Variable FZD7 expression in colorectal cancers indicates regulation by the tumour microenvironment

Authors


Abstract

Recent evidence shows that a sub-population of Wnt/β-catenin target genes is specifically induced in different tissue contexts. FZD7 is a putative Wnt/β-catenin target gene and although it is highly expressed in well-differentiated colorectal cancer tumour cells, its expression is decreased in de-differentiated tumour cells at the invasive front despite elevated Wnt/β-catenin signalling in this area. This variable expression of FZD7 implicates additional regulation by the microenvironment; however, this has not been investigated. To begin to elucidate the role of extracellular matrix in regulating FZD7 expression, we generated a FZD7 promoter reporter and analysed FZD7 promoter activity in colorectal cancer cells grown on different matrices. We demonstrate that the FZD7 promoter is regulated by β-catenin in colorectal cancer cells and observed decreased promoter activity in cells grown on fibronectin but not collagen I or collagen IV. Thus, expression of FZD7 in colorectal cancer may be regulated by fibronectin in the microenvironment. Developmental Dynamics 239:311–317, 2010. © 2009 Wiley-Liss, Inc.

INTRODUCTION

The coupling of Wnt growth factors and their receptors, the seven-pass transmembrane Frizzled (FZD) molecules, activates the transcription of gene programs that are critical for normal gut development and homeostatic control of the adult intestine. Wnt/β-catenin signalling activates different gene programs in the different cell types of the adult epithelium (for, e.g., Paneth cell differentiation, transit amplifying cell, or stem cell programs) (Clevers,2006). Wnt/β-catenin signalling is tightly controlled in the normal intestine. Loss of homeostatic control in the crypt through constitutive activation of Wnt/β-catenin signalling results in the persistence of the crypt stem cell gene program (van de Wetering et al.,2002) and subsequent adenoma (polyp) formation. Adenomas can then progress to carcinoma and metastasis (Kinzler and Vogelstein,1996; Brabletz et al.,2005b; Clevers,2006).

Constitutive activation of the Wnt/β-catenin pathway appears to be a necessary initiating event in the genesis of most colorectal cancers. Constitutive activation occurs through genetic mutation of one of the downstream components of the signalling pathway. This is most commonly the APC tumour suppressor gene (Korinek et al.,1997) and less often CTNNB1 (encoding β-catenin) (Korinek et al.,1997; Morin et al.,1997) or AXIN2 (encoding axin-2) (Liu et al.,2000). Additionally, epigenetic silencing of naturally occurring inhibitors of Wnt-FZD interaction results in further activation of the Wnt/β-catenin pathway (reviewed in Vincan and Barker,2008). Inhibitors of the Wnt-FZD receptor complex, such as the secreted FZD-related proteins (sFRP) (Caldwell et al.,2004; Suzuki et al.,2004), Dickkopf (DKK) (Aguilera et al.,2006) and Wnt inhibitory factor (WIF) (Taniguchi et al.,2005), are bonafide tumour suppressors in colorectal cancer. Notably, silencing of Wnt-FZD inhibitors is an early event, often preceding APC mutation (Caldwell et al.,2004; Suzuki et al.,2004). This implies that signalling from the Wnt-FZD ligand-receptor complex is required to complement downstream mutations in the same pathway in the genesis and evolution of colorectal cancer. Moreover, epigenetic regulation of Wnt-FZD inhibitors implies dynamic regulation of the pathway.

Indeed, Wnt/β-catenin signalling is dynamically regulated within colorectal carcinomas. Nuclear localisation of β-catenin, the hallmark of active Wnt/β-catenin signalling, is dramatically upregulated in the de-differentiated cells at invasive areas of colorectal carcinomas (Brabletz et al.,2001). This transition towards a more mesenchymal phenotype, termed epithelial to mesenchymal transition (EMT), is thought to initiate tumour cell invasion and metastasis. Several Wnt/β-catenin signalling target genes associated with tumour cell invasion are upregulated at the invasive front (Brabletz et al.,2005a). In contrast, although FZD7 is overexpressed in colorectal cancer (Sagara et al.,1998; Vincan et al.,2007b; Ueno et al.,2008) and is a putative Wnt/β-catenin target gene, its expression is downregulated in de-differentiated tumour cells (Vincan et al.,2007b). This variable expression indicates potential regulation by the tumour cell microenvironment.

Indirect evidence indicates that FZD7 is a Wnt/β-catenin target gene. FZD7 is upregulated in response to Wn3a in human embryonic carcinoma cells (Willert et al.,2002) and is upregulated in certain Wnt-driven gene programs, especially stem-cell gene programs (Gregorieff et al.,2005; Assou et al.,2007; Dormeyer et al.,2008; Melchior et al.,2008; Vijayaragavan et al.,2009). However, FZD7 has not been more directly demonstrated to be a Wnt/β-catenin target gene. To begin to understand possible mechanisms of FZD7 regulation, we generated a human FZD7 promoter reporter to investigate transcriptional regulation. We show that FZD7 reporter activity is increased in a dose-dependent manner by β-catenin. On the converse, FZD7 reporter activity is decreased by knock-down of β-catenin. Moreover, FZD7 reporter activity is partially down-regulated by fibronectin but not by collagen I or collagen IV, which indicates that some extracellular matrix (ECM) components at the invasive front may influence transcriptional regulation of FZD7.

RESULTS AND DISCUSSION

FZD7 Expression Is Decreased at the Invasive Front of Colorectal Cancer

Wnt/β-catenin signalling is deregulated in colorectal cancer through genetic mutation of the APC gene or, less frequently, β-catenin or Axin (Morin et al.,1997; Clevers,2006). As these mutations were shown to lead to constitutive activation of the pathway (Morin et al.,1997; Clevers,2006), and the hallmark of active Wnt/β-catenin signalling is nuclear localisation of β-catenin, it follows then that β-catenin should be nuclear in all tumour cells. However, immunohistochemical analysis of colorectal cancers reveals a very variable intracellular distribution of β-catenin that tracks closely with tumour cell differentiation status. In the moderate to well-differentiated colorectal tumour at the tumour centre, β-catenin is predominantly membranous, although some β-catenin must be in the nucleus since TCF/β-catenin transcription is constitutively activated in these cells through genetic mutation (Clevers,2006). In contrast, cytoplasmic and nuclear β-catenin is dramatically increased in the de-differentiated mesenchymal tumour cells at invasive areas of the carcinomas (Brabletz et al.,2001). Notably, tumour cells with nuclear β-catenin undergo cell cycle arrest (Jung et al.,2001; Wassermann et al.,2009), while in contrast, their expression of factors associated with cell invasion and migration [for, e.g., MMP7 and lamini-5 γ2 chain (Lamγ2)] are up-regulated (Brabletz et al.,2005b).

FZD7 is overexpressed in diverse human cancers (reviewed in Vincan,2004; Vincan and Barker,2008) and, as we have shown previously (Vincan et al.,2007b), FZD7 is highly expressed in colorectal cancer cells at the tumour centre (Fig. 1E) but is down-regulated in the tumour cells that are interspersed with the stroma at the invasive front (Fig. 1B, I). As expected, the invasive tumour cells show strong nuclear β-catenin localisation (Fig. 1A, G) and cell cycle arrest (Fig. 1C, H) when compared to cells at the tumour centre (Fig. 1D and F, respectively). Decreased FZD7 expression at the invasive front was a surprising result as several lines of evidence indicate that FZD7 is a Wnt/β-catenin target gene. FZD7 expression is induced by Wnt3a (Willert et al.,2002) and it features in Wnt-driven gene programs (e.g., stem cell gene programs) (Assou et al.,2007). However, we observed decreased FZD7 expression despite active Wnt/β-catenin signalling (Fig. 1). Notably, in the intestinal epithelium, FZD7 expression tracks closely with cycling cells with active Wnt/β-catenin signalling (Gregorieff et al.,2005; Mariadason et al.,2005). It would appear then that in colorectal cancer, FZD7 expression also tracks with the cycling tumour cells with active (but sub-maximal when compared to the invasive front) Wnt/β-catenin signalling (Fig. 1E, F), and the variable distribution of FZD7 in colorectal cancer indicates that FZD7 expression may be regulated by factors in the tumour microenvironment. This prompted us to generate a FZD7 promoter reporter to enable the investigation of transcriptional regulation of FZD7 expression in the context of colorectal cancer.

Figure 1.

FZD7 expression is down-regulated at the invasive front of tumours. Immunohistochemical detection of β-catenin (A, D, G), FZD7 (B, E, I), and Ki-67 (C, F, H) in serial colorectal carcinoma sections. Sections in A to F are from one carcinoma, while G to I are from another. The tubular patterning of the carcinoma is disrupted at the invasive front where nuclear localisation of β-catenin is increased (A), while FZD7 (B) and Ki-67 (C) are decreased (boxed area is shown at higher magnification in a–c, respectively) when compared to expression at the tumour centre (E and F, respectively). In contrast, β-catenin is mostly in the cytoplasm and the cell membrane at the tumour centre (D). The insets in D and E show higher magnification of the boxed areas. Similarly, tumour cells interspersed with the stoma show nuclear localisation β-catenin (G), while Ki-67 (H) and FZD7 (I) are decreased (the inset in I is FZD7 expression in more differentiated tumour cells in the same tissue).

FZD7 Promoter Activity Is Regulated by β-Catenin

We searched for potential TCF-binding elements (TBE) in the promoter/enhancer sequence of human FZD7. Three sites comprising the TBE motif (5′-WWCAAAG-3′) were identified (nt -1,095/-1,088, nt -902/-896, n-285/-278). The functional relevance of TBE motives in the human FZD7 promoter was investigated in luciferase reporter assays using the full-length promoter we generated (Fig. 2A). To test regulation of the FZD7 promoter-reporter (pFz7-prom) by β-catenin, HeLa cells were co-transfected with a TCF4 expression plasmid, the pFz7-prom or the synthetic β-catenin/TCF4 reporter plasmid, TOPflash (Korinek et al.,1997), and a β-catenin expression plasmid at the given amounts (Fig. 2). β-catenin induced dose-dependent activation of the pFz7-prom and TOPflash reporters (Fig. 2B). The level of transactivation relative to control plasmids (FOPflash or pGL) is shown in Figure 2C. Similarly, the pFz7-prom reporter was active in human colorectal cancer cell lines SW480 and DLD1 that have a constitutively active β-catenin/TCF4 pathway (Fig. 3A). Thus, endogenous factors in these cell lines can activate FZD7 promoter activity. On the converse, introduction of siRNA to knockdown β-catenin expression in DLD1 (Fig. 3A, B) and SW480 (Fig. 3A, C) cells resulted in decreased FZD7 promoter activity. Notably, the decrease in TOPflash promoter activity after siRNA depletion of β-catenin was more marked than the decrease in FZD7 promoter activity in both cell lines. This may indicate additional regulation of the FZD7 promoter by factors other than β-catenin/TCF4. Specific siRNA-mediated knockdown of β-catenin was confirmed by standard immunoblot (Fig. 3D), while FZD7 expression in DLD1 and SW480 cells was confirmed by quantitative RT-PCR (Fig. 3E). Taken together, these data indicate that the FZD7 promoter is regulated by β-catenin/TCF4 in colorectal cancer cells and is constitutively active in these cells. Although reporter activity was comparable in the SW480 and DLD1 cells (Fig. 3A), FZD7 expression in SW480 was approximately threefold higher than the DLD1 cells. The mechanism underlying this difference is not known but may indicate more complex transcriptional regulation than is revealed by the Fz7 reporter.

Figure 2.

Up-regulation of FZD7 promoter activity by β-catenin. A: Scheme of the human FZD7 promoter reporter plasmid, comprising nucleotides −1,116 to +41 (relative to FZD7 transcription initiation site). The promoter region contains three TCF binding elements (TCF); corresponding sequences and locations are indicated. B, C: HeLa cells were transfected with the FZD7 promoter reporter (Fz7 prom) or TOPflash, as well as with expression vectors for TCF4 and β-catenin or empty vectors, using increasing amounts of the β-catenin plasmid as indicated. FZD7 promoter activity and TOPflash are increased by β-catenin in a dose-dependent manner (B). The fold increase in reporter activity relative to cells transfected without β-catenin (transactivation) is shown in C. Bars indicate SE of a representative experiment performed in triplicate.

Figure 3.

Down-regulation of FZD7 promoter activity by siRNA-mediated knock-down of β-catenin. DLD-1 and SW480 cells were transfected with the FZD7 promoter reporter (Fz7prom) (A–C) or TOPflash (B, C) and siRNA targeting β-catenin or GFP as control. FZD7 promoter activity and TOPflash are decreased by β-catenin siRNA in both DLD-1 and SW480 cells. Bars indicate SE of a representative experiment performed in duplicate. D: Specific knockdown of β-catenin was shown by standard immunoblot of indicated cell lysates (5 μg per lane) detected with a mAb against β-catenin and an anti β-actin mAb as control. E: Expression of FZD7 in SW480 and DLD1 cells was shown by quantitative RT-PCR. Bars indicate SE of three cell preparations (SW480 and DLD1) or a representative sample (SW620) assayed in triplicate PCR reactions.

FZD7 Promoter Activity Is Partially Down-Regulated on Fibronectin

Our gain-and-loss-of-function experiments indicate the FZD7 promoter responds to β-catenin/TCF4 (Figs. 2, 3), which was at odds with the decreased FZD7 protein expression we observed at the invasive front of colorectal cancer tissues (Fig. 1). This prompted us to investigate the effect extracellular matrix components might have on the FZD7 promoter. SW480 cells were seeded on tissue culture plates pre-coated with fibronectin, collagen I, or collagen IV and the effect this had on pFz7-prom and TOPflash reporter activity was assessed. We observed a modest decrease in pFz7-prom (Fig. 4A) reporter activity when the cells were grown on fibronectin but not collagen or collagen IV. Fibronectin is itself a β-catenin/TCF4 target gene (Gradl et al.,1999) and is upregulated at the invasive front (Kirchner and Brabletz,2000). The collagens are also upregulated at the invasive front (Hlubek et al.,2007). However, these data indicate that fibronectin but not collagen could potentially contribute to the transcriptional regulation of FZD7 expression in colorectal cancer.

Figure 4.

Down-regulation of FZD7 promoter and TOPflash activity by fibronectin. SW480 cells were grown in tissue culture plates (Nil) or plates pre-coated with fibronectin (FN), collagen I (Col I), or collagen IV (Col IV) and transfected with the FZD7 promoter reporter (Fz7prom) (A), TOPflash (B) or control plasmids pGL (pBasic) (A) and FOPflash (B). FZD7 promoter activity and TOPflash are decreased by fibronectin but not collagen I or collagen IV. Bars indicate SE of a representative experiment performed in duplicate (*P < 0.05).

The decreased TOPflash reporter activity seen on fibronectin (Fig. 4B) confirms that TCF4/β-catenin target genes can be negatively regulated by fibronectin. This is consistent with the decreased FZD7 reporter activity on fibronectin. As for any β-catenin/TCF4 target gene at the invasive front or other tissue contexts, the ultimate outcome of transcriptional regulation is the sum result of all contributing factors. In the case of FZD7, this negative regulation appears to translate to decreased expression. In contrast; for invasive front β-catenin/TCF4 target genes (e.g., MMP7, Lamγ2), the net outcome is increased expression. Consequently, only a sub-population of β-catenin/TCF4 target genes is induced in any one context.

Potential Role of FZD7 in Colorectal Cancer

Our transfection experiments with a reporter plasmid containing the human FZD7 promoter indicate that FZD7 is a β-catenin/TCF4 target gene. We provide direct evidence that FZD7 is transcriptionally regulated by β-catenin/TCF4. Furthermore, we demonstrate experimentally that FZD7 promoter activity is adaptable and may be dependent on the microenvironment surrounding the tumour cells. Our results implicate a role for fibronectin but not collagen in this regulation. We have shown previously that FZD7 appears to direct carcinoma tubular patterning (Vincan et al.,2007a,b), a finding that is consistent with confined FZD7 protein expression in moderate to well-differentiated cells at the tumour centre. Given that fibronectin is involved in FZD7-mediated cell migration (Munoz et al.,2006), negative regulation of FZD7 expression at the invasive front by fibronectin would provide a complementary feedback. The regulation of FZD7 by the microenvironment might be relevant during the process of metastatic dissemination when the gene is transiently turned down in invasive cells to allow escape from tubular patterning (Fig. 5). Receptors other than FZD7 presumably transmit Wnt signals at the invasive front.

Figure 5.

FZD7 is a context-defined β-catenin/TCF target gene in colorectal cancer. Many β-catenin/TCF target genes have been identified; however, only a subset of genes is induced in a given context. Our data indicate that expression of FZD7 is negatively regulated at the invasive front of colorectal carcinomas. Negative regulation at the invasive front is consistent with the putative role of FZD7 in tubular patterning in this carcinoma.

EXPERIMENTAL PROCEDURES

Tissue Specimens and Immunohistochemistry

Formalin-fixed, paraffin-embedded samples of colorectal carcinomas from patients who underwent surgery without additional treatments were retrieved from the archives of the Department of Pathology, University of Erlangen-Nürnberg. Immunohistochemistry on human tumours was done as previously described (Vincan et al.,2007b). Briefly, serial 3-μM sections were immunostained after antigen retrieval by heating in TRIS/EDTA buffer. The following antibodies (Abs) were used: isotype control mouse monoclonal Ab, rabbit polyclonal control Ab, Envision Plus rabbit and mouse detection kit were from Dako Corporation (Glostrup, Denmark); anti-β-catenin from BD Biosciences (San Jose, CA); anti-Ki-67 from Chemicon International (Temecula, CA); rabbit anti-FZD7 N-terminal epitope from Abcam (Cambridge, UK).

DNA Clones

The human FZD7 promoter-reporter plasmid (pGL-Sec. fl.) was constructed as previously described (Hlubek et al.,2006; Schmalhofer et al.,2008). Briefly, for construction of the human FZD7 promoter reporter plasmid, nucleotides −1,116 to +41 (relative to FZD7 transcriptional start site) were amplified from BAC DNA clone no. RP11-107N15 (Imagenes, Berlin, Germany) using following primers: FZD7-prom-MluI 5′-AACTACGCGTGCACCAGGAAGAGGAACAAA-3′ and FZD7-prom-HindIII 5′-CAATAAGCTTGTG- CTCTCAGCCTGAGGAGT-3′. Subsequently, the resulting fragment was subcloned into pGL3basic (Promega, Mannheim, Germany). The synthetic siRNA specific for β-catenin or GFP (Thermo Electron, Ulm, Germany) have been previously described (Hlubek et al.,2006). The pcDNA/hTCF4 plasmid was from Bert Vogelstein (John Hopkins University, Baltimore, MD) and the pcDNA/hβ-catenin plasmid and the TOPflash and FOPflash TK plasmids were from Hans Clevers (Hubrecht Institute, Utrecht, The Netherlands).

Cell Culture, Transient Transfection, Reporter Assays, Immunoblot and Quantitative RT-PCR

DLD-1 and SW480 colorectal carcinoma cell lines and HeLa epithelial cells lines were purchased from ATCC (Manassas, VA). Standard cell culture, transient transfections, immunoblots, transient short interfering RNA (siRNA)-mediated knockdown and quantitative real-time reverse transcription (RT)-PCR were carried out as previously described (Brabletz et al.,1999,2001; Hlubek et al.,2001,2004; Vincan et al.,2007b). Antibodies used for immunoblot were a mAb against β-catenin (BD Biosciences) and an anti β-actin mAb as control (Sigma-Aldrich, St. Louis, MO).

For the reporter assays, cells were transfected with the indicated plasmids, harvested 48 hr later and reporter activity assayed using the Promega dual luciferase kit (Promega, Madison, WI) as previously described (Vincan et al.,2007b; Spaderna et al.,2008). Luciferase activity was normalised with Renilla activity of a co-transfected pCMV Renilla construct (Promega) for control of cell number and transfection efficiency. Where indicated, cells were seeded into tissue culture wells pre-coated with matrix. Wells were coated with either fibronectin, collagen type I, collagen type IV, and blocked with bovine serum albumin (BSA) as previously described (Pouliot and Burgess,2000).

Acknowledgements

We thank Hans Clevers and Bert Vogelstein for plasmids. This work was supported by funding from the Cancer Council of Victoria (E.V.) and NHMRC (E.V./T.B.).

Ancillary