IGFBP7, a novel tumor stroma marker, with growth-promoting effects in colon cancer through a paracrine tumor–stroma interaction

Abstract

The activated tumor stroma participates in many processes that control tumorigenesis, including tumor cell growth, invasion and metastasis. Cancer-associated fibroblasts (CAFs) represent the major cellular component of the stroma and are the main source for connective tissue components of the extracellular matrix and various classes of proteolytic enzymes. The signaling pathways involved in the interactions between tumor and stromal cells and the molecular characteristics that distinguish normal ‘resting’ fibroblasts from cancer-associated or ‘-activated’ fibroblasts remain poorly defined. Recent studies emphasized the prognostic and therapeutic significance of CAF-related molecular signatures and a number of those genes have been shown to serve as putative therapeutic targets. We have used immuno-laser capture microdissection and whole-genome Affymetrix GeneChip analysis to obtain transcriptional signatures from the activated tumor stroma of colon carcinomas that were compared with normal resting colonic fibroblasts. Several members of the Wnt-signaling pathway and gene sets related to hypoxia, epithelial-to-mesenchymal transition (EMT) and transforming growth factor-β (TGFβ) pathway activation were induced in CAFs. The putative TGFβ-target IGFBP7 was identified as a tumor stroma marker of epithelial cancers and as a tumor antigen in mesenchyme-derived sarcomas. We show here that in contrast to its tumor-suppressor function in epithelial cells, IGFPB7 can promote anchorage-independent growth in malignant mesenchymal cells and in epithelial cells with an EMT phenotype when IGFBP7 is expressed by the tumor cells themselves and can induce colony formation in colon cancer cells co-cultured with IGFBP7-expressing CAFs by a paracrine tumor–stroma interaction.

Introduction

Carcinomas consist of complex mixtures of neoplastic epithelial cells and non-neoplastic cells, collectively referred to as ‘tumor stroma’. The tumor stroma, which in some carcinomas makes up more than 80% of the tumor mass, is composed of blood (BECs) and lymphatic endothelial cells (LECs), infiltrating immune cells, pericytes and specialized fibroblasts, termed cancer-associated fibroblasts (CAFs), embedded in a network of extracellular matrix proteins. During the past years, it has become increasingly evident that far from being a mere bystander, the tumor stroma participates in many of the processes that control tumorigenesis.1, 2, 3 CAFs, which are the main source for extracellular matrix-degrading enzymes, including cysteine-, serine- and matrix-metalloproteinases, were shown to act as key players in remodeling the tumor microenviroment and to be essential for the local spread of tumor cells into the adjacent normal tissues and the formation of distant metastasis.4,5 In addition, CAFs have been shown to be involved in the formation of resistant tumors. However, the molecular characteristics that distinguish a normal ‘resting’ fibroblast from a cancer-associated or ‘activated’ fibroblast remained poorly defined. Presently, CAFs are defined by morphological characteristics and by the expression of specific sets of markers, including fibroblast activation protein alpha (FAPα),6 alpha-smooth-muscle actin (SMA),7 fibroblast-specific protein 1 (FSP1/S100A4)8 or platelet derived growth factor receptor beta (PDGFRβ).9 This molecular heterogeneity has been linked to the diverse origin of CAFs, which have been reported to be derived from resident local fibroblasts, from bone marrow-derived progenitor cells or from transformed epithelial cells, which have undergone an epithelial-to-mesenchymal transition (EMT) during tumorigenesis.10,11 An increasing number of translational studies have recently emphasized the prognostic significance of different CAF-related molecular signatures12, 13, 14, 15 and clinical studies targeting CAFs in human cancers have been proposed.16, 17, 18 We have used immuno-laser capture microdissection (iLCM)19 and whole-genome Affymetrix GeneChip analysis (Affymetrix, Santa Clara, CA, USA) to obtain transcriptional signatures from the tumor cells and the activated tumor stroma that were compared with the expression profiles from normal colonic epithelium and normal fibroblasts, derived from the same patients. Induced genes included several members of the Wnt-signaling pathway or collagen cross-linking enzymes such as lysyl oxidases. Moreover, increased expression of gene sets related to hypoxia, EMT and transforming growth factor-β (TGFβ) pathway activation were found in CAFs vs their normal counterparts. We have identified a putative TGFβ target gene,20 IGFBP7, as a tumor stroma marker of epithelial cancers that can also act as a tumor antigen in mesenchyme-derived tumors such as fibrosarcomas.

We demonstrate that in addition to its reported tumor-suppressor function, IGFPB721,22 can promote anchorage-independent growth in malignant mesenchymal cells and in epithelial cells with an EMT-phenotype when IGFBP7 is expressed by the tumor cells themselves. Moreover, IGFBP7 can induce colony formation in colon cancer cells when co-cultured with IGFBP7-expressing CAFs by a secondary paracrine tumor–stroma interaction.

Results

CAFs show increased expression levels of genes related to hypoxia, EMT and TGFβ pathway activation

To define the molecular differences between cancer-associated and ‘normal’ resting fibroblasts in colon cancer, iLCM19 was conducted on a set of colon cancer samples together with their normal counterparts. RNA extracted from the captured tumor cells, the tumor stroma and the normal stroma was processed for whole-genome Affymetrix GeneChip hybridization (Supplementary Figure S1). After normalization and bioinformatic analysis, the three compartments (tumor cells, tumor stroma and normal stroma) could be clearly separated by hierarchical cluster analysis (Figure 1a) and principal component analysis (data not shown). A total of 1299 genes was differently expressed between the tumor stroma and the normal stroma, with 627 of these genes significantly upregulated in the tumor stroma (false discovery rate P<0.01). Genes upregulated in the tumor stroma vs normal stroma contained well-established tumor stroma markers, such as PDGFRB, FGFR1, matrix metalloproteinase 2 (data not shown) or FAPα, the marker used to capture the activated fibroblasts from the tissue samples, demonstrating the high purity of our iLCM approach (Figure 1b). Furthermore, induced genes included several members of the Wnt-signaling pathway such as secreted frizzled-related protein 2 (SFRP2), WNT2, WNT5A or Wnt-1-induced secreted protein 1 (Wisp-1). The upregulation of Wisp-1 and its putative binding partner biglycan on the RNA and protein levels were confirmed in validation assays performed in independent colorectal cancer samples (Figures 1b and c).

Figure 1
figure1

Genes and gene sets induced in the tumor stroma. (a) Hierarchical cluster analysis of tumor cell, tumor stroma and normal stroma samples. A total of 1299 genes was differentially expressed between the tumor stroma and the normal stroma. Heat-map colors represent mean-centered fold change expression in log-space. (b) Well-characterized tumor stroma markers, such as FAPα, Biglycan or Wnt-1-induced protein 1 (Wisp-1), were found to be significantly induced in the tumor stroma compartment vs the normal stroma. The expression levels are indicated by whisker box plots, the bold centerline indicates the median; the box represents the interquartile range (IQR). Whiskers extend to 1.5 times the IQR. NS, normal stroma; TC, tumor cells; TS, tumor stroma. (c) Immunohistochemical staining demonstrates the induction of these genes at the protein level.

For the further identification of tumor stroma-specific genes sets, we used gene set-enrichment analysis with two collections derived from the Molecular Signature Database (Broad Institute); c2, the curated-gene sets, and c5 the gene ontology-gene sets.23 Among the gene ontology data sets most significantly enriched (false discovery rate q<0.25) were those involved in extracellular matrix deposition, metalloendopeptidase activity, cell migration, insulin receptor signaling, response to hypoxia, cell–cell adhesion, vascular development, cytokine activity, response to wounding and cell–matrix adhesion (Table 1). In addition, we identified 192 significantly enriched curated gene sets, including gene sets related to hypoxia, EMT, interleukin-6 (IL-6) and TGFβ pathway activation (Table 2).

Table 1 GSEA, GO gene sets
Table 2 GSEA, curated gene sets

Hypoxia-related gene sets contained angiogenic growth factors such as VEGFC or angiopoetin-like 4. Other genes in that category encoded collagens (COL1A2, COL4A1, COL5A1, COL9A1 and COL18A1) and their modifying enzymes (lysyl oxidase, lysyl oxidase-like 2), suggesting that the collagen biosynthesis in the stromal compartment undergoes multiple hypoxia-induced changes. Interesting examples for enriched genes related to IL-6 treatment comprised several pro-inflammatory cytokines such as CXCL3 or IL-6 itself. Enriched gene sets involved in EMT included the mesenchymal marker N-Cadherin (CDH2), several tumor-promoting matrix metalloproteinases (MMP2 and MMP12) and extracellular matrix proteins implicated in invasion and metastasis, such as tenascin C, laminin B1 or secreted protein acidic and rich in cysteine. Many of the EMT-related genes were also found among TGFβ-induced gene sets, as were genes encoding fibrillar collagens (COL1A1, COL1A2, COL3A1 and COL5A2) and members of the insulin-like growth factor-binding protein family, namely IGFBP3, IGFBP5 and IGFBP7.20 IGFBP7 appeared as one of the most significantly induced tumor stroma markers in our screen (tumor stroma vs normal stroma, P<2,14E-05; Figure 2a).

Figure 2
figure2

Induction of IGFBP7 in the tumor stroma. (a) The whisker box plot shows the induction of IGFBP7 in the tumor stroma (NS, normal stroma; TC, tumor cells; TS, tumor stroma). (b) As demonstrated by immunohistochemical staining on representative examples of colon and lung carcinomas (non-small-cell lung cancer (NSCLC)), IGFBP7 is expressed in cancer-associated fibroblasts (CAFs) and tumor vessels in those epithelial cancer samples. In soft tissue sarcomas, such as MFH (malignant fibrous histiocytoma), IGFBP7 is expressed by the malignant mesenchyme-derived cells. Paraffin sections were stained with the avidin-biotin immunoperoxidase (ABC) method and counterstained with hematoxylin (blue). Size bars represent 100 μm. (c) Double labeling of IGFBP7 (red)/EndoGlyx a marker of endothelial cells (green) and IGFBP7 (red)/NG2 a marker of pericytes (green) on sections of a colon adenocarcinoma are shown. IGFBP7 expression is observed in tumor endothelial cells and surrounding mural cells (yellow signal due to co-localization of both markers). (d) Blood (BECs) and lymphatic (LECs) human dermal microvascular endothelial cells and freshly isolated colon CAFs are positive for IGFBP7. CD31, Prox1 and vimentin were used as established markers for the individual cell types. DAPI, 4',6-diamidino-2-phenylindole.

IGFBP7 is a tumor stroma marker of activated fibroblasts and endothelial cells in epithelial cancers

Immunohistochemical studies on tissue samples from colorectal cancer validated our RNA expression data and showed selective expression of IGFBP7 in tumor-associated vessels and in subsets of activated fibroblasts in the tumor stroma (Figure 2b and Supplementary Figure S2). Stainings on a variety of other epithelial cancers revealed that IGFBP7 is frequently induced in the stromal compartment of solid tumors including non-small-cell lung cancer, pancreatic, ovarian and breast carcinomas (Figure 2b). In any of the epithelial cancers examined IGFBP7 was expressed by the tumor cells. In contrast, IGFBP7 was expressed by the malignant tumor cells and the tumor stroma in soft tissue sarcomas, most prominently in the tumor-associated vasculature (Figure 2b). This pattern of expression closely resembles that of two other tumor-stroma markers FAPα and Endosialin. In double-labeling studies, IGFBP7 co-localized with EndoGlyx, a marker for BECs,24 and the pericyte marker NG2,25 indicating that IGFBP7 is a marker of endothelial cells and the surrounding mural cells (Figure 2c). To ascertain whether IGFBP7 is a marker of BECs and/or LECs, we have tested primary BECs and LECs isolated from human dermal microvascular endothelial cells for IGFBP7 expression. IGFBP7 was present both in BECs and LECs as demonstrated by its co-expression with CD31 and Prox1 markers of BECs and LECs, respectively (Figure 2d). In addition, freshly isolated colon CAFs were shown to express IGFBP7 and vimentin, a marker of mesenchymal differentiation (Figure 2d). Taken together, IGFBP7 appeared as a marker for blood and lymphatic vessels, fibroblasts and tumor cells of mesenchymal origin.

IGFBP7 is expressed by malignant mesenchyme-derived cells and malignant epithelial cells with a mesenchymal phenotype

IGFBP7 expression was observed in malignant mesenchymal cells in soft tissue sarcomas by immunohistochemistry (Figure 2b). This expression pattern was further confirmed by immunoblotting analysis in the fibrosarcoma cell line HT1080 (Figure 3a). Moreover, expression of IGFBP7 was observed only in cancer cell lines, which have undergone EMT such as Caki-1 and SW480 cells. Colon cancer cell lines, with epithelial phenotypes such as LS174T, HT29 and DLD-1 lacked IGFBP7 expression (Figure 3a). Expression of E-cadherin and vimentin was used to substantiate the morphological distinction between the mesenchymal and epithelial phenotype in the carcinoma cell lines (Figure 3a). Matriptase, a type II transmembrane serine protease (MT-SP1)26 known to cleave IGFBP7,27 is expressed at high levels in cells with an epithelial phenotype and is absent in cells with a mesenchymal phenotype (Figure 3a). Our results indicate that IGFBP7 is expressed by the tumor stromal fibroblasts, endothelial cells, mesenchymal tumors and by malignant epithelial cells that have acquired a mesenchymal phenotype as a result of EMT. To further substantiate this observation, we examined whether the loss of E-cadherin, which was previously shown to induce EMT,28 leads to an induction of IGFBP7 expression. Therefore, E-cadherin was knocked down in DLD-1 cells via stable short hairpin RNA (shRNA) expression and verified by western blotting (Figure 3b). Phenotypically, the E-cadherin knockdown cells displayed a more mesenchymal phenotype in comparison to the control cells (Supplementary Figure S3). In addition, β-catenin protein was redistributed from the membrane into the nucleus, indicating β-catenin activation in cells lacking E-cadherin (Figure 3b). Most importantly the induction of the mesenchymal phenotype correlated with a significant induction of IGFBP7 expression (Figure 3b).

Figure 3
figure3

IGFBP7, a marker for malignant mesenchyme-derived tumor cells and cells that have undergone EMT, suppressed colony formation of epithelial cells. (a) As demonstrated by western blotting, IGFBP7 is expressed by malignant mesenchyme-derived cells (HT1080) and epithelial cells with a mesenchymal phenotype (Caki-1, SW480), whereas tumor cells with an epithelial phenotype (LS174T, HT29, DLD-1, Colo205) are IGFBP7 negative. E-cadherin and vimentin were used as markers to determine the phenotype of the cell lines. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a loading control. Soluble IGFBP7 was detected in the supernatant, vimentin (Vim), E-cadherin (E-Cad), matriptase and GAPDH in the corresponding cell lysates. (b) Western blotting demonstrated shRNA-mediated loss of E-Cad in DLD-1 cells. DLD-1-shEcad cells showed induction of IGFBP7 mRNA expression by quantitative PCR relative to GAPDH. In DLD-1-shEcad cells, β-catenin redistribution into the nucleus is shown by immunofluorescence. Size bar: 25 μm. (c) Conditioned medium from IGFBP7 overexpressing DLD-1 cells contained two IGFBP7 bands, representing a native 33-kD form and a 26-kD form, which potentially results from proteolytic cleavage by matriptase (MT1-SP1). (d) Soft agar colony formation was significantly reduced in DLD-1 cells overexpressing IGFBP7, when compared with mock transfected cells. (e) Colonies were stained with crystal violet and counted with ImageJ, the graphs represent the average results of three independent experiments. P according to Student's t-test.

Expression of IGFBP7 reduces the anchorage-independent growth ability of epithelial tumor cells

To examine how IGFBP7 affects the anchorage-independent growth abilities of epithelial tumor cells, IGFBP7 was overexpressed in DLD-1 cells. These cells were shown to express E-cadherin, matriptase and to lack endogenous IGFBP7 expression (Figure 3a). In conditioned medium of IGFBP7 overexpressing cells (DLD-1-IGFBP7), the unprocessed (33 kD) and cleaved (26 kD) IGFBP7 were detected (Figure 3c) as described.27 Next, we compared the abilities of mock and IGFBP7-transfected DLD-1 cells to form colonies in soft agar culture (Figure 3d). The number of colonies formed by DLD-1-IGFBP7 cells was reduced by half compared with the colonies formed by the DLD-1/mock cells (Figure 3e).

Loss of IGFBP7 impairs the anchorage-independent growth ability of mesenchymal tumor cells

To establish the potential role of IGFBP7 in malignant mesenchyme-derived cells and epithelial tumor cells after EMT, IGFBP7 shRNA-mediated knockdown experiments were carried out and analyzed by qPCR and western blotting (Figure 4a and Supplementary Figure S4). In HT1080 fibrosarcoma cells, shRNA-mediated knockdown of IGFBP7 resulted in >90% reduction on the protein level in cell supernatants (Figure 4a). The IGFBP7 knockdown had no effect on cell cycle progression (Supplementary Figure S5) and migration (Supplementary Figure 6a) in two-dimensional cultures and displayed only a minor nonsignificant reduction of invasion (Supplementary Figure S6b). Next, we analyzed the anchorage-independent growth of the IGFBP7 knockdown and control cells by comparing their ability to grow and form colonies in soft agar culture. The number of colonies after two weeks in soft-agar culture was reduced more than twofold in IGFBP7-depleted HT1080 (Figures 4a and c), Caki-1 and SW480 cells (data not shown). The ability of mammalian cells to proliferate anchorage independently often correlates with their ability to form tumors in vivo. In line with these observations, we could demonstrate a significant delay in tumor growth utilizing IGFBP7 knockdown HT1080 cells in a xenograft experiment. At day 14 after subcutaneous implantation, these cells exhibit pronounced growth retardation as compared with non-modified HT1080 control cells (Figure 4d).

Figure 4
figure4

Loss of IGFBP7 in sarcoma cells suppressed colony formation and xenograft tumor growth. (a) In HT1080 cells, shRNA-mediated knockdown resulted in a significant IGFBP7 protein reduction in the supernatant (SN). (b) As illustrated with a colony formation assay in soft agar, IGFBP7 knockdown results in a significant reduction of anchorage-independent growth of HT1080 cells. Colonies were stained with crystal violet and counted; size bar: 50 μm (c) the graphs represent the average results of three independent experiments. P according to Student́s t-test. (d) Xenografts were established by injection of 1.107 HT-1080-shC and HT-1080-shIGFBP7 cells in both flanks of SCID mice (n=3) and tumor growth was measured for 14 days. As indicated by the respective single and median tumor volumes, a delayed tumor take of the IGFBP7 knockdown cells was observed in comparison to the control cells. GAPDH, glyceraldehyde 3-phosphate dehydrogenase.

IGFBP7 is a TGFβ but not a Wnt target in human colonic fibroblasts and IGFBP7 expression in these cells promotes colony formation of epithelial DLD-1 cells

Next, we determined whether IGFBP7 is a TGFβ target gene in human colonic fibroblasts as described for glioblastoma cells.20 High baseline levels of IGFBP7 mRNA were present in CCD18-Co human colon myofibroblasts (data not shown) and were further increased by TGFβ treatment (Supplementary Figure S7a). A Smad3 reporter served as control for TGFβ response. IGFBP7 was not under control of the canonical Wnt-signaling in CCD18-Co, as demonstrated by Wnt3a treatment. A 7TGP reporter construct was used to demonstrate that canonical signaling was initiated by Wnt3a treatment in the fibroblasts (Supplementary Figure S7b).

As IGFBP7 is expressed in stromal cells in colon carcinomas samples, we next attempted to study its putative effects during the tumor–stroma interactions. For this purpose, we evaluated the influence of fibroblasts on anchorage-independent growth of the colon cancer cells (DLD-1) in a co-culture assay. We used a modified soft agar colony assay, in which a confluent layer of primary colon CAFs (CAF329) with endogenous expression of IGFBP7 (CAF3-shC) or stable shRNA-mediated knockdown (CAF3-shIGFBP7) were co-cultivated for 3 weeks with DLD-1 cells embedded in soft agar (Figure 5a). IGFBP7 mRNA was reduced by 90% in the CAF3-shIGFBP7 as compared with CAF3-shC (Figure 5b). DLD-1 cells formed small, medium and large sized colonies under both conditions (Figure 5c) and the CAF3 at the bottom (CAF3-shC and CAF3-shIGFBP7) were alive and equally confluent after 3 weeks of co-culture (Figure 5d). Twice as many DLD-1 colonies formed in the presence of CAF3-shC as compared with a conventional soft agar assay without stromal fibroblasts present at the bottom (Figures 5e and f). Interestingly, the DLD-1 colon cancer colonies were reduced significantly in co-culture with CAF-shIGFBP7 (Figures 5e and f). Same results were obtained with CCD18-Co/DLD-1 co-cultures, where IGFBP7 was also knocked down in the colon fibroblasts (Supplementary Figure S8). These results indicated that autocrine IGFBP7 signaling in the stromal fibroblasts could override the tumor-suppressive function of IGFBP7 on epithelial cells alone. We hypothesized that IGFBP7 signaling in the fibroblasts changed the paracrine crosstalk to the carcinoma cells by altering the secretome of the tumor microenvironment. Indeed, cytokine profiling (Figure 6a) revealed decreased C5/C5a and sICAM-1 levels produced by CAF3-shIGFBP7, whereas granulocyte-macrophage colony-stimulating factor was increased and IL-6 (among others) was highly expressed but did not vary (Figure 6b and Supplementary Figure S9). A detailed analysis of secreted factors and their impact on tumor cell growth will be investigated in further studies.

Figure 5
figure5

IGFBP7 expression in stromal fibroblasts supports anchorage-independent growth of colon cancer cells. (a) Experimental setup to monitor stromal effects on anchorage-independent growth of carcinoma cells. (b) Primary human cancer-associated fibroblasts (CAF3) were transduced with IGFBP7 shRNA (shIGFBP7) or non-targeting control (shC) via lentiviral infection. The reduction of IGFBP7 mRNA is shown using quantitative PCR. (c) DLD-1 cancer cells formed colonies of different sizes after 3 weeks of culture in soft agar on confluent CAF3-shC and CAF3-shIGFBP7 fibroblasts. (d) Fibrobalsts were viable and of the same density throughout the incubation period. Representative images of the colonies (col.) and the fibroblasts (fibros) are shown after 3 weeks. (e) Crystal violet staining revealed reduced DLD-1 colony formation in the presence of CAF3-shIGFBP7 as compared with CAF3-shC. (f) Quantification of three independent experiments revealed a significant reduction in DLD-1 colonies upon co-culture with CAF3-shIGFBP7 as compared with the controls.

Figure 6
figure6

Cytokine profiling of conditioned medium from CAF-shC and CAF-shIGFBP7 and survival rates in human colon cancer patients. (a) Cytokine arrays developed on X-ray film at 30 s exposure time. (b) Densitometric quantification of selected cytokines/factors, which were decreased (complement 5/5a, C5/C5a; soluble intracellular adhesion molecule-1; sICAM-1) or increased upon IGFBP7 knockdown in the fibroblasts (CAF3-shIGFBP7) as compared with non-targeting controls (CAF3-shC). IL-6 is shown as example for a non-regulated cytokine. EGM: normal fibroblast growth medium alone. Bars represent means, error bars are s.d., n=4. P-values: Student’s t-test. (c) 177 patient samples were analyzed with the Oncomine Premium Research Edition to correlate the overall survival follow-up time vs mRNA expression levels of IGFBP7. The patient collective was divided into four groups with lowest (cyan, group 1, 45 patients), intermediate (green, group 2, 44 patients; orange, group 3, 44 patients) and high (magenta, group 4, 44 patients). mRNA levels are shown as log2 median centered intensity. (d) Survival of the four groups is depicted in Kaplan–Meier plots.

High IGFBP7 expression in human colon cancer correlates with bad prognosis

A data set of 177 non-dissected (that is, stroma containing) colon cancer patient samples30 was analyzed using the Oncomine Premium Research Edition database31 to correlate mRNA expression levels of IGFBP7 vs the overall survival. High IGFBP7 expression present in colon cancers (Figure 6c) correlated with reduced overall patient survival (Figure 6d).

Matriptase expression in tumor cells has no effect on IGFBP7-mediated cell growth

As proteolytic cleavage of IGFBP7 has been reported to change some of its biological effects, we aimed to investigate the functional consequence of IGFBP7 cleavage in the stroma-tumor crosstalk. Based on the fact that HT1080 cells express non-truncated IGFBP7 (Figure 3a) and secrete it into the medium, whereas DLD-1 cells do not express IGFBP7 (Figure 3a), we incubated DLD-1 cells with the conditioned medium from HT1080 cells containing non-cleaved IGFBP7 (Supplementary Figure S10a). DLD-1 cells were lentivirally transduced with matriptase shRNA for stable knockdown (Supplementary Figure S10b). The effect of IGFBP7 present in HT1080 conditioned medium was tested on DLD-1 cells either endogenously expressing matriptase (ST14) or not (ST14 knockdown; Supplementary Figure S10c). Increase in DLD-1 cell number was monitored using automated live cell image analysis. No differences in cell number were observed over time between matriptase expressing and knockdown cells in response to IGFBP7 containing conditioned medium (Supplementary Figure S10d).

Discussion

We have analyzed the molecular differences between CAFs and normal resting fibroblasts by gene expression profiling from microdissected colon cancer and corresponding normal tissues. Interestingly, many of the genes identified in our study have been reported in studies performed in vitro including the ‘wound response signature’ of fibroblasts in response to serum stimulation,12 a hypoxia-associated response32 as well as the signatures obtained from co-cultures of cancer cells and fibroblasts cell lines of different origins.33,34 Similar expression signatures have also been obtained using serial analysis of gene expression on antibody-sorted stromal components from breast cancer samples35 or laser capture microdissection in breast cancer and basal cell carcinoma of the skin.36

In our iLCM screen, CAFs showed increased expression of gene sets related to hypoxia, EMT and TGFβ pathway activation. TGFβ-induced gene sets contained several members of the insulin-like growth factor-binding proteins family, including IGFBP-3, -5 and -7. IGFBP-3 upregulation in the tumor stroma of prostate cancers has recently been demonstrated by global gene expression profiling following laser capture microdissection (LCM)37 and a role as mediator for tumor–stroma interactions has been suggested in this type of tumor.38 Both IGFBP-3 and -5 have been implicated in matrix deposition during fibrosis,39 which contributes to the early stages of malignant transformation. Moreover, IGFBP3 increases drug tolerance of tumor cells by promoting IGF-1R signaling.40 IGFBP7 belongs to a group of low-affinity IGF binders, which have been implicated in several biological roles independent of their IGF-binding ability.41 IGFBP7 is postulated to function as a tumor suppressor in carcinomas.21,42,43 For example, IGFBP7 displays certain tumor-suppressive functions in colon cancer44 and hepatocellular carcinoma.26,43 Approximately 25% of patients with hepatocellular carcinoma harbor a deletion of the IGFBP7 gene locus (LOH) and the IGFBP7 promotor is frequently methylated in lung and prostate cancer.45,46 The underlying mechanism of the putative tumor-suppressor function of IGFBP7 has not been elucidated in detail so far. A recent study suggests that in epithelial cells IGFBP7 might act as an IGF-1/2 antagonist that can block IGF-1R activation by binding to the receptor, thereby inhibiting cell growth and survival.22

However, there are also several reports that demonstrate that IGFBP7 can contribute to tumor progression. In glioblastomas, IGFBP7 promotes cancer cell growth and migration, and high IGFBP7 expression is correlated with decreased patient survival.47 Moreover, IGFBP7 promotes angiogenesis in these tumors and is induced in the tumor endothelial cells.20 IGFBP7 is elevated in invasive prostate neoplasms48 and in colon cancer.49 IGFBP7 has been shown to be part of a predictive signature of aggressive inflammatory breast cancer50 and to be associated with poor prognosis in colon carcinomas, when expressed by the tumor cells at the invasive front.51

Therefore, these results suggest a more complex mode of action of IGFBP7 in cancer development than previously assumed. In this study, we have found a potential mechanism, which can at least partially explain these discrepancies. First, we show that IGFBP7 reduced anchorage-independent colony formation and xenograft tumor growth when overexpressed in tumor cells exhibiting an epithelial phenotype such as the colon cancer cells (DLD-1) as demonstrated earlier.52 This phenotype is evident, when IGFBP7 function is analyzed on epithelial cells monocultures. However, we have demonstrated that IGFBP7 is a tumor stroma marker in several epithelial tumor types that is expressed by activated fibroblasts and tumor-associated vessels. Moreover, we provide first evidence that IGFBP7 is selectively expressed in malignant mesenchyme-derived tumor cells and in epithelial cells with a mesenchymal (EMT) phenotype, where IGFBP7 promotes anchorage-independent growth. These results are supported by the finding that IGFBP7 expression is induced in DLD-1 E-cadherin knockdown cells and by similar observations in a breast cancer model system.28 Thus, our findings suggest that the function of IGFBP7 might be dependent on the cancer type and/or dependent on the differentiation state of epithelial cancers. Most importantly, we show for the first time that stromal expression of IGFBP7 contributes to the anchorage-independent growth of carcinoma cells with an epithelial phenotype. This is most likely due to secondary paracrine factors, which are induced in the stromal cells by IGFBP7. Further experiments are required to establish the role of IGFBP7 in the tumor stroma during tumor progression. A study of colorectal cancer analyzed by immunohistochemistry51 and our analysis of mRNA expression in colon cancer samples with available clinical follow-up support the hypothesis that high IGFBP7 expression levels in tumors are associated with poor outcome in colon cancer.

Interestingly, the proteolytic processing of IGFBP7 by the type II transmembrane serine protease matriptase27 appears to be of minor importance in tumor-stroma crosstalk function of IGFBP7, as knockdown of matriptase in the tumor cells did not show differences in cell number in response to stromal IGFBP7 provided in the medium.

In summary, we demonstrated that in addition to its tumor-suppressor function, IGFBP7 can promote anchorage-independent growth of malignant mesenchymal cells and of epithelial cells with an EMT-phenotype when IGFBP7 is expressed by the tumor cells themselves. Expression of IGFBP7 in tumor-associated fibroblasts can also promote colony formation when epithelial tumor cells are co-cultured with IGFBP7-expressing CAFs by secondary paracrine tumor–stroma interactions.

Materials and methods

Tissues and iLCM

Human tumor samples from patients with colorectal cancer were obtained from the Department of Pathology, Medical University of Vienna. The samples were collected in accordance with the guidelines of the institutional ethics committee. Matched pairs of tumor and normal colonic mucosa were snap frozen within 30 min after surgical resection. Five-micrometer sections were stained with hematoxylin and eosin to assess tissue preservation and for histopathological evaluation. The samples were analyzed for their expression of FAPα followed by laser capture on a PixCell IIe System (Arcturus Engineering Inc., Mountain View, CA, USA) as previously described.19

RNA processing and global gene expression profiling

Total RNA was extracted from the captured cells with the Arcturus Pico Pure RNA Isolation Kit (Arcturus Engineering Inc.) and amplified and labeled with the MassageAmpII-Biotin Enhanced Kit (Ambion, Austin, TX, USA). Fragmented antisense-RNA (15 μg) was used for hybridization of the Human Genome U133 Plus 2.0 GeneChip arrays (Affymetrix). The arrays were hybridized and scanned using standard Affymetrix protocols. Microarray data were normalized using the Robust Multi-Array Analysis as implemented in Bioconductor.53,54 All analyses were performed with log2-transformed data. Hypothesis tests were performed using a modified t statistics with an empirical Bayes approach as implemented in Bioconductor LIMMA package.55 P-values were adjusted by the false discovery rate method of Benjamini and Hochberg.56 The 200 genes with the best F-statistic values were selected for hierarchical clustering using the Ward method. For gene set-enrichment analysis, we used two gene set collections from the Molecular Signature Database provided by the Broad Institute, namely the curated gene sets (C2) and the GO gene sets (C5).23 The core genes of statistically significant genes were used to calculate principal component analysis plots.

Cell lines

All cell lines were obtained from American Type Culture Collection: DLD-1 (CCL-221), HT1080 (CCL-121), Caki-1 (HTB-46), or SW480 (CCL-228), HT29 (HTB-38) and LS174T (CL-188), respectively.

Immunohistochemistry/immunofluorescence

Analysis of IGFBP7 expression in tissue samples was performed on paraffin sections using the avidin-biotin complex immunoperoxidase method6 with a primary goat anti-IGFBP7 antibody obtained from R&D Systems (Minneapolis, MA, USA, Cat. # AF 1334) at 0.5 μg/ml. Epitope retrieval was carried out in proteinase K solution (20 μg/ml in TE Buffer; pH 8.0). Co-localization studies were done on acetone/methanol-fixed frozen sections or on cells on chamber slides by immunofluorescence. The following antibodies were used: monoclonal antibody F19 1:50,6 monoclonal antibody H572 1:20,24 rabbit anti-podoplanin 1:1000,57 goat anti-NG2 (R&D Systems) 1:100, monoclonal antibody V9 mouse anti-vimentin (Invitrogen, Carlsbad, CA, USA) 1:1000, rabbit anti-β-catenin 1:1000 (ab6302, Abcam, Cambridge, UK). Detection was performed with the following secondary antibodies: AlexaFluor 594 donkey anti-goat; AlexaFluor 594 goat anti-rabbit and AlexaFluor 488 goat anti-rabbit from Molecular Probes (Invitrogen).

Generation of stable IGFBP7/E-cadherin knockdowns

For shRNA-mediated IGFBP7 knockdown, the constructs (TRC0000077943 to TRC0000077947 in pLK0.1) developed by the RNAi consortium (TRC) from Open Biosystems (Thermo Fisher Scientific, Waltham, MA, USA) were packaged into lentiviral particles and used to infect HT1080, Caki-1 or SW480 cell lines. The colon cancer cell line DLD-1was used for E-cadherin knockdown using the SMARTvector 2.0 lentiviral particles from Dharmacon (Thermo Fisher Scientific) as described by the manufacturer. For all constructs and cell lines, selection was carried out with puromycin (Invitrogen) at 1 μg/ml final concentration. All work was done according to local biosafety regulations.

IGFBP7 overexpression

IGFBP7 Lentifect lentiviral particles (GeneCopoeia, Rockville, MD, USA) were used to overexpress the protein in DLD-1 cells. The infection and selection process was carried out as described in the previous section.

Real-time PCR

Total RNA was extracted using the QiagenRNeasy Mini Kit (Qiagen Hilden, Germany) following the manufacturer's instruction and first strand cDNA was synthesized with the Applied Biosystems' High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Carlsbad, CA, USA). TaqMan probes and primers for IGFBP7 and glyceraldehyde 3-phosphate dehydrogenase were obtained from Applied Biosystems. TaqMan PCR was done with an ABI PRISM 7000 Sequence Detection System (Applied Biosystems) according to the manufacturer’s instructions. The relative expression of IGFBP7 mRNA was normalized to the amount of glyceraldehyde 3-phosphate dehydrogenase in the same cDNA using the comparative CT method described by the manufacturer.

Immunoblotting

Conditioned media obtained from Caki-1, HT1080, SW480, LS174T, HT29 and DLD-1 cells were concentrated with a centricon centrifugal filter device (Millipore, Billerica, MA, USA) and used for IGFBP7 detection by immunoblotting with goat anti-IGFBP7 antibody (R&D Systems) at 0.1 μg/ml. In addition, the following antibodies were used for immunoblotting: V9 mouse anti-vimentin 1:10.000 and HECD-1 mouse anti-E-Cadherin 1:1000 (both from Abcam); rabbit anti-matriptase 1:1000 from Bethyl Laboratories (Montgomery, TX, USA); rabbit anti-glyceraldehyde 3-phosphate dehydrogenase 1:25.000 from Trevigen (Gaithersburg, MD, USA).

Soft agar colony formation assay

Anchorage-independent cell growth was analyzed using the colony formation assay in soft agar culture. Single cells were suspended in standard medium containing 0.4% (w/v) low-melting agarose (Invitrogen) and plated at a cell density of 3 × 104 cells/dish on six-well plates containing solidified 1.2% agar. After 14 days of incubation, the cell colonies were fixed in acetone/methanol and stained with 0.005% crystal violet. The stained cell colonies were photographed and counted with ImageJ. Anchorage-independent cancer cell growth in response to stromal cells was analyzed using a modified soft agar assay. CAFs were seeded at a density of 7 × 104 cells per dish on a 12-well plate in EGM containing 1 μg/ml puromycin. After overnight attachment, cells were washed twice with 1x phosphate-buffered saline and overlaid by plain Dulbecco’s modified Eagle’s medium containing 1.2% (w/v) low-melting agarose (Sigma-Aldrich, St Louis, MO, USA). After solidification of the base agar, DLD-1 single cells were suspended in plain Dulbecco’s modified Eagle’s medium containing 0.4% (w/v) low-melting agarose and plated at a cell density of 2 × 103 cells per well. To ease the paracrine crosstalk between stromal and tumor cells, 2 ml of Dulbecco’s modified Eagle’s medium containing 1% fetal calf serum was added to each well. After 14 days of incubation, colonies were fixed in 4% paraformaldehyde (PFA) and stained with 0.01% crystal violet. The stained cell colonies were photographed and counted with ImageJ.

Xenografts

HT1080 cells were harvested and suspended in phosphate-buffered saline/1% bovine serum albumin. 100 μl of the cell suspension containing 108 cells were injected into the flanks of SCID mice (Charles River, Wilmington, MA, USA). Tumor sizes were measured every 2 days after the tumors became visible.

Cytokine profiling of stromal fibroblasts

The cytokine profile was established following the instruction manual (Human Cytokine Array Kit, R&D Systems). In brief, CAFs (CAF3-shC and CAF3-shIGFBP7) were grown to confluency and fresh EGM2-MV (EGM) was added. After 72 h, media were removed, sterile filtered and 1ml was used for profiling. ImageJ was used for densitometry analysis.

IGFBP7 mRNA expression: data mining

Gene expression data from human colon cancer samples30 were analyzed using the Oncomine Premium Research Edition database to correlate the overall survival with the mRNA expression levels of IGFBP7. The patients were classified into four categories according to their IGFBP7 expression levels: low (group 1, 45 patients), intermediate low (group 2, 44 patients), intermediate high (group 3, 44 patients) and high (group 4, 44 patients). Details of the standardized normalization techniques and statistical calculations can be found on the Oncomine website (https://www.oncomine.com;31).

References

  1. 1

    Bhowmick NA, Chytil A, Plieth D, Gorska AE, Dumont N, Shappell S et al. TGF-beta signaling in fibroblasts modulates the oncogenic potential of adjacent epithelia. Science 2004; 303: 848–851.

  2. 2

    Mueller MM, Fusenig NE . Friends or foes - bipolar effects of the tumour stroma in cancer. Nat Rev Cancer 2004; 4: 839–849.

  3. 3

    Rasanen K, Vaheri A . Activation of fibroblasts in cancer stroma. Exp Cell Res 2010; 316: 2713–2722.

  4. 4

    Kunz-Schughart LA, Knuechel R . Tumor-associated fibroblasts (part I): Active stromal participants in tumor development and progression? Histol Histopathol 2002; 17: 599–621.

  5. 5

    Joyce JA, Pollard JW . Microenvironmental regulation of metastasis. Nat Rev Cancer 2009; 9: 239–252.

  6. 6

    Garin-Chesa P, Old LJ, Rettig WJ . Cell surface glycoprotein of reactive stromal fibroblasts as a potential antibody target in human epithelial cancers. Proc Natl Acad Sci USA 1990; 87: 7235–7239.

  7. 7

    Desmouliere A, Guyot C, Gabbiani G . The stroma reaction myofibroblast: a key player in the control of tumor cell behavior. Int J Dev Biol 2004; 48: 509–517.

  8. 8

    Grum-Schwensen B, Klingelhofer J, Berg CH, El-Naaman C, Grigorian M, Lukanidin E et al. Suppression of tumor development and metastasis formation in mice lacking the S100A4(mts1) gene. Cancer research 2005; 65: 3772–3780.

  9. 9

    Ostman A, Heldin CH . PDGF receptors as targets in tumor treatment. Adv Cancer Res 2007; 97: 247–274.

  10. 10

    Orimo A, Weinberg RA . Heterogeneity of stromal fibroblasts in tumors. Cancer Biol Ther 2007; 6: 618–619.

  11. 11

    Mani SA, Guo W, Liao MJ, Eaton EN, Ayyanan A, Zhou AY et al. The epithelial-mesenchymal transition generates cells with properties of stem cells. Cell 2008; 133: 704–715.

  12. 12

    Chang HY, Sneddon JB, Alizadeh AA, Sood R, West RB, Montgomery K et al. Gene expression signature of fibroblast serum response predicts human cancer progression: similarities between tumors and wounds. PLoS Biol 2004; 2: E7.

  13. 13

    Finak G, Bertos N, Pepin F, Sadekova S, Souleimanova M, Zhao H et al. Stromal gene expression predicts clinical outcome in breast cancer. Nat Med 2008; 14: 518–527.

  14. 14

    Navab R, Strumpf D, Bandarchi B, Zhu CQ, Pintilie M, Ramnarine VR et al. Prognostic gene-expression signature of carcinoma-associated fibroblasts in non-small cell lung cancer. Proc Natl Acad Sci USA 2011; 108: 7160–7165.

  15. 15

    Herrera M, Herrera A, Dominguez G, Silva J, Garcia V, Garcia JM et al. Cancer-Associated Fibroblast and M2 Macrophage markers together predict outcome in colorectal cancer patients. Cancer Sci 2013; 104: 437–444.

  16. 16

    Ostermann E, Garin-Chesa P, Heider KH, Kalat M, Lamche H, Puri C et al. Effective immunoconjugate therapy in cancer models targeting a serine protease of tumor fibroblasts. Clin Cancer Res 2008; 14: 4584–4592.

  17. 17

    Santos AM, Jung J, Aziz N, Kissil JL, Pure E . Targeting fibroblast activation protein inhibits tumor stromagenesis and growth in mice. J Clin Invest 2009; 119: 3613–3625.

  18. 18

    Pure E . The road to integrative cancer therapies: emergence of a tumor-associated fibroblast protease as a potential therapeutic target in cancer. Expert Opin Ther Targets 2009; 13: 967–973.

  19. 19

    Rupp C, Dolznig H, Puri C, Schweifer N, Sommergruber W, Kraut N et al. Laser capture microdissection of epithelial cancers guided by antibodies against fibroblast activation protein and endosialin. Diagn Mol Pathol 2006; 15: 35–42.

  20. 20

    Pen A, Moreno MJ, Durocher Y, Deb-Rinker P, Stanimirovic DB . Glioblastoma-secreted factors induce IGFBP7 and angiogenesis by modulating Smad-2-dependent TGF-beta signaling. Oncogene 2008; 27: 6834–6844.

  21. 21

    Ruan W, Xu E, Xu F, Ma Y, Deng H, Huang Q et al. IGFBP7 plays a potential tumor suppressor role in colorectal carcinogenesis. Cancer Biol Ther 2007; 6: 354–359.

  22. 22

    Evdokimova V, Tognon CE, Benatar T, Yang W, Krutikov K, Pollak M et al. IGFBP7 binds to the IGF-1 receptor and blocks its activation by insulin-like growth factors. Sci Signal 2012; 5 ra92.

  23. 23

    Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci USA 2005; 102: 15545–15550.

  24. 24

    Christian S, Ahorn H, Novatchkova M, Garin-Chesa P, Park JE, Weber G et al. Molecular cloning and characterization of EndoGlyx-1, an EMILIN-like multisubunit glycoprotein of vascular endothelium. J Biol Chem 2001; 276: 48588–48595.

  25. 25

    Ozerdem U, Grako KA, Dahlin-Huppe K, Monosov E, Stallcup WB . NG2 proteoglycan is expressed exclusively by mural cells during vascular morphogenesis. Dev Dyn 2001; 222: 218–227.

  26. 26

    Ahmed S, Jin X, Yagi M, Yasuda C, Sato Y, Higashi S et al. Identification of membrane-bound serine proteinase matriptase as processing enzyme of insulin-like growth factor binding protein-related protein-1 (IGFBP-rP1/angiomodulin/mac25). FEBS J 2006; 273: 615–627.

  27. 27

    Ahmed S, Yamamoto K, Sato Y, Ogawa T, Herrmann A, Higashi S et al. Proteolytic processing of IGFBP-related protein-1 (TAF/angiomodulin/mac25) modulates its biological activity. Biochem Biophys Res Commun 2003; 310: 612–618.

  28. 28

    Onder TT, Gupta PB, Mani SA, Yang J, Lander ES, Weinberg RA . Loss of E-cadherin promotes metastasis via multiple downstream transcriptional pathways. Cancer Res 2008; 68: 3645–3654.

  29. 29

    Dolznig H, Rupp C, Puri C, Haslinger C, Schweifer N, Wieser E et al. Modeling colon adenocarcinomas in vitro a 3D co-culture system induces cancer-relevant pathways upon tumor cell and stromal fibroblast interaction. Am J Pathol 2011; 179: 487–501.

  30. 30

    Smith JJ, Deane NG, Wu F, Merchant NB, Zhang B, Jiang A et al. Experimentally derived metastasis gene expression profile predicts recurrence and death in patients with colon cancer. Gastroenterology 2010; 138: 958–968.

  31. 31

    Rhodes DR, Yu J, Shanker K, Deshpande N, Varambally R, Ghosh D et al. ONCOMINE: a cancer microarray database and integrated data-mining platform. Neoplasia 2004; 6: 1–6.

  32. 32

    Chi JT, Wang Z, Nuyten DS, Rodriguez EH, Schaner ME, Salim A et al. Gene expression programs in response to hypoxia: cell type specificity and prognostic significance in human cancers. PLoS Med 2006; 3: e47.

  33. 33

    Sato N, Maehara N, Goggins M . Gene expression profiling of tumor-stromal interactions between pancreatic cancer cells and stromal fibroblasts. Cancer Res 2004; 64: 6950–6956.

  34. 34

    Gallagher PG, Bao Y, Prorock A, Zigrino P, Nischt R, Politi V et al. Gene expression profiling reveals cross-talk between melanoma and fibroblasts: implications for host-tumor interactions in metastasis. Cancer Res 2005; 65: 4134–4146.

  35. 35

    Allinen M, Beroukhim R, Cai L, Brennan C, Lahti-Domenici J, Huang H et al. Molecular characterization of the tumor microenvironment in breast cancer. Cancer Cell 2004; 6: 17–32.

  36. 36

    Casey T, Bond J, Tighe S, Hunter T, Lintault L, Patel O et al. Molecular signatures suggest a major role for stromal cells in development of invasive breast cancer. Breast Cancer Res Treat 2009; 114: 47–62.

  37. 37

    Gregg JL, Brown KE, Mintz EM, Piontkivska H, Fraizer GC . Analysis of gene expression in prostate cancer epithelial and interstitial stromal cells using laser capture microdissection. BMC Cancer 2010; 10: 165.

  38. 38

    Massoner P, Haag P, Seifarth C, Jurgeit A, Rogatsch H, Doppler W et al. Insulin-like growth factor binding protein-3 (IGFBP-3) in the prostate and in prostate cancer: local production, distribution and secretion pattern indicate a role in stromal-epithelial interaction. Prostate 2008; 68: 1165–1178.

  39. 39

    Pilewski JM, Liu L, Henry AC, Knauer AV, Feghali-Bostwick CA . Insulin-like growth factor binding proteins 3 and 5 are overexpressed in idiopathic pulmonary fibrosis and contribute to extracellular matrix deposition. Am J Pathol 2005; 166: 399–407.

  40. 40

    Sharma SV, Lee DY, Li B, Quinlan MP, Takahashi F, Maheswaran S et al. A chromatin-mediated reversible drug-tolerant state in cancer cell subpopulations. Cell 2010; 141: 69–80.

  41. 41

    Firth SM, Baxter RC . Cellular actions of the insulin-like growth factor binding proteins. Endocr Rev 2002; 23: 824–854.

  42. 42

    Vizioli MG, Sensi M, Miranda C, Cleris L, Formelli F, Anania MC et al. IGFBP7: an oncosuppressor gene in thyroid carcinogenesis. Oncogene 2010; 29: 3835–3844.

  43. 43

    Tomimaru Y, Eguchi H, Wada H, Kobayashi S, Marubashi S, Tanemura M et al. IGFBP7 downregulation is associated with tumor progression and clinical outcome in hepatocellular carcinoma. Int J Cancer 2012; 130: 319–327.

  44. 44

    Lin J, Lai M, Huang Q, Ruan W, Ma Y, Cui J . Reactivation of IGFBP7 by DNA demethylation inhibits human colon cancer cell growth in vitro. Cancer Biol Ther 2008; 7: 1896–1900.

  45. 45

    Chen Y, Cui T, Knosel T, Yang L, Zoller K, Petersen I . IGFBP7 is a p53 target gene inactivated in human lung cancer by DNA hypermethylation. Lung Cancer 2011; 73: 38–44.

  46. 46

    Sullivan L, Murphy TM, Barrett C, Loftus B, Thornhill J, Lawler M et al. IGFBP7 promoter methylation and gene expression analysis in prostate cancer. J Urol 2012; 188: 1354–1360.

  47. 47

    Jiang W, Xiang C, Cazacu S, Brodie C, Mikkelsen T . Insulin-like growth factor binding protein 7 mediates glioma cell growth and migration. Neoplasia 2008; 10: 1335–1342.

  48. 48

    Degeorges A, Wang F, Frierson HF Jr., Seth A, Chung LW, Sikes RA . Human prostate cancer expresses the low affinity insulin-like growth factor binding protein IGFBP-rP1. Cancer Res 1999; 59: 2787–2790.

  49. 49

    van Beijnum JR, Dings RP, van der Linden E, Zwaans BM, Ramaekers FC, Mayo KH et al. Gene expression of tumor angiogenesis dissected: specific targeting of colon cancer angiogenic vasculature. Blood 2006; 108: 2339–2348.

  50. 50

    Bieche I, Lerebours F, Tozlu S, Espie M, Marty M, Lidereau R . Molecular profiling of inflammatory breast cancer: identification of a poor-prognosis gene expression signature. Clin Cancer Res 2004; 10: 6789–6795.

  51. 51

    Adachi Y, Itoh F, Yamamoto H, Arimura Y, Kikkawa-Okabe Y, Miyazaki K et al. Expression of angiomodulin (tumor-derived adhesion factor/mac25) in invading tumor cells correlates with poor prognosis in human colorectal cancer. Int J Cancer 2001; 95: 216–222.

  52. 52

    Sato Y, Chen Z, Miyazaki K . Strong suppression of tumor growth by insulin-like growth factor-binding protein-related protein 1/tumor-derived cell adhesion factor/mac25. Cancer Sci 2007; 98: 1055–1063.

  53. 53

    Irizarry RA, Bolstad BM, Collin F, Cope LM, Hobbs B, Speed TP . Summaries of Affymetrix GeneChip probe level data. Nucleic Acids Res 2003; 31: e15.

  54. 54

    Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S et al. Bioconductor: open software development for computational biology and bioinformatics. Genome Biol 2004; 5: R80.

  55. 55

    Smyth GK . Linear models and empirical bayes methods for assessing differential expression in microarray experiments. Stat Appl Genet Mol Biol 2004; 3: Article3.

  56. 56

    Benjamini Y, Drai D, Elmer G, Kafkafi N, Golani I . Controlling the false discovery rate in behavior genetics research. Behav Brain Res 2001; 125: 279–284.

  57. 57

    Breiteneder-Geleff S, Soleiman A, Kowalski H, Horvat R, Amann G, Kriehuber E et al. Angiosarcomas express mixed endothelial phenotypes of blood and lymphatic capillaries: podoplanin as a specific marker for lymphatic endothelium. Am J Pathol 1999; 154: 385–394.

Download references

Acknowledgements

This work was supported by Boehringer Ingelheim Austria. We are grateful to Christina Puri, Daniela Milovanovic, Oliver Bergner and Jakob Schnabl for help with immunohistochemistry, cell culture and in vitro assays.

Author information

Correspondence to H Dolznig or P Garin-Chesa.

Ethics declarations

Competing interests

The authors declare no conflict of interest.

Additional information

Supplementary Information accompanies this paper on the Oncogene website

Supplementary information

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Rupp, C., Scherzer, M., Rudisch, A. et al. IGFBP7, a novel tumor stroma marker, with growth-promoting effects in colon cancer through a paracrine tumor–stroma interaction. Oncogene 34, 815–825 (2015). https://doi.org/10.1038/onc.2014.18

Download citation

Further reading