SNAI1 Expression in Colon Cancer Related with CDH1 and VDR Downregulation in Normal Adjacent Tissue

Oncogene (2009) 28, 4375–4385 & 2009 Macmillan Publishers Limited All rights reserved 0950-9232/09 $32.00 www.nature.com/onc ORIGINAL ARTICLE SNAI1 expression in colon cancer related with CDH1 and VDR downregulation in normal adjacent tissue

C Penã1,7, JM Garcıá1,7, MJ Larriba2, R Barderas3,IGo´mez1, M Herrera1, V Garcıá1, J Silva1, GDomıńguez1, R Rodrı´guez4, J Cuevas5, AG de Herreros6, JI Casal3, A Munõz2 and F Bonilla1

1Department of Medical Oncology, Hospital Universitario Puerta de Hierro Majadahonda, Majadahonda, Madrid, Spain; 2Instituto de Investigaciones Biome´dicas ‘Alberto Sols’, Consejo Superior de Investigaciones Cientı´ficas-Universidad Autońoma de Madrid, Madrid, Spain; 3Functional Proteomics Laboratory, Centro de Investigaciones Biolo´gicas (CIB-CSIC), Madrid, Spain; 4Department of Pathology, Hospital Virgen de la Salud, Toledo, Spain; 5Department of Pathology, Hospital General Universitario, Guadalajara, Spain and 6Unitat de Biologıá Cellular i Molecular, Institut Municipal d’Investigacio´ Me`dica, Universitat Pompeu Fabra, Barcelona, Spain

SNAI1, ZEB1, E-cadherin (CDH1), and vitamin D Introduction receptor (VDR) genes regulate the epithelial–mesenchymal transition (EMT) that initiates the invasion Numerous studies have focused on the function of process of many tumor cells. We hypothesized that this genetic abnormalities in various cancers, but very few of process could also affect the behavior of normal cells these analyzed their importance in normal tissue adjacent to the tumor. To verify this hypothesis, the adjacent to tumors. Loss of heterozygosity (LOH) or expression level of these genes was determined by allele imbalances (AI) at various chromosomal loci have quantitative RT–PCR in tumor, normal adjacent, and been observed in normal tissue adjacent to tumor in normal distant tissues from 32 colorectal cancer (CC) breast and bladder cancer patients (Deng et al., 1996; patients. In addition, we extended the study to human Larson et al., 1998, 2002; Lakhani et al., 1999; Forsti HaCaT normal keratinocytes and SW480-ADH colon et al., 2001; Kurose et al., 2001; Li et al., 2002). cancer cells co-cultured with SW480-ADH cells over- Moreover, other studies have reported gene and expressing the mouse Snai1 gene. Of 18 CC cases with chromosome alterations in normal tissues, for example SNAI1 expression in tumor tissue, five also had SNAI1 TP53 mutations in normal epidermis of patients with in normal adjacent tissue (NAT). Expression of SNAI1 non-melanoma skin cancers (Ren et al., 1996) and in in tumor tissue correlated with downregulation of CDH1 histologically normal tissue from the upper aero- and VDR genes in both tumor (P ¼ 0.047 and P ¼ 0.014, digestive tract of cancer patients (Waridel et al., 1997), respectively) and NAT lacking SNAI1 expression or cytogenetic abnormalities and growth factor receptor (P ¼ 0.054 and P ¼ 0.003). ZEB1 expression was directly overexpression in normal bronchial epithelium of lung related to VDR expression in tumor tissue (r ¼ 0.39; cancer patients (Sozzi et al., 1991). Recently, the P ¼ 0.027) and inversely to CDH1 in NAT (r ¼À0.46; importance of normal myoepithelial cells in breast P ¼ 0.010). CDH1 and VDR were also downregulated in cancer progression has been shown (Hu et al., 2008). SW480-ADH and MaCaT cells, respectively, when they These studies proved the existence of alterations in were co-cultured with Snai1-expressing cells. Further- normal tissues of cancer patients and suggested that more, cytokine analysis showed differences in the condi- these alterations might predispose individuals to local tioned media obtained from the two cell types. These recurrence or may belong to the early stages of results indicate that histologically normal tissue adjacent pathogenesis, in which case, they may suffer further to tumor tissue expressing the EMT-inducing gene SNAI1 complex and multiple genetic changes during tumor shows alterations in the expression of epithelial differ- growth (Sozzi et al., 1991; Deng et al., 1996; Ren et al., entiation genes such as CDH1 and VDR. 1996; Larson et al., 1998, 2002; Lakhani et al., 1999; Oncogene (2009) 28, 4375–4385; doi:10.1038/onc.2009.285; Forsti et al., 2001; Li et al., 2002). published online 5 October 2009 SNAI1 is a zinc-finger transcription factor that induces epithelial–mesenchymal transition (EMT) (Nieto, 2002), Keywords: colorectal cancer; epithelial–mesenchymal a fundamental process in the early stages of carcinoma transition; SNAI1; CDH1; normal adjacent tissue invasion and metastasis (Vleminckx et al., 1991; Perl et al., 1998; Peinado et al., 2007). SNAI1 expression in epithelial cells promotes the acquisition of a fibro- Correspondence: Dr F Bonilla, Department of Medical Oncology, blastoid morphotype with invasive and migratory Hospital Universitario Puerta de Hierro Majadahonda, C/Joaquin properties and characteristic changes in gene expression Rodrigo, 2, Majadahonda, Madrid E-28222, Spain. (Cano et al., 2000; Guaita et al., 2002). Importantly, E-mail: [email protected] SNAI1 represses the CDH1 gene (Batlle et al., 2000), 7These authors contributed equally to this work. Received 25 January 2009; revised 30 June 2009; accepted 13 August which encodes E-cadherin, the main component of 2009; published online 5 October 2009 adherent junctions responsible for the maintenance of EMT genes in normal tissue adjacent to tumor CPenã et al 4376 the adhesive and polarized phenotype of epithelial cells Table 1 Minimum, maximum, and median of VDR, ZEB1, and (Perez-Moreno et al., 2003). CDH1 downregulation is CDH1 expression levels in tumor and normal adjacent tissue, as one of the hallmarks of EMT. The transcription factor against normal distant tissue ZEB1 also downregulates CDH1 in cultured cells and Tumor Normal adjacent cooperates with SNAI1 in inducing EMT (Grooteclaes and Frisch, 2000; Guaita et al., 2002; Peinado et al., VDR ZEB1 CDH1 VDR ZEB1 CDH1 2007). In contrast, vitamin D (1a,25-dihydroxyvitamin Minimum 0.03 0.01 0.03 0.17 0.05 0.01 D3) induces the expression of CDH1 and an epithelial Maximum 100 10 100 100 33.33 100 phenotype in human colon cells expressing vitamin D Median 0.72 0.52 1.16 1.45 1.30 1.03 receptor (VDR) (Palmer et al., 2001). In addition, we showed that SNAI1 represses VDR expression in tumor cell lines and that its gene expression in human tumors is associated with CDH1 and VDR downregulation tissues counterpart (Figure 1). Finally, SNAI1 protein (Palmer et al., 2004; Pena et al., 2005). was none detected in any of the normal tissues analyzed. As some genetic alterations occur before the development of histologically abnormal tissue in cancer Study of the relation between gene expression levels patients, and because of the interplay between SNAI1, A statistically significant association was observed ZEB1, CDH1 and VDR in the control of EMT, we between the presence of SNAI1 and the downregulation hypothesized that expression of EMT genes (SNAI1, of CDH1 and VDR in tumor tissues: geometric averages ZEB1) in tumors could affect expression of epithelial of the VDR and CDH1 expression levels in those cases genes (CDH1, VDR) in normal adjacent tissue (NAT). with SNAI1 presence in tumor tissue were, respectively, Paracrine regulation by the tumor or other mechanisms 0.52 and 0.83, compared with 2.04 and 3.57 when could drive changes in the overall structure of normal SNAI1 P P tissue surrounding the tumor and facilitate tumor was not detected ( ¼ 0.014 and ¼ 0.047, ANOVA) (Figure 2). growth or invasion. It should not be forgotten that Remarkably, presence of SNAI1 in tumor tissue but tissue remodeling occurs continuously during tumoro- genesis (Herzig and Christofori, 2002). To carry out this not in NAT was associated with downregulation of VDR and CDH1 in NAT (all of them SNAI1 negatives). study, we analyzed the expression of these four genes in Thus, of 13 cases with SNAI1 expression in tumor tissue the tumor tissue and in histologically normal tissue but not in NAT, the geometric averages of VDR and adjacent to carcinoma in a series of 32 colorectal cancer CDH1 in NAT tissue were 0.85 and 0.84, respectively, (CC) patients. Some patients with SNAI1 expression in against 4.75 and 3.38 in the 14 cases without SNAI1 tumors also had SNAI1 expression in NAT. In addition, expression in tumor tissue (P 0.003 and P 0.054, we found a correlation between expression of SNAI1 in ¼ ¼ the tumor and VDR and CDH1 downregulation in both ANOVA) (Figure 2). No association between the presence of SNAI1 in tumor and NAT (not expressing SNAI1). tumor tissue and the expression of ZEB1 in tumor or in NAT was observed (P ¼ 0.351 and P ¼ 0.273, respectively, ANOVA) (Figure 2). The study of the expression data of VDR, CDH1, and Results ZEB1 genes in tumor and NAT showed some correla- Gene expression analysis of patient samples tions. As expected, a correlation between expression levels of CDH1 and VDR was observed in both tumor Expression of SNAI1, CDH1, VDR, and ZEB1 genes (r 0.58; P 0.001) and NAT (r 0.54; P 0.001). In was studied in tumor (T) and histologically NAT in a ¼ o ¼ ¼ addition, direct correlation between VDR and ZEB1 series of 32 CC patients and related to expression found (r 0.39; P 0.027) was found in tumor tissue, but not in normal tissue distant from the tumor (N). ¼ ¼ in NAT. In contrast, when ZEB1 and CDH1 were SNAI1 was expressed in 18 out of 32 tumor samples (56.3%) and five of these patients also showed SNAI1 related, inverse correlation in the NAT was observed (r 0.46; P 0.010), but not in the tumor tissue expression in the NAT (15.6%). These patients were ¼À ¼ (Figure 3). excluded from the subsequent analysis to avoid mis- interpretations of the results obtained. None of the normal distant tissues showed SNAI1 expression. The Analysis of K-RAS exon 1 mutation and LOH and minimum, maximum, and median of VDR, ZEB1,and microsatellite instability at 17q21 region CDH1 expression in tumor and NAT are shown in To characterize at the molecular level, the histologically Table 1. normal tissue adjacent to tumor, we studied K-RAS We detected SNAI1 protein, by immunohistochem- mutation and LOH and microsatellite instability in istry, in the entire SNAI1 mRNA positive tumor tissues 17q21. Mutations in K-RAS exon 1 were found in 10 out tested, but never in the NAT tissues, despite that was of 32 tumor samples (31.3%). However, none of the possible to detect SNAI1 mRNA in them (Figure 1). samples of NAT or distant tissue showed mutations Moreover, perceptible decrease in E-cadherin protein (Figure 4). level in NAT tissue was observed in those cases with Two of the five cases with SNAI1 expression in NAT positive SNAI1 expression in their respective tumor (40%) had mutations in the exon 1 of K-RAS in the

Oncogene EMT genes in normal tissue adjacent to tumor CPen˜a et al 4377

N NAT T

Figure 1 Immunohistochemical analysis of SNAI1 (a) and E-cadherin (b) protein in the normal distant tissue (N), normal adjacent tissue (NAT) and tumor tissue (T) from one patient in which SNAI1 mRNA was detected both in NAT and T counterpart tissues. Despite SNAI1, mRNA was detected in T and NAT samples, SNAI1 protein was only detected in tumor tissue. A clear decrease in E-cadherin expression regarding normal tissue is observed both in T and NAT samples.

0 SNAI1 + - + - + - + - + - + - CDH1 VDR ZEB1 CDH1 VDR ZEB1 p = 0.047 p = 0.014 p = 0.351 p = 0.054 p = 0.003 p = 0.704

T Tissue NAT Tissue Figure 2 Geometric average of the ratios (T/N and NAT/N) of CDH1, VDR, and ZEB1 expression in tumor and NATs for SNAI1 presence in tumor tissue.

tumor tissue, but none of them had in the NAT or Mock cells mixed with different proportions of SW480- distant tissue. ADH-Snai1 (0, 10, 25 and 75%). There were 3.9-fold We observed 4 out of 20 (20%) informative cases with decreases in CDH1 levels in SW480-ADH-Snai1 versus LOH in 17q21 in 32 tumor samples. In addition, SW480-ADH-Mock cells, and much >75% of SW480- microsatellite instability was found in another patient. ADH-Snai1 cells contaminating the SW480-ADH- Similar to K-RAS mutations, neither NAT nor N tissues Mock cells were required to obtain a notable decrease had LOH or microsatellite instability in 17q21 (Figure 4). in CDH1 levels (Figure 4). To quantify the proportion of contaminating cells with low CDH1 expression levels present in a population of normal cells with high CDH1 expression levels, Paracrine assays with human colon cancer cells needed to justify the loss of CDH1 observed in our The results suggested that SNAI1-expressing tumor cells study, we extracted the mRNA from SW480-ADH- modify the expression of CDH1 and VDR in adjacent

Oncogene EMT genes in normal tissue adjacent to tumor CPen˜a et al 4378 1.5 1.0

1.0 0.5 0.5 0.0 0.0 -0.5 -0.5 expression expression -1.0 -1.0 VDR VDR -1.5 -1.5 r = 0.58 r = 0.54 -2.0 -2.0 p < 0.001 p = 0.001 -2.5 -2.5 -4 -3 -2 -1 0 1 2 -4 -3 -2 -1 0 1 2 3 CDH1 expression CDH1 expression

1.5 1.0 1.0 0.5 0.5 0.0 0.0 -0.5 -0.5 expression expression -1.0 -1.0 VDR VDR -1.5 r = 0.39 -1.5 r = 0.27 -2.0 p = 0.027 p = 0.157 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 ZEB1 expression ZEB1 expression

2 2.5 2.0 1 1.5 0 1.0 0.5 -1 expression expression 0.0 -2 -0.5 CDH1 CDH1 -1.0 -3 r = 0.15 r = -0.46 p = 0.425 -1.5 p = 0.010 -4 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 ZEB1 expression ZEB1 expression Figure 3 Relation between expression levels of VDR, CDH1, and ZEB1 genes [log R(T/N) and log R(NAT/N)] in tumor tissue (a–c) and in normal tissue adjacent to tumor (d–f).

cells without SNAI1 expression. To analyze this regulation was observed independently of the position hypothesis, we studied the expression of VDR, ZEB1, of SW480-ADH-Mock cells in the Transwell (upper or and SNAI1 in human colon cancer SW480-ADH-Mock lower compartment) (Figure 5, Ib). The expression of cells and human normal kerotinocytes (HaCaT) co- VDR, ZEB1, and SNAI1 was the same in SW480-ADH- cultured with SW480-ADH-Snai1 cells stably expressing Mock cells cultured alone or co-cultured with SW480- Snai1. Co-cultures were performed in Transwells, which ADH-Snai1 cells (Supplementary Table 2A; Supple- allow physical separation between the cell types (upper mentary Materials). In contrast VDR, but not CDH1, and lower compartments). All the experiments included downregulation was observed in the case of HaCaT cells controls in which SW480-ADH-Mock cells were seeded at 72 h (P ¼ 0.008) (Supplementary Figure 1). in both compartments. In addition, CDH1 downregulation was observed in SW480-ADH-Mock cells co-cultured with SW480- SW480-ADH-Mock cells incubated for 4, 8, 24, or 30 h ADH-Snai1 cells for 72 or 96 h showed downregulation with 24 h-conditioned medium from SW480-ADH- of CDH1 expression (Figure 5, Ia). CDH1 down- Snai1 cells. The highest downregulation was found at

Oncogene EMT genes in normal tissue adjacent to tumor CPenã et al 4379 Snai1 stably transfected cells and with SW480-ADH- Mock as control (Figure 6a and b) (Figure 6a). After incubation, membranes were scanned and each cytokine was quantified by densitometry. Relative cytokine intensities were normalized in relation to control spots on the same membrane. A significant increase was observed in SW480-ADH-Snai1-conditioned medium for several cytokines. We observed a 9.5-fold upregulation change in MIP-3 a, together with significant increases in IL-6, GRO-a, MCP-3, and GCP-2, which were also upregulated in SW480-ADH-Snai1 cells, with an approximately 2.5-fold change from SW480-ADH- Mock (Figure 6b). In contrast, other EMT-related factors such as TGFB1 did not show any alteration.

N NAT T Proteomic assay The aforementioned observations suggest that SW480- CDH1 levels 3 ADH-Snai1 cells supply paracrine signals that induce CDH1 downregulation in adjacent cells. To understand 2.5 this process better, we compared the proteoma of the 2 conditioned medium from SW480-ADH-Mock and 1.5 SW480-ADH-Snai1 cells by 2D-DIGE and Peptide 1 Mass Fingerprinting searching for differences in 0.5 secreted proteins. We identiﬁed 22 differently expressed 0 proteins (Table 2). Though none of these proteins have 0%10% 25% 75% 100% an acknowledged paracrine function, further studies are Figure 4 Photograph showing the LOH at D17S855 marker of the needed to substantiate whether any of them contributes 17q21 region in patient 2 (a) and K-RAS mutation at exon 1 in to the downregulation of CDH1. patient 4 after SSCP study (b). In both patients, the alterations appeared in tumor (T) tissue, but not in normal tissue (N) or in normal tissue adjacent to tumor (NAT). (c) CDH1 levels in SW480- ADH-Mock cells mixed with different proportion of SW480-ADH- Discussion Snai1 (0, 10, 25, 75, and 100%). Tumor invasion can be viewed as a deregulation in the 24 h of incubation (Figure 5, IIa). In another series of proper sorting of cell populations, causing an alteration experiments, SW480-ADH-Mock cells were incubated of normal tissue boundaries (Brown et al., 1999; for 24 h with medium of SW480-ADH-Snai1 or SW480- Hanahan and Weinberg, 2000; Park et al., 2000). It is ADH-Mock cells conditioned for 2, 4, 8, or 24 h. Similar generally accepted that physiological and malignant results were obtained and the greatest CDH1 down- invasion use similar molecular mechanisms (Liotta regulation was achieved with 8 h-conditioned SW480- et al., 1991). Cross-talk between mesenchyme and ADH-Snai1 medium (Figure 5, IIb). epithelium has been considered as a driver of differ- CDH1 downregulation was not observed when entiation and development (Aboseif et al., 1999; Nilsson conditioned medium from SW480-ADH-Snai1 cells and Skinner, 2001) and has been viewed as involved in was heated (60 1C, 5 min) for protein inactivation the promotion of epithelial transformation (Aboseif (ratio of CDH1 expression in SW480-ADH-Mock cells et al., 1999; Olumi et al., 1999; Li et al., 2000). Altered incubated with conditioned medium from SW480- cell–cell and cell–substratum signals may release normal ADH-Snai1 versus expression in cells incubated with tissue constraints, thereby enabling malignant cells to medium from SW480-ADH-Mock cells ¼ 1.03; standard migrate across tissue boundaries (Liotta et al., 1991; deviation ¼ 0.13; P ¼ 0.744). Hanahan and Weinberg, 2000). Unlike results with CDH1, incubation of SW480- The objective of our study was to search for ADH-Mock cells with conditioned medium from alterations that may favor tumor progression and SW480-ADH-Snai1 did not change VDR, ZEB1,or invasion in the expression levels of EMT-related genes SNAI1 expression levels (Supplementary Table 2B; in the normal tissue adjacent to tumor. The few such Supplementary Materials). studies performed were mainly directed to the analysis of genetic or chromosome alterations, such as LOH and allelic instability at various chromosomal loci or TP53 Regulation of cytokines by Snai1 mutations. Our study shows deregulation of genes We further investigated whether Snai1 mediates the involved in EMT in NAT that is associated with tumor production and release of cytokines, growth factors, and characteristics. It could be argued that these results are other immuno-modulators by using a human cytokine masked by the presence of contaminating tumor cells array with conditioned medium from SW480-ADH- in NAT. Nevertheless, we believe that this is highly

Oncogene EMT genes in normal tissue adjacent to tumor CPen˜a et al 4380

1.2 1.4 p = 0.216 p = 0.006 p = 0.010 p = 0.978 p = 0.016 p = 0.003 1 1.2 1 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0 0 48 h 72 h 96 h 48 h 72 h 96 h

1.2 1.2 p = 0.058 p = 0.029 p = 0.006 p = 0.185 p = 0.142 p = 0.054 p = 0.014 p = 0.419 1 1 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2 0 0 4 h 8 h 24 h 30 h 4 h 8 h 24 h 30 h Figure 5 Arithmetic average of CDH1 expression levels in (I) SW480-ADH-Mock cells seeded in the lower compartment of transwells and co-cultured during the times indicated with SW480-ADH-Mock cells (black columns) or SW480-ADH-Snai1 cells (gray columns) on the upper compartment (a) or vice versa (b). (II) SW480-ADH-Mock cells incubated for the times indicated with 24 h-conditioned medium from SW480-ADH-Mock cells (black columns) or SW480-ADH-Snai1 cells (gray columns) (a) or incubated for 24 h with medium conditioned for different times by SW480-ADH-Mock cells (black columns) or SW480-ADH-Snai1 cells (gray columns) (b).

ADH-GFP ADH-SN

Table Cytokine array. Proteins overexpressed in ADH-SN respect to ADH-GFP conditioned cell culture media. Row Column Cytokine CytokineFull Name Median * ADH-SN/ADH-GFP Error A9 GM-CSF Granulocyte-macrophage colony-stimulating factor 1.42 0.13 A 11 GRO-α Growth-regulated alpha protein 2.48 0.10 B8 IL-6 Interleukin-6 2.57 0.10 B 10 IL-8 Interleukin-8 1.82 0.06 C5 MCP-1 Monocyte chemoattractant protein-1 1.62 0.23 C7 MCP-3 Monocyte chemoattractant protein-3 2.56 0.25 C8 MCSF Macrophage colony-stimulating factor 1 1.42 0.04 E3 VEGF Vascular endothelial growth factor 1.83 0.14 F7 GCP-2 Granulocyte chemotactic protein 2 2.68 0.31 G8 MIF Macrophage migration inhibitory factor 1.38 0.04 G 9 MIP-3α Macrophage inflammatory protein 3 alpha 9.53 3.55 G 10 NAP-2 Neutrophil-activating peptide 2 1.65 0.18

Figure 6 Cytokine antibody array of conditioned medium from SW480-ADH-Mock and SW480-ADH-Snai1 stably transfected cells. (a) The membranes were blotted with equal amounts of conditioned medium from SW480-ADH-Mock or SW480-ADH-Snai1 cells. (b) Table showing the differential upregulation of human cytokines. The experiment was carried out in duplicate. The media fold change and the error of the experiments calculated after quantiﬁcation of the cytokine arrays are also shown in the table.

improbable, as minor contamination not detectable by low CDH1 expressing cells were required to achieve histological analysis should not justify the observed these results. It is very difﬁcult to believe that this high changes in a quantitative study, especially as we are proportion of tumor cells invading the NAT tissue is not looking at downregulation of particular genes such as detected by the pathologist. Moreover, if contaminating CDH1. On the other hand, there were 3.7-fold decreases tumor cells affect a quantitative analysis, their presence in CDH1 levels in NAT SNAI1 positive tissues versus should also be detected by a qualitative analysis such as NAT SNAI1-negative tissues; >75% of contaminating K-RAS mutation or allelic instability analysis.

Oncogene EMT genes in normal tissue adjacent to tumor CPenã et al 4381 Table 2 Protein content of conditioned media (24 h incubation) from SW480-ADH-Mock or SW480-ADH-Snai1 cells analyzed by 2D-DIGE and identified by peptide mass fingerprinting using MALDI-TOF-TOF MS No. Name t-test Average ratioa

1 YWHAZ protein (Homo sapiens) 0.019 À1.4 2 Tropomyosin isoform; TPMSK3; tropomyosin 3 0.0071 1.52 3 Triosephosphate isomerase 0.024 1.57 4 Serum albumin precursor 0.0047 1.52 5 Serum albumin precursor 0.0018 1.64 6 Proteasome subunit alpha type 6 0.046 1.21 7 Nucleophosmin (NPM) 0.0036 1.1 8 Nucleophosmin (Homo sapiens) 0.042 1.15 9 Nucleolin (Protein C23) 0.0024 À1.39 10 Keratin type I and type II cytoeskeletal 0.00034 À1.7 11 Keratin type I cytoeskeletal 18 0.01 À1.59 12 Keratin type I and type II cytoeskeletal 0.0037 À1.95 13 Keratin type I cytoeskeletal 9 0.02 1.55 14 Keratin type I cytoeskeletal 18 0.0022 À2.96 15 Keratin 10 0.0072 2.31 16 K1C16; K2C1; K1C14 0.041 1.54 17 Heterogeneous nuclear ribonucleoproteins 0.00095 À2.09 18 Cytokeratin 9 0.0036 1.95 19 Alternative splicing factor ASF-3 human 0.00039 1.48 20 Alpha actinin 4 0.0029 À1.69 21 ACTB protein; beta actin 0.0071 À1.22 22 Hypothetical protein 0.00091 1.37 aSW480-ADH-Snai1/SW480-ADH-Mock.

We suggest that the observed deregulation of SNAI1, ments with tumor cell lines showed that Snai1-expres- CDH1, VDR,andZEB1 genes in the normal tissue sing cells could paracrinally downregulate the boundaries may result from the paracrine interaction expression of CDH1 and VDR in neighboring cells between malignant and histological normal epithelium. without changing the endogenous level of SNAI1. Moreover, it has been described that changes in stromal Human cytokine array showed several cytokines differ- cell population and remodeling of the extracellular entially released by tumor cells expressing SNAI1; most matrix/basement membrane results in a breakdown of of these released cytokines have a major function in normal tissue boundaries and seems to be a necessary monocyte-macrophage recruitment and pro-inflamma- stage in the local invasion of tumor cells (Liotta et al., tory activities, opening an interesting link between 1991; Vleminckx et al., 1991; Werb, 1997; Condeelis SNAI1 and inflammation, which deserves further and Segall, 2003; Sahai and Marshall, 2003; Wolf investigation. Likewise, it was recently described that et al., 2003). SNAI1 proteins influence the behavior of Snai1-negative As expected, the expression of the two epithelial neighboring cells in zebrafish embryo (Blanco et al., genes, VDR and CDH1, are strongly associated. 2007). Although we performed a proteomic analysis Interestingly, when SNAI1 was expressed in tumor looking for differences between conditioned media from tissue, both CDH1 and VDR were downregulated not SW480-ADH-Mock and SW480-ADH-Snai1, we did only in tumors but also in NAT. This relation may be a not identify proteins with an acknowledged paracrine direct effect from tumor to normal adjacent cells, a function. However, several deregulated proteins of secondary effect because of specific characteristics unknown function were characterized. conferred on the tumor by SNAI1 expression or related We found an inverse relation between the expression to other unknown alterations. For example, changes in of ZEB1 and CDH1 genes in NAT, which was not secreted matrix proteins by tumor cells could modulate observed in tumor tissue. It is possible that, in SNAI1- the expression of CDH1 in adjacent cells through the expressing tumors, the effects of ZEB1 on CDH1 were activation of the integrin-linked kinase (Lochter et al., masked by SNAI1 effects, as this factor is a more potent 1997; Wu et al., 1998; Menke et al., 2001). This effect repressor of CDH1 transcription (Guaita et al., 2002). In might be mediated by SNAI1 activation in adjacent addition, these two factors may cooperate in CDH1 cells, as SNAI1 transcription is stimulated by integrin- repression or act on parallel pathways. This hypothesis linked kinase activation in colon cells (Tan et al., 2001). is supported by our earlier studies with a large series of This hypothesis is supported by the fact that SNAI1 colorectal carcinoma, in which we found a loss of was detected in five normal tissues adjacent to relation between SNAI1 expression and CDH1 down- tumor in our series. Alternatively, as SNAI1-expressing regulation when ZEB1 expression was high (Pena et al., tumors are more invasive (Cano et al., 2000; Guaita 2005). In addition, it should be remembered that the et al., 2002), they are more prone to disrupt the boun- ability of ZEB proteins to work as transcriptional daries established with neighboring tissues. Otherwise, repressors or activators can be modulated by acetyla- although the mechanism remains unknown, the experi- tion, a process that selectively allows ZEB proteins to

Oncogene EMT genes in normal tissue adjacent to tumor CPenã et al 4382 interact with co-repressors or co-activations such as RNA was extracted from tumor and normal (both adjacent p300 or CtBP (Postigo et al., 2003). Different expression to and distant from tumor) samples with the RNeasy Mini Kit levels of ZEB1 cofactors, p300 and CtBP, in tumor or (Qiagen Inc., Hilden, Germany), according to manufacturer’s normal tissue may explain the different repressive protocol. The RNA extracted was quantified with a Nano- activity on CDH1, and even the positive relationship Drop ND-1000 Spectrophotometer (NanoDrop Technologies detected in these tissues with VDR expression. It should Inc., Wilmington, DE, USA). also be remarked that transfection of ZEB1 into SW620 colon carcinoma cells results in the upregulation of Real-time PCR endogenous VDR (Lazarova et al., 2001). In fact, we As SNAI1 RNA was not detected in normal distant tissues, SNAI1 expression in tumor and NAT was only valued as earlier observed an inverse correlation between ZEB1 presence or absence. A SNAI1 retrogene, SNA1P, located at and CDH1 expression in tumor tissue when the cofactor chromosome 2q34-X is very similar to SNAI1 (Locascio et al., CtBP was overexpressed and a stronger direct correla- 2002). Expression of SNA1P in epithelial cell lines induces EMT tion between ZEB1 and VDR when the cofactor p300 and represses CDH1 expression, albeit less strongly than SNAI1 was overexpressed (Pena et al., 2006). In summary, the does (Locascio et al., 2002). To amplify SNAI1, primers were regulation of CDH1 and VDR seems to be different in designed from a region with 16 bp differences between SNAI1 tumor and NAT. and SNA1P. Amplification of SNA1P was thus avoided, which SNAI1 and ZEB1 genes are expressed by mesenchy- was checked by sequencing of the PCR fragments obtained. mal cells (Batlle et al., 2000; Cano et al., 2000) and, Primers used are shown in Supplementary Table 1. presumably, by stromal cells. In this context, the results VDR, CDH1, and ZEB1 mRNA levels were calculated in the tumor and in normal adjacent and distant tissues by a of SNAI1 or ZEB1 deregulation observed in tumor relative quantification approach, in which the amounts in the tissue might be influenced by the expression of these targets were expressed in relation to the geometric average of genes in non-epithelial cells, masking the expression in three reference housekeeping genes: TATA binding protein, epithelial tumor cells. However, similar changes to those succinate dehydrogenase complex subunit A, and ubiquitin C observed in the tumors were detected when using (Palmer et al., 2004). The relative concentrations of target and epithelial enriched cell populations in an earlier study reference genes were calculated by interpolation, using a (Pena et al., 2005). Therefore, we conclude that standard curve of each gene generated with a serial dilution of alterations found in the expression of these genes in a cDNA prepared from RNA extracted from SW480-ADH the whole tumor samples of this series correspond to cells. The expression level of a target gene in a patient was carcinoma cells. calculated as the ratio of its expression in T or NAT versus its expression in N (T/N or NAT/N). Primers used are shown in As for CDH1 expression, we observed in an earlier Supplementary Table 1. study that the quantification of CDH1 RNA in the For the synthesis of the first strand of cDNA, 400 ng of total whole biopsy is a good marker of E-cadherin expression RNA were retro-transcribed using the Gold RNA PCR Core and that non-significant intra-tumor differences Kit (PE Biosystems, Foster City, CA, USA), following the were found in E-cadherin staining (Pena et al., 2005), manufacturer’s instructions. Random hexamers were used for in contrast with other authors’ findings (Brabletz cDNA synthesis. Real-time PCR was performed in a Light- et al., 2001). Cycler apparatus (Roche Diagnostics, Mannheim, Germany) Our data add novel information about the behavior of using the LightCycler-FastStartPLUS DNA Master SYBR tumor-normal tissue boundaries. We suggest that Green I Kit (Roche Diagnostics). tumor-adjacent tissue considered as histologically and molecularly normal undergoes alterations in gene Analysis of K-RAS exon 1 mutation and LOH and microsatellite expression that may influence tumor progression instability at 17q21 region and phenotypic features. These results support the need PCR amplification of K-RAS gene and D17S855 microsatellite marker located at intron 20 of BRCA1 (17q21 region) was for further study of the signals involved in tumor- carried out from 100 ng of genomic DNA as template, using stromal cross-talk, an approach that may bring Ampli Taq DNA polymerase (Perkin-Elmer, Roche Molecular therapeutic benefit. Systems Inc., Branchbur, NJ, USA). The allelic band intensity on the gels was detected by non-radioisotopic means with a commercially available silver staining method (Oto et al., Materials and methods 1993). Primers used for amplification of K-RAS exon 1, containing codons 12 and 13, and for the 17q21 region were Patients, tumor samples, and RNA extraction shown in Table 1 (Supplementary Materials). The bands with This study, approved by the Research Ethics Board of our different mobility shift pattern were sequenced in an ABI Prism hospital, was based on a consecutive series of 32 patients who 377 DNA sequencer apparatus (PE Applied Biosystems, Foster underwent surgery for CC between January 2004 and January City, CA, USA). 2005. Samples of tumor and of normal adjacent (taken 0.5 cm from the outer tumor margin) and normal distant (taken at Co-culture and conditioned medium assays least 3 cm from the outer tumor margin) tissues were obtained SW480-ADH (Tomita et al., 1992; Baulida et al., 1999) human sequentially, immediately after surgery, snap-frozen in liquid colon cancer cells stably expressing mouse Snai1 cDNA nitrogen and stored at À80 1C until processing. (SW480-ADH-Snai1) or an empty vector (SW480-ADH- All formalin-fixed paraffin-embedded normal and tumor Mock) were generated by retrovirus-mediated gene transfer, tissues were examined histologically by two independent as described earlier (Palmer et al., 2004). Cells were cultured in pathologists to ensure that each type of tissue was correctly Dulbecco’s modified Eagle medium (Gibco Life Technologies, assigned to its corresponding group. Gergy-Pontoise, France) containing 10% heat-inactivated fetal

Oncogene EMT genes in normal tissue adjacent to tumor CPenã et al 4383 calf serum, 2 mML-glutamine, penicillin (100 units/ml), control. The normal tissues of all of them were used as a streptomycin (100 ng/ml), and fungizone (0.25 mg/ml) at 37 1C negative control. in a 5% CO2-humidified atmosphere. Immunophenotypic analysis was performed according to Co-cultures of SW480-ADH-Mock, HaCaT, and SW480- standard procedures, with overnight incubation in the presence ADH-Snai1 cells, physically separated, were grown in six-well of anti-E-cadherin DAKO antibody. Immunodetection was plates using Transwells Permeable Supports (0.4 mm pore size, performed with peroxidase-labeled streptavidin biotin (LSA; high pore density, Becton Dickinson, Le Pont-de-Claix, Dako, Glostrup, Denmark) using diaminobenzidine chroma- France). RNA was extracted from SW480-ADH-Mock and gen as substrate. All immunostaining was performed using the HaCaT cells after the indicated times of co-culture. RNA TechMate 500 (Dako) automatic immunostaining device. obtained from SW480-ADH-Mock cells co-cultured seeded in SNAI1 immunohistochemical staining was performed as the two transwell chambers were always used as a control. All follows: 2.5-mm-thick paraffin-embedded complete tissue sec- the experiments included controls in which SW480-ADH- tions were cut onto Dako slides (Dako), and subsequently Mock cells were seeded in both compartments. dewaxed, rehydrated, and subjected to antigen retrieval using To obtain conditioned mediums, the normal medium of the Dako’s PT Link with 10 mM of sodium citrate, at pH 9, and SW480-ADH-Mock or SW480-ADH-Snai1 cells were replaced heated at 95 1C. The slides were cooled and treated with by half the volume of fetal calf serum-free medium as peroxidase-blocking solution (Dako) for 5 min. Sections were incubation was continued for the indicated times. The medium then immunostained with SNAIL (EC3 clone) (Franci et al., was then filtered and 10 Â concentrated using Amicon Ultra 2006) hybridoma supernatant by the two-stage peroxidase- technology (Centrifugal filter devices, regenerated cellulose based EnVision technique (Dako), counterstained with hema- 10 000 MWCO, Millipore Corporation, Billerica, MA, USA). toxylin, and mounted. Incubations, either omitting the specific SW480-ADH-Mock cells were incubated with 10 Â concen- antibody or containing unrelated antibodies, were used as a trated conditioned medium, diluted 1:4 in normal medium for control of the technique. the times indicated. Four repetitions were carried out for each experiment. Data analysis As the distribution of the gene expression values was not normal distributed (Kolmogozov-Smvinov test), to carry out Human protein cytokine array the statistical analysis, we normalized data distribution by SW480-ADH-Mock and SW480-ADH-Snai1 stably trans- using log . We used the geometric average of the T/N and fected cells were cultured at 37 1C in an atmosphere of 5% 10 NAT/N, instead of the arithmetical one, to describe the CO for 48 h in Dulbecco’s modified Eagle medium with 10% 2 expression gene data because of the non-normal distribution of fetal calf serum. Then, they were washed with PBS and the data. Expression levels of CDH1 and VDR were contrasted cultured for 24 h under the same conditions in Dulbecco’s with the presence or absence of SNAI1 expression in different modified Eagle medium fetal calf serum-free medium. The tissue samples by ANOVA. CDH1, VDR, and ZEB1 expres- conditioned medium was harvested, centrifuged at 1500 g to sion levels were studied by the Pearson’s test. Two-tailed remove cell debris and filtered. Conditioned medium from each P-values 0.05 were considered statistically significant. Statis- stable cell type was prepared and incubated with membranes p tical analysis used the SPSS statistical package, version 13.0. containing an array of 80 human cytokine antibodies. The human protein cytokine array V membranes (RayBiotech, Norcross, GA, USA) were blocked in blocking buffer for 1 h, and then 2 ml of conditioned medium from each culture cell Abbreviations was individually (physically separated) added and incubated at 4 1C overnight. The membranes were then treated and AI, allele imbalance; CC, colorectal cancer; EMT, epithelial– analyzed according to the manufacturer’s instructions. The mesenchymal transition; LOH, loss of heterozygosity; N, experiment was repeated twice, with excellent reproducibility. normal; NAT, normal adjacent tissue; T, tumor; VDR, vitamin D receptor. Proteomic assays Protein content of conditioned media (24 h incubation) from SW480-ADH-Mock or SW480-ADH-Snai1 cells was analyzed Conflict of interest by 2D-DIGE and the spots of interest were excised and identified by Peptide Mass Fingerprinting using MALDI- The authors declare no conflict of interest. TOF-TOF MS (Text 1, Supplementary Materials).

Immunohistochemistry Acknowledgements E-cadherin and SNAI1 protein were analyzed by immunohistochemistry in several samples taken from normal, tumor, We thank M Eaude for help with the English paper. This study and NAT tissues of the ﬁve patients, in which SNAI1 mRNA was supported by the grants SAF2007-60431, CAM: was detected. At the same time, one patient, in which SNAI1 S-GEN/0266/2006, ISCIII-RETIC RD06/0020 and /0009, mRNA was only detected in tumor tissue, was used as a and a grant from the Accion Transversal del Cancer (ISCIII).

References

Aboseif S, El-Sakka A, Young P, Cunha G. (1999). Mesenchymal Batlle E, Sancho E, Franci C, Dominguez D, Monfar M, Baulida J et al. reprogramming of adult human epithelial differentiation. Differ- (2000). The transcription factor snail is a repressor of E-cadherin gene entiation 65: 113–118. expression in epithelial tumour cells. Nat Cell Biol 2: 84–89.

Oncogene EMT genes in normal tissue adjacent to tumor CPenã et al 4384 Baulida J, Batlle E, Garcia De HA. (1999). Adenomatous polyposis Liotta LA, Steeg PS, Stetler-Stevenson WG. (1991). Cancer metastasis coli protein (APC)-independent regulation of beta-catenin/Tcf-4 and angiogenesis: an imbalance of positive and negative regulation. mediated transcription in intestinal cells. Biochem J 344(Pt 2): Cell 64: 327–336. 565–570. Locascio A, Vega S, de Frutos CA, Manzanares M, Nieto MA. (2002). Blanco MJ, Barrallo-Gimeno A, Acloque H, Reyes AE, Tada M, Biological potential of a functional human SNAIL retrogene. J Biol Allende ML et al. (2007). Snail1a and Snail1b cooperate in the Chem 277: 38803–38809. anterior migration of the axial mesendoderm in the zebrafish Lochter A, Galosy S, Muschler J, Freedman N, Werb Z, Bissell MJ. embryo. Development 134: 4073–4081. (1997). Matrix metalloproteinase stromelysin-1 triggers a cascade of Brabletz T, Jung A, Reu S, Porzner M, Hlubek F, Kunz-Schughart LA molecular alterations that leads to stable epithelial-to-mesenchymal et al. (2001). Variable beta-catenin expression in colorectal cancers conversion and a premalignant phenotype in mammary epithelial indicates tumor progression driven by the tumor environment. Proc cells. J Cell Biol 139: 1861–1872. Natl Acad Sci USA 98: 10356–10361. Menke A, Philippi C, Vogelmann R, Seidel B, Lutz MP, Adler G et al. Brown LF, Guidi AJ, Schnitt SJ, Van De WL, Iruela-Arispe ML, Yeo (2001). Down-regulation of E-cadherin gene expression by collagen TK et al. (1999). Vascular stroma formation in carcinoma in situ, type I and type III in pancreatic cancer cell lines. Cancer Res 61: invasive carcinoma, and metastatic carcinoma of the breast. Clin 3508–3517. Cancer Res 5: 1041–1056. Nieto MA. (2002). The snail superfamily of zinc-finger transcription Cano A, Perez-Moreno MA, Rodrigo I, Locascio A, Blanco MJ, del factors. Nat Rev Mol Cell Biol 3: 155–166. Barrio MG et al. (2000). The transcription factor snail controls Nilsson E, Skinner MK. (2001). Cellular interactions that control epithelial-mesenchymal transitions by repressing E-cadherin expres- primordial follicle development and folliculogenesis. J Soc Gynecol sion. Nat Cell Biol 2: 76–83. Investig 8: S17–S20. Condeelis J, Segall JE. (2003). Intravital imaging of cell movement in Olumi AF, Grossfeld GD, Hayward SW, Carroll PR, Tlsty TD, tumours. Nat Rev Cancer 3: 921–930. Cunha GR. (1999). Carcinoma-associated fibroblasts direct tumor Deng G, Lu Y, Zlotnikov G, Thor AD, Smith HS. (1996). Loss of progression of initiated human prostatic epithelium. Cancer Res 59: heterozygosity in normal tissue adjacent to breast carcinomas. 5002–5011. Science 274: 2057–2059. Oto M, Miyake S, Yuasa Y. (1993). Optimization of nonradioisotopic Forsti A, Louhelainen J, Soderberg M, Wijkstrom H, Hemminki K. single strand conformation polymorphism analysis with a conven- (2001). Loss of heterozygosity in tumour-adjacent normal tissue of tional minislab gel electrophoresis apparatus. Anal Biochem 213: 19–22. breast and bladder cancer. Eur J Cancer 37: 1372–1380. Palmer HG, Gonzalez-Sancho JM, Espada J, Berciano MT, Puig I, Franci C, Takkunen M, Dave N, Alameda F, Gomez S, Rodriguez R Baulida J et al. (2001). Vitamin D(3) promotes the differentiation of et al. (2006). Expression of Snail protein in tumor-stroma interface. colon carcinoma cells by the induction of E-cadherin and the Oncogene 25: 5134–5144. inhibition of beta-catenin signaling. J Cell Biol 154: 369–387. Grooteclaes ML, Frisch SM. (2000). Evidence for a function of CtBP Palmer HG, Larriba MJ, Garcia JM, Ordonez-Moran P, Pena C, Peiro in epithelial gene regulation and anoikis. Oncogene 19: 3823–3828. S et al. (2004). The transcription factor SNAIL represses vitamin D Guaita S, Puig I, Franci C, Garrido M, Dominguez D, Batlle E et al. receptor expression and responsiveness in human colon cancer. Nat (2002). Snail induction of epithelial to mesenchymal transition in Med 10: 917–919. tumor cells is accompanied by MUC1 repression and ZEB1 Park CC, Bissell MJ, Barcellos-Hoff MH. (2000). The influence of the expression. J Biol Chem 277: 39209–39216. microenvironment on the malignant phenotype. Mol Med Today 6: Hanahan D, Weinberg RA. (2000). The hallmarks of cancer. Cell 100: 324–329. 57–70. Peinado H, Olmeda D, Cano A. (2007). Snail, Zeb and bHLH factors Herzig M, Christofori G. (2002). Recent advances in cancer research: in tumour progression: an alliance against the epithelial phenotype? mouse models of tumorigenesis. Biochim Biophys Acta 1602: 97–113. Nat Rev Cancer 7: 415–428. Hu M, Yao J, Carroll DK, Weremowicz S, Chen H, Carrasco D et al. Pena C, Garcia JM, Garcia V, Silva J, Dominguez G, Rodriguez R (2008). Regulation of in situ to invasive breast carcinoma transition. et al. (2006). The expression levels of the transcriptional regulators Cancer Cell 13: 394–406. p300 and CtBP modulate the correlations between SNAIL, ZEB1, Kurose K, Hoshaw-Woodard S, Adeyinka A, Lemeshow S, Watson E-cadherin and vitamin D receptor in human colon carcinomas. Int PH, Eng C. (2001). Genetic model of multi-step breast carcinogen- J Cancer 119: 2098–2104. esis involving the epithelium and stroma: clues to tumour- Pena C, Garcia JM, Silva J, Garcia V, Rodriguez R, Alonso I et al. microenvironment interactions. Hum Mol Genet 10: 1907–1913. (2005). E-cadherin and vitamin D receptor regulation by SNAIL Lakhani SR, Chaggar R, Davies S, Jones C, Collins N, Odel C et al. and ZEB1 in colon cancer: clinicopathological correlations. Hum (1999). Genetic alterations in ‘normal’ luminal and myoepithelial Mol Genet 14: 3361–3370. cells of the breast. J Pathol 189: 496–503. Perez-Moreno M, Jamora C, Fuchs E. (2003). Sticky business: Larson PS, de las MA, Bennett SR, Cupples LA, Rosenberg CL. orchestrating cellular signals at adherens junctions. Cell 112: 535–548. (2002). Loss of heterozygosity or allele imbalance in histologically Perl AK, Wilgenbus P, Dahl U, Semb H, Christofori G. (1998). A normal breast epithelium is distinct from loss of heterozygosity causal role for E-cadherin in the transition from adenoma to or allele imbalance in co-existing carcinomas. Am J Pathol 161: carcinoma. Nature 392: 190–193. 283–290. Postigo AA, Depp JL, Taylor JJ, Kroll KL. (2003). Regulation of Larson PS, de las MA, Cupples LA, Huang K, Rosenberg CL. (1998). Smad signaling through a differential recruitment of coactivators Genetically abnormal clones in histologically normal breast tissue. and corepressors by ZEB proteins. EMBO J 22: 2453–2462. Am J Pathol 152: 1591–1598. Ren ZP, Hedrum A, Ponten F, Nister M, Ahmadian A, Lundeberg J Lazarova DL, Bordonaro M, Sartorelli AC. (2001). Transcriptional et al. (1996). Human epidermal cancer and accompanying pre- regulation of the vitamin D(3) receptor gene by ZEB. Cell Growth cursors have identical p53 mutations different from p53 mutations Differ 12: 319–326. in adjacent areas of clonally expanded non-neoplastic keratinocytes. Li Y, Liu W, Hayward SW, Cunha GR, Baskin LS. (2000). Plasticity Oncogene 12: 765–773. of the urothelial phenotype: effects of gastro-intestinal mesenchyme/ Sahai E, Marshall CJ. (2003). Differing modes of tumour cell invasion stroma and implications for urinary tract reconstruction. have distinct requirements for Rho/ROCK signalling and extra- Differentiation 66: 126–135. cellular proteolysis. Nat Cell Biol 5: 711–719. Li Z, Moore DH, Meng ZH, Ljung BM, Gray JW, Dairkee SH. Sozzi G, Miozzo M, Tagliabue E, Calderone C, Lombardi L, Pilotti S (2002). Increased risk of local recurrence is associated with allelic et al. (1991). Cytogenetic abnormalities and overexpression of loss in normal lobules of breast cancer patients. Cancer Res 62: receptors for growth factors in normal bronchial epithelium and 1000–1003. tumor samples of lung cancer patients. Cancer Res 51: 400–404.

Oncogene EMT genes in normal tissue adjacent to tumor CPen˜a et al 4385 Tan C, Costello P, Sanghera J, Dominguez D, Baulida J, De Herreros Waridel F, Estreicher A, Bron L, Flaman JM, Fontolliet C, Monnier P AG et al. (2001). Inhibition of integrin linked kinase (ILK) et al. (1997). Field cancerisation and polyclonal p53 mutation in the suppresses beta-catenin-Lef/Tcf-dependent transcription and upper aero-digestive tract. Oncogene 14: 163–169. expression of the E-cadherin repressor, snail, in APCÀ/À human Werb Z. (1997). ECM and cell surface proteolysis: regulating cellular colon carcinoma cells. Oncogene 20: 133–140. ecology. Cell 91: 439–442. Tomita N, Jiang W, Hibshoosh H, Warburton D, Kahn SM, Wolf K, Mazo I, Leung H, Engelke K, von Andrian UH, Deryugina Weinstein IB. (1992). Isolation and characterization of a highly EI et al. (2003). Compensation mechanism in tumor cell migration: malignant variant of the SW480 human colon cancer cell line. mesenchymal-amoeboid transition after blocking of pericellular Cancer Res 52: 6840–6847. proteolysis. J Cell Biol 160: 267–277. Vleminckx K, Vakaet Jr L, Mareel M, Fiers W, van RF. (1991). Wu C, Keightley SY, Leung-Hagesteijn C, Radeva G, Coppolino M, Genetic manipulation of E-cadherin expression by epithelial Goicoechea S et al. (1998). Integrin-linked protein kinase regulates tumor cells reveals an invasion suppressor role. Cell 66: ﬁbronectin matrix assembly, E-cadherin expression, and tumori- 107–119. genicity. J Biol Chem 273: 528–536.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene