothelial migration, and a small proportion will go onto proliferate to form secondary colonies [
2‐
4]. Matrix Metalloproteinases (MMPs) have been implicated as the central mediators of ECM remodeling throughout the metastatic cascade, particularly Membrane Type-1 MMP (MT1-MMP), MMP-2, and −9 [
5‐
7]. Collectively, these MMPs play a key role in the degradation of the major structural components of the basement membrane (BM), which separates the primary tumour from the endothelium and is the first main structural barrier to metastasis [
8,
9]. It was long thought that the role of MMPs was to mediate movement of cancer cells through the BM via their ability to degrade ECM components; however, MMPs are now known to be more pleiotropic than previously thought [
10], participating in a multitude of cell processes during cancer progression [
11,
12], and perhaps also acting as transcription factors [
13‐
15],
MT1-MMP (MMP-14) is multifunctional MMP that affects cell behavior via proteolytic and non-proteolytic mechanisms [
16,
17]. As a result of its proteolytic activity, MT1-MMP can cleave ECM and non-ECM substrates [
5,
18], and participates in activation of the pro-forms of MMP-2 and MMP-9 [
19,
20]. MT1-MMP also signals through the ERK and AKT pathways [
21,
22] and mediates HIF-1α stabilization [
23] via non-proteolytic mechanisms. Moreover, it has been shown that MT1-MMP overexpression also increases the migratory ability of breast cancer cells independent of its proteolytic activity [
24,
25]. The mechanism underlying this increase in migratory potential remains poorly understood, with evidence suggesting that it involves ERK activation and cleavage of the cell adhesion molecule CD44 [
26,
27]. However, reported data reveal contradictions regarding the requirement for TIMP-2 in increasing migratory potential, as some report the necessity for TIMP-2 to increase migration of cancer cells overexpressing MT1-MMP [
24,
25], whereas others did not using similar cell models [
26‐
28]. Other experimental inconsistencies regarding the source of MMPs, whether they are predominately stromal or cancer cell derived [
29], and their role in mediating cancer cell growth and invasion in vivo are also under intense debate [
7,
9,
30,
31].
To address this, we overexpressed MT1-MMP in MCF-7 breast cancer cells, which represent early stage breast cancer [
32] and are naturally MT1-MMP deficient [
33,
34], and for these reasons are commonly used by others to understand the effects of MT1-MMP overexpression [
21,
24,
25,
27,
28]. We determined that high levels of MT1-MMP expression did not correlate with migratory potential, and rather that low levels of MT1-MMP are optimal to enhance migration. Similarly, we showed that overexpression of MT1-MMP in MDA-MB-231 cells, which are naturally invasive and express endogenous MT1-MMP [
32,
34,
35], decreased their migratory potential and viability rather than enhanced it. These observations are consistent across platforms, both ex vivo with 3D cell culture [
36], and with in vivo tumourigenesis and cancer cell extravasation assays [
37], which all demonstrate that low MT1-MMP expression is optimal to induce a protrusive phenotype, increased invasiveness and metastatic capability in vivo
. Additionally, we analyzed the level of MT1-MMP protein in human 21 T breast cancer cell lines, which represent a progression from atypical ductal hyperplasia (ADH) to invasive mammary carcinoma (IMC), to show that the metastatic cell line produces little MT1-MMP protein, consistent with our conclusions using MCF-7 and MDA-MB 231 breast cancer cells. This “low MT1-MMP” migratory phenotype is accompanied by concomitant levels of TIMP-2, thus reconciling many conflicting studies on proteolytic factors in primary human tumours.
Methods
Cell culture
MCF-7, MDA-MB 231 and HS578t human breast cancer cell lines were obtained from the American Type Culture Collection (Manassas, VA). Cells were maintained in DMEM/F-12 media (Thermo Fisher) supplemented with 10 % FBS, 100 IU/ml penicillin, 100 μg/ml streptomycin, and incubated at 37 °C and 5 % CO2.
cDNA clones and reagents
Human MT1-MMP (sc116990), TIMP-2 (sc118083) and MMP-2 (sc321560) cDNA clones were purchased from Origene and subcloned into the vector pcDNA 3.3 (Thermo Fisher). The generation of the ALA + TIMP-2 cDNA construct in pcDNA 3.3 is described in Walsh et al. [
38]. The following reagents were used: Recombinant human TIMP-2 and 4-aminophenylmercuric acetate (APMA) (Sigma-Aldrich), BB-94 (Batimastat), U-0126, and AKT inhibitor IV (Santa Cruz), and Furin inhibitor II (Millipore).
Antibodies
For immunoblot analysis, the following primary antibodies were used: MT1-MMP (1:1000, AB6004, Millipore); MT1-MMP (1:1000, AB51074, Abcam); Phospho-ERK1/2 (1:2000, D13.14.4E), ERK1/2 (1:2000, 137 F5) (Cell Signaling Technology); TIMP-2 (1:1000, 3A4), β-Actin (1:1000, C4), and phospho-histone-3 (PH3) (1:5000, C1513) (Santa Cruz). Goat anti-mouse IgG (H + L) (Bio-Rad) and goat anti-rabbit IgG (H + L) (Thermo Fisher) HRP conjugates were used as secondary antibodies (1:10000). For immunofluorescence analysis we used MT1-MMP antibody AB6004 (1:200), and anti-rabbit-IgG-Alexa488 or Alexa594 (Thermo Fisher) as secondary antibodies (1:400).
Transfection and generation of stable cell lines
MCF-7 and MDA-MB 231 cells were seeded at a density of 5×105 cells/ml and incubated for 24 h. Following incubation, cells were transfected with Lipofectamine 2000 (Thermo Fisher) according to the manufacturer’s instructions. For transient transfection experiments, cells were incubated for 24 h after transfection and then utilized for experiments.
Stable cell lines were generated by transfection of cells with the respective cDNAs in the vector pcDNA 3.3, which contains a neomycin mammalian selection marker. Following transfection, cells were split 1:1000 and incubated in media containing 1 mg/ml G-418 (VWR). Individual colonies were selected after four weeks of incubation in selection media and expanded to assay for the levels of MT1-MMP by qPCR and immunoblotting. Stable cells lines expressing an shRNA sequence targeting MT1-MMP in the vector pRS (TR311445, Origene) were generated in the same manner expect using puromycin (2 μg/ml) as the selection antibiotic.
For zsGreen infection, cells were seeded at ~ 40 % density in a 6-well cell culture dish in 3 ml of media with a final concentration of 8 μg/ml polybrene and infected with 250 μL of virus. For virus production, the pLVX-ZsGreen1-N1 lentiviral plasmid was used. Twenty-four hours post-infection, the media containing virus was removed and replaced with puromycin selection media (2 μg/ml) for three days of incubation to select for infected cells.
Generation of MMP-2, TIMP-2 and ALA + TIMP-2 conditioned media (CM)
Conditioned media (CM) containing high levels of MMP-2, TIMP-2, and ALA + TIMP-2 protein was created by transfecting MCF-7 cells with cDNA constructs coding for the respective proteins. Following a 24-h incubation post-transfection, transfected cells were washed with phosphate buffered saline (PBS) and incubated in DMEM/F12 media without FBS for 24 h. The serum-free CM was then collected, aliquoted and stored for later use. Conditioned media from mock-transfected cells was used as a control.
Quantitative real-time PCR
RNA was collected from cells using the RNeasy Kit (Qiagen) and cDNA was synthesized from 1 μg of RNA using qScript cDNA supermix (Quanta). MT1-MMP mRNA levels were assayed by qPCR using PerfeCta SYBR Green Supermix (Quanta) and a CXF connect real time system with CFX manager software (Bio-Rad). mRNA levels were quantified by the ΔΔCT method and are displayed as fold change relative to parental MCF-7 cells. The level of GAPDH mRNA was used as the internal control. Primers are as follows: MT1-MMP; F: gcagaagttttacggcttgca, R: tcgaacattggccttgatctc, GAPDH; F: acccactcctccacctttga, R: ctgttgctgtagccaaattcgt [
33].
Immunoblotting
Post-incubation cells were washed 3× with PBS (pH = 7.2) and disrupted using lysis buffer (150 mM NaCl, 1 % NP-40, 0.5 % NaDC, 0.1 % SDS, 50 mM Tris pH 8.0) supplemented with protease/phosphate inhibitor (Thermo Scientific). Cell lysates were homogenized by sonication and 15 μg aliquots were analyzed by immunoblotting with MT1-MMP, TIMP-2, pH-3, β-actin, pERK1/2 or ERK1/2 primary antibodies, followed by incubation with the appropriate secondary HRP-conjugated antibody and detection using SuperSignal West Pico chemiluminescent substrate (Thermo Fisher). Protein lysate from human 21 T cell lines were acquired as previously described [
39].
Gelatin zymography and reverse zymography
Gelatin zymography and reverse zymography was done using samples of serum-free media (15 μl) as described previously [
40]. To assess the proMMP-2 activation ability of MCF-7 cells, which endogenously express very low levels of MMP-2 [
34], proMMP-2 CM was added to cells at a dilution of 30 μl proMMP-2 CM/mL SF media. For reverse zymography, serum-free media conditioned for 24 h by HS578t cells was used as the source of active MMP-2 within the gel, as these cells naturally express high levels of MMP-2 [
33].
Immunofluorescence
Samples for fluorescent gelatin degradation and 3D culture experiments were fixed and prepared for immunofluorescence according to their respective protocols (see below). MT1-MMP primary antibody (AB6004) was detected using anti-rabbit IgG Alexa594 or Alexa488 for Oregon green-488 gelatin degradation or 3D culture experiments respectively. F-actin was stained with Alexa633 phalloidin (1:100, Thermo Fisher) and nuclei with 4′,6-diamidino-2-phenylindole (1 μg/ml, BioShop Canada). Samples were imaged using a Nikon A1R+ confocal microscope (1.2 au) with a 20× dry or 60× oil-immersion lens and presented using NIS Elements software.
Transwell assays
The migratory potential of cells was measured using 24-well 8 μm pour transwell inserts (Corning Costar). Cells (2×10
4) were seeded on the upper chamber of the transwell in serum-free media and allowed to migrate towards the bottom chamber which was placed in DMEM/F-12 media supplemented with 10 % FBS. Migration assays were done with uncoated transwell inserts, whereas invasion assays were done with inserts coated with 20 % Matrigel. Cells that migrated to the lower chamber of the transwell insert were quantified as described previously [
41]. Migrated cells were normalized to MCF-7/MDA-MB 231 parental cells in control conditions and are presented as a mean percentage ± SEM.
Scratch closure migration assay
Cells were seeded at a density of 1×106 cells in a 35 mm cell culture dish and allowed to form a monolayer for 24 h. Following incubation, media was removed and a scratch was made down the middle of the cell culture dish with a 100 μl pipette tip. Cells were washed 3× with PBS (pH 7.2) to remove cell debris and then incubated with fresh media. After 2 h, 10 images were captured down the length of the scratch that represent the ‘initial’ size of the scratch for that sample. Every 24 h for 3 days, the same area of the scratch was imaged to examine how cells migrated to close the scratch. Scratch closure was quantified using ImageJ 2.0.0 software by measuring the width of the scratch each day and normalizing it to the initial size of the scratch. Scratch closure is presented as a mean percentage of the initial scratch size ± SEM.
Celltiter96® proliferation assay
Cells were seeded in triplicate (5000 cells/well) onto a 96-well cell culture dish in media supplemented with 10 % FBS or serum-free media. Immediately after seeding, 20 ul of Celltiter96® AQueous One Solution (Promega) was added in triplicate to each sample to obtain an initial reading and ensure that the different cell lines were seeded equally. The OD at 490 nm was measured using a Bio-Rad model 3550-UV microplate reader after incubating the cells with the substrate for 2 h. The substrate was added at the indicated day intervals and measured in the same way as the initial measurement to create a growth curve over that span, or to compare the effect of chemical inhibitors on cell viability during prolonged serum-free incubation.
Fluorescent gelatin degradation assay
Coverslips were coated with either Oregon Green-488 gelatin (for IF analysis) or gelatin labeled using an Alexa594 protein labeling kit (for live imaging analysis) (Thermo Fisher) and used to analyze ECM degradation as per Martin et al. [
42]. Immunofluorescence conditions are as described above.
Three-dimensional (3D) cell culture
Cells (2.5×10
4) were embedded in Matrigel (Corning Costar) and processed for immunofluorescence as per Cvetkovic et al. [
43]. To quantify colony morphologies observed in 3D culture, five random 50 μm Z-stacks (2 μm step size) were acquired using DIC microscopy at 10× magnification every day for 5 days post-embedding. For each individual Z-stack, invasive features (disseminations and protrusions, see text) were blindly counted and normalized to the number of circular colonies per field of view. Immunofluorescence procedure was done as described above using 3 % BSA/PBS as the blocking buffer. Single cells (marked by DAPI), F-actin disseminations, F-actin protrusions and zsGreen protrusions were blindly counted from 20× 3D volumes and normalized to the number of circular colonies per field of view.
Live-imaging and timelapse movies
Cells were embedded in Matrigel or seeded on Alexa594 gelatin coverslips as described above and placed in a live imaging chamber mounted on the stage of a Leica DM16000 B fluorescent microscope. To analyze the relationship between ECM degradation and migration, cells stably expressing zsGreen were incubated on Alexa594 gelatin coverslips and imaged at the same stage position at 10× magnification every 10 min for 20 h. These images were then compiled into timelapse movies using ImageJ. The zsGreen channel timelapse movies were used to quantify the migration of individual cells in an automated manner using the ADAPT plugin for ImageJ [
44]. The ADAPT plugin (v1.146) was used with default conditions and the trajectory visualization output was used to group cells according to the distance migrated. The Alexa594 gelatin channel was used to manually quantify the percentage of cells that had degraded the underlying gelatin at different time points. To visualize the dynamics of cells in 3D culture, cells were embedded in Matrigel and z-stacks (100 μm, 5 μm step size) were acquired at 20× magnification every 30 min for 72 h. Focal panels showing colony features with the greatest clarity were isolated and compiled into timelapse movies using ImageJ.
Avian embryo CAM implantation and extravasation efficiency assay
For implantation experiments, the superficial layer of the CAM of day 9 chicken embryos was removed to expose the underlying capillaries [
45]. MDA-MB 231, MCF-7 or MCF-7 MT1-MMP cell lines stably expressing zsGreen were then resuspended in Matrigel and pipetted onto the exposed capillaries (5×10
5 cells in 10 μl Matrigel/embryo). Eight days post-implantation the resulting tumour was imaged using a fluorescent stereoscope to examine vascularization of the tumour shown by the presence of non-fluorescent CAM vessels within the zsGreen tumour.
MCF-7 and MCF-7 MT1-MMP cell lines stably expressing zsGreen were used for intravenous injection into the CAM vasculature of day 14 chicken embryos and extravasation efficiency was quantified 24 h post-injection as per Kim et al. [
37].
Densitometry analysis
Quantitative analysis of immunoblots was done using QuantityOne software (Bio-Rad). Band intensity was obtained for the pERK and total ERK signal of each sample from three independent experiments. ERK activation is presented as a ratio between the pERK and the total ERK band intensity within each sample normalized to MCF-7 cells under control conditions.
Statistics
Statistical analysis and graphing was performed using GraphPad Prism version 6.0 (GraphPad software, La Jolla, CA, USA). Data is presented as mean ± SEM. One-way ANOVA followed by Tukey’s post-hoc test was used unless otherwise indicated in the figure legend. Significant differences denoted by asterisks are shown versus MCF-7 or MDA-MB 231 parental cells in control conditions unless otherwise shown in figure. Different levels of statistical significance are denoted by a different number of asterisks and are as follows: ****p ≤ 0.0001, ***p ≤ 0.001, **p ≤ 0.01, *p ≤ 0.05. All experiments were repeated in triplicate with comparable results.
Discussion
In this study, we utilized overexpression of functional MT1-MMP in MCF-7 and MDA-MB 231 breast cancer cells and demonstrated how overexpression corresponds to proMMP-2 activation and ECM degradation, but inversely correlates to migratory potential, protrusive phenotype in 3D culture, and tumorigenic features in vivo. Instead we showed that high overexpression of MT1-MMP negatively affects cell viability, and causes an abnormal loss of colony structure and cell fragment release in 3D culture that translates to decreased tumorigenic potential in vivo. We also demonstrated using the human 21 T cell lines mammary tumour progression series that breast cancer cells which mimic an invasive mammary carcinoma (IMC) are better represented by low, rather than high, levels of MT1-MMP protein. Our data is at odds with the notion that high MT1-MMP expression is crucial for tumour progression, as numerous studies report that MT1-MMP overexpression is associated with enhanced migratory ability and tumourigenecity [
17,
26,
27,
33,
34,
54], although there is also evidence in agreement with our study which shows that high MT1-MMP overexpression is insufficient to increase metastasis of human cancer cells [
31]. Additionally, despite the mounting evidence demonstrating that MMPs have a direct role in basement membrane degradation in vitro
, there is no evidence to date that shows this is the case in vivo
, and instead such studies suggest that the role of MMPs may not be related to ECM degradation but rather proteolysis of non-ECM substrates (Reviewed in [
9]).
Here, using MCF-7 clonal cell lines stably expressing untagged MT1-MMP, we show that migration, as shown by a scratch closure assay and by time-lapse microscopy of cells on fluorescent substrate, is dependent on levels of MT1-MMP, with high levels decreasing migratory ability and low levels promoting it. We also demonstrate that migration changes seen using transwells involve a substantial cell viability component, where high MT1-MMP overexpression negatively affects viability and low levels enhances it. This viability difference likely contributed to the magnitude of migration augmentation between transwell and other assays. Studies that show an increase in cancer cell migration as a result of MT1-MMP overexpression demonstrate migration enhancement ranging from 50 -500 % [
26‐
28], whereas others show a requirement of TIMP-2 for MT1-MMP mediated migration enhancement [
24,
25]. These reports do not agree on a specific mechanism, except general ERK activation. In contrast, there are studies demonstrating that MT1-MMP overexpression does not increase migration of breast cancer cells [
55], and also that MT1-MMP overexpression decreases ERK activation in cancer cells [
56]. In the latter study, the authors demonstrated that MT1-MMP overexpression in various cancer cell lines, including MCF-7 cells, downregulates ERK activation and migration in response to FGF-2, which is consistent with our findings using cells that express high levels of MT1-MMP. Other studies have provided strong mechanistic evidence that MT1-MMP is involved in apoptosis protection [
21] and viability enhancement via HIF1α stabilization [
23,
57], which are in line with our observations that MT1-MMP enhances viability during serum-free incubation.
A striking finding of our in vitro analysis regarding the relationship between MT1-MMP expression and migration was that the most migratory cells were the ones which had a low MT1-MMP:high TIMP-2 ratio. Of the MCF-7 MT1-MMP cell lines used, C3 and C3 SH 1 cells displayed low MT1-MMP levels (11 and 1.8 fold change vs parental MCF-7 cells, respectively), and their migration ability was greatly enhanced when the levels of TIMP-2 increased, especially C3 SH 1 cells. MDA-MB 231 MT1-MMP cell lines displayed the same trend whereby MT1-MMP overexpression with no change in TIMP-2 expression (data not shown) shifted the ratio in favor of excess MT1-MMP, and thereby causing a decrease in migratory potential. Similarly, analysis of the natural migration potential of MCF-7, MDA-MB 231, and HS578t cells is consistent with this relationship to TIMP-2, as HS578t cells were the most migratory and displayed the highest level of TIMP-2 relative to MT1-MMP. Noteworthy in this analysis was also the observation that MDA-MB 231 cells displayed the highest ERK activation, but were not the most migratory. This was consistent with our comparison of MCF-7 C2 and C3 cells, whereby C2 cells showed the highest ERK activation but were less migratory than C3 cells, which displayed comparatively lower ERK activation even in the presence of high levels of TIMP-2.
Although TIMP-2 is a natural MMP inhibitor and as such has attracted therapeutic interest along with synthetic MMP inhibitors [
10,
58‐
62], neither have shown value in clinical trials [
63]. Instead, some have suggested that high TIMP-2 levels may promote tumourigenecity [
24,
64,
65], which has been strengthened by the association of high TIMP-2 levels with poor prognosis in various human cancers, including breast [
66‐
71]. Although this association between an MMP inhibitor and poor cancer prognosis may be paradoxical, we describe here that while TIMP-2 regulation of MT1-MMP activity is complex, as exemplified by MDA-MB 231 and HS578t cells, high TIMP-2:low MT1-MMP ratios in these cells correlate with their migratory potential, and also with their lack of gelatin degradation and inhibited proMMP activation ability. Despite the fact that MDA-MB 231 and HS578t cells naturally express MT1-MMP, MMP-2, and −9, the extracellular gelatinases are predominately found in their pro-forms yet to be activated. As MT1-MMP activity is pivotal in gelatinase activation [
20], this indicates that MT1-MMP present in these cells is inhibited, likely by TIMP-2. This is consistent with lack of gelatinase activity of C3 cells, and in stark contrast to C1 and C2 cells, suggesting that cells with non-physiologically high MT1-MMP exhibit excessive proteolysis which may be counterproductive to migration and cell viability. This is corroborated both with the rescued serum free viability and migration of C1 cells as a result of BB94 treatment, and with the role of TIMP-2 in mediating survivability under serum free conditions as shown by others [
21]. Since MT1-MMP is a proteolytic enzyme that can cleave and alter the function of many ECM and non-ECM proteins crucial for proper cell behavior [
18], it is logical that such a potent protease with wide substrate specificity would be under tight control by TIMP-2 to appropriately mediate cell behaviour.
Maden and Bugge (2015) analyzed the last two decades of literature to examine if there was a consensus regarding the cellular source of MMPs (including MT1-MMP) in human cancers and whether they were predominately stromal or cancer cell derived. These authors noted that publications were widely inconsistent in regards to the cellular source of MMPs, particularly when immunodetection was involved. Only when in situ hybridization was used was there a consensus seen that MMPs were likely stromal cell derived [
29]. The authors proposed reasons for these difficulties, one being that there is likely inherently low expression of MMPs in cancer cells compared to stromal cells making immunodetection technically challenging.
We believe that unreliable immunodetection reagents (discussed in [
72]) is a major contributing reason as to why there is such inconsistency when assessing the abundance of MMPs in human cancers and their value as prognostic markers. In our study, we initially experienced difficulties assessing immunoblots for MT1-MMP, which could only be correctly interpreted after examining the immunological banding pattern for MT1-MMP expressing MCF-7 and MDAMB-231 cell lines and probing with two different primary antibodies against MT1-MMP (Additional file
15: Figure S5). To strengthen the idea that improper immunodetection of MT1-MMP protein can lead to incorrect conclusions, we highlight our (lack of) immunodetection of MT1-MMP in MCF-7 breast cancer cells. We strongly believe, as suggested by others, that MCF-7 cells are MT1-MMP deficient [
24,
33], particularly because it has been shown that the MT1-MMP promoter in these cells is hypermethylated and thus transcriptionally repressed [
73]. Yet despite this observation, published studies claim to detect both pro- and active MT1-MMP protein via immunoblot in these cells [
34,
74], which could be due to incorrect identification of the cell line used for experimentation, or lack of stringency when conducting immunodetection. As can be seen from our immunoblot data, usage of a polyclonal antibody against MT1-MMP resulted in a non-specific signal that could easily be misinterpreted as pro- and active MT1-MMP in MCF-7 cells. Furthermore, in the study done by Köhrmann et al., although the authors reported MT1-MMP protein in MCF-7 cells, they were not able to detect MT1-MMP protein from clinical samples via immunoblot. However, they were able to detect MT1-MMP protein in tissue sections from tumour samples and not from normal tissues. Studies that are internally inconsistent regarding MT1-MMP protein detection, and which contain clinical samples, can create confounding conclusions regarding the role of MMPs in cancer and are in agreement with the observations of Madsen and Bugge regarding the discrepancies of the source of MMPs in human cancer. Additionally, visualizing MT1-MMP protein at a cellular level using immunofluorescence may also lead to similar immunodetection problems. Lodillinsky et al. recently implicated the p63/MT1-MMP axis in the transition from in situ to metastatic breast cancer, reporting that MT1-MMP protein is present during BM invasion of MCF10DCIS.com xenografts. However, with the knowledge that MT1-MMP is localized to distinct specialized regions of the cell membrane to initiate invasion (invadopodia), we question such immunofluorescence data that show MT1-MMP protein is present throughout the cell membrane of every cell in the xenograft, regardless of whether it is in physical proximity to invade the BM.
To assess the physiological relevance of observed levels of MT1-MMP expression in this study, we compared these to recently reported MT1-MMP mRNA levels in malignant and non-malignant human breast tissues. The reported increase in MT1-MMP mRNA levels between malignant tissues and normal tissues ranged between ~1.7 and ~3.2 fold [
75‐
77]. Similarly, a pioneering study used MDA-MB 231 variants that produced constitutively active scr kinase, which is known to be upregulated during cancer progression [
35]. These MDA-MB 231 variants generated significantly more MT1-MMP containing invadopodia. Analysis of MT1-MMP mRNA changes between control and constitutively active src kinase cells demonstrated a ~1.8 fold change increase in MT1-MMP mRNA, which the authors describe as a mechanistically meaningful increase in MT1-MMP expression. Therefore, the physiological relevance of extreme changes in expression levels, such as our ~17,000 fold change in MT1-MMP mRNA seen in our transient transfectants, or ~1500 fold change in stable cell lines, would be difficult to reconcile with primary human breast cancers which have a ~1.7 to 3.2 fold change in MT1-MMP mRNA expression. Additionally, in line with our idea that immunological reagents of MT1-MMP may be unreliable, if normal non-malignant tissues do not contain detectable levels of MT1-MMP [
17,
34] and cancerous tissues demonstrate only a ~1.7 to 3.2 fold increase in MT1-MMP mRNA, then is it reasonable that a transcriptional increase of that magnitude would be difficult to immunodetect at the protein level.
Consistent with the conclusion of the importance of low levels of MT1-MMP expression are the observations seen in our physiologically relevant ex vivo and in vivo experiments, whereby MCF-7 cells expressing low levels of MT1-MMP (C3) demonstrated behavior consistent with the role of MT1-MMP during cancer progression. C3 cells have increased protrusive morphology in 3D culture and this is contrasted with MCF-7 cells overexpressing high levels of MT1-MMP which show reduced protrusive ability and increased cell fragmentation. Similarly, C3 cells were tumorigenic and showed metastatic potential in vivo, unlike C1 and C2 cells, which is consistent with studies that knock down MT1-MMP expression and show inhibition of these parameters. Additionally, analysis of MT1-MMP protein levels in the 21 T cell lines mammary tumour progression series demonstrated that breast cancer cells which represent ADH and DCIS mammary tumours produce high levels of active MT1-MMP protein, whereas invasive 21MT-1 IMC cells produce undetectable levels of MT1-MMP, an observation that is consistent with our findings using MCF-7 C3 cells and our overall conclusion that low levels of MT1-MMP may better represent metastatic cancer. A similar study using the HMT-3522 epithelial cell series yielded results consistent with our analysis of MT1-MMP levels in 21 T cells, as these authors analyzed microarray data to show that MMP-9, −13,-15 and −17, but not MT1-MMP, were functionally significant in the acquisition of invasiveness [
78].
Interestingly, the observation that DCIS 21 T cells produced high levels of active MT1-MMP in comparison to their IMC counterparts is similar to the 3D culture phenotype of our MDA-MB 231 cells overexpressing MT1-MMP. In our study, parental MDA-MB 231 cells, which are naturally invasive, readily form irregular networks in 3D culture in contrast to non-invasive cell lines that partially maintain polarity and form acini similar to the TDLU in the human breast (eg MCF-7 cells). It was surprising to us to that overexpression of MT1-MMP in invasive MDA-MB 231 cells reverted their phenotype in 3D culture towards a DCIS-like morphology where the ability to form networks in matrigel 3D culture was restricted and a higher proportion of these cells retain acini-like colonies. The reversion of MDA-MB 231 cells to a DCIS-like phenotype in 3D culture as a result of MT1-MMP overexpression is consistent with our analysis of MT1-MMP protein levels in the 21 T cell lines whereby DCIS (21NT) cells produce more active MT1-MMP and predominately form acini in 3D culture, and IMC (21MT-1) cells produce less MT1-MMP protein and display invasive 3D behavior [
39], similar to our MT1-MMP MDA-MB 231 cells and parental MDA-MB 231 cells, respectively.
Taken together, this study shows that a physiologically relevant increase in MT1-MMP expression is at best represented by a 1.8 to 11-fold change compared to normal tissue. Additionally, although abnormally high levels of MT1-MMP overexpression may not reflect those seen in primary breast cancers, there is still mechanistic value in this approach, as utilized in this study to demonstrate the constancy of the TIMP-2 mediated activation of proMMP-2 by MT1-MMP. With this work we want to challenge the long-standing view that MMPs, particularly MT1-MMP, exert their role in cancer progression as proteases that predominately degrade ECM components to allow cancer cell invasion, and instead suggest a subtle role for MT1-MMP in tumour progression as metastatic cancer appears to be better represented by low levels of inhibited MT1-MMP protein.