Extracellular matrix (ECM) disruption by matrix metalloproteinases (MMPs) is one of the key events in metastasis. MMPs are regulated not only by their natural inhibitors, tissue inhibitors of MMPs (TIMPs), but also at the post-transcriptional level by microRNAs (miRNAs). One of these MMPs is MMP-2, which may be involved in prostate cancer (PCa) progression and metastasis [1, 2].

However, there is evidence that interstitial collagen may be involved in metastasis, indicating an active role for the desmoplastic reaction observed in several cancers. Increased production of several types of collagens has been reported: type II and IV collagens were observed in osteosarcoma [3], collagen type V was produced at elevated levels by fibrosarcoma cells compared with its production in normal muscle cells [4], and increased production of collagens I and III was observed in ovarian carcinoma [5]. Additionally, researchers have reported that collagen expression can facilitate neoplastic cell spreading [6].

The COL1A1 and COL3A1 genes encode the alpha-1 chains of collagen types 1 and 3, respectively, which are present in most connective tissues. Type 1 collagen is present in almost 70% of the extracellular bone matrix. Previously, Steele et al. [7] reported that a single miRNA (miR-29b) regulates MMP-2, COL1A1 and COL3A1 genes, although an assay to evaluate metastasis was not employed. Subsequently, Ru et al. showed that miR-29b overexpression in PCa cell lines limits metastasis, but this study did not focus on collagen genes or MMP-2 and finally Yan et al. [8] employed only LnCaP cells to report that miR-29b upregulation inhibits metastasis and that MMP-2 was not involved in this issue. Therefore, the debate about the relationship between MMP-2, miR-29b, collagen genes and metastases still persists in PCa. Thus, the aim of the present study was to evaluate in vitro whether transfection of PCa cell lines with miR-29b affects metastasis through modification of collagen and MMP-2 gene expression.

Desmoplasia refers to interstitial collagen production in cancer. Collagen types I and III are two of the most important molecules that support tissue structure in solid organs. Decades ago, Waisman et al. [10] observed that the production of extracellular material was an important feature of granulosa-theca tumors, and the culture of chondrosarcoma cells provided evidence of increased collagen production compared to that in normal cells [11].

Richards et al. [12] studied the growth of normal and neoplastic mouse mammary cells on three different substrates: plastic, rat tail, and rat tail collagen. Increased cell growth was reported when collagen was used as a substrate, primarily because the cells produced a structure that resembled collagen IV and laminin. Further, these authors provided evidence that the cells could actively produce collagen, because CIS-OH-proline (which blocks collagen production) inhibited cell growth.

As epithelial cells change their environment, they acquire functions of stromal cells, and collagen production may be one consequence of this transformation [13]. Another example of the interaction between malignant cells and the extracellular environment was provided by Noel et al., who showed higher collagen production in cocultures of MCF7 malignant cells and fibroblasts on a substrate containing type I collagen [14]. However, these authors hypothesized that the collagen was produced not by malignant cells but by fibroblasts. One of the hypotheses regarding the collagen production in neoplastic diseases was that this event functions as an attempt to construct a barrier that can halt tumor spread. However, Hewitt et al. [15] reported that intense collagen production occurred at the center of colon malignancies, in contrast with that observed at the edge where the invasion occurred. These authors, as well as a study published by other authors [14], also considered collagen production by stromal cells. In pancreatic cancer, the desmoplastic reaction occurs with increased production of collagen types I and III, which are primarily stimulated by TGF-β from granulocytes, suggesting an interaction between malignant epithelial cells and the stroma [16]. In renal neoplasms, the myofibroblasts identified in the tumor capsule have high type I collagen mRNA expression, likely also induced by TGF-β [17].

Kaupilla et al. [5] investigated whether fibroblasts or epithelial cells were responsible for the production of collagen types I and III in ovarian cystadenomas and cystadenocarcinomas. These authors demonstrated through in situ hybridization that poorly differentiated tumors could produce type I and III collagens. These authors also highlighted that collagen mRNA does not always translate into fibrils, except in ovarian cancer.

In PCa, there is evidence of collagenous micronodule deposition in biopsy and prostatectomy samples [18]. Further, intense desmoplastic activity was observed in intermediate and high-grade PCa, which also showed increased expression of vimentin, IGF-1, MMP-2, FGF-2, c-Myc, PSCA and Era [19], and an intense reactive stroma has been reported to distinguish benign from malignant prostatic tissue [20]. Such findings highlight the importance of extracellular collagen as a microenvironment component that can enhance metastasis. In other words, changes in the ECM created by tissue fibrosis can enhance tumor progression [21].

Previous studies have shown that malignant cells have reduced levels of miR-29b [22], and other studies have demonstrated reduced miR-29b expression in prostate cell lines and human prostate adenocarcinoma tissues [23, 24].

miR-29b inhibition leads to increased expression of the collagen I, III and V genes in PCa, as reported by Steele et al. [7], and the present results are consistent with these findings. However, in the present study, we found that miR-29b inhibition leads to increased collagen gene expression and further augments metastasis. We would expect that the increased cell invasion in the Matrigel assay was due to MMP-2 overexpression promoted by miR-29b inhibition, because this miRNA was previously shown to control MMP-2. However, because we did not observe MMP-2 mRNA expression in the examined cell lines, it is possible that invasion occurred with collagen gene expression independently of MMP-2. One possible explanation was provided by the results of other studies showing that MMP-2 is primarily expressed by stromal cells, as evidenced by the low to absent MMP-2 levels found in conditioned medium of PCa epithelial cells determined by zymography, whereas stromal cells exhibited higher levels of this protein [25]. Another study conducted by Yan et al. reported decreased cell invasion after miR-29b overexpression only with LncaP cells and without modification of MMP-2 levels. Then, our results obtained with another cell line (DU145) are in line with them [8].

A limitation of the present study is that these results could be improved with ELISA or western blotting to confirm at the protein level the increased collagen production induced by miR-29b downregulation in PCa cells. Another question raised in the present study is whether COL1A1/COL3A1 forms fibrils or if these collagens act as molecules that activate the metastasis pathway. Independently, the observed behavior of PCa cells after miR-29b inhibition indicated increased metastatic potential.

Thus, the present study showed that miR-29b inhibition increases cell invasion, and the expression of COL1A1 and COL3A1 genes in PC3 and DU145 prostate cell lines is independent of MMP-2. Thus, collagen expression may play an active role in metastasis physiology.

Blockade of miR-29b leads to increased mRNA expression of type I (COL1A1) and type III (COL3A1) collagens and enhances the invasion of PC3 and DU145 cells lines in vitro. This finding provides a possible role for the desmoplastic reaction. This mechanism was independent of the expression of MMP-2, and therapies that restore miR-29b levels might be promising alternatives for controlling the metastatic pathway in PCa.