The transcription factor Snail controls epithelial ... - Nature

articles

The transcription factor Snail controls epithelial–mesenchymal transitions by repressing E-cadherin expression Amparo Cano*¶, Mirna A. Pérez-Moreno*†, Isabel Rodrigo*†‡, Annamaria Locascio†§, María J. Blanco§, Marta G. del Barrio§, Francisco Portillo* and M. Angela Nieto§# *Instituto de Investigaciones Biomédicas (CSIC-UAM), Arturo Duperier 4, 28029 Madrid, Spain §Instituto Cajal (CSIC), Doctor Arce 37, 28002 Madrid, Spain ‡Present address: Institute of Mammalian Genetics, GSF Research Centre, Ingolstaedter Landstrasse 1, 85764 Neuherberg, Germany ¶e-mail: [email protected] #e-mail: [email protected] †These authors contributed equally to this work

The Snail family of transcription factors has previously been implicated in the differentiation of epithelial cells into mesenchymal cells (epithelial–mesenchymal transitions) during embryonic development. Epithelial–mesenchymal transitions are also determinants of the progression of carcinomas, occurring concomitantly with the cellular acquisition of migratory properties following downregulation of expression of the adhesion protein E-cadherin. Here we show that mouse Snail is a strong repressor of transcription of the E-cadherin gene. Epithelial cells that ectopically express Snail adopt a fibroblastoid phenotype and acquire tumorigenic and invasive properties. Endogenous Snail protein is present in invasive mouse and human carcinoma cell lines and tumours in which E-cadherin expression has been lost. Therefore, the same molecules are used to trigger epithelial–mesenchymal transitions during embryonic development and in tumour progression. Snail may thus be considered as a marker for malignancy, opening up new avenues for the design of specific anti-invasive drugs.

E

76

a

1.5 Relative activit y

-cadherin-mediated cell–cell adhesion is essential for the formation and maintenance of embryonic epithelia1–4 and, subsequently, in the adult organism for maintenance of the homeostasis and architecture of epithelial tissues5,6. During development, E-cadherin expression is under strict spatiotemporal control and its downregulation is essential for certain morphogenetic movements within the embryo, many of which involve epithelial– mesenchymal transitions (EMTs)7,8. Slug, a transcription factor of the Snail family, is implicated in the EMTs required for the emigration of the neural crest from the neural tube and of the early mesoderm from the primitive streak in the avian embryo9. These phenotypic transitions are reminiscent of those that take place during the acquisition of the invasive phenotype in tumours of epithelial origin. Indeed, the process of invasion of epithelial cells frequently involves EMTs that occur concomitantly with the downregulation of E-cadherin expression and the acquisition of migratory properties. Functional perturbations of E-cadherin and/or catenins have been associated with the dedifferentiation/aggressiveness of tumours (reviewed in refs 10, 11) and even implicated in the transition from adenomas to invasive carcinomas12. Therefore, Ecadherin is thought to be an invasion-suppressor gene13–16 and its loss is considered to be diagnostic of a poor clinical prognosis. Thus, a knowledge of the molecular mechanisms that control its expression or function is of prime importance in understanding the process of tumour invasion. Expression of the E-cadherin gene is regulated by several elements located in the 5′ proximal sequences of its promoter17–20. Among them, an E-pal element containing two adjacent E-boxes has been identified in the mouse E-cadherin promoter. This E-pal element acts as a repressor in undifferentiated normal cells and transformed E-cadherin-deficient cells (that is, cells that do not express E-cadherin), in which it can overcome the effect of other positive factors acting in the proximal promoter21. However, a moderate positive regulatory role for the E-pal has also been observed in E-cadherin-expressing cells17,22,23. The nature of the transcription factor(s) that interact with this element, and the cor-

MCA3D

PDV

1

0.5 0

pcDNA3 mSnail

b -RT 0

wt-178 + – – +

Mock 12 24

mE-pal + – – +

48

wt-178 + – – +

Time (h)

E-cadherin

E-cadherin

Snail

Snail

GAPDH

GAPDH

mE-pal + – – +

-RT 0

Snail 12 24

48

Figure 1 Snail represses the activity of the E-cadherin promoter in epidermal cell lines. a, MCA3D and PDV cells were transfected with the wild-type (wt -178) or mutant (mE-pal) E-cadherin promoter fused to the CAT reporter gene in the presence of 1 µg control pcDNA3 or pcDNA3-Snail vector. 2 µg CMV-Luc reporter plasmid21 was co-transfected in each sample to normalize the transfection efficiency. Luciferase and CAT activities were assayed 24 h after transfection 21. The activity of the promoter constructs is expressed relative to that of the wt -178 construct in the presence of pcDNA3 for each cell line. Results represent the mean ± s.d. of two independent experiments. b, PDV cells were transfected with 8 µg control pMT-CB6 (Mock) or pMT-CB6-Snail (Snail) vectors; 16 h later, 100 µM ZnSO4 was added to the cells in fresh culture medium. Cells were collected at the time points indicated and analysed by RT-PCR for E-cadherin and Snail expression. The expression of GAPDH was analysed in the same samples as a control for the amount of cDNA present in each sample. The –RT lane shows the results of amplification in the absence of reverse transcriptase.

© 2000 Macmillan Magazines Ltd NATURE CELL BIOLOGY | VOL 2 | FEBRUARY 2000 | cellbio.nature.com

articles responding sequence in the human E-cadherin promoter, has so far remained elusive. Members of the Snail gene family are good candidates for transcription factors that interact with E-pal, as Slug has been implicated in EMTs during avian development9. Furthermore, the DNA-binding specificity of Snail-family members conforms to an E2-box consensus sequence present in the right half of the E-pal element in the mouse E-cadherin promoter and in the corresponding region of the human promoter17,19,24–26. Here we show that mouse Snail is a strong repressor of E-cadherin transcription that specifically interacts with the E-pal element of the mouse E-cadherin promoter through its E2-box sequence. Snail overexpression in different epithelial cells leads to a dramatic conversion towards a fibroblastic phenotype at the same time that E-cadherin expression is lost and tumorigenic and invasive/migratory properties are acquired. Furthermore, Snail is expressed in Ecadherin-deficient murine and human carcinoma cell lines and tumours and in dedifferentiated and invasive regions of carcinomas, but is absent from well differentiated, non-invasive mouse and human carcinomas.

E-cadherin

Snail

a

c nc

nc

al

a

al pm

8.5d.p.c.

d

e

pnc

e

np

pm

f

nc nc pe

g

e h

ps

i

em

em ps

j

ov k

Results Snail is a direct repressor of E-cadherin expression. To identify transcription factors that interact with the E-pal element we adopted a one-hybrid approach, using the oligomerized mouse Epal sequence (positions –90 to –70 of the gene) as bait and an NIH 3T3 complementary DNA library fused to the Gal4 activation domain as prey. By analysing transcription of a reporter gene required for growth under histidine-deficient conditions, we isolated 130 clones that interact with the construct carrying the wildtype E-pal element and which did not recognize an oligomerized mutant element. This mutant element carries two point mutations (GC to TT) that abolish the E2 box and derepress the proximal mouse E-cadherin promoter in E-cadherin-deficient cells22,23. Sequencing of the isolated clones showed that 49% encoded partial or complete mouse Snail cDNAs27,28, whereas a single clone coded for a partial sequence of mouse Slug29–31. These results are in agreement with the reported DNA-binding specificity of mouse Snail and members of the Drosophila Snail family, whose preferred sequence is a perfect match to the E2 box of the E-pal element in the E-cadherin promoter (GCAGGTG)24–26. We confirmed the specific binding of Snail to the E-pal element by electrophoretic mobility-shift assays, using the wild-type E-pal or oligonucleotides with a mutant E2 box together with a recombinant glutathione-Stransferase (GST)–Snail fusion protein (data not shown). To investigate the effect of Snail as a transcription factor in the context of the proximal E-cadherin promoter (position –178 to +92), we co-expressed the complete cDNA sequence of mouse Snail together with E-cadherin-promoter reporter constructs in mouse epidermal keratinocyte cell lines characterized by their high level of E-cadherin expression and E-cadherin promoter activity21,23,32. Cotransfection of Snail into E-cadherin-expressing MCA3D and PDV cells induced a strong repression (95% and 75%, respectively) of the wild-type E-cadherin promoter (Fig. 1a) but not of the E-cadherin promoter carrying the mutated E2 box described above. Similar results were obtained in other epithelial cell lines, such as mammary EpH4 cells, in which a 45% inhibition of the wild-type promoter was detected in the presence of Snail (data not shown). In E-cadherin-deficient HaCa4 cells, Snail also showed a strong specific repression (90%) of the transfected wild-type E-cadherin promoter (data not shown), although the absolute activity of the wild-type promoter construct in HaCa4 cells was much lower than that exhibited by MCA3D or PDV cells, as reported previously23. These results indicate that Snail acts as a repressor of the E-pal element of the Ecadherin promoter in both E-cadherin-positive and -deficient cells. To analyse the effect of Snail on endogenous E-cadherin expression over time, we transiently transfected PDV cells with the pMTCB6-Snail vector, in which Snail expression is under the control of

Slug

b

ov nc

9.5d.p.c.

E-cadherin Snail pm m l pm

e

pe ba

ba

hg

a

Figure 2 Expression of E-cadherin and Snail-family members in mouse embryos. Whole-mount in situ hybridization of 8.5-d.p.c. mouse embryos (a–c) and transverse paraffin sections of the same embryos taken at the level of the hindbrain (d–f) and of the primitive streak (g–i). j–m, Transverse sections of the hindbrain (j, k) and tail region (l, m) of 9.5-d.p.c. embryos. a, d, g, j, l, E-cadherin expression; b, e, h, k, m, Snail expression; c, f, i, Slug expression. Observe the expression of E-cadherin in the ectoderm up to the boundary with the neural plate (d, arrowhead). High levels of Snail transcripts can be detected at the edges of the neural plate (e); these high levels of Snail transcripts correspond to premigratory crest cells undergoing EMTs (white asterisk). Also note that Snail expression is maintained in neural crest cells (nc) while migrating (e). No expression of Slug is observed in the EMT region (white asterisk in f), although transcripts are detected in migratory neural crest cells (nc) in their way to the branchial arches (f). A similar situation is observed in the primitive-streak area. E-cadherin expression is detected in the ectoderm and downregulated in the epiblast of this region (g), which shows low levels of Snail expression (h, arrowhead). High levels of Snail transcripts are seen during the EMTs that occur concomitantly with ingression through the primitive streak (h, black asterisk). No expression of Slug is detected at the primitive streak (i, black asterisk) nor in the early mesoderm. The inverse correlation shown for Ecadherin and Snail transcripts can be also clearly observed at later stages of development (j–m). a, amnion; al, allantois; ba, branchial arch; e, ectoderm; em, early mesoderm; hg, hind gut; nc, neural crest; np, neural plate; ov, otic vesicle; pe, pharyngeal endoderm; pm, paraxial mesoderm; pnc, premigratory neural crest; ps, primitive streak. Scale bar indicates 200 µm.

the Zn2+-inducible methallothionein promoter (Fig. 1b). Strong expression of Snail messenger RNA was visible 12 h after induction and increased for up to 48 h. At the same time, we detected a 25% decrease in endogenous E-cadherin mRNA 12 h after induction, and a 56% reduction after 24 h (Fig. 1b). Considering an average transfection efficiency of about 30% of the cell population, the 56% decrease in overall E-cadherin transcription indicates a dramatic repressive effect in the transfected cells that follows Snail expression. Taken together, these results validate those obtained with the one-hybrid screening and show that Snail is a direct repressor of Ecadherin transcription, acting through binding to the E2 box present in the right-hand site of the E-pal element in the mouse Ecadherin promoter. Snail and E-cadherin expression during development. The expression of the Snail and Slug genes during early embryonic develop-


77

articles E-cadherin

Plakoglobin

a

b

c

d

e

f

g

h

MCA3D

PDV

Plakoglobin

c

(Snail) plasmid (b, d, f, h). The presence of E-cadherin (a, b, e, f) or plakoglobin (c, d, g, h) was analysed 48 h after transfection by immunofluorescence. The photographs show laser confocal microscopic images of the corresponding cultures.

Desmoplakin

d

Vimentin

e

Fibronectin

f

m

Snail

E-cadherin

b

Snail

Mock

Figure 3 Transient expression of Snail in epidermal keratinocytes induces the loss of E-cadherin and plakoglobin, loss of cell–cell adhesion and the appearance of membrane extensions. a–h, MCA3D (a–d) and PDV (e–h) cells were transfected with empty pcDNA3 vector (Mock) (a, c, e, g) or pcDNA3-Snail

a

Mock

Snail

-RT

Mock

Snail

Mock

h

i

j

k

l

Mock

g

n

Snail

Snail

GAPDH

E-cadherin Plakoglobin Vimentin Fibronectin

Figure 4 Stable transfection of Snail into MDCK cells induces an epithelial– mesenchymal conversion concomitantly with the loss of epithelial markers and the gain of mesenchymal markers. a, g, Phase-contrast images of living, subconfluent cultures of a mock-transfected clone (a) and a Snail-transfected clone (g). b–f, h–l, Laser confocal microscopic images showing E-cadherin (b, h), plakoglobin (c, i), desmoplakin (I and II) (d, j), vimentin (e, k) and fibronectin (f, l),

ment in the chick and mouse embryo has been described9,27,28,30,33. Their expression is dynamic. With regard to the processes with which we are concerned, chicken Slug and mouse Snail transcripts are present in undifferentiated mesoderm and in tissues undergoing EMTs, namely the precursors of the neural crest cells and the primitive streak. We compared the expression of E-cadherin with that of Snail and Slug during early mouse development (Fig. 2). At 7.5 (data not shown), 8.5 and 9.5 days post-coitum (d.p.c.), the expression of E-cadherin was inversely correlated with that of Snail, being found in the embryonic and extraembryonic ectoderm and in epithelial derivatives of the endoderm, but completely absent from all mesoderm and regions undergoing EMTs (Fig. 2a, d, g, j). In particular, the complementary pattern of expression of E-cadherin 78

immunofluorescence. m, The presence of Snail transcripts in mock- and Snailtransfected clones was analysed by RT-PCR. The expression of GAPDH was analysed in the same samples as a control for the amount of cDNA present in each sample. The –RT lane shows the results of amplification in the absence of reverse transcriptase. n, Western blot analyses of the indicated proteins in mock- and Snailtransfected clones.

and Snail could be clearly observed in the primitive streak at 8.5 d.p.c. (Fig. 2g, h) and in different regions of condensing mesenchyme at 9.5 d.p.c. (Fig. 2j–m). In contrast, no expression of mouse Slug was found in regions undergoing EMTs (Fig. 2f, i), and its expression pattern did not show any relationship to that of E-cadherin (Fig. 2). These data are in agreement with the hypothesis that Snail represses E-cadherin expression in regions undergoing EMTs during embryonic development. Snail induces a fibroblastic conversion in epithelial cells. To gain further insight into the role of Snail in E-cadherin downregulation and its involvement in EMTs, we ectopically expressed Snail in several epithelial cell lines. We first analysed transient expression of Snail in epidermal MCA3D and PDV keratinocyte cells, taking


a

CarB

Mock

HaCa4

c

PDV

b

-RT

a

MCA3D

articles

– – –

+ – –

++ + +

++ + +

E-cadherin

e

d

f P-cadherin

120

g

h

Snail

i

Slug

80 40

GAPDH

Snail

advantage of their ability to grow as isolated islands and to organize strong E-cadherin-mediated cell–cell contacts even at low density, as well as their poor motility32,34,35. Ectopic Snail expression induced the loss of E-cadherin at cell–cell contacts within 24–48 h of transfection in both cell types (Fig. 3b, f). Cytoplasmic aggregates could be detected in some transfected PDV cells (Fig. 3f), probably as a result of the internalization of the remaining E-cadherin molecules; E-cadherin is very stable in this cell type36. In addition, the expression of plakoglobin, which associates with E-cadherin and desmosomes, was also reduced in the two transfected epidermal cell lines (Fig. 3d, h). Concomitant with these changes, the morphology of the Snail-transfected cells was profoundly altered. We observed abundant membrane extensions and long filaments resembling filopodia in both epidermal cell lines (Fig. 3b, d, f, h). Similar morphological changes were observed in PDV cells after transient transfection with the pMT-CB6-Snail vector (data not shown). To extend our studies of the effects of Snail expression to other epithelial cell types, we stably transfected Snail into the well characterized MDCK cell line, which exhibits a prototypical epithelial phenotype and forms a polarized monolayer in culture. This phenotype was not affected by expression of the control vector in six independent isolated clones (Fig. 4a), which maintained the expression of E-cadherin (Fig. 4b), plakoglobin (Fig. 4c) and desmoplakins (Fig. 4d) in organized cell junctions. However, stable expression of Snail induced a dramatic conversion to a dedifferentiated fibroblastic phenotype. We confirmed the expression of Snail in the different transfectant clones by reverse transcription withpolymerase chain reaction (RT-PCR) analysis (Fig. 4m). MDCK cells transfected with Snail lost their ability to grow as a monolayer and to show contact-mediated growth inhibition. Instead, they formed networks of cells crossing over each other with extremely long membrane extensions (Fig. 4g). Immunofluorescence analysis of E-cadherin, plakoglobin and desmoplakins showed that Snailtransfected MDCK cells downregulated expression of E-cadherin and desmoplakins (Fig. 4h, j) and exhibited a disorganized plakoglobin distribution (Fig. 4i). Identical results were obtained in ten independently isolated transfectant clones. We also analysed

A375P

HT29P

LoVo

b

MCF7

Figure 5 Snail induces a migratory and invasive phenotype in epithelial cells. a–f, The motility/migratory behaviour of mock-transfected (a–c) and Snailtransfected (d–f) MDCK cells was analysed in an in vitro wound model. Subconfluent cultures of the mock clones and Snail-transfected clones were gently scratched with a pipette tip to produce a wound. Photographs of the cultures were taken immediately after the incision (a, d) and after 10 h (b, e) and 20 h (c, f) in culture. g–i, The invasive properties of mock- and Snail-transfected MDCK cells were analysed in an invasion assay on collagen-type-IV gels using the modified Boyden chamber. g, The cells present in the lower compartment and those adhered to the lower surface of the filter were collected and counted under a light microscope. h, i, Cells adhered to the lower surface of the filters were fixed and their nuclei stained with DAPI and visualized under a fluorescence microscope. In all cases, duplicate samples were analysed.

Tumorigenic Invasive Metastatic

MDA

Snail

T24

Mock Mock

-RT

Number of migrated cells

Snail

+ – –

+ + +

+ – –

+ + +

+ – –

+ – –

E-cadherin Snail GAPDH Tumorigenic Invasive Metastatic

Figure 6 Endogenous Snail is present in mouse and human invasive cell lines. a, Endogenous expression of Snail and E-cadherin is inversely correlated in normal and transformed epidermal keratinocyte cell lines. The expression of E-cadherin, Pcadherin, Snail and Slug was analysed by RT-PCR in a panel of mouse epidermal cells, ranging from well differentiated non-tumorigenic cells to dedifferentiated and highly aggressive spindle carcinoma cells. Snail was amplified only in the cell lines that showed repressed E-cadherin expression, whereas Slug was amplified in all cells analysed. P-cadherin was present in all the analysed lines with the exception of CarB cells, a spindle carcinoma cell line. b, The inverse correlation between E-cadherin and Snail expression was also observed in human carcinoma cell lines of various aetiologies. These cell lines were analysed for the presence of human E-cadherin and Snail transcripts by RT-PCR. E-cadherin was highly expressed in cell lines obtained from differentiated carcinomas (MCF7, HT29P and LoVo cells) and absent from dedifferentiated invasive/metastatic cell lines (MDA-MB435S and A375P cells) and from the T24 cell line, which carries a methylated E-cadherin promoter that prevents its expression. In contrast, Snail was highly expressed in the invasive cell lines and absent from those obtained from differentiated carcinomas.

MDCK transfectants for the presence of mesenchymal markers such as vimentin and fibronectin. Mock-transfected MDCK cells expressed low levels of vimentin, as reported for parental MDCK cells37 (Fig. 4e), whereas Snail-transfected MDCK cells showed much higher levels of this protein, which was organized in marked intermediate filaments all over the cells (Fig. 4k). In addition, mock-transfected cells showed a weak punctuate staining of fibronectin on the cell membrane (Fig. 4f) whereas Snail-transfected cells exhibited much higher levels of this protein, which was organized in an extracellular network (Fig. 4l) typical of fibroblastic cells35. Quantitative western-blot analyses confirmed the immunofluorescence results (Fig. 4n). The transition triggered by Snail in MDCK cells led us to analyse the tumorigenic and invasive/migratory properties of control and Snail-transfected cells. We first analysed the migratory properties of the transfectants in a wound assay. Snail transfectants exhibited a highly migratory behaviour and were able to invade a wound in the


79

articles E-cadherin

Snail

a

b

f

g

b

E-cadherin

Snail

Haematoxylin

c

d

e

h

i

j

PDV

CarB

k

l

m

DMBA TPA

H/E

Ecad:Sna

H/E

Figure 7 Snail expression in mouse epidermal tumours is associated with spindle-cell carcinomas and invasive areas of squamous-cell carcinomas. a–j, Tumours induced in nu/nu mice by injection of PDV cells (a–e) or CarB cells (f–j) were serially sectioned at 70 µm and hybridized with probes for mouse E-cadherin (a, c, f, h) or Snail (b, d, g, i). Adjacent sections of each tumour were stained with haematoxylin after paraffin embedding and sectioning at 10 µm to show the cellular morphology better (e, j). Low-power (a, b, f, g) and high-power (c–e, h–j) photographs of these tumours are shown. E-cadherin was highly expressed in well differentiated squamous-cell carcinomas (produced by injection of PDV cells), in which no Snail

expression was detected. Spindle-cell carcinomas (produced by injection of CarB cells) showed a fully dedifferentiated histology, in agreement with their complete absence of E-cadherin transcripts. These carcinomas expressed high levels of Snail. k–m, Double in situ hybridization for Snail (blue) and E-cadherin (red/brown) transcripts in chemically induced (with DMBA and TPA) squamous-cell carcinomas shows Snail expression in invasive areas (bounded by arrows in l, m), which have lost E-cadherin expression. Note the breakdown of the basement membrane in the invasive area. H/E, haematoxylin-and-eosin staining.

culture 8–10 h after the incision was made (Fig. 5e), by which time the mock-transfected cells had barely started to move towards the wound (Fig. 5b). In contrast to control transfectants, Snail transfectants had completely healed the wound after 20 h (Fig. 5c, f). We observed no differences in the growth rates of the mock- and Snailtransfected clones in culture. We further analysed the invasive capacity of the Snail transfectants in an invasion assay on gels composed of collagen type IV. Snail-transfected cells invaded and migrated through the collagen gels (Fig. 5g, i), whereas control cells were not invasive in this assay (Fig. 5g, h). In addition, subcutaneous injection of two independent Snail-transfected MDCK clones into athymic nu/nu mice gave rise to tumours at all injection sites (6 out of 6, per clone), each of which reached an external diameter of 1 cm at 2 weeks after injection. MDCK control clones remained non-tumorigenic for at least 2 months. Thus Snail expression in epithelial cells induces both EMTs and the acquisition of tumorigenic and invasive properties. Snail expression in tumour invasion. As the EMT process that occurs during embryonic development and is triggered by Snail in epithelial cell lines is reminiscent of that taking place during the acquisition of the invasive phenotype in epithelial tumours, we analysed the endogenous expression of Snail and E-cadherin in a panel of mouse epidermal keratinocyte cell lines. The analysis included epithelial (MCA3D) cells through to fully dedifferentiated cells with the spindle phenotype (CarB cells); the cells analysed differ in their tumorigenic, metastatic and invasive properties16,32,35,38 (Fig. 6a). We detected Snail mRNA in cell types that were highly invasive and metastatic, whereas it was completely absent from non-invasive epithelial cell lines. Furthermore, we detected a strong inverse correlation between Snail and E-cadherin expression in the different cell lines (Fig. 6a). In contrast, Slug expression was observed in all

cell lines analysed, regardless of the phenotype, and expression of Pcadherin (which encodes the other major cadherin expressed by epidermal keratinocytes32) did not show any apparent correlation with the expression of either Snail or Slug. To further investigate the relationship between E-cadherin and Snail, we analysed expression of both genes in a panel of human carcinoma cell lines. The carcinoma cell lines chosen included epithelial and dedifferentiated cells derived from carcinomas of various aetiologies, including breast (differentiated MCF7 and dedifferentiated MDA-MB435S cells), colon (differentiated HT29P and LoVo cells), bladder (differentiated T24 cells) and melanomas (A375P cells). RT-PCR analysis of E-cadherin and Snail expression in the different cell lines showed a strong inverse correlation between the expression of both mRNAs (Fig. 6b). We detected high levels of Snail mRNA in E-cadherin-deficient dedifferentiated MDAMB435S (MDA) and A375P cells, and either no expression or almost undetectable levels in E-cadherin-positive epithelial MCF7, HT29P and LoVo cells. The bladder transitional-cell carcinoma T24 cell line, which shows downregulated E-cadherin expression as a result of hypermethylation of the E-cadherin promoter39, did not exhibit Snail mRNA expression (Fig. 6b). We also analysed endogenous Snail and E-cadherin expression in epidermal tumours induced by PDV or CarB cells, as well as in chemically induced mouse skin tumours. We observed extensive Ecadherin expression in well differentiated squamous-cell carcinomas induced by PDV cells, with higher levels of transcript being associated with the differentiated and keratinized areas of the tumour (Fig. 7a, c). In contrast, E-cadherin transcripts were completely absent in spindle-cell carcinomas induced by CarB cells (Fig. 7f, h). Expression of Snail in the same tumours was inverse to that of E-cadherin. Snail was expressed at high levels in spindle-cell car-

80


articles E-cadherin

Snail

E-cadherin

Snail

H/E

a

b

c

d

e

f

g

h

i

j

k

l

m

n

o

MDA

A375P

LoVo

Figure 8 Endogenous Snail is expressed in human invasive tumours. The expression of E-cadherin (a, c, f, h, k, m) and Snail (b, d, g, i, l, n) was analysed by in situ hybridization in human primary tumours obtained after xenografting of different human cell lines. The first two columns show-low power photographs of tumour sections; the remaining columns show the corresponding high-power photographs.

e, j, o, Haematoxylin-and-eosin (H/E) staining of the different tumours. High levels of Snail transcripts were detected in the invasive tumours (b, d, MDA-MB435S; g, i, A375P), whereas no Snail expression was observed in the differentiated tumours (l, n, LoVo). As in the mouse tumours, an inverse correlation between E-cadherin and Snail expression was observed.

cinomas (CarB; Fig. 7g, i) whereas no expression was seen in the differentiated carcinomas (PDV; Fig. 7b, d). Double in situ hybridization analyses carried out in mouse skin primary tumours that were chemically induced by the DMBA/TPA (7,12-dimethyl-benz (a) anthracene/12-O-tetradecanoylphorbol13-acetate) protocol40 showed an inverse correlation between Ecadherin and Snail expression (Fig. 7l). The areas of Snail expression corresponded to invasive areas of the tumour, which had lost E-cadherin expression (Fig. 7l). Breakdown of the basement membrane was easily apparent in these invasive areas (Fig. 7l, m; area surrounded by arrows). Identical results were obtained in four independent primary tumours analysed. We also analysed the expression of both genes in primary human tumours obtained from xenografts of different carcinoma cell lines, using specific human probes in in situ hybridization studies (Fig. 8). Again, there was an inverse correlation between E-cadherin and Snail expression. Snail mRNA was abundant in dedifferentiated breast carcinomas (MDA; Fig. 8b, d) and melanomas (A375P; Fig. 8g, i) in which E-cadherin mRNA was completely absent (Fig. 8a, c, f, h), whereas it was undetectable in differentiated gastric carcinomas (LoVo; Fig. 8l, n) that maintained high levels of E-cadherin mRNA (Fig. 8k, m). Taken together, the results obtained with the murine and human carcinoma cell lines and tumours provide strong evidence that Snail is involved in the downregulation of E-cadherin transcription that occurs during the progression of malignant tumours.

determine the silencing of the E-cadherin gene in dedifferentiated transformed cells, and that such repression is capable of overcoming the activation of the E-cadherin gene by other, positively acting factors that interact with the GC-rich proximal region or the CCAAT box21. These results are also in agreement with the phenotype of Drosophila snail mutants, which fail to eliminate DE-cadherin-mediated adhesion during gastrulation41. Stable expression of Snail in epithelial MDCK cells induces a dramatic phenotypic transition concomitantly with the loss of Ecadherin expression and an increase in the expression of mesenchymal markers. The organization of the mesenchymal markers (vimentin and fibronectin) is reminiscent of that in prototypic fibroblastic cells. This phenotypic change is also accompanied by the acquisition of tumorigenic and invasive/migratory properties. Another Snail-family member, Slug, has a role in EMTs during avian development9. Loss of Slug function prevents EMTs both at gastrulation and during emigration of the neural crest9. We have proposed that the roles of Slug and Snail are inverted between chick and mouse, indicating that Snail, rather than Slug, might be a candidate for inducing EMTs in mammals30. The gain-of-function experiments described here confirm that mouse Snail is in fact able to drive the EMTs associated with the acquisition of the migratory and invasive phenotype, and provide insight into the mechanism by which Snail triggers this transition, namely, the repression of E-cadherin expression. It is worth mentioning here that, in our onehybrid screening, we identified only one Slug clone, but 64 Snail clones. However, we cannot eliminate the possibility that Slug, although not involved in E-cadherin repression, might participate in EMTs in mouse, possibly by contributing to the maintenance of the dedifferentiated phenotype42, as suggested by its expression in the already migratory neural crest cells30,31 (Fig. 2f). This suggestion is compatible with the finding that mouse Slug participates in desmosome dissociation in rat bladder epithelial cells29. Interestingly, all the Snail clones that bound to the E-pal element in our one-hybrid screen contained the four zinc fingers present in this family member27, indicating that all four zinc fingers are required for proper binding to the E2 box of the E-pal element. The fact that Snail can trigger EMTs in differentiated epithelial cells indicates that Snail is not only able to block cell–cell contacts through the inhibition of epithelial genes, but may also positively affect the expression of other genes involved in the control of motility and migration. That is, Snail may also act as a transcriptional activator26 of other genes required for the acquisition of the fibroblastic phenotype. In fact, previous studies of mouse epidermal

Discussion The stromal invasion of epithelial tumour cells represents the first step in the metastatic cascade of carcinomas. Different experimental approaches have established that disruption of the E-cadherinmediated adhesion system is a major event in the transition from non-invasive tumours to invasive malignant carcinomas11,12. An understanding of the molecular mechanisms that lead to downregulation of E-cadherin during tumour progression has been long awaited. Analysis of the regulation of E-cadherin expression supports the idea that there are repressor mechanisms involving specific E-boxes located in proximal elements of both the mouse and the human promoters17,19–23. Using the one-hybrid system, analysis of E-cadherin promoter activity and ectopic expression assays, we have identified the transcription factor Snail as a direct and strong repressor of E-cadherin expression. These results confirm our hypothesis that repression through the E-pal element helps to


81

articles keratinocytes have shown that a block in E-cadherin expression by stable antisense transfection, although able to lead to the acquisition of invasive and metastatic properties, is not sufficient to induce a full EMT16. Moreover, transfection of E-cadherin is not sufficient to revert the fibroblastic phenotype of spindle carcinoma cells (such as CarB cells, which we have shown here to express Snail) to the epithelial character, even in the presence of other epithelial molecules such as plakoglobin36,43. It is also noteworthy that particular Drosophila mutant snail alleles produce an intermediate cellular phenotype. These cells exhibit both mesodermal and ectodermal markers44, again indicating that Snail may regulate a battery of genes involved in the loss of the epithelial phenotype and the gain of the mesenchymal character, whose regulation might be independently affected in the mutants. In this sense, Snail may be considered to be the regulator of EMTs proposed in ref. 45. We have shown here that EMTs are required for the acquisition of an invasive phenotype and occur in epithelial cell lines that express Snail ectopically. These data, together with the observation that this transcription factor is endogenously expressed in E-cadherin-deficient malignant cells, dedifferentiated invasive areas of carcinomas and fully dedifferentiated tumours, indicate that Snail is an important regulator of invasiveness during tumour progression. This hypothesis is reinforced by the fact that the link between Snail expression and E-cadherin downregulation has not been observed in human cell lines in which silencing of E-cadherin is achieved by epigenetic mechanisms, such as hypermethylation of the promoter. Finally, our results show that the same molecules and mechanisms operate in EMTs during embryonic development9,46 and in the adult10,11, under both physiological and pathological conditions. This link between development and tumorigenesis opens up new avenues in cancer research, allowing the use of experimentally amenable systems and the use of Snail as a marker of tumour invasion. Our results also pave the way for the design of specific anti-invasive drugs. h

Methods

Analysis of the E-cadherin promoter.

MCA3D and PDV cells were co-transfected with 5 µg of the –178 construct or the mE-pal construct fused to the CAT reporter gene22 and 1 µg of pcDNA3-Snail or control pcDNA3 plasmids. CAT activity was assayed as described23, with the activity normalized to that of the wild-type promoter.

RT-PCR analysis. Poly(A)+ mRNA was isolated from the different cell lines using the Microfast Track isolation kit (Invitrogen). RT-PCR was carried out as described30 with specific primers. Mouse and human PCR products, except for that of human Snail, were obtained after 30 cycles of amplification with an annealing temperature of 60–65 °C. Amplification of human Snail required two rounds of 30 cycles each, the second round being performed on the product of the first one with internal specific primers and an annealing temperature of 68 °C. Sequences of the specific primers are available upon request.

Immunofluorescence and western blots. Staining for the different markers was performed on cells fixed in methanol or 3.7% paraformaldehyde as described34,35. Anti-E-cadherin antibody (ECCD-2; 1:200 dilution), anti-plakoglobin monoclonal antibody (1:200 dilution; Transduction Laboratories), anti-desmoplakin-I and -II monoclonal antibodies (1:5 dilution; Boehringer Mannheim), anti-vimentin monoclonal antibodies (1:200 dilution; Dako) and anti-fibronectin rabbit polyclonal antibodies 35 (1:100 dilution) were used with the appropriate secondary antibodies (1:100 dilution). Preparations were visualized using a Leica confocal microscope. Western blot analyses were carried out with the indicated antibodies as described34,35.

Migration and invasion assays.

Cells were seeded in T6-well culture dishes at a density of 3 × 105 cells per well. A wound was incised 24 h later in the central area of the confluent culture, which was incubated for a further 20 h after careful washing to remove detached cells and addition of fresh medium. Cultures were observed at timely intervals and phase-contrast pictures were taken of the wounded area using an inverted Zeiss Axiovert microscope. Invasion assays on collagen-type-IV gels were carried out using the two-compartment Boyden chamber in duplicate samples as described16, except that we seeded 4 × 105 cells on the top of the filters in the upper chamber. The cells in the lower compartment and those in the lower surface of the filter were collected after 8 h of incubation and counted. In a parallel experiment, the nuclei of the cells present in the lower surface of the filters were stained with 4,6-diamidinophenylindole (DAPI) after fixing in methanol and careful removal of the cells present in the upper surface of the filters.

Histological analysis and in situ hybridization of embryos and tumour sections.

Plasmid constructions and one-hybrid screening. A multimerized version (six copies) of the wild-type or a mutant E-pal element of the E-cadherin promoter17,22,23 was inserted into the reporter plasmid pHISi (Clontech), which drives the expression of the HIS3 gene. The recombinant plasmids were integrated into the URA3 chromosomal locus of the yeast strain YM4271 (Clontech) and the yeast strain containing the E-palHIS reporter gene was transformed with an NIH 3T3 cDNA library fused to the Gal4 activation domain into the pACT2 expression vector (Clontech). The transformants (3 × 106 independent clones) were tested for growth on medium lacking histidine in the presence of 20 mM 3-aminotriazole. 300 positive clones were picked out and their plasmids isolated from the yeast. To eliminate false positives, these plasmids were separately introduced into yeast cells containing either the E-palHIS or the mE-palHIS3 reporter genes. Plasmids that conferred expression of the reporter gene in only the E-palHIS host (130 out of 300 putative positives) were chosen for further analysis. All of the 130 positive inserts were linked in-frame to the Gal4 activation domain and 49% of them carried cDNA inserts encoding Snail. The complete cDNA sequence was subcloned into the pcDNA3 vector (Invitrogen) and the pMT-CB6 vector47 under the control of the cytomegalovirus and the sheep metallothionein I promoters, respectively.

Cell culture and generation of tumours. The origin, tumorigenic properties and expression of E-cadherin in the murine keratinocyte cell lines MCA3D, PDV, HaCa4 and CarB have been described32,34,43 and are summarized in Fig. 6a. Human cell lines derived from differentiated colon carcinomas (HT29P and LoVo), mammary adenocarcinomas (MCF7 and MDA-MB435S), bladder transitional-cell carcinoma (T24) and melanoma (A375P) and primary xenografted tumours were provided by A. Fabra. The characteristics of these human cell lines are summarized in Fig. 6b. Cells were grown in DMEM (CarB, MDCK-II and NIH 3T3 cells), Ham’s F12 medium supplemented with a complete set of amino acids (MCA3D, PDV and HaCa4 cells) or DMEM:Ham’s F12 medium (1:1, Gibco; human cell lines), supplemented with 10 µg ml–1 insulin for the mammary cells. Tumours were induced in athymic male nu/nu mice by subcutaneous injection (for MDCK, Snail-transfected MDCK, PDV and CarB cells) or orthotopic injection (A375P, LoVo and MDAMB435S cells) (1 × 106 cells per injection site) as described32. Animals were obtained from the animalproduction unit of the IFA-CREDO factory (Lyon, France) and maintained in sterile conditions according to institutional guidelines. Injected animals were observed every 2 days and killed when the tumours reached an external diameter of 1.5–2.0 cm. Mouse skin tumours were also induced by the two-stage DMBA/TPA protocol in the back skin of Balb-C mice as described40. Tumours were excised and immediately frozen in isopenthane-cooled liquid nitrogen and maintained at –70 °C until further analysis.

Transient and stable transfections. Transfections were carried out as described23, but using lipofectamine plus (Life Technologies) according to the manufacturer’s instructions. Stable transfectants were generated from MDCK cells after selection with 400 µg ml–1 G418. Ten and six independent clones were isolated from pcDNA3-Snail and from

82

control pcDNA3 transfections, respectively. Transient transfections were carried out with pcDNA3mSnail and pMT-CB6-mSnail vectors and their corresponding controls. PDV cells were grown in F-75 flasks; 16 h after transfection (pMT-CB6-mSnail and pMT-CB6 control vectors), 100 µM ZnSO4 was added to the cultures to induce the expression of the metallothionein I promoter. Cells were collected at the indicated times and analysed for E-cadherin and Snail expression by RT-PCR. For immunofluorescence analysis, MCA3D and PDV cells were grown on coverslips in 6-cm cell-culture dishes; 24–48 h after transfection (pcDNA3-mSnail and pcDNA3 control vectors), they were fixed in cold methanol (–20 °C) for 30 s and washed several times in PBS.

Mouse embryos were obtained from natural matings of Balb-C (Harlan) and SJL (Jackson) mouse strains. Ages were determined as days post-coitum (d.p.c.), the day on which the vaginal plug was detected being designated 0.5 d.p.c. Mouse and human tumours were gelatin-embedded and sectioned in a vibratome to obtain 70-µm slices. The odd-numbered sections were used for in situ hybridization and the even-numbered sections were subsequently paraffin-embedded, sectioned at 10 µm and stained with haematoxylin and eosin. Digoxigenin-labelled antisense riboprobes were prepared and used for in situ hybridization in whole embryos or free-floating vibratome slices as described48. The mouse E-cadherin probe corresponded to nucleotides 1,600–3,100 of the complete cDNA sequence and the mouse Snail and Slug probes were as described30. The final human Snail and E-cadherin PCR products described above and corresponding to the sequences extending from positions 695 to 1,297 and positions 3,205 to 3,735, respectively, of the translation-initiation codon were subcloned in the pGEM-T vector (Promega) and used as templates with which to synthesize digoxigenin-labelled probes. Double labelling was done by simultaneous hybridization with two probes. The mouse Snail probe was labelled with fluorescein-UTP (Boeringher Mannheim) and the E-cadherin probe was labelled with digoxigenin-UTP (Boeringher Mannheim). After hybridization, the slices were incubated with alkaline-phosphatase-conjugated anti-digoxigenin and anti-fluorescein antibodies. The alkaline-phosphatase activity was detected by incubation with the following substrates: NBT/ BCIP for Snail and INT/BCIP for E-cadherin (both substrates from Boeringher Mannheim) according to the manufacturer’s instructions. Sections were then cleared in 50% glycerol in PBS and mounted in the same solution containing 0.02% sodium azide. The slices were photographed with a Leica DMR microscope under Nomarski optics. RECEIVED 30 NOVEMBER 1999; REVISED 20 DECEMBER 1999; ACCEPTED 20 DECEMBER 1999; PUBLISHED 13 JANUARY 2000.

1. Bursdal, C. A., Damsky, C. H. & Pedersen, R. A. The role of E-cadherin and integrins in mesoderm differentiation and migration at the mammalian primitive streak. Development 118, 829–844 (1993). 2. Levine, E., Lee, C. H., Kintner, C. & Gumbiner, B. M. Selective disruption of E-cadherin function in early Xenopus embryos by a dominant negative mutant. Development 120, 901–909 (1994). 3. Larue, L., Ohsugi, M., Hirchenhain, J. & Kemler, R. E-cadherin null mutant embryos fail to form a trophoectoderm epithelium. Proc. Natl Acad. Sci. USA 91, 8263–8267 (1994). 4. Riethmacher, D., Brinkmann, V. & Birchmeier, C. A targeted mutation in the mouse E-cadherin gene results in defective preimplantation development. Proc. Natl Acad. Sci. USA 92, 855–859 (1995). 5. Gumbiner, B. M. Cell adhesion: the molecular basis of tissue architecture and morphogenesis. Cell 84, 345–357 (1996). 6. Wheelock, M. J. & Jensen, P. J. Regulation of keratinocyte intercellular junction organization and epidermal morphogenesis by E-cadherin. J. Cell Biol. 117, 415–425 (1992). 7. Takeichi, M. Morphogenetic roles of classic cadherins. Curr. Opin. Cell Biol. 7, 619–627 (1995). 8. Huber, O., Bierkamp, C. & Kemler, R. Cadherins and catenins in development. Curr. Opin. Cell Biol.


articles 8, 685–691 (1996). 9. Nieto, M. A., Sargent, M. G., Wilkinson, D. G. & Cooke, J. Control of cell behaviour during vertebrate development by Slug, a zinc finger gene. Science 264, 835–839 (1994). 10. Takeichi, M. Cadherins in cancer: implications for invasion and metastasis. Curr. Opin. Cell Biol. 5, 806–811 (1993). 11. Birchmeier, W. & Behrens, J. Cadherin expression in carcinomas: role in the formation of cell junctions and the prevention of invasiveness. Biochim. Biophys. Acta 1198, 11–26 (1994). 12. Perl, A. K., Wilgenbus, P., Dahl, U., Semb, H. & Christofori, G. A causal role for E-cadherin in the transition from adenoma to carcinoma. Nature 392, 190–193 (1998). 13. Frixen, U. H. et al. E-cadherin-mediated cell-cell adhesion prevents invasiveness of human carcinoma cells. J. Cell Biol. 113, 173–185 (1991). 14. Vleminckx, K., Vakaet, L. J., Mareel, M., Fiers, W. & Van Roy, F. Genetic manipulation of E-cadherin expression by epithelial tumor cells reveals an invasion suppressor role. Cell 66, 107–119 (1991). 15. Miyaki, M. et al. Increased cell-substratum adhesion, and decreased gelatinase secretion and cell growth, induced by E-cadherin transfection of human colon carcinoma cells. Oncogene 11, 2547– 2552 (1995). 16. Llorens, A. et al. Downregulation of E-cadherin in mouse skin carcinoma cells enhances a migratory and invasive phenotype linked to matrix metalloproteinase-9 gelatinase expression. Lab. Invest. 78, 1–12 (1998). 17. Behrens, J., Löwrick, O., Klein, H. L. & Birchmeier, W. The E-cadherin promoter: functional analysis of a GC-rich region and an epithelial cell-specific palindromic regulatory element. Proc. Natl Acad. Sci. USA 88, 11495–11499 (1991). 18. Ringwald, M., Baribault, H., Schmidt, C. & Kemler, R. The structure of the gene coding for the mouse cell adhesion molecule uvomorulin. Nucleic Acids Res. 19, 6533–6539 (1991). 19. Bussemakers, M. J., Giroldi, L. A., van Bokhoven, A. & Schalken, J. A. Transcriptional regulation of the human E-cadherin gene in human prostate cancer cell lines: characterization of the human Ecadherin gene promoter. Biochem. Biophys. Res. Commun. 203, 1284–1290 (1994). 20. Giroldi, L. A. et al. Role of E-boxes in the repression of E-cadherin expression. Biochem. Biophys. Res. Commun. 241, 453–458 (1997). 21. Rodrigo, I., Cato, A. C. B. & Cano, A. Regulation of E-cadherin gene expression during tumor progression: the role of a new Ets-binding site and the E-pal element. Exp. Cell Res. 248, 358–371 (1999). 22. Hennig, G., Löwrick, O., Birchmeier, W. & Behrens, J. Mechanisms identified in the transcriptional control of epithelial gene expression. J. Biol. Chem. 271, 595–602 (1996). 23. Faraldo, M. L. M., Rodrigo, I., Behrens, J., Birchmeier, W. & Cano, A. Analysis of the E-cadherin and P-cadherin promoters in murine keratinocyte cell lines from different stages of mouse skin carcinogenesis. Mol. Carcinogen. 20, 33–47 (1997). 24. Mauhin, V., Lutz, Y., Dennefeld, C. & Alberga, A. Definition of the DNA-binding site repertoire for the Drososphila transcription factor SNAIL. Nucleic Acids Res. 21, 3951–3957 (1993). 25. Fuse, N., Hirose, S. & Hayashi, S. Diploidy of Drosophila imaginal cells is maintained by a transcriptional repressor encoded by escargot. Genes Dev. 8, 2270–2281 (1994). 26. Nakayama, H., Scott, I. C. & Cross, J. C. The transition to endoreduplication in trophoblast giant cells is regulated by the mSna zinc finger transcription factor. Dev. Biol. 199, 150–163 (1998). 27. Nieto, M. A., Bennet, M. F., Sargent, M. G. & Wilkinson, D. G. Cloning and developmental expression of Sna, a murine homologue of the Drosophila snail gene. Development 116, 227–237 (1992). 28. Smith D. E, Del Amo, F. F. & Gridley, T. Isolation of Sna, a mouse gene homologous to the Drosophila genes snail and escargot: its expression pattern suggests multiple roles during postimplantation development. Development 116, 1033–1039 (1992). 29. Savagner, P., Yamada, K. M. & Thiery, J. P. The zinc finger protein Slug causes desmosome dissociation, an initial and necessary step in growth factor-induced epithelial-mesenchymal transition. J. Cell. Biol. 137, 1403–1419 (1997). 30. Sefton, M., Sánchez, S. & Nieto M. A. Conserved and divergent roles for members of the Snail family of transcription factors in the chick and mouse embryo. Development 125, 3111–3121 (1998).

31. Jiang, R., Lan, Y., Norton, C. R., Sundberg, J. P. & Gridley, T. The Slug gene is not essential for mesoderm or neural crest development in mice. Dev. Biol. 198, 277–285 (1998). 32. Navarro, P. et al. A role for the E-cadherin cell-cell adhesion molecule in tumor progression of mouse epidermal carcinogenesis. J. Cell Biol. 115, 517–533 (1991). 33. Isaac, A., Sargent, M. G. & Cooke, J. Control of vertebrate left-right asymmetry by a Snail-related zinc finger gene. Science 275, 1301–1304 (1997). 34. Lozano, E. & Cano, A. Cadherin/catenin complexes in murine epidermal keratinocytes: E-cadherin complexes containing either β-catenin or plakoglobin contribute to stable cell-cell contacts. Cell Adhes. Commun. 6, 51–67 (1998). 35. Gómez, M., Navarro, P. & Cano, A. Cell adhesion and tumor progression in mouse skin carcinogenesis: increased synthesis and organization of fibronectin is associated with the undifferentiated spindle phenotype. Invasion Metastasis 14, 17–26 (1994). 36. Lozano, E. & Cano, A. Induction of mutual stabilization and retardation of tumor growth by coexpression of plakoglobin and E-cadherin in mouse skin spindle carcinoma cells. Mol. Carcinogen. 21, 273–287 (1998). 37. Hay, E. D. An overview of epithelio-mesenchymal transformation. Acta Anat. 154, 8–20 (1995). 38. Frontelo, P. et al. Transforming growth factor β1 induces squamous carcinoma cell variants with increased metastatic abilities and a disorganized cytoskeleton. Exp. Cell Res. 244, 420–432 (1998). 39. Yoshiura, K. et al. Silencing of the E-cadherin invasion-suppressor gene by CpG methylation in human carcinomas. Proc. Natl Acad. Sci. USA 92, 7416–7419 (1995). 40. Cano, A. et al. Expression pattern of the cell adhesion molecules E-cadherin, P-cadherin and α6β4 integrin is altered in pre-malignant skin tumors of p53-deficient mice. Int. J. Cancer 65, 254–262 (1996). 41. Oda, H., Tsukita, S. & Takeichi, M. Dynamic behavior of the cadherin-based cell-cell adhesion system during Drosophila gastrulation. Dev. Biol. 203, 435–450 (1998). 42. Ros, M., Sefton, M. & Nieto, M. A. Slug, a zinc finger gene previously implicated in the early patterning of the mesoderm and the neural crest, is also involved in chick limb development. Development 124, 1821–1829 (1997). 43. Navarro, P., Lozano, E. & Cano, A. Expression of E- or P-cadherin is not sufficient to modify the morphology and the tumorigenic behavior of murine spindle carcinoma cells. Possible involvement of plakoglobin. J. Cell Sci. 105, 923–934 (1993). 44. Hemavathy, K., Meng, X. & Ip, Y. T. Differential regulation of gastrulation and neuroectodermal gene expression by Snail in the Drosophila embryo. Development 124, 3683–3691(1997). 45. Hay, E. D. Epithelial-mesenchymal transitions. Semin. Dev. Biol. 1, 347–356 (1990). 46. Duband, J. L., Monier, F., Delannet, M. & Newgreen, D. Epithelium-mesenchyme transition during neural crest development. Acta Anat. 154, 63–78 (1995). 47. Cook, D. M., Hinkes, M. T., Bernfield, M. & Rauscher F. J. III Transcriptional activation of the syndecan-1 promoter by the Wilms’ tumor protein WT1. Oncogene 13, 1789–1799 (1996). 48. Nieto, M. A., Patel, K. & Wilkinson, D. G. In situ hybridisation analysis of chick embryos in whole mount and tissue sections. Methods Cell Biol. 51, 220–235 (1996). ACKNOWLEDGEMENTS We thank G. Dhont for help with the one-hybrid screen; J.J. Arredondo for helpful advice in the design of this screen; A. Fabra for the human cell lines and tumours; M. Manzanares for help with the isolation of the human probe; M. Quintanilla for help in the tumorigenic assays; C. Bailón for assistance with confocal microscopy; C. Martinez for providing mouse embryos; A. Montes for technical assistance; M. Takeichi for the E-cadherin probe and ECCD-2 antibody; J. Behrens for E-cadherin promoter constructs; F.J. Rauscher for the pMT-CB6 vector; and M. Sefton for critical reading of the manuscript and editorial assistance. This work has been supported by the Spanish Ministry of Culture (grants DGICYT-PM950024 and PM98-0125 to M.A.N., SAF95-0818 and SAF98-0085-C03-01 to A.C., and PB97-0054 to F.P.), the Comunidad Autónoma de Madrid (grant 08.1/0020/97 to A.C. and M.A.N.) and the EU (grant FMRX-CT96-0065 to M.A.N). Correspondence and requests for materials should be addressed to M.A.N. or A.C.


83