Stromelysin-3 over-expression enhances tumourigenesis in MCF-7 and MDA-MB-231 breast cancer cell lines: involvement of the IGF-1 signalling pathway

Background Stromelysin-3 (ST-3) is over-expressed in the majority of human carcinomas including breast carcinoma. Due to its known effect in promoting tumour formation, but its impeding effect on metastasis, a dual role of ST-3 in tumour progression, depending on the cellular grade of dedifferentiation, was hypothesized. Methods The present study was designed to investigate the influence of ST-3 in vivo and in vitro on the oestrogen-dependent, non-invasive MCF-7 breast carcinoma cell line as well as on the oestrogen-independent, invasive MDA-MB-231 breast carcinoma cell line. Therefore an orthotopic human xenograft tumour model in nude mice, as well as a 3D matrigel cell culture system, were employed. Results Using both in vitro and in vivo techniques, we have demonstrated that over-expression of ST-3 in MCF-7 and MDA-MB-231 cells leads to both increased cell numbers and tumour volumes. This observation was dependent upon the presence of growth factors. In particular, the enhanced proliferative capacity was in MCF-7/ST-3 completely and in MDA-MB-231/ST-3 cells partially dependent on the IGF-1 signalling pathway. Microarray analysis of ST-3 over-expressing cells revealed that in addition to cell proliferation, further biological processes seemed to be affected, such as cell motility and stress response. The MAPK-pathway as well as the Wnt and PI3-kinase pathways, appear to also play a potential role. Furthermore, we have demonstrated that breast cancer cell lines of different differentiation status, as well as the non-tumourigenic cell line MCF-10A, have a comparable capability to induce endogenous ST-3 expression in fibroblasts. Conclusion These data reveal that ST-3 is capable of enhancing tumourigenesis in highly differentiated "early stage" breast cancer cell lines as well as in further progressed breast cancer cell lines that have already undergone epithelial-mesenchymal transition. We propose that ST-3 induction in tumour fibroblasts leads to the stimulation of the IGF-1R pathway in carcinoma cells, thus enhancing their proliferative capacity. In addition, further different cellular processes seem to be activated by ST-3, possibly accounting for the dual role of ST-3 in tumour progression and metastasis.


Background
The human protein family of matrix metalloproteinases (MMPs) includes at least 25 members that are believed to play a role in the various stages of tumourigenesis [1,2]. Most of these zinc-dependent endopeptidases cleave more than one extracellular matrix (ECM) component and each ECM component is generally cleaved by more than one MMP. Aside from matrix degradation, MMPs play an important role in releasing growth factors, such as TGF-β or VEGF, or other biologically active factors [3,4]. The effect of over-expression or inhibition of MMPs in tumours might be complex, as these proteases are able to promote and impede tumour relevant processes such as proliferation, apoptosis, angiogenesis and metastasis [5].
Stromelysin-3 (ST-3, MMP-11) was originally identified as a breast cancer associated gene, which is over-expressed in more than 90% of invasive breast carcinomas [6]. Recent studies have shown that ST-3 over-expression is found in most human carcinomas [7]. Expression of the gene has also been detected in fibroblasts at the invasion front of tumours in close proximity to carcinoma cells [6]. Furthermore, it has been shown that ST-3 expression is induced in fibroblasts by co-culturing with breast carcinoma cells [8,9]. Although ST-3 was discovered more than a decade ago and assigned to the family of MMPs due to sequence homologies, no ECM component has yet been identified that is cleaved by full length human ST-3 [10]. However, ST-3 has been found to cleave IGFBP-1, thereby releasing IGF-1 [11]. IGF-1 is known to stimulate proliferation, enhancing survival and migration of cancer cells [12]. Despite a limited knowledge of ST-3 substrates and mode of action, an in vitro study has demonstrated that ST-3 over-expression in MCF-7 cells is capable of increasing cell survival under serum starvation conditions [13]. However, the mechanism of ST-3 activation during tumourigenesis remains unclear. Various in vivo studies reveal a ST-3 specific function in tumourigenesis e.g. ST-3 enhanced tumour take of MCF-7 cells in nude mice when injected subcutaneously [14]. Other studies, performed in an ST-3 knock out mouse model and in human xenograft models of ST-3 over-expressing MCF-7 cells demonstrate that the activation of ST-3 leads to a lower percentage of apoptotic cells in the induced tumours [15,16]. According to these results, the generation of MMTV-ras transgenic mice in an ST-3 deficient genotype results in the development of less and smaller primary tumours [17]. Interestingly, these mice developed more lung metastases. The question has therefore arisen, whether ST-3 over-expression may cause different functional effects in different stages of tumourigenesis and thereby play a dual role [17,18], or whether compensation by other MMP family members could account for the observed effect. Indeed, it has been shown that during involution, ST-1 and ST-2 is up-regulated in matrilysin knock out mice, and that mat-rilysin and ST-2 expression is elevated in ST-1 deficient mice [19].
In the present study, the influence of ST-3 on cell lines of different stages of breast cancer progression is investigated. To this end, the induction of endogenous ST-3 expression and the effect of ST-3 over-expression on the non-invasive cell line MCF-7 as well as on the highly invasive and dedifferentiated cell line MDA-MB-231 is analysed both in vitro and in vivo. Furthermore, the impact of ST-3 over-expression on the gene expression pattern of breast cancer cells was determined.

Immunocytochemistry/Immunohistochemistry
Immunocytochemistry was carried out as described previously [20]. Briefly, cells grown on cover slips were washed with PBS, fixed with 4% PFA and permeabilized in 0.2% Triton-X100/PBS. Cover slips were incubated with the primary and, after washing, with the secondary antibodies for 20 min at RT. For immunohistochemical staining, paraffin sections were dewaxed and dehydrated by graded ethanol. After blocking, sections were incubated with the primary and secondary antibodies. As primary antibodies goat α-(human vimentin) (1:30, Santa Cruz, Germany), rabbit α-(human cytokeratin) (1:100, Novocastra, Newcastle) and mouse α-(human ST-3) (1:50, Clone SL3.05, Neomarkers, CA) were used. Secondary antibodies were donkey-α-(mouse IgG) Rhodamin Red (1:100, Dianova, Germany) and donkey α-(goat IgG)Cy5 (1:400, Dianova, Germany). Stainings were performed as triple stainings on the same section. Samples were examined using a Leica TCS SL confocal microscope (Leica, Germany). ST-3 cDNA was amplified by performing a PCR and cloning  into  the  mammalian  expression  vector  pcDNA3.1V5HisTOPO (Invitrogen, Karlsruhe, Germany). Expression of ST-3 protein was confirmed by in vitro transcription/translation in reticulocyte lysate (Promega). The ST-3 or empty expression plasmid was transfected into MCF-7 and MDA-MB-231 cells. 6 μl Fugene (Boehringer, Mannheim, Germany) was mixed with 196 μl Optimem (Gibco, Karlsruhe, Germany) and incubated for 5 min at room temperature. After supplementation with 2 μg DNA and a further incubation period of 15 min, this mixture was applied to a 60% confluent 6-well plate. On the following day, transfected cells were seeded at a density of 4000 cells/cm 2 and selected by supplementation of media with G418 (750 μg/ml for MDA-MB-231 and 600 μg/ml for MCF-7). Transfection efficiencies were determined using a β-Gal plasmid and were in the range of 10-20%. Three independently derived and randomly chosen cell clones were isolated.

Western blotting
Cell culture supernatant containing 0.1% FCS was collected after 48 hours of culture and concentrated 20 times by Biomax 5 k filter (Fa. Millipore, MA). Western blotting was carried out according to manufacturer's instructions utilizing the Novex gel system (Invitrogen, Germany) and 4-12% gradient SDS-PAGE. 12 μl of the concentrated supernatants were applied and equal protein loading was confirmed by Ponceau staining after transfer onto nitrocellulose membranes. Immunoreactive proteins were detected by secondary antibody binding (donkey α-(mouse IgG)) conjugated to horseradish peroxidase, (1:50.000, Amersham Bioscience, Germany) followed by incubation with ECL-Plus (0.1 ml/cm 2 , Amersham Biosciences, Germany). The primary antibody, mouse anti-ST-3 (Clone SL3.05, Neomarkers, CA) was used at a dilution of 1:300.

Proliferation assays
Cells of either three ST-3 or three mock clones were detached by trypsin, counted and mixed in equal parts. For each well, a total number of 4.5 × 10 4 cells (i.e. 1.5 × 10 4 cells per clone) were mixed with 75 μl matrigel and applied to matrigel coated 96-well plates. Gels were allowed to solidify for 30 min at 37°C before supplementation with 100 μl of cell culture media. After 7 days of 3D culture, cell numbers were determined by means of a MTS assay (Fa. Promega, Germany). The applicability of the MTS test was determined beforehand by confirming the linearity of measurement of known cell numbers in 3D cultures (data not shown). All assays were conducted at least 3 times with triplicates.
In vivo tumourigenesis assay [8][9][10][11][12] week-old female immunocompromised NMRI nude/nude mice were obtained from Harlan Winkelmann (Borchem, Germany). 2 × 10 6 cells consisting of equal parts of either three ST-3 or mock clones were mixed with 50 μl matrigel and injected into the mammary fat pad. Growth factor depleted matrigel was prepared as described by Taub et al [21]. Tumour take and tumour growth was determined by the measurement of tumour burdens estimated from calliper measurements twice a week. The minimum tumour size was defined as 3 × 3 mm. Animals were sacrificed when tumours were larger than 1000 mm 3 . Animals receiving MCF-7 tumour cells were supplemented with 17-beta Estradiol pellets (IRA, FI) that were implanted subcutaneously one week in advance. Animals were sacrificed according to the guidelines of the "Deutsches Tierschutzgesetz".

Microarray gene expression analysis
Cells were grown for 3 days in 3D matrigel cultures as described above. RNA-isolation was carried out using Tri-Fast (peqLab, Erlangen, Germany). The integrity of resuspended total RNA was measured on a Bioanalyzer gel (Agilent Technologies). For the microarray experiment, 500 ng of total RNA was reverse transcribed with an oligo-(dT)-T7 promoter primer and Moloney murine leukemia virus-reverse transcriptase (MMLV-RT) to synthesize first and second strands of cDNA. Fluorescent antisense cRNA was synthesized with simultaneous incorporation of either cyanine 3-cytidine 5'-triphosphate (3-CTP) or cyanine 5-CTP. The labelling efficiency was verified with a Nanodrop photometer (Kisker). Before hybridisation, 2 μg of each labelled cRNA product were fragmented and mixed with control targets and hybridisation buffer according to the supplier's protocol (Agilent Technologies). Hybridisations were done overnight for ~19 h at 60°C and arrays were subsequently washed. The scanning resolution was 5-μm using a DNA microarray laser scanner (Agilent). Experiments were then repeated, performing a colour swap dye reversal. Features were extracted with an image analysis tool (Agilent Technologies, version A 6.1.1) using default settings. Data analysis was conducted on the Rosetta Inpharmatics Platform Resolver Build version 4.0. Expression patterns were identified by stringent data analysis using anticorrelation of the dye reversal ratio profiles and a 2-fold expression cut-off. Combining the first and the second criteria of analysis, data points were filtered out with a low p value (p < 0.01). To conduct functional categorizing, all genes with a minimum of two-fold absolute changes were submitted to the Database for Annotation, Visualization and Integrated Discovery (DAVID [22]). Pathway mapping analysis for these genes was done by BioRag (Bioresource for array genes [23]).

Statistical analysis
For all statistical analyses, a significance level of p ≤ 0.05 was employed. To test differences in means of cell numbers and tumour volumes, a two sided T-test was used. Differences of tumour take curves were evaluated by means of a log rank test. To conduct these statistical analyses, the software package SPSS 12.0 was employed.

Induction of ST-3 expression in co-cultures of breast epithelial cells and fibroblasts
The expression of ST-3 in fibroblasts of breast cancer tissue was confirmed by co-staining with marker proteins for fibroblasts and epithelial cells ( Figure 1A

Characterisation of ST-3 over-expressing clones
To investigate the functional consequences of such ST-3 over-expression, the non-invasive cell line MCF-7, as well as the invasive cell line MDA-MB-231, were stable transfected with ST-3. FISH analysis revealed a single copy integration of each of the ST-3 and mock transfected clones, as well as their clonality (data not shown). Each separate clone displayed a different site of integration. Western blot analysis proved that the over-expressed ST-3 protein displayed similar molecular weight patterns in MCF-7 and MDA-MB-231 clones, corresponding to the published molecular weight of ST-3 gene products (figure 2) [24].

Functional analysis in vivo
The tumourigenic potency of transfected cells was investigated in an orthotopic xenograft model in nude mice. As demonstrated in Figure 3

Analysis of ST-3 influence on gene expression pattern
To investigate potential genes and pathways involved in an ST-3 response apart from IGF-1R signalling, transcriptional profiling of ST-3 over-expressing MDA-MB-231 cells in comparison to mock transfected cells was employed. Of the total of 1190 significantly regulated genes the overall number of genes induced by ST-3 was larger than the number of genes that showed a down regulation (911 versus 279). In all, 708 differentially expressed genes could be clustered according to their involvement in biological processes by gene ontology analysis resulting in 15 functional categories ( Figure 6). Fourteen of these categories, including signal transduction, cell proliferation, cell motility, and regulators of cell metabolism revealed significant changes in expression (p < 0.05, category "response to extracellular stimulus": p = 0.06; Figure 6). Examining the expression of MMP genes revealed an up-regulation of MMP-28 in ST-3 overexpressing cells (NM_032950, 2.9 fold). Up-regulation of the ST-3 gene was not been detected at the mRNA level. Since the enhanced protein production was clearly shown by Western blotting (Figure 2), this was probably due to technical reasons. Pathway mapping analyses showed the following signalling pathways that were mainly affected: the MAPK-signalling pathway, pathways involved in cell cycle progression, Wnt-pathway, and the phosphatidylinositol signalling system ( Table 1). The expression of small GTPases of the Rho family, RhoC and cdc42 were all up-regulated (2.4 and 5.7 fold) by ST-3 as well as the adapter protein SHC3 (2.2 fold), which has been shown to be involved in PI3-kinase activation promoting cell sur-vival [25]. Transcripts coding for cell cycle regulators required for G1/S transition, such as cyclin dependent kinase 6 and 4 (CDK6, CDK4), CDK1 (also known as cell division cycle 2, CDC2) and CyclinB (CCNB1) were upregulated, indicating a positive role for ST-3 during cell cycle progression. A number of components of the MAPK signalling pathway were also up-regulated (MAP4K2, MAP4K1, MAP2K4) as well as PDGFRB (but not the IGF-R1), which activates the MAPK-and the Akt pathways indicating a role for growth factor mediated cellular response. However, the MAPK pathway downstream transcription factor AP-1 subunit jun (here v-jun sarcoma virus 17 oncogene homologue, AC: BC002646) and MAP3K2 were down regulated whereas a negative regulator of cell cycle p27KIP1 (CDKN1B, cyclin dependent kinase inhibitor1B) was up-regulated. ST-3 was also found to up-regulate PKN2 (protein kinaseN2, also known as PRK2) and the PI3-kinase catalytic subunit p110 (PIK3CA). The latter is known to promote cell proliferation via downstream activation of p70S6-kinase [26]. PI3-kinase also leads to cell motility (via Rac1 activation) and cell survival (via Akt activation). Furthermore, the anti-apoptotic Bcl-2, which was up-regulated 2.4 fold, points towards a role for ST-3 in cell survival processes. PKN2 was recently shown to be part of a ternary Rho/ PDK1 protein complex enhancing Akt activation in a PI3kinase dependent manner [27]. Components of the Wnt  signalling pathway, such as FZD7 (frizzled homologue7) and its downstream target DVL3 (dishevelled, dsh) were up-regulated. These genes are known to be involved in tumour cell proliferation and invasion via activation of the transcription factor TCF/LEF1. However, CTBP-1, an inhibitor of this transcription factor, was up-regulated. These results indicate that ST-3 activation induces synthesis of intracellular signalling components characteristic for downstream activation in response to growth factor receptor activation, promoting cell cycle progression and migration. Interestingly, no transcripts for soluble growth factors (e.g. IGF) were induced by ST-3 over-expression.

Discussion
It has been shown in the present study that ST-3 induction by an immortalized, non-tumourigenic cell line is similar to the induction by tumourigenic breast cancer cell lines. The stimulation of ST-3 expression in breast cancer fibroblasts by different breast tumour cell lines was recently described on mRNA and protein level and is in line with our results [8,9]. Cell-cell contacts seem to be beneficial, but not essential to induce ST-3 expression, since ST-3 was induced in direct co-cultures and to a lower extent in indirect co-cultures [8]. PKC seems to be involved in the induction of ST-3 expression in normal pulmonary fibroblasts by non-small cell lung cancer cells [28]. The observation that non-tumourigenic cells are also able to induce ST-3 expression might indicate that the association of ST-3 expression and invasive properties of carcinomas is a consequence of invasive epithelial cells coming into contact with fibroblasts rather than being the cause of invasion. The observation, both by our group as well as others, that ST-3 over-expression does not lead to an enhanced invasiveness of cells (data not shown and [14]) provides further strength to this hypothesis. Normal as well as tumor derived breast epithelial cells expressed low amounts of ST-3. An epithelial expression of ST-3 was already described in gastric and pancreatic carcinoma [29,30] and seems to also occur in breast tissue.
ST-3 significantly enhanced tumour take as well as tumour growth rates. Thus, the results indicate that ST-3 is not only involved in tumour formation, as it was postulated by previous studies [14], but additionally in tumour progression. This was true for MCF-7, the breast cancer cell line also used in former in vivo experiments [14,31], and for MDA-MB-231 cells. MCF-7 is a rather well differentiated tumour cell line, reflected by its oestrogendependent growth. This cell line is non-invasive with characteristics of epithelial cells (e.g. E-cadherin expression and a lack of ST-3 expression). In contrast, MDA-MB-231 is a tumour cell line that has undergone epithelialmesenchymal transition and thus acquired fibroblastoid characteristics (e.g. morphology, invasiveness, lack of E-Cadherin and expression of low amounts of ST-3). MDA-MB-231 cell growth is independent of the presence of oestrogen. Thus, ST-3 seems to have similar effects on breast cancer cells derived from more differentiated tumours and on cells from more aggressive and dedifferentiated lesions.

ST-3 induced increase of tumour volume is dependent on the presence of growth factors in vivo
The described in vitro results indicate that ST-3 overexpression leads to an accelerated cell proliferation. Indeed, ST-3 protein expression has been found to be upregulated in human invasive breast carcinomas and in coexpression to clinically-used proliferation markers [32]. Since ST-3 has been shown to be capable of releasing IGF-1 bound to IGFBP-1, thereby stimulating proliferation of BAF/3 cells [11], this signalling pathway was investigated in the present study. The IGF-1 signalling cascade seemed to be the only or predominant pathway responsible for the ST-3 effect in the MCF-7 cell line, possibly via the ERK/ MAPK pathway. It was described that ST-3 over-expression in MCF-7 cells results in the activation of p42/p44 MAP kinase [13]. In contrast, additional pathways seem to be involved in MDA-MB-231 cells. In this respect, it is notable that there seems to be cross-talk between IGF-1 and the estrogen receptor [33].

ST-3 over-expression results in an increased cell number in vitro
Transcriptional profiling revealed that cancer related processes such as cell proliferation, cell motility and cell metabolism are affected by ST-3 over-expression. Most notably, cell proliferation and cell cycle progression seem to be influenced, an observation that has been confirmed by our in vitro model. ST-3 leads to an up-regulation of the small GTPases of the Rho family, RhoC and cdc42, which have an important role in cancer progression, including proliferation, invasion, and metastasis of breast tumours [34]. Cdc42 and RhoC are induced in response to integrin activation suggesting a role for ST-3 in mediating cellmatrix adhesion, a prerequisite for cell migration. Rho GTPases also play a role in the induction of MMP expression [35][36][37] and, conversely, blocking Rho function can inhibit transcription of MMPs [38]. Whether this is also true for ST-3 activated Rho GTPases still needs to be proven. Interestingly, MMP-28 expression was induced in ST-3 overexpressing cells. This MMP has been described to be involved in EMT in lung carcinoma cells [39] and could additionally contribute to the observed effect.
However, no induction of gene expression of growth factors could be detected in the presence of ST-3, suggesting that ST-3 might not be involved in de novo growth factor synthesis, which is crucial for autocrine activation of signalling pathways as a characteristic feature for tumour cells [40]. Moreover, ST-3 seems to be involved in the release and activation of soluble growth factors bound by ECM components. This hypothesis is also supported by our in vivo data and the finding that ST-3 can activate extracellular IGF bound to IGFBP-1, which strengthened its definition as a MMP [11]. Thus, our results suggest that ST-3 enhances rather than initiates tumourigenesis by either activating ECM-bound signaling components or at least by synergistic effects with ECM containing growth factors such as IGF-1. Our gene expression data also support a role in ST-3 activation of Wnt-signalling components, such as FZD7 and DVL3, which is in line with the activation of the Wnt pathway by ST-1/MMP-3 via E-cad-  herin cleavage, resulting in EMT and thus enhanced invasive properties of mammary tumour cells [41].

Conclusion
This study shows that ST-3 expression in fibroblasts is induced not only by epithelial cells of tumour origin, but also by non-transformed cells. Results from an orthotopic in vivo model as well as from in vitro 3D assays have demonstrated that ST-3 leads to an enhanced tumourigenic potency in cell lines of different dedifferentiation status. The mechanisms by which ST-3 exerts its function seem to involve activation of IGF-1 signalling and potentially MAPK and PI3-kinase pathways. Clearly, further studies are needed to confirm and investigate the impact of ST-3 on these pathways, but our data provide some novel clues for the mechanisms of ST-3 activation in breast cancer cells.
GK: participated in study and experiment design, carried out co-cultures, Western blots and immunocytochemistry, generated stable transfected clones, established in vitro 3D assay, drafted the manuscript MR: designed, performed and analysed animal experiments MT: designed, performed and analysed microarray analysis ND: performed and analysed in vitro cell culture assays