Cancer progression by breast tumors with Pit-1-overexpression is blocked by inhibition of metalloproteinase (MMP)-13

The human breast adenocarcinoma cell lines MCF-7 and MDA-MB-231 were obtained from the European Collection of Cell Cultures (ECCC, Salisbury, UK). These cell lines were grown in 100-mm Petri dishes in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 100 U/ml penicillin, and 100 μg/ml streptomycin in an air-CO₂ (95:5) atmosphere at 37°C. Confluent cells were washed twice with phosphate-buffered saline (PBS) and harvested by a brief incubation with trypsin- ethylenediaminetetraacetic acid (EDTA) solution (Sigma-Aldrich, St. Louis, MO, USA). Geneticin (G418), culture medium, and sera were purchased from Invitrogen Life Technologies, Carlsbad, CA, USA. Immobilon-P membranes were from Millipore (Merck Millipore, Billerica, MA, USA). Mytomycin C, MTT, puromycin, and hygromycin B were from Sigma-Aldrich.

Transient transfection, Pit-1 knockdown, and stable transfection of Pit-1 into MCF-7 cells were performed as previously described [19],[20]. The Pit-1-overexpressing MCF-7 cells were then transfected with pBABE-puro-Luc vector and 48 hours later treated with 2.5 μg/μl of puromycin to select clones (MCF-7-hPit-1-luc cells). The shpLKO.1-MMP-1 and shpLKO.1-MMP-13 lentiviral vectors containing two different short hairpin RNA (shRNA) sequences for MMP-1 and MMP-13 were obtained from Thermo Fisher Scientific (Waltham, MA, USA) (see Additional file 1 and Figure S1A in Additional file 2).

The pLKO.1-puro non-target shRNA control containing a shRNA (shControl) was obtained from Sigma-Aldrich. MCF-7-hPit-1-luc-shControl, MCF-7-hPit-1-luc-shMMP-1 and MCF-7-hPit-1-luc-shMMP-13 cells were obtained through infection with shControl, shMMP-1, and shMMP-13 virus particles, respectively. Briefly, cell culture medium was replaced by a medium without FBS and containing 5 μg/ml of polybrene.

After 4 hours, culture medium was again replaced (DMEM plus 10% FBS), and lentiviral particles (Mission pLKO.1-puro non-target shRNA, Mission-MMP-1 shRNA, and Mission MMP-13 shRNA transduction particles, Sigma-Aldrich) were added and incubated at 37°C for 24 hours. Pit-1 knockdown was carried out using two different Pit-1 small interfering RNA (siRNA) (Pit-1 siRNA-1 and Pit-1 siRNA-2), as previously described (19). A scrambled siRNA was employed as control. Sequences of siRNAs are detailed in Additional file 1. The proximal promoter regions of the human MMP-1 and MMP-13 genes were synthesized by PCR and the product subcloned into the Xho I and Hind III site of the pGL2-basic plasmid (see Additional file 1). Site-directed mutagenesis was performed with the QuikChange kit from Stratagene (Agilent Technologies, Santa Clara, CA, USA). The mutagenized oligonucleotide primers were as follows (mutagenized bases on the sense strand are identified by lowercase letters): 5′-GATTGCCTAGTCT- ATgacTAGCTAATCAAG-3′ and 5′-CCAGGACCCCTGtcgaCATCTTGAATGG-3′ for MMP-1 and MMP-13, respectively. The newly constructed mutant plasmids were designated pGL2-B-MMP-1-1633/-1MUT and pGL2-B-MMP-13-145/+2MUT.

For luciferase assays, MCF-7 cells were transfected in 6-well plates containing 6 μl of jetPEI Polyplus transfection reagent (PolyPlusTransfection, Illkirch, France), 2 μg of pRc-RSV or pRSV-hPit-1, 1 μg of each reporter plasmid, and 50 ng of pRL-TK-Renilla (as transfection control) for 48 hours. The cells were lysed in buffer (100 μl lysis buffer, Promega Corporation, Madison, WI, USA) and luciferase activity was then measured in a Mithras LB 940 apparatus (Berthold Technologies, Bad Wildbad, Germany).

Total RNA was isolated from the cell lines using TRIzol (Invitrogen). cDNA was synthesized with Transcriptor First Strand cDNA Synthesis Kit (Roche Diagnostics, Basel, Switzerland), and reactions of quantitative real-time PCR were done using iQ SYBR Green Supermix (Bio-Rad Laboratories, Alcobendas, Spain) on iCycler equipment (7500 PCR Systems, Applied Biosystems, Life Technologies, Carlsbad, CA, USA). The Pit-1, MMP-1, MMP-13 and 18S samples were denatured at 94°C for 10 sec, annealed at 58, 59, 59 and 60°C, respectively, for 10 sec and extended at 72°C for 10 sec, for a total of 33, 35, 35 and 30 cycles, respectively.

The samples were quantified using Sequence Detection Software 1.4 (Applied Biosystems), with 18S as normalization control. The oligonucleotide sequences are described in Additional file 1.

Cell proliferation experiments were carried out using MTT assay. MCF-7-hPit-1-luc cells, MCF-7-hPit-1-luc-shControl cells, MCF-7-hPit-1-luc-shMMP-1 cells, or MCF-7-hPit-1-luc-shMMP-13 cells (2.5 × 10⁴ cells/ml) were seeded in a volume of 0.5 ml in 24-well tissue culture plates. The absorbance of the samples was recorded 48 hours after transfection at 590 nm in a multiwell plate reader (LB 940 Mithras, Berthold Technologies). Results were plotted as the mean ± SD values of quadruplicates from at least two independent experiments.

Western blotting was carried out as described elsewhere [19]. Briefly, 60 μg of total protein was subjected to SDS-PAGE electrophoresis. Proteins were transferred to a nitrocellulose membrane, blocked, and immunolabeled overnight at 4°C with a primary antibody (detailed in Additional file 1). Then, the membrane was washed three times with PBS-Tween-20, and incubated with the appropriate secondary antibody for 1 hour. The signal was detected with the Pierce enhanced chemiluminescence (ECL) Western blotting substrate (Thermo Fisher Scientific, Rockford, IL, USA), and visualized by placing the blot in contact with standard X-ray film. Relative protein expression was quantified using the ImageJ software (National Institutes of Health, Bethesda, MD, USA) in at least three different blots, and values correspond to mean values of fold-change in relation to beta-actin values.

Chromatin immunoprecipitation (ChIP) assays were performed using the Upstate protocol as described previously [20]. Diluted soluble chromatin fractions were immunoprecipitated with 1 μg polyclonal anti-Pit-1 antibody (Santa Cruz Biotechnologies, Heidelberg, Germany), or control human immunoglobulin G (IgG) (Sigma-Aldrich). The histone-DNA crosslinks were reversed by 4-hour incubation at 65°C. PCR was used to analyze the DNA fragments from ChIP assays. Primer sequences are detailed in Additional file 1.

To perform the wound-healing assay, cells were seeded in 60 mm plates and allowed to reach confluence. Wounding was created using plastic pipette tip, and the cells were serum starved and treated with mitomycin C for 24 and 48 hours. Images were captured by an Olympus DP72 camera (Olympus, Tokyo, Japan), and the distance between the wound edges was measured. Cell invasion assay was performed in BD BioCoat matrigel invasion chambers according to the manufacturer’s instructions (BD Biosciences, San Agustin de Gualix, Spain), as previously described [19]. Uncoated porous filters (8-μm pore size) were used for estimating cell migration, and matrigel-precoated filters were used for examining cell invasion. Values for cell invasion were expressed as the mean number of cells per field over four fields per filter for triplicate experiments.

Experimental metastasis assays were done as previously described [19]. Briefly, 1 × 10⁶ MCF-7-hPit-1-luc cells (n = 9, controls), MCF-7-hPit-1-luc-shControl cells (n = 8, controls), MCF-7-hPit-1-luc-shMMP-1cells (n = 8), or MCF-7-hPit-1-luc-shMMP-13 cells (n = 8) in 0.15 ml of PBS and matrigel (50:50, BD Biosciences) were injected into the mammary fat pad. At day 24 for control mice or day 33 for MMP-1 and MMP-13 knockdown mice, orthotopic primary mammary tumors were measured (as described below), and removed under anesthesia. Seventeen days later (day 41 after cell injection for controls, and day 50 for MMP-1 and MMP-13 knockdown mice) mice were sacrificed, and lungs removed and examined for metastasis. Xenografts were visualized by luminescence at days 10 (all groups), 24 (Pit-1-overexpressed mice), and 33 (Pit-1-overexpressed and MMP-1 or MMP-13 knockdown) using the In Vivo Imaging System (IVIS, Caliper Life Sciences, Alameda, CA, USA). An intensity map was obtained using the Living Image software (Caliper Life Sciences). The software uses a color-based scale to represent the intensity of each pixel (ranging from blue for low to red for high). Lung micrometastasis was explored in paraffin sections by hematoxylin and eosin (H&E) staining and cytokeratin 7 (CK7) immunostaining. Immunohistochemistry studies were performed in an Autostainer Link 48 (Dako, Glostrup, Denmark). FLEX Ready-to-Use Primary Antibodies to CK7, CK19, and ki-67 (Dako) were used. For detection we used EnVision FLEX/HRP (Dako). The number of metastatic foci was counted in lung after staining with H&E and immunostaining with CK7, and the size of each foci was evaluated by measuring its diameter (in μm) using the Olympus DP-Soft morphometry program in an Olympus DX51 microscope. For automated ki-67 scoring the ACIS III (Automated Cellular Imaging Systems, Dako-Agilent Technologies, Carpinteria, CA, USA) was used. The ACIS III system scans the slide and is capable of differentiating positive and negative nuclei. Six representative areas were selected in each section and the system generated an average score.

One hundred and ten patients with invasive breast cancer (without distant metastasis at the time of initial diagnoses) treated at Fundación Hospital de Jove of Gijón (Spain), between 1990 and 2003, were selected based on the availability of clinical history and a minimum 5-year follow-up. The clinicopathological characteristics of patients and their tumors are shown in Table S1 in Additional file 3. Women were treated according to our institutional guidelines. The study adhered to national regulations and was approved by our regional Ethics and Investigation Committee (Comité Ético de Investigación Clínica Regional del Principado de Asturias). Breast carcinoma tissue samples were obtained at the time of surgery. Prior informed consent was obtained from patients. Routinely fixed (overnight in 10% buffered formalin), paraffin-embedded tumor samples stored in our pathology laboratories were used. Histopathologically representative tumor areas without necrosis were defined on H&E-stained sections. Serial 5-μm sections were consecutively cut with a microtome (Leica Microsystems, Barcelona, Spain) and transferred to adhesive-coated slides. Imunohistochemistry was done on these sections using a TechMate TM50 autostainer (Dako) as previously described [19]. A polyclonal anti-Pit-1 (Santa Cruz Biotechnology), monoclonal anti-MMP-1 (NeoMarkers, Fremont, CA, USA) and anti-MMP-13 (Santa Cruz Biotechnology) antibodies were used. To enhance antigen retrieval, tissue sections were treated in a PT-Link™ (Dako) at 97°C for 20 minutes, in citrate buffer pH 6.1 for MMP-1 and in Tris-EDTA buffer pH 9 for MMP-13, and then washed in PBS. Endogenous peroxidase activity was blocked by incubating the slides in peroxidase-blocking solution (Dako) for 5 minutes. The EnVision Detection Kit (Dako) was used as the staining detection system. Sections were counterstained with hematoxylin, dehydrated with ethanol, and permanently coverslipped. For each antibody preparation, the location of immunoreactivity, percentage of reactive area and intensity were determined. All cases were semiquantified for each protein-stained area. An image analysis system with the Olympus BX51 microscope and soft analysis (analySIS™, soft imaging system) were used as follows: tumor sections were stained with antibodies according to the method explained above and counterstained with hematoxylin. There were different optical thresholds for both stains. Each slide was scanned with a 400X power objective and four fields were selected per case to determine protein-reactive areas. The computer program selected and traced a line around antibody-reactive areas (higher optical threshold: red spots), with the remaining, nonstained areas (hematoxylin-stained tissue with lower optical threshold) standing out as a blue background. Area ratios of stained (red) versus nonstained (blue) were determined for all fields. To evaluate immunostaining intensity we used a numeric score from 0 to 3: 0 = no reactivity; 1 = weak reactivity; 2 = moderate reactivity; and 3 = intense reactivity. Using an Excel spreadsheet, the score of one field was obtained by multiplying the intensity score (I) by the percentage of reactivity area (PA) (total score: I × PA). In addition, for each tumor the mean score of the four fields evaluated was calculated. We also evaluated the immunohistochemical staining exclusively in cancerous cells or in stromal cells (mononuclear inflammatory cells, (MICs)- and fibroblast-like cells), and every evaluated field contained at least 10 stromal cells. We considered immunostaining to be positive when at least 10% of cells showed positivity. We distinguished stromal cells from cancer cells because the latter are larger in size, and because fibroblasts are spindle-shaped whereas mononuclear inflammatory cells are rounded. Moreover, while cancer cells are arranged forming either acinar or trabecular patterns, stromal cells are spread.

Values are expressed as mean ± standard deviation (SD). Means were compared using two-tailed Student’s t test or one-way ANOVA, with the Tukey-Kramer multiple comparison test for post hoc comparisons. After analyzing the human tumor distribution of score values by the Kolmogorov-Smirnov test, nonparametric methods were used to analyze the data. Immunostaining score values for each protein were expressed as a median (range). Correlation between score values was calculated by using the Spearman correlation test. Comparison of immunostaining values between groups was done with the Mann-Whitney or Kruskal-Wallis tests. Statistical results were corrected applying Bonferroni’s correction. P values of less than 0.05 were considered statistically significant. The PASW Statistics 18 program was used for all calculations (SPSS Inc, Chicago, IL, USA).

To evaluate the effect of Pit-1 on MMP-1 and MMP-13 mRNA expression, we carried out a real-time PCR. Pit-1 overexpression in MCF-7 cells using the pRSV-hPit-1 vector induced a significant increase in both MMP-1 (P <0.01) and MMP-13 (P <0.001) mRNA signal in relation to controls (Figure 1A). Our data indicated that Pit-1 overexpression increased MMP-1 and MMP-13 mRNA between two and three times, respectively. Given that MMP protein levels in MCF-7 cell extracts are low, MMP-1 and MMP-13 protein expression was evaluated in MDA-MB-231 cells by Western blot. Figure 1B shows increased and decreased MMP-1 and MMP-13 protein expression after Pit-1 overexpression and Pit-1 knockdown, respectively. Three different blots were quantified by densitometry, and the results are shown in Figure 1C.

Using the Transcription Element Search Software program (TESS, University of Pennsylvania, Philadelphia, PA, USA [21]), several putative Pit-1 binding sites were found in the MMP-1 and MMP-13 promoter (Figure 2A-C). To determine whether Pit-1 binds to the MMP-1 and MMP-13 genes, a ChIP assay was carried out using gene-specific primers (see Additional file 1). Specific binding of Pit-1 to the A, B, and D sites of the MMP-1 promoter was observed in pRSV-hPit-1-overexpressed MCF-7 cells, indicating that Pit-1 binds to the MMP-1 promoter gene in vivo (Figure 2B). A ChIP assay was also carried out to evaluate Pit-1 binding to the MMP-13 promoter (Figure 2D). Our data indicated specific Pit-1 binding to regions A, B, C, D, and E in the MMP-13 promoter, suggesting MMP-13 regulation by Pit-1, such as occurs with the MMP-1 gene. Transfection reporter assays (Figure 3A-B) indicate that Pit-1 transcriptionally regulates the MMP-1 and the MMP-13 gene. Specific deletions of the MMP-1 promoter demonstrate that Pit-1 binds to a region comprised between −1633/-1 bp from the start transcription site and this region is necessary for Pit-1-induced MMP-1 transcription. The functionality of the proximal Pit-1 response element was tested by mutation in the context of the pGL3B-MMP-1-1633/-1 construct (see Additional file 1). Specifically, three different mutations were introduced into the MMP-1-1633/-1 sequence at a position −104/-101 from the transcription start site (pGL3B-MMP-1-1633/-1MUT). This construct was cotransfected into MCF-7 cells with the pRc/RSV or with the pRSV-hPit-1 vector. As shown in Figure 3A, the response to Pit-1 was completely wiped out in cells transfected with the mutant construct. Similarly, we also evaluated transcriptional regulation of the MMP-13 promoter by Pit-1 using several constructs containing specific deletions of the pGL3B-MMP-13-1548/+2 construct. As observed in Figure 3B, the response to Pit-1 overexpression in MCF-7 cells gradually decreased, depending on the size of the MMP-13 promoter. Mutation of a Pit-1 binding site in the position −35/-31 bp in the pGL2-B-MMP-13-145/+2MUT construct (see Additional file 1) reduced transcriptional activation as compared with the wild-type construct (pGL2-B-MMP-13-145/+2). In summary, our data suggest that Pit-1 regulates MMP-1 and MMP-13 expression by binding to their gene promoter region.

To evaluate the effect of MMP-1 and MMP-13 knockdown on motility and invasion in MCF-7 cells with Pit-1 overexpression and in MDA-MB-231 cells with Pit-1 knockdown, we carried out wound-healing, migration, and invasion assays. As previously demonstrated, Pit-1 induces a significant increase in MCF-7 cell migration [19]. This was also observed in the present study, as demonstrated by the absence of wound after Pit-1 overexpression (Figure 4A and Figure S2 in Additional file 4). Knockdown of both MMP-1 and MMP-13 (two independent hairpins) significantly (P <0.005) reduced cell motility in Pit-1-overexpressing MCF-7 cells (Figure 4A, and Figure S2 in Additional file 4). In MDA-MB-231 cells, which have higher basal Pit-1 expression levels than MCF-7 cells, Pit-1 knockdown significantly reduced cell motility (Figure 4B and Figure S3 in Additional file 5). However, in Pit-1 knocked-down cells, neither MMP-1(1) and (2) nor MMP-13(1) and (2) knockdown significantly modified migration as compared to cells with MMP-1(1) and (2) and MMP-13(1) and (2) knockdown alone (Figure 4B, and Figure S3 in Additional file 5). We next explored whether MMP-1 and MMP-13 knockdown affected migration and invasion in Pit-1-overexpressing MCF-7 cells in the matrigel invasion assay. Figures 4C-F shows that knockdown of either MMP-1 or MMP-13 led to a significant (P <0.0001) decrease in both migration and invasion capacity of Pit-1-overexpressed MCF-7 cells, as compared with only Pit-1-overexpressed MCF-7 cells. We also studied the effect of MMP-1 and MMP-13 knockdown on three-dimensional growth. MDA-MB-231 cells were cultured in matrigel, in which the cells form spherical structures. However, reduced levels of MMP-1 or MMP-13 expression did not modify three-dimensional growth of MDA-MB-231 cells, as compared with control cells (Figure S1 B-C in Additional file 2).

It has been shown that MCF-7 cells present very low levels of metastasis [22],[23]. However, overexpression of Pit-1 in this cell line significantly increases metastasis in lung [19]. Therefore, we tested the potential to metastasize in vivo of MCF-7 cells stably transfected with Pit-1 alone or with Pit-1 and either MMP-1 or MMP-13 knockdown. SCID female mice were injected in the mammary fat pad with either MCF-7-hPit-1-luc cells (n = 9), MCF-7-hPit-1-luc-shControl cells (n = 8), MCF-7-hPit-1-luc-shMMP-1(1) cells (n = 8), or MCF-7-hPit-1-luc-shMMP-13(1) cells (n = 8). Twenty-four days after MCF-7-hPit-1-luc and MCF-7-hPit-1-luc-shControl cell injection, and 33 days after MCF-7-hPit-1-luc-shMMP-1 and -shMMP-13 injection, primary mammary tumors were excised under anesthesia in both groups of animals (Figure 5A). Our data indicated that knockdown of MMP-1 and MMP-13 reduced tumor growth in Pit-1-overexpressing mice. In fact, tumor growth at day 10 was significantly lower in mice injected with MCF-7-hPit-1-luc-shMMP-1 (P = 0.0011) and -shMMP-13 (P = 0.0013) cells as compared to controls (Figure 5B). At day 24, considerable tumor growth was observed in controls, while almost no tumor growth was observed in MMP-1 and MMP-13 knockdown mice (data not shown). Even though tumors in knockdown mice were allowed to continue growing until day 33, only very slight growth was found (Figure 5B). To evaluate why the growth of tumors in mice with Pit-1 overexpression was faster than in mice with Pit-1 overexpression plus MMP-1 or MMP-13 knockdown, we evaluated the ki-67 proliferation marker expression in tumors by immunohistochemistry. As shown in Figure 5C, tumors from mice injected with MCF-7-hPit-1-luc and MCF-7-hPit-1-luc-shControl cells had a larger proliferative area (high ki-67 expression, 83.6 ± 13.7% and 93.0 ± 3.7%, respectively) and a larger necrotic area than those injected with MCF-7-hPit-1-luc-shMMP-1 and -shMMP-13, which showed low ki-67 expression (47.2 ± 19.6% and 36.2 ± 26.3%, respectively). This suggests that the small tumor size in MMP-1 and MMP-13 knockdown mice could be due to low cell proliferation. An example of whole tumor size stained with ki-67 is shown in Figure S4A in Additional file 6. In addition, we carried out an in vitro MTT assay using MCF-7 cells to evaluate cell proliferation. Our data show a significant (P <0.001) decrease in cell proliferation at 48 hours in cells with Pit-1 overexpression and MMP-1 or MMP-13 knockdown as compared to cells with Pit-1 overexpression alone (Figure S4B in Additional file 6).

After excision of tumors, mice lived until day 41 (Pit-1 and Pit + shControl) and 50 (Pit-1 + shMMP-1 or shMMP-13) after injection of MCF-7 cells. Mice were then sacrificed and lungs were removed, fixed, paraffin-embedded, sectioned, and stained with H&E and specific antibodies for immunohistochemistry analysis. Five out of the nine mice injected with Pit-1-overexpressing cells (controls), six out of the eight mice injected with Pit-1-overexpressing cells + shControl, and four out of the eight mice injected with Pit-1-overexpressing + shMMP-1 transfected cells developed micrometastases in lung, while none of the eight mice injected with the MCF-7 cells transfected with the pTRE2-hPit-1-Luc-shMMP-13 showed micrometastases in lungs (Figure 5D). Micrometastases in lungs showed immunopositivity for CK-7, and CK-19 (Figure 5E). These antibodies react only with human and not with mouse cytokeratins. Neither size (Pit-1: 166.2 ± 174.1 μm, Pit-1 + shControl: 183.7 ± 79.2 μm, and Pit-1 + shMMP-1: 185.5 ± 132.1 μm) nor number of metastases (Table S2 in Additional file 7) were significant among groups of mice.

To further evaluate the clinical value of MMP-1, MMP-13 and Pit-1 expression in human breast tissue, these proteins were analyzed by immunohistochemistry in 110 invasive ductal carcinomas of the breast. Representative examples of Pit-1, MMP-1 and MMP-13 immunostaining in tumors are shown in Figure 6A. Pit-1 protein expression was detected mainly in the nuclei of tumor or control epithelial cells, whereas MMP-1 and MMP-13 expression had a cytoplasmic location in all positive cases. MMP-1 and MMP-13 expression was also evaluated in stromal cells (that is fibroblasts, and mononuclear inflammatory cells, MICs) and correlated with global Pit-1 expression. Although Pit-1 expression was not quantified by cell group, it was indeed observed in all three cell groups. An example of Pit-1, MMP-1, and MMP-13 expression in tumor and stromal cells is shown in Figure 6B. The percentage of positive cells for each protein was always higher than 50% in positive case for each cell type. A total of 89 tumors (80.9%) stained positively for Pit-1, with clear differences in intensity and percentage of stained cells. MMP-1 expression was detected in 105 tumors (95.4%) and MMP-13 in 84 tumors (76.4%). The median score value was 35.4 (range 0 to 194.5) for Pit-1, 119 (0 to 206.2) for MMP-1, and 42 (0 to 135.8) for MMP-13. Distribution of Pit-1, MMP-1 and MMP-13 score values are shown in Figure S5 in Additional file 8. In relation to global expression (score values) of Pit-1, MMP-1 and MMP-13, our result showed a direct correlation between Pit-1 score values and MMP-1 (r sub B = 0.242, P = 0.011) and MMP-13 (r sub B = 0.199, P = 0.041) (Figure 6C). In addition, we observed that MMP-1 and MMP-13 global expression was higher in Pit-1-positive tumors compared with Pit-1-negative tumors (median (range): MMP-1: 123.7 (0 to 206.2) vs. 60.4 (0 to 151.8), P = 0.014; MMP-13: 43.9 (0 to 135.8) vs. 0 (0 to 67.9), P = 0.034). A significant positive correlation between Pit-1 and both MMP-1 and MMP-13 was also found in human breast tumor datasets [24]-[26].

Our results also showed significant associations between proteins in terms of cellular type. Score values of Pit-1 were significantly higher in tumors with MMP-1-positive fibroblasts and MMP-13-positive fibroblasts than in tumors with MMP-1-negative fibroblasts and MMP-13-negative fibroblasts (P = 0.017 and P = 0.029, respectively) (Table 1). Likewise, score values of Pit-1 were significantly higher in tumors with MMP-1-positive cancerous cells and MMP-13-positive cancerous cells than in tumors with MMP-1- or MMP-13-negative cancerous cells (P <0.001) (Table 1).

In addition, we evaluated the potential association between score values from Pit-1, MMP-1 and MMP-13 and relapse-free survival in all patients included in the present study. The determination of optimal cutoff values for Pit-1 and MMP-1 in breast tumors was done for predicting recurrence. P values obtained for each cutoff value are plotted against the value itself (Figure S6A-B in Additional file 9). This statistical analysis showed no significant association between score values of MMP-13 and recurrence (data not shown). However, our data showed that high score values for Pit-1 and MMP-1 were significantly associated with recurrence. This analysis led us to define a score value of 22 for Pit-1 (χ 2 = 4.71, P = 0.03) and of 130 for MMP-1 (χ 2 = 6.52, P = 0.011) as optimal cutoff points. These values identified 73 (66.4%) and 40 (36.4%) patients, respectively, with a high probability of recurrence (Figure S6C-D in Additional file 9). We also determined the relapse-free survival curves for patients with breast carcinomas based on the combination of these optimal cutoff points. Both high Pit-1 and MMP-1 expression identified a patient subgroup (n = 31, 28.2%) with the highest probability of recurrence (P = 0.004) (Figure S6E in Additional file 9). The relationship between Pit-1, MMP-13 expression and prognosis was determined in MICs. Breast tumors containing MICs that were positive for Pit-1 and MMP13 (n = 24, 21.8%) had a high probability of recurrence (P = 0.010) (Figure S6F in Additional file 9).