ETV4 transcription factor and MMP13 metalloprotease are interplaying actors of breast tumorigenesis

The TAC murine mammary epithelial cell line [32, 33] was cultured on collagen-coated plates in Gibco high-glucose DMEM (Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 10% FCS, penicillin (110 IU/ml), and streptomycin (110 μg/ml). A wild-type mouse mammary tumor (MMT) cell line (ATCC® CCL-51™; American Type Culture Collection, Manassas, VA, USA) was cultured in DMEM supplemented with 10% (vol/vol) FBS, gentamicin (100 IU/ml), and nonessential amino acids (Gibco; Thermo Fisher Scientific). The MCF10A cell line (ATCC® CRL-10317™; American Type Culture Collection) was propagated in DMEM/F-12 medium (Gibco; Thermo Fisher Scientific) supplemented with 5% horse serum, 20 ng/ml epidermal growth factor, 100 ng/ml cholera toxin, 500 ng/ml hydrocortisone, and 0.01 mg/ml insulin.

pTracer-ETV4 and pLPCX-ETV4-V5 plasmids were described previously [8, 12, 32]. pMX-MMP13 was generated by PCR amplification of the mouse MMP13 complementary DNA (cDNA) and cloned into the pMX-Puro retroviral vector. pRS-shMMP13 retroviral plasmid was kindly provided by S. Meierjohann [34]. pGL3 constructs containing different parts of the MMP13 promoter were kindly provided by J. M. Davidson [35]. Mutations to proximal ETS and activator protein (AP)-1 binding site in MMP13 promoter were made using the QuikChange® II XL Site-Directed Mutagenesis Kit (Agilent Technologies, Santa Clara, CA, USA). The proximal ETS site was changed from GGAA to CCAA, and the proximal AP-1 site was changed from TGACT to GCACT. The sequence of all promoter constructs was verified by DNA sequencing.

HEK293GP packaging cells (Clontech) (3 × 10⁶) were transfected with pLPCX, pMX, or pRS retroviral constructs, and 1.2 × 10⁶ MMT cells or 1 × 10⁶ MCF-10A cells per 100-mm dish were incubated with supernatant as previously described [16]. The selection procedure was started the next day using puromycin (Life Technologies, Carlsbad, CA, USA).

TAC cells overexpressing ETV4 after retroviral infection (ETV4) and control mock-infected TAC cells (ctrl) were previously described [8]. MMT cells infected with pLPCX retroviral vector (Ctrl) were used as a control for MMT cells overexpressing ETV4 after retroviral infection with the pLPCX-ETV4 vector (ETV4). MMT cells infected with pMX retroviral vector (Ctrl) were used as a control for MMT cells overexpressing MMP13 after retroviral infection with the pMX-MMP13 vector (MMP13).

MMT cells infected with pRS retroviral vector (shCtrl) were used as a control for MMT cells overexpressing shMMP13 after retroviral infection with the pRS-shMMP13 vector (shMMP13). MMT-ETV4 cells (overexpressing ETV4) infected with pRS retroviral vector (ETV4 + shCtrl) were used as a control for MMT-ETV4 cells overexpressing shMMP13 after retroviral infection with the pRS-shMMP13 vector (ETV4 + shMMP13).

Total RNA was extracted using the RNeasy Mini Kit according to the manufacturer’s instructions (Qiagen, Hilden, Germany). Total RNA (1 μg) was reverse-transcribed using the high-capacity cDNA reverse transcription kit (Life Technologies). Specific gene expression was determined by real-time PCR using the Fast SYBR® Green Master Mix (Life Technologies) and the Mx3005P qPCR system (Agilent Technologies). The results were analyzed with the comparative cycle threshold method normalized to cyclophilin A and compared with a comparator sample. The nucleotide sequences of the primers used were as follows: MMP13-F (5′-TCCCTGCCCCTTCCCTATGG-3′) and MMP13-R (5′-CTCGGAGCCTGTCAACTGTGG-3′) for the MMP13 gene (PCR product of 173 bp), ETV4-F (5′-CCGCTCGCTGCGATACTATT-3′) and ETV4-R (5′-CGGTCAAACTCAGCCTTCAGA-3′) for the ETV4 gene (PCR product of 162 bp), and PPIA-F (5′-GGGAACCGTTTGTGTTTGGT-3′) and PPIA-R (5′-TGTGCCAGGGTGGTGACTTT-3′) for PPIA gene.

TAC cells (3 × 10⁴) were seeded in 12-well plates and transfected with ExGen 500 (Euromedex, Strasbourg, France) and 250 ng of DNA (200 ng of expression vector, 25 ng of firefly luciferase reporter vector, 25 ng of the Renilla luciferase pRL-TK; Promega, Madison, WI, USA), according to a protocol previously described [12].

TAC pLNCX, pLNCX-ETV4 and MMT pLPCX, pLPCX ETV4 were fixed, lysed, and used for chromatin immunoprecipitation (ChIP) with an anti-ETV4 immunoglobulin G (IgG) (sc-113x; Santa Cruz Biotechnology, Dallas, TX, USA) or a nonrelevant antibody normal rabbit IgG (sc-2027), as described in Additional file 1. Detection of specific DNA regions was performed by PCR using the MMP13 gene promoter region and CCND2 gene promoter region. The nucleotide sequences of the primers used were as follows: MMP13 forward: 5′-TCCATTTCCCTCAGATTCTGCCAC-3′ and MMP13 reverse: 5′- TCTCTCCTTCCCAGGGCAAGCAT-3′ for the MMP13 gene (PCR product of 164 bp) and CCND2 forward: 5′-GAGAGGGAGGGAAAGATTGAAAGGA-3′ and CCND2 reverse: 5′- AGGTGGGCGAGCGGAGCCTCAAG-3′ for the CCND2 gene (PCR product of 212 bp).

MMP13 and oligonucleotides used for RNA interference were purchased from Dharmacon (SMARTpool ON-TARGETplus MMP13 J-047459, Dharmacon, Lafayette, CO, USA) and consisted of a mix of four siRNAs: J-047459-12-GGCCCAUACAGUUUGAAUA/J-04745911-AGACUAUGGACAAAGAUUA/J-047459-10-UCAAAUGGUCCCAAACGAA/J-047459-9-CUGCGACUCUUGCGGGAAU. Control oligonucleotide consisted of ON-TARGETplus nontargeting siRNA#1 (D-001810-01-UGGUUUACAUGUCGACUAA control siRNA). MMT cells (3.5 × 10⁵) were seeded in six-well plates for reverse transfection with 75 pmol of each siRNA and 5 μl of Lipofectamine® 2000 reagent (Thermo Fisher Scientific) as recommended by the manufacturer. Cells were incubated for another 24 hours under standard conditions before being assayed.

Cells were lysed in buffer made of 150 mM NaCl, 50 mM Tris-HCl, pH 7.5, 1% Nonidet P-40 (vol/vol), 1 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, and 10 g/ml aprotinin. After scraping, cellular debris was removed by centrifugation at 10,000 × g for 5 minutes. Protein concentrations were determined by using a Bradford assay. For the supernatant, a subconfluent culture was grown for 24 hours in serum-free medium, then the supernatant was centrifuged at 3000 × g for 3 minutes. Whole-cell extracts (50 μg) or 1:20 of supernatants were separated in precast gels (Mini-PROTEAN® TGX Stain-Free™; Bio-Rad Laboratories, Hercules, CA, USA) gel and transferred onto nitrocellulose membranes (Trans-Blot Turbo Transfer System; Bio-Rad Laboratories). After blocking with Tris-buffered saline, 0.1% Tween, and 3% bovine serum albumin (BSA), the membrane was probed with the primary and secondary antibodies. The enzymatic activity was detected using an Amersham enhanced chemiluminescence kit (GE Healthcare Life Sciences, Marblehead, MA, USA). Equal transfer of proteins from the gel was controlled by using the stain-free system of the gel and the membrane as well as by using an anti-GAPDH antibody. We used anti-ETV4 1:500 (GTX114393; GeneTex, Irvine, CA, USA), anti-MMP13 1:500 (18165-1-AP; Proteintech, Rosemont, IL, USA), anti-GAPDH 1:1000 (6C5-sc-32233; Santa Cruz Biotechnology), and secondary antimouse or antirabbit antibodies coupled to horseradish peroxidase (HRP) (GE Healthcare Life Sciences).

Gelatin zymography was used to determine the activity of MMP13. The supernatant from subconfluent serum-free culture medium was collected, and the cells were removed by centrifugation. Then, 40 μl of sample were loaded onto a 10% precast polyacrylamide gel with 0.1% of gelatin (Bio-Rad Laboratories). After electrophoresis, the gels were renatured by soaking for 30 minutes at room temperature in 2.5% Triton X-100. The gels were then incubated in a developing buffer (50 mM Tris, 200 mM NaCl, 5 mM CaCl₂, 0.02% Brij-35 [MilliporeSigma], pH 7.5) overnight at 37 °C. The gels were stained with Coomassie Brilliant Blue R-250 and destained in demineralized water. The transparent bands of gelatinolytic activity were visualized as clear bands against the blue-stained gelatin background.

Stable MMT cells (1.5 × 10⁴; pLPCX/pLPCX-ETV4-pMX/pMX-MMP13-pRS/pRS-shMMP13 and MMT-ETV4-pRS/MMT-ETV4-pRS-shMMP13) were seeded in six-well plates. The cells were trypsinized and counted after 10, 35, 55, 80, and 100 hours using an Invitrogen Tali™ Image-based Cytometer (Thermo Fisher Scientific). Each time point was counted three times.

Stable MMT cells (3 × 10⁵; pLPCX/pLPCX-ETV4-pMX/pMX-MMP13-pRS/pRS-shMMP13 and MMT-ETV4-pRS/MMT-ETV4-pRS-shMMP13) were seeded in 500 μl of medium mixed with 1 ml of 0.35% agar in growth medium (DMEM [Thermo Fisher Scientific] with 10% FBS). The cell suspension was cast onto 12-well plates with 500 μl of 0.65% agar in growth medium, which was used as an underlay. Growth medium was added onto the agar layer and changed weekly. Colonies were photographed after 10 days using a light microscope (Axio Vert A.1; Carl Zeiss Microscopy, Jena, Germany).

Boyden chamber cell migration was assayed using a cell culture chamber insert system (BD Biosciences, San Jose, CA, USA) with an 8-μm polyethylene terephthalate (PET) membrane. Stable MMT cells (4 × 10⁴; pLPCX/pLPCX-ETV4-pMX/pMX-MMP13-pRS/pRS-shMMP13 and MMT-ETV4-pRS/MMT-ETV4-pRS-shMMP13) were seeded in the upper chamber in DMEM with 10% FBS. The same medium was added in the lower chamber. After 18 hours, cells that did not cross the membrane were scraped off the upper side of the membrane with a cotton swab. Cells that had migrated to the lower side were fixed with methanol at − 20 °C and stained with Hoechst 33258 (MilliporeSigma). The membrane was excised from its support and mounted on a glass side with Dako Glycergel mounting medium (Agilent Technologies). Cells were photographed using a light microscope (Axio Vert A.1) and counted using ImageJ software (National Institutes of Health, Bethesda, MD, USA).

Boyden chamber cell invasion was assayed using a cell culture chamber insert system (Corning® BioCoat™ Growth Factor Reduced Matrigel® Invasion Chamber; Corning Life Sciences, Corning, NY, USA) with an 8-μm PET membrane coated with Matrigel®. Stable MMT cells (8 × 10⁴; pLPCX/pLPCX-ETV4-pMX/pMX-MMP13-pRS/pRS-shMMP13 and MMT-ETV4-pRS/MMT-ETV4-pRS-shMMP13) were seeded in the upper chamber in DMEM with 0% FBS and 0.1% BSA. DMEM with 5% FBS was added in the lower chamber. After 36 hours, cells that did not cross the membrane were scraped off the upper side of the membrane with a cotton swab. Cells that had migrated to the lower side were fixed with methanol at − 20 °C and stained with Hoechst 33258. The membrane was excised from its support and mounted on a glass side with Dako Glycergel mounting medium. Cells were photographed using a light microscope (Axio Vert A.1) and counted using ImageJ software.

MMT pLPCX-ETV4-pRS and MMT pLPCX-ETV4-pRS-shMMP13 cells were trypsinized, then suspended in PBS (5 × 10⁶ cells/ml). Cells (5 × 10⁵) were injected subcutaneously into the inguinal flank of 6–7-week-old female severe combined immunodeficiency (SCID)-deficient mice. A total of six mice per condition were used in three independent experiments. Tumor size was assessed by measuring the length and width of tumors every 3–4 days. Tumor volume was estimated using the formula: (length × width²)/2. The experiments were stopped when the largest tumors reached the critical size of about 10% of the mouse’s weight in accordance with the ethical approval form, thus meaning between 2 and 3 weeks postinjection. At the time the mice were killed, the tumors were removed, fixed in 4% paraformaldehyde, and embedded in paraffin. Results are expressed as the mean tumor volume for each experimental group. All animal procedures were conducted with the approval of and in compliance with the guidelines of the Nord Pas de Calais Regional for Ethical Animal Care and Use Committee (CEEA-003243.01).

IHC of paraffin-embedded mouse tumor tissue sections was performed using anti-ETV4 (1:100, GTX100812; GeneTex), anti-MMP13 (1:100, 18165-1-AP; Proteintech), anti-Ki67 (1:200, ab15580; Abcam, Cambridge, UK), and anti-cleaved caspase 3 (1:100, Asp175, catalogue no. 9661; Cell Signaling Technology, Danvers, MA, USA). Sections were incubated with secondary HRP-conjugated antibody. Counterstaining was performed using Mayer’s hematoxylin (Merck, Darmstadt, Germany). Imaging was carried out using ZEN Blue imaging software (Carl Zeiss Microscopy).

Samples of 456 primary unilateral invasive breast tumors excised from women managed at Curie Institute-René Huguenin Hospital (St. Cloud, France) from 1978 to 2008 were analyzed. Immediately after biopsy or surgery, the tumor samples were stored in liquid nitrogen until messenger RNA (mRNA) extraction. Tumor samples were considered suitable for our study if the proportion of tumor cells exceeded 70%.

All patients (mean age 61.7 years, range 31–91 years) met the following criteria: primary unilateral nonmetastatic breast carcinoma for which complete clinical, histological, and biological data were available; no radiotherapy or chemotherapy before surgery; and full follow-up at Curie Institute-René Huguenin Hospital. Treatment consisted of modified radical mastectomy in 278 cases (63.6%) and breast-conserving surgery plus locoregional radiotherapy in 159 cases (36.4%) (information available for only 437 cases). The patients underwent a physical examination and routine chest radiography every 3 months for 2 years, then annually. Mammograms were done annually. Adjuvant therapy was administered to 369 patients, consisting of chemotherapy alone in 91 cases, hormone therapy alone in 176 cases, and both treatments in 102 cases. The histological type and the number of positive axillary nodes were established at the time of surgery. The malignancy of infiltrating carcinomas was scored according to the Scarff-Bloom-Richardson (SBR) histoprognostic grading system. Hormone receptor (HR) [estrogen receptor (ERα), progesterone receptor (PR)] and human epidermal growth factor receptor 2 (ERBB2) status were determined at the protein level by using biochemical methods (dextran-coated charcoal method, enzyme immunoassay, or IHC) and confirmed by qPCR assays as described in Additional file 1.

The population was divided into four groups according to HR (ERα and PR) and ERBB2 status, as follows: two luminal subtypes [HR+ (ERα + or PR+)/ERBB2+ (n = 54)] and [HR+ (ERα + or PR+)/ERBB2− (n = 289)]; an ERBB2+ subtype [HR− (ERα− and PR−)/ERBB2+ (n = 45)] and a triple-negative subtype [HR− (ERα− and PR−)/ERBB2− (n = 68)]. The median follow-up was 8.9 years (range 130 days to 33.2 years). One hundred eighty-one patients had a metastasis. Clinicopathological characteristics of patients in relation to metastasis-free survival (MFS) are provided in Additional file 2: Table S1. Ten specimens of adjacent normal breast tissue from patients with breast cancer or normal breast tissue from women undergoing cosmetic breast surgery were used as sources of normal mRNA.

For human statistical analysis, relationships between mRNA expression of genes and clinical parameters were identified using nonparametric tests, the χ² nonparametric test (relationship between two qualitative parameters), and the Kruskal-Wallis H test (relationship between one quantitative parameter and two or more qualitative parameters). Differences were considered significant at confidence levels greater than 95% (P < 0.05). MFS was determined as the interval between initial diagnosis and detection of the first metastasis.

To visualize the efficacy of MMP13 and ETV4 mRNA levels for discriminating between two populations (patients who developed/did not develop metastases) in the absence of an arbitrary cutoff value, data were summarized in an ROC curve. The AUC was calculated as a single measure to discriminate efficacy. The population was divided into four patient subgroups according to MMP13 ROC curve value in the series of 456 breast cancer samples and then according to ETV4 ROC curve value in the high and low MMP13 mRNA expression level subpopulations. Finally, the very small subgroup (n = 16) of high MMP13/low ETV4 mRNA expression level was merged with the subgroup (n = 66) of low MMP13/low ETV4 mRNA expression level to obtain a unique group of 82 patients with low ETV4 mRNA expression level. Survival distributions were estimated by the Kaplan-Meier method, and the significance of differences between survival rates was ascertained with the log-rank test. The Cox proportional hazards regression model was used to assess prognostic significance in multivariate analysis [36]. We also analyzed an independent dataset of breast tumors for which microarray data were publicly available (Netherlands Cancer Institute [NKI], n = 295; http://​ccb.​nki.​nl/​data/​).

We previously described ETV4-regulated genes in mammary tumorigenic MMT cells following ETV4 inhibition [16]. Among them, MMPs such as MMPs 1, 2, 3, 9, and 14 were shown to be slightly regulated [16]. However, our attention was focused on MMP13, which was identified as a potentially interesting ETV4 target gene through large-scale transcriptomic analysis that we performed on these MMT cells [16], thereafter completed with transcriptomic analysis performed with ETV4-overexpressing and ETV4-repressing TAC cells (unpublished data). In these latter analyses, we found that ETV4 positively modulates MMP13 gene expression (18.12-fold, P = 0.0014 following ETV4 overexpression; 0.16-fold, P = 0.02 following ETV4 inhibition). In order to characterize the regulation of MMP13 expression by ETV4, we used the mammary epithelial TAC cell line overexpressing ETV4 [8] as well as the mammary cancerous MMT cell line and the breast epithelial MCF10A cell line, engineered to overexpress a V5-tagged ETV4 protein after a retroviral infection (MMT-ETV4 and MCF10A-ETV4) (Fig. 1a and c and Additional file 3: Figure S1a and c). Overexpression of ETV4 upregulates MMP13 mRNA and protein expression in the mouse TAC, MMT (Fig. 1b and d) and human MCF10A cells (Additional file 3: Figure S1b and d). Moreover, the secretion of the active form of MMP13 is increased in the supernatant of ETV4-overexpressing cells, as shown by Western blotting (Fig. 1e) and zymography (Fig. 1f).

We next completed these data by analyzing the ETV4-regulated MMP13 promoter. TAC cells were transfected with various MMP13 gene promoter fragments cloned into a luciferase reporter vector and ETV4 expression vector or control vector. Our results indicate that the MMP13 promoter (− 1800) is active in TAC cells. Moreover, ETV4 transactivates the MMP13 gene promoter region spanning 1800 bp downstream from the translation start codon (MMP13; − 1800) (Fig. 2a and b). To delineate the responsive elements driving the promoter activity, we tested various deletion constructs (Fig. 2a) and identified a region of 91 bp (MMP13–91), which displays an optimal promoter activity as well as induction by ETV4 (Fig. 2b). This region contains a putative ETS binding site (EBS) and a putative AP-1 binding site. These two sites are very highly conserved among mouse, human, and rabbit [37, 38] (Additional file 4: Figure S2), and the EBS was previously described to be important in MMP13 gene promoter activity [38]. In fact, the mutation of the EBS site in the MMP13 − 91 and MMP13 − 391 fragments reduced by half the transactivation by ETV4 (Fig. 2b). Moreover, AP-1 synergized the ETV4-induced transactivation effect, and the AP-1 site is required for this activity (Fig. 2c). We thereafter evidenced ETV4 recruitment to this chromatin region in TAC and MMT cells by ChIP using an antibody directed against ETV4, and, as a positive control, we analyzed the binding of ETV4 at the cyclin D2 promoter, as previously described [8] (Fig. 2d). It is noteworthy that the same results were obtained in TAC and MMT cells that overexpress ETV4 (Additional file 5: Figure S3). Therefore, MMP13 is an ETV4 target gene in TAC and MMT mammary epithelial cells with AP-1 as a likely coactivator.

Proliferation and migration assays showed that MMT cells overexpressing ETV4 display enhanced proliferation and migration abilities as determined using a Boyden chamber (Fig. 3a and b). Similar results were obtained with or without treatment with Mitomycin C, an inhibitor of proliferation, indicating that the effect on cell migration was not a consequence of an increase in cell number (data not shown). Moreover, invasion assay in a Matrigel®-overlaid Boyden chamber and in a clonogenic assay revealed that ETV4 significantly increases invasion (Fig. 3c) and anchorage-independent growth (Fig. 3d) in vitro. These data confirm that ETV4 is an important actor of the cellular abilities (proliferation, migration, invasion, anchorage-independent growth) involved in the tumorigenic properties of MMT cells.

Next, we evaluated the role of MMP13 during MMT cell migration, invasion, or clonogenicity by using MMT cells in which MMP13 is overexpressed (MMT-MMP13) or knocked down by shRNA (MMT-shMMP13). MMP13 overexpression or repression was confirmed by qPCR (Additional file 6: Figure S4a and b), Western blotting (Additional file 6: Figure S4c) and/or zymography (Additional file 6: Figure S4d). As shown in Fig. 4, MMP13 overexpression increases cell proliferation (Fig. 4a), migration (Fig. 4b), and anchorage-independent growth (Fig. 4c). As expected, MMP13 repression leads to a reduction in cell proliferation (Fig. 4d), cell migration (Fig. 4e), and anchorage-independent cell growth (Fig. 4f). Thus, similarly to ETV4, but with a weaker effect, MMP13 is an inducer of cancer cell proliferation, migration, and invasion.

In order to determine if MMP13 participates in ETV4-regulated cancer cell properties, we repressed MMP13 in the ETV4-overexpressing MMT cell line. To that end, we established the MMT-ETV4 + shMMP13 cell line, which expresses an MMP13-shRNA construct allowing for a significant reduction in MMP13 mRNA expression and subsequently a reduction in MMP13 metalloprotease activity (Fig. 5a and b). Importantly, as determined by qPCR and Western blotting, ETV4 mRNA and protein expression remains unchanged in these cells (Additional file 7: Figure S5a and b). The repression of MMP13 in the ETV4-overexpressing MMT cells drastically decreases their proliferation (by 60% at 100 hours after the beginning of the experiment) (Fig. 5c) and significantly reduces cell migration (twofold decrease compared with ETV4 + shCtrl) (Fig. 5d), cell invasion (twofold decrease compared with ETV4 + shCtrl) (Fig. 5e), and anchorage-independent growth (2.5-fold increase compared with ETV4 + shCtrl) (Fig. 5f). This was confirmed by transient transfection of a siRNA directed against MMP13 in the MMT-ETV4-overexpressing cells, which led to a 60% decrease in MMP13 expression (Additional file 8: Figure S6a) and a significant reduction in anchorage-independent cell growth (Additional file 8: Figure S6b). Altogether, these results show that MMP13 acts as a relay of ETV4 to control mammary cancer cells’ tumorigenic abilities.

To investigate whether MMP13 expression is necessary for the induction of tumors by ETV4 in vivo, MMT-ETV4 + shCtrl and MMT-ETV4 + shMMP13 cells were injected into the inguinal flanks of immunocompromised mice, and tumor growth was evaluated every 3–4 days (Fig. 6). Three days postinjection, all of the mice that received an injection of MMT-ETV4 + shCtrl cells showed a palpable tumor, whereas none could be detected at this stage in the group that received an injection of MMT-ETV4 + shMMP13 cells. By day 6 postinjection, all MMT-injected mice developed palpable tumors. However, a 3–4-day measurement of tumor size over the course of 10 more days (until animals were killed) indicated that MMP13 expression was required for optimal tumor growth because MMP13-shRNA-expressing ETV4 cells are, on average, twofold smaller than controls. Immunocytochemistry performed on paraffin-embedded mouse tumor tissue sections showed an equivalent Ki-67 expression in tumors from ETV4 + shCtrl and ETV4 + shMMP13 cells, which all show proliferative activity. Cleaved caspase 3 expression showed that apoptosis events are present in both conditions, to a slightly greater extent in the ETV4 + shMMP13 tumors, according to their slow growth. ETV4 expression is, as expected, equivalent in both ETV4-expressing cell-derived tumors. In contrast, MMP13 expression decreased in ETV4 + shMMP13-derived tumors, thus confirming the suitable MMP13 regulation (here a repression) in these in vivo assays (Fig. 6b). Therefore, these data bring out that MMP13 is a mediator of ETV4 tumorigenic activity in MMT cancer cells.

In order to corroborate the relevance of the phenotypic and mouse in vivo data and to explore the link between ETV4 and MMP13 in human breast cancer, we assessed MMP13 and ETV4 mRNA expression levels in a series of 456 primary unilateral invasive primary breast tumors from patients with known clinical and pathological status and long-term outcome. We used a log-rank test to identify relationships between MFS and MMP13 and/or ETV4 expression. Tumors with the highest levels of MMP13 mRNA (n = 135 [29.6%]) were significantly associated with poor MFS (P = 0.00016), which was not the case for ETV4-expressing tumors (Additional file 9: Figure S7a and b). This result was confirmed in the NKI breast cancer cohort (Additional file 10: Figure S8b and c). Combined analysis (as described in the “Patients and samples for MMP13 and ETV4 expression” subsection of the Methods section above) of MMP13 and ETV4 mRNA expression levels defined three separate prognostic groups of 82 (Low-ETV4), 255 (High-ETV4/Low-MMP13), and 119 (High-ETV4/High-MMP13) patients with significantly different survival (P = 0.000041) (Fig. 7). The patients with the poorest prognosis were observed in the subgroup of 119 of 456 (26.1%) patients characterized by association of high MMP13 and high ETV4 mRNA expression levels. These data were also confirmed in the NKI breast cancer cohort (P = 0.0013) (Additional file 10: Figure S8a). Multivariate analysis using a Cox proportional hazards model was performed to assess the prognostic value for MFS of the parameters found to be significant in univariate analysis (i.e., SBR histological grade, lymph node status, macroscopic tumor size, PR status [Additional file 2: Table S1] and combined MMP13 and ETV4 mRNA levels). The prognostic significance of the lymph node status (P = 0.000016), macroscopic tumor size (P = 0.0028), and combined MMP13 and ETV4 mRNA level was maintained (Additional file 11: Table S2).

We sought links between the three prognostic groups and classical clinicopathological parameters in breast cancer (Table 1). Using HR (ERα and PR) and ERBB2 status, we also subdivided the total population (n = 456) into four breast cancer molecular subtypes: HR+/ERBB2+ (n = 54), HR+/ERBB2− (n = 289), HR−/ERBB2+ (n = 45) and HR−/ERBB2− (n = 68). High MMP13 and ETV4 mRNA expression levels were associated with negative ER status (P = 0.00067) and the HR−/ERBB2+ subtype (P = 0.0015), two parameters associated with breast cancer aggressiveness (Table 1). We did not observe a correlation between the three prognostic groups and mutations of PIK3CA, which is the most frequently mutated oncogene in breast cancer (P = 0.96), as well as mRNA level of the MKI67 gene, which encodes for the proliferation-related Ki-67 antigen (P = 0.073).