In this report, we identified the regulation of MMP1/9 and cathepsin D by c-Myb as a novel mechanism of the matrix-specific breast cancer cell invasion. c-Myb has been recently identified as a regulator of tumor cell motility and invasion, and the Slug transcription factor was described as the mediator of the c-Myb-induced mesenchymal-like phenotype in neuroblastoma, colon carcinoma, and embryonic kidney cells . The TGFβ-induced EMT and invasion of estrogen receptor-positive breast cancer cells was also found to be dependent on c-Myb expression . In our study, we confirmed the regulatory role of c-Myb in control of migration and invasion of breast carcinoma cells as well. We also observed upregulation of Slug in the MYBup variants of MDA-MB-231 cells, but we did not detect any c-Myb-dependent changes in expression of either epithelial or mesenchymal cell markers, such as E-cadherin, N-cadherin, or vimentin (Additional file 5). This may presumably result from the mesenchymal-like phenotype of the parental MDA-MB-231 cells. We demonstrated that c-Myb activates transcription of cathepsin D in a Slug-independent manner (Additional file 6). This implies that EMT is not an exclusive mode of c-Myb control over cell migration/invasion. In hepatocellular tumor cells, c-Myb stimulated migration/invasion via osteopontin secretion . Activation of the osteopontin promoter by c-Myb was also reported in melanomas . However, expression of the osteopontin gene was not deregulated in the c-myb-overexpressing MDA-MB-231 cells (data not shown) implying that c-Myb uses another mechanism to control migration/invasion of breast cancer cells.
There are two main types of ECM in vertebrates: BM and the stromal/interstitial matrix. Components of the BMs include collagen IV, laminin, perlecan, and nidogen. Stromal/interstitial matrices that form the majority of the body connective tissues are composed primarily of fibrillar collagen I. As both BM and stromal matrices represent steric barriers to cell migration, matrix remodeling is a critical prerequisite for metastasis formation. BM extract/Matrigel is frequently used to assess cell invasive capacity in vitro. Matrigel invasion in vitro is considered to simulate the penetration of BMs underlying epithelial cells or blood vessels by tumor cells in vivo. We documented that c-Myb induces the invasion of MDA-MB-231 cells through Matrigel. Conversely, cancer cell invasion through stromal/interstitial matrices composed primarily of fibrillar collagen I was found to be critical for metastasis in several tumor types [22, 23]. We found that c-Myb does not activate cell invasion through the collagen I barrier. Modulation of cell invasion by ECM components was described previously [30–32]. Different invasion modes reflecting the specificity of substrates may be attributed to the different requirements on proteolytic systems [22, 33]. Therefore, we analyzed the expression of candidate proteases in c-myb-overexpressing and control MDA-MB-231 cells and confirmed the differential expression of MMP1, MMP9, and cathepsin D. We confirmed the recent observation by Bhattarai that the c-Myb upregulates MMP9 in breast cancer cells . Our demonstration that c-Myb modulates MMP1 has not been described yet. While cathepsin D and MMP9 were upregulated, the expression of MMP1 was considerably reduced in the MYBup variants. c-Myb-induced Matrigel invasion is sensitive to the broad spectrum MMP inhibitor GM6001 (Ilomastat; Additional file 3). As MMP9 is the only MMP found to be upregulated in MYBup cells and collagen IV, the major component of the BM/Matrigel, is the main substrate for MMP9, we hypothesize that apart from cathepsin D, the MMP9 is an effector of the c-Myb-induced Matrigel invasion. MYBup cells exhibiting high MMP9 activity could transverse the Matrigel matrices more efficiently than controls, even in the absence of MMP1. MMP1 (interstitial collagenase) predominantly cleaves collagen I; therefore, the lack of MMP1 in MDA-MB-231MYBup cells might compromise their penetration through the collagen I matrix. Cathepsin D might secure partial invasion of these cells to collagen I substrates , as shown by reduced collagen I penetration of MDA-MB-231MYBup cells with silenced cathepsin D expression (Figure 5).
We propose cathepsin D as a novel downstream mediator of the c-Myb migration/invasion-promoting function. Cathepsin D was suggested as one of the c-Myb-target genes in MCF7 breast cancer cells previously . Our results confirmed this observation in MDA-MB-231 cells. We identified multiple putative Myb-binding sites in the human cathepsin D gene promoter using TESS software (Additional file 7). There are conflicting results regarding participation of the cathepsin D in the control of breast cancer cell invasion and metastasis [37–44]. Johnson et al. observed that breast cancer cell invasion did not reflect various release rates of cathepsin D from different subclones of MCF7 cells . Similarly, Glondu et al. modulated cathepsin D expression using antisense inhibition without any effect on invasiveness of breast cancer cells in vitro . In contrast, suppression of cathepsin D by antisense oligonucleotides and shRNA in MCF7 and MDA-MB-231 cells, respectively, associated with reduction of their invasion through Matrigel was observed by others [43, 44]. The effects of siRNA-mediated silencing of cathepsin D in MDA-MB-231MYBup cells described in our study provide further support to the studies documenting the regulatory function of cathepsin D in breast cancer cell migration and invasiveness. Several hypotheses have been raised concerning the mechanism how cathepsin D exerts its effects in tumors. They include facilitated release of growth factors, degradation of the extracellular matrix to permit invasion of the tumor cells and proteolytic activity-independent stimulation of the tumor cells via protein-binding activity of cathepsin D. We observed that unlike siRNA-mediated suppression of cathepsin D expression, inhibition of its activity using pepstatin A blocked neither migration nor invasion of MDA-MB-231MYBup cells (data not shown). This implies that it is the protein-binding activity of cathepsin D that may be involved in stimulation of the tumor cells. Our results correspond with studies documenting that catalytically inactive mutants of cathepsin D stimulate cell invasion [45, 46]. Recently, the LRP1 cell surface receptor was identified as a binding partner for pro-cathepsin D in fibroblasts . Breast cancer cells express LRP1  and the LRP-induced stimulation of cancer cell motility and invasion was described elsewhere [49, 50]. Therefore, we can hypothesize that cathepsin D enhances migration/invasion of breast cancer cells via LRP1 signaling.
Despite the upregulation of MMP9 mRNA/protein observed in the cells overexpressing c-myb, transient transfection studies demonstrated no transactivation of the reporter gene derived from the MMP9 promoter by c-Myb. Similarly, c-Myb did not repress transcription from the MMP1 promoter, although the level of MMP1 decreased in the MYBup cells. c-Myb was demonstrated to act as a transcriptional transactivator/repressor through specific binding to the promoter regions of target genes in numerous studies. However, there are also reports demonstrating that c-Myb can affect gene expression via indirect mechanisms [51, 52]. We studied the stability of the MMP9/1 transcripts using actinomycin D and found that c-Myb does not affect the stability of MMP9/1 mRNAs (data not shown). We hypothesize that structural and functional differences between transiently transfected plasmid DNA and genomic templates, such as inefficient chromatinization, might explain why c-Myb failed to transactivate/repress the MMP9/1 promoters . There are reports indicating that the Myb-induced chromatin binding and remodeling are essential for the transactivation of its target genes [54, 55].
We demonstrated that murine c-Myb regulates migration/invasion of mouse 4T1 cells in vitro. 4T1 cells were employed as orthotopic mammary tumor model because they effectively metastasize and display metastatic characteristics similar to those observed in cancer patients . Surprisingly, the c-myb-overexpressing 4T1 cells injected into the mammary fat pads of BALB/c mice exhibited delayed tumor growth and no formation of spontaneous pulmonary metastases. Previous studies described that the myb genes can function either as oncogenes or as tumor suppressors in different cellular contexts  and there are conflicting results concerning the c-myb function in breast cancer. Oncogenic role of c-Myb was documented by c-myb knockdown in estrogen receptor (ER)-positive breast cancer cell lines resulting in block of estrogen-dependent proliferation  and TGFβ-induced invasiveness in vitro . Miao et al. have recently published data documenting that established human breast cancer xenografts do not advance when c-Myb is knocked down using shRNA . The tissue-specific deletion of c-myb also interferes with mammary tumorigenesis in mouse mammary tumor virus (MMTV)-NEU and MMTV-PyMT mice . On the other hand, there is also report that c-myb depletion increases the cell growth and tumorigenesis of MCF7 breast cancer cells both in vitro and in vivo  documenting that c-Myb can also act as tumor suppressor. The data based on human breast cancer microarray expression analysis in vivo showed that high c-Myb expression is associated with a good outcome and high differentiation status of the tumors . Similarly, a unique subgroup of estrogen receptor-positive human breast cancers with 100% overall survival, no metastatic potential, and high c-myb expression has been described recently . Our study of ectopic c-myb overexpression in a mouse orthotopic tumor model supports the view of c-Myb as tumor-suppressor in breast cancer. The c-Myb-controlled expression of Hep27 gene was suggested as a mechanism how c-Myb exerts its tumor-suppressing function in ER-positive breast cancer [17, 60]. Hep27 inhibits Mdm2 thereby stabilizes p53 . Therefore, the c-Myb-Hep27-Mdm2-p53 signaling pathway may have functional significance for ER- and p53wt-positive breast cancers cells . Interestingly, good prognosis of patients with ER-negative basal-like subtype of breast tumors with frequently mutated p53 was also associated with high c-myb expression . To date, the function of c-Myb in metastasis formation in mouse models has not been clarified. It was demonstrated, however, that c-Myb promotes the bone marrow homing of leukemic cells . Our study showed that overexpressed c-Myb can suppress the formation of pulmonary metastases in a mouse model of mammary carcinoma. We propose that interstitial collagenase may be one of the mediators of the tumor- and metastasis-suppressing function of c-Myb in ER-negative breast cancer cells because interstitial collagenase was clearly downregulated in MDA-MB-231MYBup cells (Figure 4), 4T1MYBup cells (Additional file 8) and in the c-myb- overexpressing tumors (Additional file 8). A number of observations support this hypothesis. Upregulation of MMP1 was previously associated with advanced stages of breast cancer, and it was thus suggested as a predictive marker for the development of invasive disease . The shRNA-mediated silencing of MMP1 inhibited growth of MDA-MB-231 cells orthotopically implanted into the mammary fat pads of nude mice . The requirement for MMP1 for breast tumor growth was partly attributed to breakdown of the fibrous stroma within the mammary fat pad and partly to the liberation of growth factors present in ECM . In addition, the MMP1/protease-activated receptor 1 signaling axis promoting mammary tumor growth and metastasis was identified [64–66]. In addition, Gupta et al. reported that MMP1 silencing in combination with epiregulin (EREG), cyclooxygenase 2 (COX2), or MMP2 delayed tumor progression . Interestingly, MMP1, EREG, and COX2 are parts of the clinically validated lung metastasis signature . This signature comprises genes marking and mediating breast cancer metastasis to the lungs. MMP1 belongs to the family of genes with dual functions conferring both breast tumorigenicity and lung metastagenicity [68, 69].
Moreover, MMP1 has been implicated in pulmonary extravasation [67, 70]. The barriers to metastasis are distinct in different organs [71, 72]. The lung vascular endothelial junctions act as barriers limiting cell passage. In contrast, the bone marrow and liver vasculature consists of capillary vascular channels possessing a discontinuous endothelium [71, 73]. Therefore, lung metastases may require robust extravasation such as those provided by MMP1, EREG, COX2, and MMP2 [67, 71]. In contrast, the requirement for effective extravasation is not principal for bone and liver metastasis [71, 73, 74]. These findings correspond with our results revealing organ-specific differences in the metastatic ability of c-myb-overexpressing cells. The deficit of MMP1 function in the c-myb-overexpressing cells may contribute to the selective disadvantage of these cells in lung colonization. However, they can still initiate metastasis in bone and liver.
We demonstrated that enhanced migration and invasion of c-myb-overexpressing breast cancer cells through Matrigel in vitro does not imply increased metastatic capacity. Inconsistency in proinvasive behavior of tumor cells in vitro and metastatic potential in mouse xenograft models in vivo was also described previously [75, 76]. It was postulated that the capacity of tumor cells to metastasize is determined not only by the inherent characteristics of the cancer cells, such as motility and in vitro invasiveness, but it is also modulated by the cancer cell microenvironment, such as ECM deposition, the presence of other proteases, cytokines, growth factors and adaptor proteins [75–77]. Apparently, there are tissue-specific factors in the host that participate in the control of cancer cell invasiveness .