STC1 Expression By Cancer-Associated Fibroblasts Drives Metastasis of Colorectal Cancer Free

Studies on metastasis have traditionally focused on properties of the malignant cells. However, recent studies in tumor biology have shown that the tumor microenvironment exerts major influence on tumor behavior, including the metastatic process (1–3). Several inflammatory cell types of the tumor microenvironment have been shown to affect the metastatic capacity of the malignant cells (4). Also, vascular characteristics, determined by properties of endothelial cells and pericytes, contribute to the formation of a more or less metastasis-permissive microenvironment (5). Moreover, factors derived from cancer-associated fibroblasts (CAF) have been shown to promote metastasis (6).

CAFs constitute a functionally important component of tumor stroma in many types of cancer (7, 8). Clinical data indicate that carcinomas with desmoplastic stroma, consisting of fibroblast cells and extracellular matrix, are associated with a poorer prognosis, which is consistent with the idea that CAF-derived factors stimulate invasion and metastases. Clinical associations between activated CAFs and metastasis have also been shown through prognostic fibroblast-derived gene signatures, including the core serum response signature derived from serum-stimulated fibroblasts and stroma-derived breast and lung cancer gene signatures (9–11).

CAF-derived secreted proteins include growth factors, cytokines, chemokines, extracellular matrix proteins, proteases, protease inhibitors, and lipid products, which may be involved in proinvasive effect of CAFs (7, 8). Production of these factors is generally induced by growth factors that activate and recruit CAFs. Growth factors linked to CAF activation include members of the TGF-β and hedgehog families, and chemokines including CXCL14 and CCL3 (12–15).

Members of the PDGF family have also been identified as key regulators of CAFs (16, 17). Paracrine activation of PDGF receptors on fibroblasts acts as a potent signal for tumor stroma recruitment (18–20). Antitumor effects have been shown following targeting of stromal PDGF receptors (21, 22). PDGF-activated fibroblasts have also been shown, in animal tumor models, to exert a negative influence on tumor drug uptake, presumably through effects on interstitial fluid pressure (23–27).

Recent analyses of primary tumors have established that stromal PDGF receptor expression is associated with worse prognosis in breast and prostate cancer (28, 29). Similar findings have been made in studies of smaller series of colorectal cancer (21). These clinical findings indicate that PDGF receptor–activated fibroblasts are involved in cross-talk with malignant cells that promotes metastasis. However, there have been few experimental analyses of the potential proinvasive effects of PDGF-activated fibroblasts, and the identity of potential PDGF-induced prometastatic factors remains unknown.

The aim of this study was to analyze the effects of PDGF-activated fibroblasts on migration and invasion of cancer cells and to identify novel proinvasive factors induced by PDGF activation of fibroblasts.

Collectively, the studies show promigratory and -invasive effects of PDGF-stimulated fibroblasts. Importantly, stanniocalcin 1 (STC1), a poorly characterized, secreted, and hypoxia-regulated protein, was identified as a fibroblast-derived mediator of the PDGF-dependent stimulatory in vitro effects. Fibroblast-derived STC1 was also shown in animal studies to contribute to prometastatic effects of fibroblasts.

Colorectal tumor cell lines LIM1215, SW620, HT29 and HCT116, immortalized fibroblast BJhTERT and wild-type, and STC1−/− mouse embryonic fibroblasts (MEF) were cultured in Dulbecco's Modified Eagle's Medium (DMEM; Gibco Life Technologies) containing 1% heat-inactivated fetal calf serum (FCS), 2 mmol/L l-glutamine, penicillin (100 U/mL), and streptomycin (100 ng/mL) at 37°C in a 5% CO₂ humidified atmosphere. Cells were stimulated with PDGFBB (Peprotech) for 5 minutes at 37°C, and the expression of PDGF β-receptor and its activation were analyzed by Western blotting using a specific PDGF β-receptor antibody (Cell Signaling) and an antibody against phospho-tyrosine (phospho-Y99, sc-7020, Santa Cruz).

LIM1215, SW620, HT29, and BJhTERT cell lines were obtained from commercial provider (Cell Bank Australia and American Type Culture Collection). HCT116 Cell lines were purchased from the Cancer Cell Line Repository of the Parc de Recerca Biomèdica de Barcelona (Barcelona, Spain). STC1 (+/+) and (−/−) MEFs were provided by Roger R. Reddel. These cells were derived from previously described wild-type and knockout STC1 C57BL/6 mice (30). MEFs were used at low passages and without experimentally induced or spontaneous immortalization.

Colon cancer cells were cocultured with PDGF-stimulated or nonstimulated fibroblasts on 8.0 μm pore Transwells (Costar, Corning Incorporated) without coating for migration assays, or coated with Matrigel matrix (BD Biosciences) for invasion assays. Before coculture, epithelial cells were labeled with a molecular probe (Cell Tracker green CMFDA C2925, Invitrogen), according to the manufacturer's instructions, to distinguish them from fibroblasts. After 48 hours, epithelial cells that had reached the lower surface of the filter were recovered by trypsin treatment and counted by fluorescence with WALLAC plate reader (excitation at 485 nm and emission at 535 nm) by interpolation using a standard curve. For migration and invasion studies with physical separation, stimulated or nonstimulated fibroblasts were seeded in the bottom compartment of the Transwell and the labeled colon cells in the top one.

Analyses were conducted largely using standard procedures. Details are given in Supplementary Information.

BJhTERT cells were seeded under standard conditions and transfected with either control or STC1 siRNA (Dharmacon RNA Technologies) using DharmaFect transfection reagent (Dharmacon RNA Technologies) according to the manufacturer's instructions. The cells were then cocultured with colon cells for invasion experiments as described above.

The tissue microarray was derived from a population-based cohort of 320 patients with colorectal cancer. Details about patients and technical aspects related to the generation of the tissue microarray is provided in Supplementary Information.

The animal experiments were conducted in accordance with National guidelines and approved by the Stockholm North Ethical Committee on Animal Experiments. The animal experiments and tissue analyses largely followed previously published procedures. For details, see Supplementary Information.

Immunohistochemistry analyses, and analyses of mitotic rate, followed established and previously published procedures. For details, see Supplementary Information.

Sense and antisense RNA probes were generated using the PCR primer sequences detailed in Supplementary Information Methods. Labeled RNA probes were generated by in vitro transcription from PCR products in the presence of digoxigenin–UTP. For in situ hybridization, tissues were digested by Proteinase K (Roche) and incubated with digoxigenin-labeled RNA probe. High-stringency washing was conducted after hybridzation, followed by detection of the bound probe by chromogenic anti-digoxigenin antibody immunohistochemistry. Blocking of various endogenous enzymes was conducted throughout the entire process.

Fresh tissue from primary tumors of patients operated for colorectal cancer at Linköping University Hospital (Linköping, Sweden) was used for the propagation of primary CAFs. Primary colon cancer CAFs were propagated from minced tumor tissue pieces in RPMI-1640 (Fisher Scientific), supplemented with 20% FCS (Invitrogen), 10 mmol/L HEPES (Invitrogen), 30 μg/mL gentamicin (Invitrogen), and 2.5 μg/mL Fungizone (Invitrogen). Subcultivation and experiments were carried out in DMEM (# 21969, Gibco) with 1% or 10% FCS and penicillin/streptomycin antibiotics in standard conditions. CAFs were used at low passages (between 6 and 8) and without experimentally induced or spontaneous immortalization.

All t tests were conducted following evaluation of equality of variance or not with Levene test. Two-tailed P values ≤ 0.05 were considered statistically significant. Comparison of STC1 and PDGF β-receptor expression, or Ki67, was done with χ² test.

The colon cancer cell lines, LIM1215, SW620, HT29, and HCT116 were initially characterized with regard to expression of PDGF β-receptors and production of their PDGF ligands. As shown in Fig. 1A, none of the cell lines expressed PDGF β-receptors as determined by immunoblotting. Furthermore, none of the cultures displayed an increase in tyrosine phosphorylation following addition of exogenous PDGF. In the case of LIM1215 cells, absence of detectable receptors was also confirmed under coculture conditions (Supplementary Fig. S1A).

To measure the expression of PDGF ligands in colon cancer cells, BJhTERT fibroblasts were incubated with conditioned medium from the different colon cancer cell lines, and effects on PDGF receptor phosphorylation were analyzed. This analysis showed that the colon cancer cell lines produced different levels of PDGF ligands, with HCT116 and SW620 cells showing the highest levels of PDGF production and LIM1215 with the lowest levels (Supplementary Fig. S1B). Basal levels of migration among the different epithelial cell lines were characterized before coculture experiments. These analyses identified HCT116 as the most migratory cells and the LIM1215 cells as the least migratory cells (Supplementary Fig. S1C).

Activation of the fibroblasts with PDGF induced a significant increase in the migration of cocultured LIM1215 cells in formats where the 2 cell types were either cultured together in the upper chamber of a Transwell culture system (Fig. 1B) or when fibroblasts were separated from the cancer cells (Fig. 1C). Furthermore, a PDGF dose dependency of the promigratory effects of stimulated fibroblasts were also shown (Supplementary Fig. S1D). A potent PDGF-dependent effect was also observed on LIM1215 invasion in coculture experiments. This was shown when the 2 cell types were cultured together in the top chamber of the Transwell culture system (Fig. 1D).

As shown above, SW620 and HCT116 cells produced high levels of PDGF (Supplementary Fig. S1B). Addition of exogenous PDGF ligand in cocultured fibroblasts did not show a significant increase in the migration of HCT116 cells (Supplementary Fig. S1E). Invasion experiments were carried out, which analyzed invasion of SW620 cells when cultured with fibroblasts in the absence or presence of the AG1296 PDGF receptor tyrosine kinase inhibitor (31). As shown in Fig. 1E, SW620 cell invasion was significantly reduced upon PDGF receptor inhibition.

Together, these experiments thus show PDGF-dependent paracrine promigratory and proinvasive effects of fibroblasts on colon cancer cells.

To identify mediators involved in PDGF-dependent paracrine signaling between fibroblasts and colon cancer cells, the gene expression profile of nonstimulated and PDGF-stimulated BJhTERT fibroblasts was analyzed by microarray analysis (GEO Series accession number: GSE40720). STC1 was among the most upregulated genes encoding secreted factors (see GEO dataset). This poorly characterized, secreted hypoxia-induced protein was recently linked to ovarian cancer (32). Furthermore, earlier studies have also linked this protein to progression of colorectal cancer (33, 34), although negative effect on prosurvival signaling pathways has also been observed (35).

Quantitative real-time PCR (qRT-PCR) analyses of STC1 expression confirmed a 15-fold STC1-induction upon PDGF stimulation of fibroblasts (Fig. 2A). On the basis of these findings, STC1 was selected for further analyses.

Previously isolated and characterized MEFs from STC1 knockout mice (30, 35) were used to analyze the contribution of STC1 to the PDGF-induced paracrine effects. These cells were observed to have similar levels of expression, and ligand-induced phosphorylation, of PDGF β-receptor as wild-type MEFs (Fig. 2B). Previous studies have reported that proliferation rate of STC1−/− MEFs is increased as compared with wild-type MEFs (35).

Analyses using the wild-type MEFs showed an increase of invasion in LIM1215 colon cells upon coculture with PDGF-activated wild-type MEFs (Supplementary Fig. S2A). Furthermore, the wt MEFs enhanced the invasion of the PDGF-producing SW620 cells in a manner that was not affected by PDGF but was reduced by PDGF receptor inhibitors (Supplementary Fig. S2B). Together, these results support the findings of Fig. 1, showing PDGF-dependent proinvasive effects of fibroblasts on colon cancer cells.

Importantly, experiments analyzing LM1215 cell invasion, in the presence of PDGF stimulation, revealed significantly lower LIM1215 migration in the STC1−/− MEF cocultures, as compared with that observed in wild-type MEF cocultures (Fig. 2C). Experiments using SW620 and HCT116 cells also showed that loss of STC1 significantly reduced the invasion-stimulatory effects of PDGF-stimulated fibroblasts (Fig. 2D and E). Similar results were observed when the promigratory effects of fibroblasts on HT29 migration were analyzed (Supplementary Fig. S2C). In some cases, the experiments with the MEF pair were also carried out in the absence of PDGF stimulation. Also, in this setting, the STC1−/− fibroblasts displayed a lower invasion-stimulatory effect than the wild-type fibroblasts, but this difference was consistently smaller than that observed in the presence of PDGF (Fig. 2C).

To obtain independent evidence for a role of STC1 in the paracrine proinvasive effects of fibroblasts, siRNA experiments were carried out using the BJhTERT fibroblasts. As shown in Fig. 2F, downregulation of STC1 dramatically reduced the proinvasive effect of BJhTERT cells in cocultures with LIM1215 cells.

Together, these experiments thus identify STC1 as a previously unrecognized important mediator of PDGF-dependent paracrine effects of fibroblasts on cancer cell migration and invasion.

PDGF β-receptor and Ki67 were analyzed by immunohistochemistry, STC1 expression was analyzed by in situ hybridization, and expression in tumor stroma and the malignant cells were scored separately in a tissue microarray of a population-based collection of human primary colorectal cancers.

In agreement with previous studies, PDGF β-receptor expression was predominantly observed in the tumor stroma, whereas variable expression of STC1 was found both in tumor stroma and in malignant cells. When results from PDGF β-receptor and STC1 analyses were combined, a significant positive association between stromal PDGF β-receptor and stromal STC1 expression was detected (P = 0.032; Fig. 3A and B). Analyses of stroma abundance, based on α-smooth-muscle-actin (ASMA) immunohistochemistry analyses, showed that the STC1 expression was not correlated to stroma abundance (Supplementary Fig. S3A).

To substantiate these findings, STC1 expression in primary cultures of CAFs isolated from colon cancers was also analyzed. Induction of STC1 was also observed in these cells following PDGF stimulation (Fig. 3C and D). Tissue analyses in 105 patients with TMA available from both primary tumor and normal mucosa revealed that the stromal expression of STC1 was significantly higher in the primary tumors than in the corresponding normal mucosa (P < 0.001). In addition, a significant positive correlation was observed between stromal STC1 expression and colon cancer cell proliferation, as determined by Ki67 analysis (Fig. 3E).

Together, these analyses thus show an upregulation of STC1 in tumor stroma, as compared with normal mucosa, and also suggest that the upregulation of STC1 in tumor stroma is related to stromal PDGF receptor signaling and cancer cell proliferation. No conclusive evidence was obtained concerning associations between STC1 expression, as presently analyzed, and survival (Supplementary Fig. S3B and S3D; see Discussion).

A previously described orthotopic colon cancer model was used to evaluate the importance of fibroblast-derived STC1 for metastasis (36). In previous studies, HCT116 cells have been shown to recapitulate critical features of colon cancer growth including invasion into the muscle layer, infiltration into lymphatic and blood vessels, and formation of distant metastases in, for example, the liver and the lung (36).

HCT116 cells were coinjected with wild-type or STC1−/− MEFs in a total of 24 animals. The take rate in both groups was similar with 9 and 8 animals forming tumors in the wild-type and STC1−/− MEF groups, respectively.

The tumors in animals coinjected with HCT116 cells and either wild-type or STC1−/− MEFs expanded from their initial location in the submucosa throughout all of the intestinal wall layers, ulcerating the mucosa and reaching the serosal surface. The local tumors gave rise to poorly differentiated adenocarcinomas in both groups. They were highly cellular, including atypical cells with pleomorphic nuclei and multinucleated cells. Tumors also formed diffuse fronts that invaded the local normal colon (Supplementary Fig. S4A and S4B).

Tumor volume was measured at necropsy. Tumors derived from coinjection of HCT116 cells and STC1−/− MEFs were significantly smaller than the tumors of the control group (Fig. 4A). The presence of fibroblast-like cells was observed in all tumors. Staining with PDGF β-receptor antibodies and cytokeratin did not reveal major differences between the tumor groups with regard to stromal–epithelial ratio (Fig. 4B and Supplementary Fig. S4C). Necrotic areas were also detected in most tumors but did not differ between the 2 experimental groups (Fig. 4B and Supplementary Fig. S4C).

Proliferation and apoptosis were evaluated in primary tumors by mitotic figures counting and caspase-3 immunohistochemistry, respectively. Tumors derived from coinjection with STC1−/− MEFs displayed reduced proliferation and also increased apoptosis (Fig. 4C–F). PCNA immunohistochemistry also confirmed the reduced proliferation in tumors derived from coinjection with STC1−/− MEFs (Supplementary Fig. S4D and S4E).

Vascular density was evaluated in primary tumors following immunohistochemical analyses using Isolectin B4 antibodies. No differences were observed between groups (Supplementary Fig. S4F and S4G).

Together, these analyses showed that the depletion of STC1 from fibroblasts decreased their ability to support orthotopic colon cancer growth through mechanisms involving reduced proliferation and increased apoptosis.

The lymphatic and hematogenous cancer cell dissemination was analyzed by macroscopic and microscopic examination of lymph nodes, peritoneum, diaphragm, pancreas, liver, and lungs. Interestingly, there was a difference in the number of affected organs such that the tumors derived from the STC1−/− group showed a significantly lower number of affected organs per mouse (Fig. 5A). Characterization of the distant metastatic lesions also revealed that the foci in the STC1−/− MEF group were significantly smaller than those in the control group (Fig. 5B).

The potential relationship between dissemination and size of primary tumor was also analyzed. Tumors were divided into 2 groups based on size (50% cut off), regardless of experimental group. This analysis did not reveal significant differences between the 9 “small primary tumor group” and the 8 “large primary tumor group” with regard to number of organs affected per mouse, or average lesion size (Supplementary Fig. S5A and S5B). Calculations of “metastatic index”, relating the number of affected organs to the size of the primary tumors, also indicated a reduced metastatic potency of tumors from the STC1−/− MEF group (Supplementary Fig. S5C).

Metastases were divided into groups of lesions formed through the peritoneal route (peritoneum and diaphragm), the lymphatic route (lymph nodes and pancreas), and the hematogenous route (liver and lungs). As shown in the Supplementary Fig. S6, this analysis indicated that the metastasis formation via all 3 routes, based on the size of foci, was reduced in the STC1−/− MEF group (Supplementary Fig. S6A–S6I).

This analysis thus showed that the STC1 status of coinjected fibroblasts determined metastatic behavior of orthotopically implanted colon cancer cells.

The possibility that STC1 status affected vascular infiltration was evaluated. Interestingly, the number and size of intravasated tumor emboli was significantly reduced in the STC1−/− MEF group (Fig. 6A and B). Neither of these parameters differed significantly when tumors were divided according to size (Supplementary Fig. S7A and S7B).

Decreased lymphatic vessel density, evaluated by LYVE-1 immunohistochemistry, was also considered as a factor contributing to the reduced metastasis in the STC1−/− MEF group. A significant decrease of lymph vessel density was observed in primary tumors with STC1−/− MEFs (Supplementary Fig. S7C and S7D). However, a significant difference in lymphatic vessel density was also observed when tumors were grouped on the basis of size (Supplementary Fig. S7E), making it difficult to evaluate whether the phenotype was secondary to the effect of STC1 on tumor size, or reflected a more direct relationship between STC1 and lymphangiogenesis.

These findings thus suggest that STC1 status of fibroblasts affects metastasis by regulating the intravasating ability of colon cancer cells.

The presence of cell displaying an EMT phenotype, the abundance of cancer stem cells, and the amount of macrophages was evaluated as additional possible factors involved in the reduction of metastases from tumors with STC1−/− MEFs.

CD44 was used as stem cell marker (Supplementary Fig. S7F–S7H). Tumors with STC1−/− or +/+ MEFs did not show changes in the CD44 immunostainning. F4–80 Immunohistochemistry was analyzed to determine the amount of macrophages in tumors (Supplementary Fig. S7I–S7K). A reduction of the macrophage number was observed in the STC1−/− MEF group. However, the analyses of size-divided tumor groups also revealed differences macrophage density, and the differences between the experimental groups could thus be secondary to the generally smaller size STC1−/− tumors, rather than a direct consequence of the fibroblast genotype.

Vimentin immunostaining, using an antibody specific for human vimentin, was used as an EMT cell marker. EMT was detected in 4 of 9 tumors in the STC1+/+ group. In contrast, no sample in the STC1−/− MEF group showed vimentin expression (P = 0.053; χ² test; Fig. 6C).

Together, these analyses suggest that STC1 directly contributes to EMT, which could be involved in the effects on metastasis to lymph nodes and distant organs.

In summary, this study shows that PDGF receptor stimulation of fibroblasts significantly increases migration and invasion of cocultured colorectal cancer cells in an STC1-dependent manner. The in vivo significance of these findings is documented by the observations that tumors with STC1−/− fibroblasts formed fewer and smaller distant metastases in a manner involving reduced intravasation of tumor cells and a reduction in cells displaying an EMT phenotype. Furthermore, clinical relevance is indicated by the observation that PDGF receptor and STC1 expression is significantly correlated in the stroma of human colorectal cancers.

A series of other studies based on tumor tissue analyses have provided findings, which support a role for STC1 in cancer progression and metastasis (32–34, 37). Increased levels of circulating STC1 mRNA have also been associated with worse prognosis in lung cancer (38).

The findings imply stromal STC1 as a prometastatic factor. Preliminary analyses of the prognostic significance of stromal STC1 expression, based on the in situ analyses, showed no prognostic effect of stromal STC1 in a multivariate analyses including tumor stage (Supplementary Fig. S3). Surprisingly, univariate analyses indicated an association between high stromal STC1 and better prognosis (Supplementary Fig. S3). These inconclusive analyses should be extended. Such studies should particularly explore STC1 expression in the invasive front of colorectal cancer. It should also be noted that the present analyses relied on in situ expression analyses of STC1, which might not reflect protein expression. This approach was selected because validation experiments, using various antibodies and cells with known STC1 expression status, failed to confirm the specificity of the available antibodies. Future studies should also address the possibility that the prognostic significance of stromal STC1 expression might be restricted to certain molecular subsets of colorectal cancer.

This study, dominated by experimental findings, presents some findings that indicate clinical relevance, including the demonstration of a significant association between PDGF β-receptor expression and STC1 in a large clinical cohort. Importantly, PDGF receptor-dependent upregulation of STC1 expression was also shown in a primary culture of colorectal CAFs. The cell culture experiments suggest that the proinvasive effects are not restricted to a particular cell line. Moreover, the orthotopic animal model used in this study has been well characterized and is recognized to model human colorectal cancer well (36).

The results from the animal experiments strongly support a stimulatory effect of fibroblast-derived STC1 on metastasis to lymph nodes and distant organs. An important topic for future studies will be to further describe the in vivo mechanism(s) that are most important for the prometastatic effects of STC1. On the basis of the coculture experiments, it is likely that a promigratory and proinvasive effect of STC1 contributes to the prometastatic in vivo results. Furthermore, the analyses of tumor tissue suggest that STC1 also enhances the ability of cancer cells to intravasate. The potential involvement of epithelial–mesenchymal transition in the STC1-dependent differences in metastasis should be further explored on the basis of the preliminary findings of the present study (Fig. 6C). In the present study, no major alterations in cell type composition of tumors from the 2 groups were observed. However, this issue should be revisited by expanded analyses of various subsets of additional tumor stromal cells, such as fibroblasts, pericytes, and neutrophils. These future analyses should also consider recent studies, which have implicated STC1, in noncancer models, as a regulator of inflammation, macrophage mobility, and vascular permeability (39–42). Interestingly, earlier studies have also shown that hypoxia, a prometastatic stimuli, also leads to the upregulation of STC1 (43).

Tumors with wild-type fibroblasts were significantly larger than the tumors with STC1−/− fibroblasts, and it could be suggested that the differences in metastasis burden between the 2 groups are secondary to differences in tumor size. However, it should be noted that when animals were divided according to the size of primary tumor, no significant differences in metastasis number or size was detected. The lack of correlation between the size of primary tumor and metastasis is also in agreement with clinical findings showing that the size of the primary tumor is not a consistent risk factor for metastasis (44).

According to preliminary results (data not shown), fibroblast-derived STC1 did not affect cancer cell proliferation in vitro. However, a potential indirect effect of STC1 on cancer cell proliferation should not be excluded since Ki67 staining of human samples showed an association with STC1 expression levels (Fig. 3E). Also, the results from the study of experimental tumors support an effect on proliferation, as a decrease in PCNA immunostaining and a reduction in the number of mitotic figures, was observed in tumors derived from HCT116 cells coinjected with STC1 (−/−) MEFs (Fig. 4D and Supplementary Fig. S4D and S4E). This effect of STC1 on in vivo proliferation should be further explored and could be secondary to the effect of interactions with other cells in the tumor microenvironment.

The accumulating evidence suggesting an important role of STC1 in cancer progression prompts further studies on the molecular mechanism(s) of action of this still poorly characterized protein. Identification of its molecular receptor and delineation of the intracellular pathways mediating the promigratory and -invasive effects are highly warranted. Analyses of STC1−/− fibroblasts have shown that these cells display resistance to apoptosis and higher levels of activation of MEK and ERK1/2 than wild-type MEFs (35). To what extent these pathways are also important in the paracrine effects described in the present study merits further investigations. A better characterization of the signaling involved in the PDGF-induced activation of STC1 is also motivated. Preliminary studies with PI3K and MEK inhibitors showed that monotreatment with either of these agents were not sufficient to significantly reduce the PDGF-induced STC1 expression (data not shown).

The findings of the present study, in general terms, show the importance of CAFs in modulating metastatic potential of cancer cells. Obvious and important implications of these findings are, first, that inhibition of fibroblast–epithelial interactions represents a strategy for interference with metastasis and, second, that careful analyses of fibroblast characteristics should be emphasized in ongoing efforts to identify markers for progression in colorectal and other cancers.

No potential conflicts of interest were disclosed.

Conception and design: C. Peña, M.V. Céspedes, H. Birgisson, P. Sandström, T. Sjöblom, R. Mangues, A. Ostman

Development of methodology: C. Peña, M.V. Céspedes, M.B. Lindh, R. Mangues, M. Augsten, A. Ostman

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Peña, M.V. Céspedes, M.B. Lindh, S. Kiflemariam, P.-H. Edqvist, H. Birgisson, L. Bojmar, K. Jirstrom, P. Sandström, A. Gallardo, R.R. Reddel, R. Mangues, M. Augsten

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C. Peña, M.V. Céspedes, M.B. Lindh, S. Kiflemariam, A. Mezheyeuski, H. Birgisson, L. Bojmar, E. Olsson, S. Veerla, A.C.-M. Chang, R. Mangues, M. Augsten, A. Ostman

Writing, review, and/or revision of the manuscript: C. Peña, M.V. Céspedes, M.B. Lindh, S. Kiflemariam, H. Birgisson, P. Sandström, T. Sjöblom, A.C.-M. Chang, R.R. Reddel, R. Mangues, A. Ostman

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P.-H. Edqvist, C. Hägglöf, H. Birgisson, K. Jirstrom, A. Ostman

Study supervision: R. Mangues, A. Ostman

The authors thank the staff at the MTC animal facility at Karolinska Institutet for expert technical assistance, Paloma Martín from Pathology department in Hospital Puerta de Hierro (Majadahonda, Madrid, Spain) for expert immunohistochemistry analyses, Jan Mulder, Sci Life Laboratory (Stockholm, Sweden) for assistance with digital microscopy expertise, and members of the Arne Östman laboratory who contributed with constructive criticism throughout this project.

A. Östman's laboratory was supported by grants from Swedish Research Council (60016002), VINNOVA, and Swedish Cancer Society (11 0371). R.R. Reddel's laboratory was supported by National Health and Medical Research Council of Australia project grant.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.