Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Epithelial–mesenchymal transition in development and cancer: role of phosphatidylinositol 3′ kinase/AKT pathways


Epithelial–mesenchymal transition (EMT) is an important process during development by which epithelial cells acquire mesenchymal, fibroblast-like properties and show reduced intercellular adhesion and increased motility. Accumulating evidence points to a critical role of EMT-like events during tumor progression and malignant transformation, endowing the incipient cancer cell with invasive and metastatic properties. Several oncogenic pathways (peptide growth factors, Src, Ras, Ets, integrin, Wnt/β-catenin and Notch) induce EMT and a critical molecular event is the downregulation of the cell adhesion molecule E-cadherin. Recently, activation of the phosphatidylinositol 3′ kinase (PI3K)/AKT axis is emerging as a central feature of EMT. In this review, we discuss the role of PI3K/AKT pathways in EMT during development and cancer with a focus on E-cadherin regulation. Interactions between PI3K/AKT and other EMT-inducing pathways are presented, along with a discussion of the therapeutic implications of modulating EMT in order to achieve cancer control.


Epithelial–mesenchymal transition (EMT) is a cellular mechanism long recognized as a central feature of normal development. Several developmental milestones, including gastrulation, neural crest formation and heart morphogenesis, rely on the plastic transition between epithelium and mesenchyme. More recent studies have revealed that similar but physiopathological transitions occur during the progression of epithelial tumors, endowing cancer cells with increased motility and invasiveness. Multiple oncogenic pathways mediated by peptide growth factors, Src, Ras, Ets, integrin, Wnt/β-catenin and Notch signaling, induce EMT.

A critical molecular feature of EMT is the downregulation of E-cadherin, a cell adhesion molecule present in the plasma membrane of most normal epithelial cells. E-cadherin acts de facto as a tumor suppressor inhibiting invasion and metastasis, and it is frequently repressed or degraded during transformation.

A variety of signal transduction pathways impinge on the regulation of E-cadherin levels or subcellular distribution. Very recently, the oncogenic serine/threonine kinase AKT (also known as PKB), a downstream effector of the phosphatidylinositol 3′ kinase (PI3K), has been shown to repress transcription of the E-cadherin gene. This transcriptional repression induces cellular responses leading to the conversion of epithelial cells into invasive mesenchymal cells. Cells producing a constitutively active form of Akt produce a transcription factor, Snail, which is known to repress expression of the E-cadherin gene (Grille et al., 2003).

It should be emphasized that a large body of literature concerning EMT is available in model organisms (Xenopus, Drosophila, sea urchin and Caenorhabditis elegans), but it is not known whether EMT in these non-mammalian species involves AKT. For this reason, we concentrate on mammals. This review focuses on EMT in development and cancer with a particular emphasis on the PI3K/AKT pathways involved in the regulation of E-cadherin, motility and invasion. Since the PI3K/AKT axis is frequently activated in human cancer (see reviews by Altomare and Testa; Majumder and Sellers, in this issue), we will also discuss potential therapeutic strategies based on inhibition of critical steps of EMT.

Definition of EMT

Epithelial and mesenchymal cells represent two of the main cell types in mammals. Epithelial cells are characterized by: (i) cohesive interactions among cells, facilitating the formation of continuous cell layers; (ii) existence of three membrane domains: apical, lateral and basal; (iii) presence of tight junctions between apical and lateral domains; (iv) apicobasal polarized distribution of the various organelles and cytoskeleton components; and (v) lack of mobility of individual epithelial cells with respect to their local environment.

Multicellular mesenchymal architectures differ from multicellular epithelial architectures in having: (i) loose or no interactions among cells, so that no continuous cell layer is formed; (ii) no clear apical and lateral membranes; (iii) no apicobasal polarized distribution of organelles and cytoskeleton components; and (iv) motile cells that may even have invasive properties. During development, certain cells can switch from an epithelial to a mesenchymal status by means of a tightly regulated process defined as the EMT, which is associated with a number of cellular and molecular events. In some cases, EMT is reversible and cells undergo the reciprocal mesenchymal to epithelial transition (MET). A schematic view of the molecular and structural features associated with epithelial and mesenchymal cells during EMT and MET is shown in Figure 1.

Figure 1

Cellular modifications associated with epithelial–mesenchymal transition (EMT). (a) Schematic of EMT. Epithelial cells (in blue) adhere to each other through adherens junctions, using E-cadherin (E-cad), and their desmosomes (tight junctions) constituted by various proteins such as desmoplakin (dp). Mesenchymal cells (in red) have a totally different morphology, they are neither adherent nor apically polarized. E-cadherin is sequestrated in perinuclear vesicles and desmoplakin is internalized. The amount of E-cadherin and desmoplakin can be greatly reduced in the mesenchymal status. The intermediate filament protein vimentin (vim) is induced in mesenchymal cells. (b–g) EMT induced by AKT in the human squamous cell carcinoma line SCC15 (reproduced from Grille et al. (2003). Cancer Res., 63, 2172–2178). Photomicrography of epithelial cells (b) and AKT-induced mesenchymal cells (c). E-cadherin (d and e) and vimentin (f and g) immunolocalization in epithelial (d and f) and AKT-induced mesenchymal cells (e and g)

EMT and evolution

During evolution, four major steps occurred in terms of cell–cell adhesion. The first step was the transition from simple eukaryotic unicellular organisms to multicellular structures by means of organized connections of individual cells. The appearance of true multicellular organisms with specialized tissue functions coincided with the appearance of two epithelial layers. The next major step was the appearance of a third cellular layer, the mesoderm, a relatively ‘loose’ (i.e. mesenchymal) tissue. The mesoderm appeared as a consequence of the emergence of EMT, allowing cells to make the transition between epithelial and mesenchymal architecture. Finally, the reverse transition (MET) occurred later with the evolution of coelomates: their embryos have a cavity (i.e. coelom) that is delimited by epithelial cells derived from mesodermal cells. Thus, evolution has led to the establishment of a series of molecular and cellular programs for reversible transitions between epithelium and mesenchyme. Such transitions are important mechanisms of mammalian development and also occur, in part or completely, during oncogenesis.

EMT and development

In mammals, interconversions between epithelium and mesenchyme occur during early embryonic development. The first MET occurs during preimplantation, with formation of the trophectoderm, and the first EMT, during gastrulation, with formation of the mesoderm. Mouse mutants studied to date have provided little information concerning the possible role of PI3K/Akt during early development, but the role of cadherin regulation in this process is well established (Larue et al., 1996).

In addition, several additional EMT conversions occur later in embryonic development. Examples include the following developmental milestones: formation of neural crest cells from the neural tube on embryonic day 8 (E8); formation of the atrial and ventricular mesenchymal septa from the endothelium during heart development on E8; formation of the sclerotome from somites on E9; formation of coronary vessel progenitor cells from the epicardium around E10–11; formation of palate mesenchymal cells from the oral epithelium on E13.5; and formation of mesenchymal cells during regression of the Mullerian tract on E15. In general, these transitions are associated with similar main molecular events but are clearly regulated differently. The role of PI3K and Akt in these processes has not yet been determined. As mentioned above, formation of the trophectoderm and mesoderm represent prototypical MET and EMT, respectively.

Trophectoderm formation

The formation of the trophectoderm, the first epithelium in mammals, occurs after segmentation of the egg. At this stage, the embryo has eight cells and resembles a small blackberry, mora in Latin, and is therefore referred to as the ‘morula’. The eight cells of the morula, the blastomeres, are very similar in shape and morphology. The cells divide and the eight-cell morula gives rise to the 16-cell compacted morula; the peripheral blastomeres gradually become flatter and polarized, and all types of intercellular junctions appear among them. These cells become the first embryonic epithelium, the trophectoderm. Trophectodermal cells are functional and transport fluid from the exterior to the interior to create a fluid-filled cavity, the blastocoele. The internal cells of the compacted morula become the inner cell mass (ICM), are totipotent, and will form the embryo itself.

Trophectoderm formation entails several cellular and molecular modifications. The cell adhesion molecule E-cadherin encoded by the Cdh1 gene plays an important role in all cells until gastrulation. In compaction and trophectoderm formation, E-cadherin of maternal origin begins to be replaced by the zygotic E-cadherin in the blastomeres at the end of the two-cell stage. Before compaction, blastomeres are already linked to each other using E-cadherin. Maternal and zygotic E-cadherin is gradually redistributed and concentrated on the basolateral side of each blastomere towards the outer surface of the future compacted morula. By the 16-cell stage, a large amount of E-cadherin accumulates on the basolateral side of the blastomere and cell–cell adhesion becomes tighter. Embryos lacking E-cadherin cannot form a functional epithelium and die at the peri-implantation stage. Interestingly, the compaction of Cdh1−/− embryos is not perturbed, indicating that maternal E-cadherin is sufficient for this process. However, zygotic E-cadherin is clearly required for blastocyst formation (Larue et al., 1994).

Although AKT is present at the time of trophectoderm formation, there is currently no evidence to implicate this protein in the formation of the first epithelium. However, AKT has been shown to have a functional role in embryo implantation in the uterus, an invasive process mediated by the trophoblast, a trophectoderm derivative. Implantation requires the activation of metalloproteinases, such as matrix metalloproteinase 9 (MMP-9), and of the tissue inhibitor of metalloproteinase-1 (TIMP-1). The activation of these enzymes is associated with cell migration, one of the characteristics of EMT. In vitro, EGF, which is produced in vivo by the uterus, induces the production of MMP-9 and TIMP-1 in trophoblast cell lines. Interestingly, the activation of these two proteins by EGF requires simultaneous induction of both the PI3K/AKT and MAPK pathways. In fact, activation of only one pathway is insufficient to induce (pseudo)implantation in this system (Qiu et al., 2004).

AKT has also been implicated in preeclampsia, a maternal hypertensive disease with proteinuria that represents a serious complication of pregnancy (Perkins et al., 2002). The pathogenesis of preeclampsia is associated with a failure of trophoblasts to invade and remodel the spiral arteries of the placenta in order to allow increased blood flow to the developing fetus. This failure is caused at least in part by trophoblast cell death due to changes in oxygenation and is antagonized by the PI3K/AKT pathway.

Mesoderm formation

On E4.0, a second epithelial layer – the primitive endoderm – is formed at the interface between the cells of the ICM and the blastocoele. The cells of the ICM then form the first embryonic epithelium – the primitive ectoderm – a pseudo-stratified epithelium consisting of long columnar cells. Highly coordinated EMT conversions, cell migration, cell proliferation and differentiation are subsequently observed during mouse gastrulation. The general organization of the embryo is laid down during this process, with the formation of a new layer, the mesoderm, between the ectoderm and the endoderm. Mesoderm formation involves a transition from the epithelial organization of the epiblast (tight and polarized) to the mesenchymal organization of the mesoderm (loose with no apicobasal polarity).

The molecular mechanisms underlying this EMT in vivo are poorly understood, because it is extremely difficult to carry out functional studies on mammalian embryos in utero. In the mouse, transformation of the epiblast epithelium into mesoderm involves a loss of E-cadherin production (Vestweber et al., 1987; Butz and Larue, 1995).

Functional studies have been carried out ex vivo and epiblast explants have been cultured in vitro. Under basal conditions, these explants form epithelial clusters of cells, show typical adhesive and nonmigratory behavior and express markers characteristic of the epiblast. Antibodies blocking E-cadherin activity at the surface of explants from the posterior epiblast disrupt the epithelial organization of the explant, allowing cells to adopt a mesenchymal organization (Vestweber and Kemler, 1984). This process involves radical changes in cell–cell and cell–extracellular matrix adhesion, acquisition of migratory properties and corresponding changes in the expression of molecular markers for mesodermal and epiblast cells. This mesenchymal state of the cells is preserved during their migration.

In the future, in order to investigate the role of AKT during gastrulation, similar experiments with ex vivo explants could be performed using antibodies directed against Akt or specific drugs inhibiting Akt. We need to emphasize that there is no direct evidence that AKT function is crucial during mesoderm formation. However, the induction of IGF and growth factors of the CFC family suggests a possible role for AKT in gastrulation. Proteins of the CFC family, such as mouse Cripto (Cr-1) have been directly implicated in the induction of EMT (Strizzi et al., 2004). Cr-1 is involved in early development, at the time of gastrulation, and we may hypothesize that it would activate Akt. Cr-1 mRNA is found in the embryonic ectoderm following implantation of the blastocyst and is temporally and spatially restricted to the epiblast cells of the primitive streak (see, for a review, Salomon et al., 2000).

IGF has been shown to induce Akt activation and IGF-II has been shown to be abundant in the nascent mesoderm of the gastrulating mouse embryo. The possible role of IGF has been investigated following the in vitro and in vivo differentiation of several biparental, parthenogenetic and androgenetic Igf2−/− murine embryonic stem (ES) cell lines (Morali et al., 2000). These cells differ in endogenous IGF-II levels, because Igf2 is an imprinted gene expressed from the paternal allele. Interestingly, these cells show parallel differences in the amount of activated AKT in most tissues of the mouse embryo (Morali et al., 2000). Specifically, the recruitment of mesodermal cells is directly dependent on the amount of IGF-II and consequently on the extent of AKT activation. The main conclusion of this study was that the binding of IGF-II to IGF1R at the surface of multipotent epithelial precursor cells increases the formation of mesodermal cells. Both the PI3K/Akt and MAPK pathways, downstream of the activated IGF1R, may be involved in mesoderm formation, possibly in a redundant or synergistic manner.

EMT and cancer

The main difference between normal development and pathological processes is that cellular and molecular events follow highly regulated spatial and temporal plans during development, whereas during transformation the order of events may be stochastic and time-independent, or particular events may be bypassed. During tumorigenesis EMT may increase the motility and invasiveness of cancer cells, and malignant transformation may be associated with signaling pathways promoting EMT (Boyer et al., 2000). Various processes associated with EMT occur during tumor progression, and these processes closely resemble those occurring in normal development. However, it should be emphasized that important differences exist between normal and physiopathological EMT. Moreover, it appears that the molecular program leading to EMT during tumor progression is characterized by the amplification of only some aspects of the full-fledged EMT in development. This may be the consequence of preferential oncogenic signaling through fewer signal transduction pathways in tumorigenesis.

AKT is frequently activated in human epithelial cancer (Cheng et al., 1992, 1996; Bellacosa et al., 1995; Yuan et al., 2000; Brognard et al., 2001 and reviewed by Altomare and Testa, in this issue; Ringel et al., 2001; Sun et al., 2001; Testa and Bellacosa, 2001). Interestingly, AKT2 activation in ovarian carcinomas has been linked to aggressive clinical behavior and a loss of the histological features of epithelial differentiation (Bellacosa et al., 1995). These findings are consistent with AKT directly affecting epithelial cell morphology, tumorigenicity, cell motility and invasiveness. However, the definitive demonstration that EMT was induced by AKT was provided by a study in which squamous cell carcinoma lines overexpressing activated mutants of AKT were shown to undergo EMT and downregulate E-cadherin (Grille et al., 2003).

E-cadherin and EMT

Both EMT and MET are dependent on E-cadherin, and cells undergoing EMT downregulate E-cadherin. This cell–cell adhesion molecule is a calcium-dependent transmembrane glycoprotein present in most epithelial cells in embryonic and adult tissues. It is required for normal embryonic development and homeostasis and an understanding of its regulation is therefore extremely important. Cadherins are generally regulated at both the mRNA and protein levels, by means of changes in subcellular distribution, translational or transcriptional events, and degradation. E-cadherin is considered to act as a tumor suppressor for two main reasons: transcription of its gene is silenced in various carcinomas, and re-expression of a native form of E-cadherin in carcinomas in vitro is sufficient to reduce the aggressiveness of tumor cells (Vleminckx et al., 1991). In agreement with the definition of tumor suppressor, germline mutations of the Cdh1 gene are associated with a syndrome of hereditary gastric and colorectal cancer (Guilford et al., 1998; Suriano et al., 2003).

In various human carcinomas, functional loss of E-cadherin may result from the production of a defective protein or transcriptional silencing due to promoter hypermethylation. In addition, the production of a defective E-cadherin protein may result from gene mutation, abnormal post-translational modifications (phosphorylation or glycosylation) or protein degradation (proteolysis). Cases of E-cadherin upregulation in tumor progression have been reported (Kang and Massague, 2004; Thiery and Morgan, 2004), specifically during intravasation and seeding of metastatic cells (see below).

In addition to promoter hypermethylation, E-cadherin transcriptional repression may result from the activation of repressors, such as Snail, Slug, Sip1 and Ets. The molecular mechanisms associated with E-cadherin internalization/sequestration and repression of the E-cadherin gene remain unclear, but AKT was recently shown to regulate levels of E-cadherin mRNA and protein (Grille et al., 2003). Two main types of consensus binding sites have been shown to downregulate E-cadherin expression: Ets sites and palindromic E-boxes (E-pal).

Moreover, during development and translocation the loss of expression of E-cadherin is often associated with the gain of expression of N-cadherin. The molecular mechanisms associated with this common switch remain unclear.

E-cadherin Ets-binding sites

The expression of c-ets-1 in breast carcinoma cell lines induces EMT, partly due to repression of the E-cadherin promoter (Gilles et al., 1997; Rodrigo et al., 1999). An Ets-binding site has been identified at position −97 in the E-cadherin promoter. The binding of Ets to this region downregulates E-cadherin promoter activity in keratinocyte cell lines producing this protein (Rodrigo et al., 1999). Interestingly, beside acting as repressors of E-cadherin transcription, Ets factors are also involved in the upregulation of key mediators of invasiveness, such as matrilysin, matrix metalloprotease, collagenase, heparanase and urokinase (reviewed in Shepherd and Hassell, 2001; Hsu et al., 2004).

E-cadherin E-boxes

E-boxes are characterized by the consensus sequence, CANNTG, and are common in genomic sequences. Three E-boxes have been identified in the human E-cadherin promoter: two upstream from the transcription start site and one in exon 1. Snail, Slug, Sip1 (also known as Zeb2) and Zeb1 bind these E-boxes and repress E-cadherin transcription.

Snail and Slug (encoded by the Snai1 and Snai2 genes, respectively) are zinc-finger proteins originally shown to be involved in mesoderm formation, together with Twist (see below). Snail expression is inversely correlated with E-cadherin transcription. Abnormal Snail production in a large series of cell lines and primary tumors is associated with aggressiveness and loss of E-cadherin expression (Birchmeier and Behrens, 1994; De Craene et al., 2005).

The overproduction of Snail or Slug induces EMT in vitro (Batlle et al., 2000). The repression of Snail RNA production is associated with E-cadherin upregulation and MET. An interesting positive–negative regulation system for E-cadherin was recently discovered (Palmer et al., 2004). E-cadherin is positively regulated by 1,25(OH)2D3 via the vitamin D receptor and Snail can repress E-cadherin and the vitamin D receptor, so the balance between vitamin D receptors and Snail may regulate E-cadherin levels (Palmer et al., 2004). Phosphorylation by the p21-activated kinase PAK1 promotes Snail nuclear retention and repressor activity (Yang et al., 2005). These examples show some facets of the complex regulation of E-cadherin during EMT.

Sip1 (also known as ZFHX1B or SMADIP1) is a member of the delta EF1/Zfh1 family of two-handed zinc-finger/homeodomain proteins. It contains a Smad-binding domain facilitating interaction with full-length Smad proteins. Sip1 may therefore modulate the transforming growth factor beta (TGFβ) signaling pathway, which is known to induce EMT (see below). Mutations in the gene encoding Sip1 have been found in many patients affected with Hirschsprung disease (Amiel and Lyonnet, 2001; Cacheux et al., 2001; Wakamatsu et al., 2001; Yamada et al., 2001; Van de Putte et al., 2003). Mice with targeted inactivation of Sip1 present clinical features of Hirschsprung disease-mental retardation syndrome (Amiel and Lyonnet, 2001; Cacheux et al., 2001; Wakamatsu et al., 2001; Yamada et al., 2001; Van de Putte et al., 2003). These mutant mice do not develop postotic vagal neural crest cells – the precursors of the enteric nervous system affected in patients with Hirschsprung disease – and display arrest in the delamination of cranial neural crest cells, which form the skeletal muscle elements of the vertebral head. In the absence of Sip1, the correct delamination of neural crest cells does not occur. The delamination of neural crest cells is a good example of EMT requiring type I cadherin downregulation.

AKT, E-cadherin and EMT

AKT was initially described as an oncogene (Staal, 1987; Bellacosa et al., 1991). Three AKT proteins have been isolated, AKT1–3, and all three display serin–threonine kinase activity (Testa and Bellacosa, 2001; Bellacosa et al., 2004). AKT2 is frequently upregulated and activated in ovarian, breast and pancreatic tumors. Akt is involved in many basic cellular processes, including cell cycle progression, cell proliferation, cell survival, metabolism and EMT (see editorial by Testa and Tsichlis, in this issue). The EMT induced by activated Akt (Grille et al., 2003) involves: loss of cell–cell adhesion, morphological changes, loss of apico-basolateral cell polarization, induction of cell motility, decrease in cell–matrix adhesion, and changes in the production or distribution of specific proteins. For instance, desmoplakin, a protein involved in the formation and maintenance of desmosomes, is internalized, and vimentin, an intermediate filament protein present in many mesenchymal cells, is induced (Figure 1). Akt also induces the production of metalloproteinases and cell invasion (Kim et al., 2001; Park et al., 2001).

At the molecular level, Akt production in epithelial cells has two major consequences for E-cadherin: transcription of the E-cadherin gene is strongly repressed, and the small amount of E-cadherin protein produced by the cell is concentrated in perinuclear organelles (Grille et al., 2003). This double regulation allows the cell to remain in a mesenchymal state during the exponential growth phase. The sequestration of E-cadherin may be associated with the AKT-mediated activation of the Rab5 protein, while the transcriptional repression of E-cadherin expression may be associated with the activation of Snail gene expression (Barbieri et al., 1998; Batlle et al., 2000; Cano et al., 2000).

Proteins activating both AKT and EMT

Various types of proteins, ligands, receptors and effectors induce EMT. The various associated signal transduction pathways converge on Akt, which represses the E-cadherin transcription. The growth factor Cripto-1 (=also known as CFC and Cr-1), hyaluronan and M-Ras have been shown to activate Akt, resulting in the induction of EMT.

Cr-1 belongs to a family of extracellular proteins playing key roles in intercellular signaling pathways during mesoderm formation in vertebrates. Cr-1 is a GPI-linked membrane protein that has been implicated as an oncogene in human carcinogenesis. CR-1 is overproduced in various types of cancer, including carcinomas of the breast, colon, stomach, pancreas, ovary and testis. Cr-1 has been shown to be involved in EMT and to activate Akt. In vivo experiments have confirmed the results obtained in vitro (Strizzi et al., 2004). Following the introduction of the Cr-1 gene under control of the MMTV promoter into mouse mammary epithelium, cells proliferated excessively and underwent EMT. The amount of active Akt was increased and the amount of E-cadherin decreased in these cells. The expression of markers of mesenchymal status, such as Snail and vimentin, increased in these cells.

Hyaluronan is a high-molecular-weight glycosaminoglycan of the extracellular matrix of nonepithelial cells and is present in connective tissue, cartilage, bone marrow and synovial fluid (Watanabe and Yamaguchi, 1996). Hyaluronan is involved in the EMT of endothelial cells during development of early heart valve and septal mesenchyme (Camenisch et al., 2002). The stimulation of hyaluronan synthesis in epithelial cells induces a transition to mesenchymal morphology. Three hyaluronan synthases have been isolated, one of which, hyaluronan synthase-2, has been found in two classical epithelial cell lines, Madin Darby canine kidney (MDCK) and MCF-10A. The production of this enzyme in these two cell lines induces EMT and stimulates the PI3K/Akt pathway (Zoltan-Jones et al., 2003). The production of E-cadherin was not followed in this study, but one can hypothesize that it would be downregulated.

M-Ras, a 27 kDa protein, is a member of the RAS superfamily of GTP-binding proteins. Ras proteins are membrane-anchored, intracellular signal transducers that are activated in a significant proportion of tumors. M-Ras is produced in large amounts in the brain and myoblasts, and in moderate amounts in various differentiated muscle cells: myotubes, skeletal muscle and heart. It is also found in the uterus and testis. M-Ras is associated with plasma membrane-bound structures, including microspikes, membrane ruffles and pseudopods, and induces the formation of peripheral microspikes and dendrites in cell culture. The wild-type form of M-Ras weakly stimulates MAPK activity, but this stimulation is enhanced by RAF1 (A-Raf) coexpression. Overexpression of a G22V or Q71L M-Ras in fibroblasts induces cellular transformation (Quilliam et al., 1999), associated with the induction of EMT and Akt activation (Ward et al., 2004).

EMT and metastasis

Invasion of surrounding tissues and metastasis to distant organs have long been recognized as features of malignancy. Recent findings indicate that both invasion and metastasis may be critically dependent on the acquisition by the incipient cancer cell of EMT features (Kang and Massague, 2004; Thiery and Morgan, 2004). Metastasis is a multistep process characterized by dissociation of tumor cells from the epithelial layer, penetration through the basement membrane into the adjacent connective tissue, intravasation, survival in the bloodstream, extravasation at a distant site and growth of metastatic cells in the distant site with stimulation of neoangiogenesis (Chambers et al., 2002). Since the first steps are characterized by increased motility and invasiveness, it has been hypothesized that they are associated with EMT (Thiery, 2002). Recent studies have confirmed this hypothesis by showing that the basic helix–loop–helix transcription factor Twist plays a critical role in the early steps of metastasis (Yang et al., 2004). Twist was previously known to be involved in mesoderm differentiation and neural crest cell migration (Thisse et al., 1987; Leptin and Grunewald, 1990; Soo et al., 2002), and was recently identified in a microarray-based screen for genes associated with metastasis to the lungs of murine mammary tumor lines injected in the mammary gland (Yang et al., 2004). Downregulation of Twist by siRNA in a highly metastatic line reduced the number of circulating tumor cells, thus indicating a role for Twist in early metastasis, including intravasation. Moreover, Twist overexpression caused an EMT of human immortalized mammary epithelial cells, characterized by downregulation of E-cadherin, β- and γ-catenin, upregulation of vimentin and fibronectin, spindle-like morphological changes and increased migration. Downregulation of E-cadherin was due to Twist repression of the E-cadherin promoter via E-box elements (Yang et al., 2004). Taken together, these findings indicate that EMT is involved in the early steps of metastasis.

The role of PI3K/AKT in TGFβ-mediated EMT

TGFβ has a double role in tumorigenesis, as a tumor suppressor and a tumor promoter (Muraoka-Cook et al., 2005). In fact, the TGFβ pathway is known to prevent epithelial cell transformation by inhibiting proliferation and inducing senescence or apoptosis. On the other hand, during late tumorigenesis, enhanced TGFβ signaling via autocrine or paracrine stimulation is associated with cancer progression, via increased motility, invasiveness and ultimately metastasis (Muraoka-Cook et al., 2005). The latter events appear to be the consequence of an EMT mediated by the PI3K/AKT axis.

Using a model of non-transformed murine mammalian cells, it was shown that TGFβ stimulates AKT phosphorylation via PI3K (Bakin et al., 2000). This stimulation resulted in a frank EMT characterized by reorganization of actin fibers (appearance of stress fibers) and delocalization of E-cadherin, the tight junction protein ZO-1 and the cell–matrix adhesion protein integrin β1 (Bakin et al., 2000); in addition, N-cadherin appeared at the cell membrane (Bhowmick et al., 2001). These molecular events accompanied a transition from cuboidal morphology to a spindle-like, elongated shape. Both molecular and cellular changes were inhibited by the PI3K inhibitor LY294002 and by a dominant-negative (kinase-inactive) AKT mutant, establishing the requirement of the PI3K/AKT axis for the TGFβ-mediated EMT. To the contrary, no effect was observed when cells were treated with a dominant-negative SMAD3 mutant or with the SMAD signaling inhibitor SMAD7, indicating that the TGFβ-mediated EMT is independent of SMAD signaling (Bhowmick et al., 2001). This, however, contradicts earlier reports that had implicated a role of the transcriptional response downstream of SMAD signaling in the TGFβ-mediated EMT (Piek et al., 1999). A possible reason for this discrepancy may be the use of different model systems that are at different stages of tumor progression.

The small GTPase RhoA (but not Rac1 or Cdc42) is activated in response to TGFβ stimulation and it, or a closely related molecule, is required for TGFβ-mediated EMT, because expression of the dominant-negative mutant N19-RhoA inhibited all the morphological and molecular features of EMT (Bhowmick et al., 2001). Interestingly, p160ROCK (ROCK-1), a serine–threonine kinase downstream of RhoA, is required in stress fiber formation, acquisition of mesenchymal morphology and enhanced migration, but is not involved in E-cadherin delocalization. On the other hand, LY294002 inhibited both the basal and TGFβ-induced migration of nontransformed mammary cells as well as two metastatic breast tumor lines, indicating that PI3K is involved in this aspect of EMT (Bakin et al., 2000; Bhowmick et al., 2001). Although the role of AKT in TGFβ-induced migration was not investigated, in view of the data obtained with squamous cell carcinoma lines (Grille et al., 2003), it is likely that AKT is one of the downstream effectors of PI3K in TGFβ-induced enhanced motility.

The mechanisms of PI3K/AKT activation by TGFβ remain to be fully elucidated. Bakin et al. showed that the dominant-negative N19 RhoA reduced AKT activation by TGFβ, while a constitutively active mutant enhanced it, thus suggesting a role of RhoA upstream of PI3K/AKT (Bakin et al., 2000). It is also possible that a complex of the PI3K regulatory subunit p85 with the TGFβ receptors I and II may mediate PI3K activation (Krymskaya et al., 1997). However, given the relatively late kinetics of AKT phosphorylation (30 min–2 h), another possibility is that activation of the PI3K/AKT axis in mammary cells is not directly mediated by the heterotrimeric type I–III TGFβ receptors, but possibly by the autocrine secretion of growth factors of the EGF family with consequent activation of receptor tyrosine kinase(s), as shown for rat hepatocytes (Murillo et al., 2005).

The gene encoding the p110 catalytic subunit of PI3K is frequently mutated in human cancer (Samuels and Velculescu, 2004; Samuels et al., 2004). Consistent with the observations presented above, the functional consequences of the PI3K activating mutations are not only promotion of cell growth and survival but also stimulation of invasiveness of cancer cells (Samuels et al., 2005).

Other pathways leading to EMT

Several other pathways may lead to EMT, in some cases, at least in part, through activation of PI3K and AKT (Figure 2). Useful models have been represented by the Nara rat bladder carcinoma line NBT-II and the MDCK cells, in which induction of EMT is associated with cell scattering, that is, the combination of reduced cell adhesion and increased motility (Boyer et al., 2000).

Figure 2

Schematic of the signal transduction pathways associated with epithelial–mesenchymal transition. End points of EMT are boxed. RTK: receptor tyrosine kinase; ROS: reactive oxygen species (see text for details)

Tyrosine kinases

Several growth factors can induce EMT by binding to receptor tyrosine kinases. We and others showed that FGF, EGF, TGFα and IGF2 can induce EMT and scattering in NBT-II cells (Gavrilovic et al., 1990; Valles et al., 1990; Morali et al., 2001). While for IGF2 EMT was associated with nuclear translocation of β-catenin (Morali et al., 2001), critical effectors downstream of the activated receptor tyrosine kinases are represented by SRC family members and RAS (Boyer et al., 2000; Thiery, 2002, and references therein). SRC phosphorylates cytoskeletal and focal adhesion proteins (FAK, Cas and paxillin), mediating the rearrangement of cell architecture and focal adhesions. SRC-induced EMT may or may not require downstream transcription factors, depending on the model system (Boyer et al., 2000).

EMT and scattering in MDCK cells is induced by binding of HGF/scatter factor to its receptor tyrosine kinase, c-Met. Dissection of the signaling pathways downstream of activated c-Met has been conducted by mutagenesis of the relevant phosphotyrosine residues phosphorylated during receptor activation. This analysis revealed that Y1349 and Y1356 are involved in scattering via the SH2 domain-based recruitment of several effectors, including Src, PI3K and Grb2 (Ponzetto et al., 1994). In particular, PI3K is required for motility.

RAS pathway

RAS activation plays a role in the EMT downstream of activated receptor tyrosine kinases. Epistatic analysis indicates that Jun and Fos are involved downstream of RAS (Boyer et al., 1997). Signaling from RAS to Jun, Fos and other transcription factors involved in EMT, such as Snail and Slug occurs via RAF, MEK – both inducers of EMT – and MAPK (Edme et al., 2002). In addition, another downstream effector of RAS in the EMT response may be PI3K itself. Finally, RAS effects the activity of the two small GTPases, Rac and Rho, possibly via PI3K. These two molecules may play a role in EMT by regulating adherens junctions, focal adhesions, myosin phosphorylation, actin stress fibers and, therefore, motility and scattering (Edme et al., 2002; Thiery, 2002).

Integrin-linked kinase (ILK) and integrin signaling

ILK is a key component of focal adhesions that binds to integrins β1–3 and is indispensable for integrin function during development (Wu, 1999). ILK is activated in a PI3K-dependent manner by cell attachment to fibronectin or insulin stimulation and its targets include AKT, with activating phosphorylation on S473, and glycogen synthase kinase 3, with inhibitory phosphorylation (Delcommenne et al., 1998). Overexpression of ILK leads to nuclear translocation of β-catenin, increased invasiveness and repression of E-cadherin (Novak et al., 1998; Wu et al., 1998) via upregulation of Snail transcription (Tan et al., 2001; Barbera et al., 2004). ILK is also involved in the TGFβ-mediated EMT of human keratinocytes (Lee et al., 2004).


The highly conserved canonical Wnt/β-catenin pathway plays important roles in development and is frequently activated in cancer. In the absence of Wnt signals, β-catenin, in a complex with APC and axin, is phosphorylated by GSK3β and targeted for ubiquitin/proteasome-mediated degradation. Wnt ligands bind Frizzled receptors and activate Disheveled, which blocks β-catenin degradation. Excess β-catenin enters the nucleus where, in association with the transcription factor TCF/LEF, it promotes the expression of several target genes.

Wnt/β-catenin signaling induces EMT during sea urchin gastrulation (Logan et al., 1999), regulates EMT during cardiac valve formation in zebrafish (Hurlstone et al., 2003) and, when deregulated, causes premature EMT of the blastocyst ectoderm in the mouse (Kemler et al., 2004). Wnt/β-catenin-induced EMT plays a role in colorectal cancer metastasis and squamous cell carcinoma progression (Taki et al., 2003; Brabletz et al., 2005).

Wnt/β-catenin induction of EMT is due to the activation of Slug, which is a direct transcriptional target of β-catenin/TCF (Conacci-Sorrell et al., 2003), and stabilization of Snail, via inhibition of its phosphorylation and degradation (Yook et al., 2005). Consistent with these observations, inhibition of GSK3 activity leads to Snail transcription, E-cadherin repression and EMT (Bachelder et al., 2005).


Recently, it has been shown that the Notch pathway, which is involved in cell fate determination during development, can induce EMT during cardiac valve development by inducing Snail and repressing VE-cadherin transcription. Similarly, overexpression of active alleles of Notch in immortalized endothelial cells resulted in an EMT due to Snail activation and VE-cadherin repression (Timmerman et al., 2004).

Rac1b and reactive oxygen species (ROS)

Very recent observations highlight a role of ROS in EMT (Radisky et al., 2005). Treatment of mouse mammary epithelial cells with MMP-3 induced the transcription of a splicing variant of Rac1, called Rac1b. In turn, Rac1b stimulated the production of ROS, which stimulated Snail expression and induction of EMT. Treatment with the anti-oxidant N-acetyl cysteine prevented Snail induction and EMT (Radisky et al., 2005). These findings link extracellular matrix, oxidative damage and EMT, and have obvious therapeutic implications.

In summary, it should be emphasized that EMT pathways are very complex (Figure 2). Most observations are conducted in in vitro models and a deep understanding of the pathways that are relevant in vivo is lacking, particularly for what concerns the role of EMT in tumorigenesis (Thiery, 2002). A critical examination of the literature reveals that in many cases, the observations are cell-specific, model-specific and context-dependent, and may result in apparently discordant findings in different cell lines: examples include the relative importance of Snail and Slug in E-cadherin repression, and that of the GTPases Rho and Rac in the assembly of adherence junctions and actin fibers (Thiery, 2002).

Interactions between the PI3K/AKT and other pathways of EMT

As mentioned above, crosstalk between the PI3K/AKT axis and tyrosine kinases, RAS and ILK pathways to EMT is likely to occur via the well-defined connections among these signal transducers (Boyer et al., 2000; Thiery, 2002, 2003; Bellacosa et al., 2004). On the other hand, interactions between PI3K/AKT and other pathways of EMT are less characterized and represent an area of active investigation. However, some interesting connections are emerging from Wnt/β-catenin and Notch signaling.

PI3K/AKT and Wnt/β-catenin signaling

While canonical Wnt and PI3/AKT signaling could converge at the level of inhibition of GSK3, there are indications that the two pathways may affect and phosphorylate different pools of GSK3 (Weston and Davis, 2001; Grille et al., 2003). Other possible connections include the stimulation of AKT activity by Wnt/Disheveled signaling (Fukumoto et al., 2001) and the converse stimulation of β-catenin transcription (and possibly phosphorylation) by AKT and 14-3-3ζ (Tian et al., 2004). However, the direct role of crosstalks between PI3K/AKT and Wnt/β-catenin signaling in EMT needs to be established.

PI3K/AKT and notch signaling

Recent data indicate that expression of the Notch ligand Jagged in human keratinocytes and cervical cancer cell lines leads to AKT phosphorylation and induces a frank PI3K-dependent EMT characterized by enhanced motility, morphological changes, E-cadherin downregulation and vimentin and fibronectin upregulation (Veeraraghavalu et al., 2005).

Modulating EMT: implications for cancer therapy

From the discussion presented above, it is clear that EMT represents a very promising therapeutic target. Inhibition of EMT may prevent or restrain invasion and metastasis. Ideally, it should be possible to inhibit the early steps of invasion and metastasis; this of course may fall within the realm of chemoprevention, because at the time of cancer diagnosis, dormant metastatic cells may already be present (Chambers et al., 2002).

Multiple molecules involved in EMT can be envisioned as targets of anti-EMT therapy. Some potential targets are receptor- and SRC-family tyrosine kinases, RAS and other small GTPases, including Rho, and the PI3K/AKT axis (see reviews by Cheng et al. and Kumar and Madison, in this issue). In fact, agents targeting these molecules are already available either as proof-of-principle agents (e.g. LY294002) or as formulations already in the clinic (IRESSA, Herceptin, farnesyltransferase inhibitors) (Bellacosa et al., 2005). More innovative approaches may include restoration of E-cadherin function and inhibition of transcription factors that repress the E-cadherin promoter, such as Snail, Slug, Sip1 and Twist.

Given the complexity of the molecular and cellular pathways leading to EMT, and the fact that tumor cells may be critically dependent on a few deregulated, oncogenic pathways (Weinstein, 2002), a tailored treatment should be aimed at the specific molecular defects resulting in EMT.

Another potential therapeutic approach involves a forced stimulation of MET, which in principle should be a very elegant way to prevent EMT (Thiery, 2002). However, the molecular basis of this process are less well known in comparison to EMT, and therefore it may take longer for the rational design of modulators of this pathway.


Accumulating evidence in recent years indicates that EMT is a critical process not only in development but also in tumorigenesis. Acquisition of EMT properties during tumor progression is associated with dissolution of epithelial integrity, increased migration, local invasion and, ultimately, metastasis. Several pathways involved in the developmental control of EMT are taken over during tumor formation by activation of oncogenic signaling and disruption of tumor suppressor networks. In this respect, given its frequent alteration in human cancer, activation of the PI3K/AKT axis with downregulation of E-cadherin expression and induction of EMT may be particularly important.

For its association with invasion and early steps of metastasis, inhibition of EMT appears a viable strategy for novel approaches of cancer therapy. However, given the complex, intertwined circuitry regulating EMT (Figure 2), it is likely that effective deployment of anti-EMT therapeutics cannot prescind from a detailed understanding of the molecular alterations of the specific tumor being targeted.


  1. Altomare DA and Testa JR . (2005). Oncogene Rev., 24, 7455–7464.

  2. Amiel J and Lyonnet S . (2001). J. Med. Genet., 38, 729–739.

  3. Bachelder RE, Yoon SO, Franci C, de Herreros AG and Mercurio AM . (2005). J. Cell Biol., 168, 29–33.

  4. Bakin AV, Tomlinson AK, Bhowmick NA, Moses HL and Arteaga CL . (2000). J. Biol. Chem., 275, 36803–36810.

  5. Barbera MJ, Puig I, Dominguez D, Julien-Grille S, Guaita-Esteruelas S, Peiro S, Baulida J, Franci C, Dedhar S, Larue L and Garcia de Herreros A . (2004). Oncogene, 23, 7345–7354.

  6. Barbieri MA, Kohn AD, Roth RA and Stahl PD . (1998). J. Biol. Chem., 273, 19367–19370.

  7. Batlle E, Sancho E, Franci C, Dominguez D, Monfar M, Baulida J and Garcia De Herreros A . (2000). Nat. Cell Biol., 2, 84–89.

  8. Bellacosa A, de Feo D, Godwin AK, Bell DW, Cheng JQ, Altomare DA, Wan M, Dubeau L, Scambia G, Masciullo V, Ferrandina G, Panici PB, Mancuso S, Neri G and Testa JR . (1995). Int. J. Cancer, 64, 280–285.

  9. Bellacosa A, Kumar CC, Di Cristofano A and Testa JR . (2005). Adv. Cancer Res., 94, 29–86.

  10. Bellacosa A, Testa JR, Moore R and Larue L . (2004). Cancer Biol. Ther., 3, 268–275.

  11. Bellacosa A, Testa JR, Staal SP and Tsichlis PN . (1991). Science, 254, 274–277.

  12. Bhowmick NA, Ghiassi M, Bakin A, Aakre M, Lundquist CA, Engel ME, Arteaga CL and Moses HL . (2001). Mol. Biol. Cell, 12, 27–36.

  13. Birchmeier W and Behrens J . (1994). Biochim. Biophys. Acta, 1198, 11–26.

  14. Boyer B, Roche S, Denoyelle M and Thiery JP . (1997). EMBO J., 16, 5904–5913.

  15. Boyer B, Valles AM and Edme N . (2000). Biochem. Pharmacol., 60, 1091–1099.

  16. Brabletz T, Hlubek F, Spaderna S, Schmalhofer O, Hiendlmeyer E, Jung A and Kirchner T . (2005). Cells Tissues Organs, 179, 56–65.

  17. Brognard J, Clark AS, Ni Y and Dennis PA . (2001). Cancer Res., 61, 3986–3997.

  18. Butz S and Larue L . (1995). Cell Adhes. Commun., 3, 337–352.

  19. Cacheux V, Dastot-Le Moal F, Kaariainen H, Bondurand N, Rintala R, Boissier B, Wilson M, Mowat D and Goossens M . (2001). Hum. Mol. Genet., 10, 1503–1510.

  20. Camenisch TD, Schroeder JA, Bradley J, Klewer SE and McDonald JA . (2002). Nat. Med., 8, 850–855.

  21. Cano A, Perez-Moreno MA, Rodrigo I, Locascio A, Blanco MJ, del Barrio MG, Portillo F and Nieto MA . (2000). Nat. Cell Biol., 2, 76–83.

  22. Chambers AF, Groom AC and MacDonald IC . (2002). Nat. Rev. Cancer, 2, 563–572.

  23. Cheng JQ, Godwin AK, Bellacosa A, Taguchi T, Franke TF, Hamilton TC, Tsichlis PN and Testa JR . (1992). Proc. Natl. Acad. Sci. USA, 89, 9267–9271.

  24. Cheng JQ, Lindsley CW, Cheng GZ, Yang H and Nicosia SV . (2005). Oncogene Rev., 24, 7482–7492.

  25. Cheng JQ, Ruggeri B, Klein WM, Sonoda G, Altomare DA, Watson DK and Testa JR . (1996). Proc. Natl. Acad. Sci. USA, 93, 3636–3641.

  26. Conacci-Sorrell M, Simcha I, Ben-Yedidia T, Blechman J, Savagner P and Ben-Ze'ev A . (2003). J. Cell Biol., 163, 847–857.

  27. De Craene B, Gilbert B, Stove C, Bruyneel E, van Roy F and Berx G . (2005). Cancer Res., 65, 6237–6244.

  28. Delcommenne M, Tan C, Gray V, Rue L, Woodgett J and Dedhar S . (1998). Proc. Natl. Acad. Sci. USA, 95, 11211–11216.

  29. Edme N, Downward J, Thiery JP and Boyer B . (2002). J. Cell Sci., 115, 2591–2601.

  30. Fukumoto S, Hsieh CM, Maemura K, Layne MD, Yet SF, Lee KH, Matsui T, Rosenzweig A, Taylor WG, Rubin JS, Perrella MA and Lee ME . (2001). J. Biol. Chem., 276, 17479–17483.

  31. Gavrilovic J, Moens G, Thiery JP and Jouanneau J . (1990). Cell Regul., 1, 1003–1014.

  32. Gilles C, Polette M, Birembaut P, Brunner N and Thompson EW . (1997). Clin. Exp. Metast., 15, 519–526.

  33. Grille SJ, Bellacosa A, Upson J, Klein-Szanto AJ, van Roy F, Lee-Kwon W, Donowitz M, Tsichlis PN and Larue L . (2003). Cancer Res., 63, 2172–2178.

  34. Guilford P, Hopkins J, Harraway J, McLeod M, McLeod N, Harawira P, Taite H, Scoular R, Miller A and Reeve AE . (1998). Nature, 392, 402–405.

  35. Hsu T, Trojanowska M and Watson DK . (2004). J. Cell. Biochem., 91, 896–903.

  36. Hurlstone AF, Haramis AP, Wienholds E, Begthel H, Korving J, Van Eeden F, Cuppen E, Zivkovic D, Plasterk RH and Clevers H . (2003). Nature, 425, 633–637.

  37. Kang Y and Massague J . (2004). Cell, 118, 277–279.

  38. Kemler R, Hierholzer A, Kanzler B, Kuppig S, Hansen K, Taketo MM, de Vries WN, Knowles BB and Solter D . (2004). Development, 131, 5817–5824.

  39. Kim D, Kim S, Koh H, Yoon SO, Chung AS, Cho KS and Chung J . (2001). FASEB J., 15, 1953–1962.

  40. Krymskaya VP, Hoffman R, Eszterhas A, Ciocca V and Panettieri Jr RA . (1997). Am. J. Physiol., 273, L1220–1227.

  41. Kumar CC and Madison V . (2005). Oncogene Rev., 24, 7493–7501.

  42. Larue L, Antos C, Butz S, Huber O, Delmas V, Dominis M and Kemler R . (1996). Development, 122, 3185–3194.

  43. Larue L, Ohsugi M, Hirchenhain J and Kemler R . (1994). Proc. Natl. Acad. Sci. USA, 91, 8263–8267.

  44. Lee YI, Kwon YJ and Joo CK . (2004). Biochem. Biophys. Res. Commun., 316, 997–1001.

  45. Leptin M and Grunewald B . (1990). Development, 110, 73–84.

  46. Logan CY, Miller JR, Ferkowicz MJ and McClay DR . (1999). Development, 126, 345–357.

  47. Majumder PK and Sellers WR . (2005). Oncogene Rev., 24, 7465–7474.

  48. Morali OG, Delmas V, Moore R, Jeanney C, Thiery JP and Larue L . (2001). Oncogene, 20, 4942–4950.

  49. Morali OG, Jouneau A, McLaughlin KJ, Thiery JP and Larue L . (2000). Dev. Biol., 227, 133–145.

  50. Muraoka-Cook RS, Dumont N and Arteaga CL . (2005). Clin. Cancer Res., 11, 937s–943s.

  51. Murillo MM, del Castillo G, Sanchez A, Fernandez M and Fabregat I . (2005). Oncogene, 24, 4580–4587.

  52. Novak A, Hsu SC, Leung-Hagesteijn C, Radeva G, Papkoff J, Montesano R, Roskelley C, Grosschedl R and Dedhar S . (1998). Proc. Natl. Acad. Sci. USA, 95, 4374–4379.

  53. Palmer HG, Larriba MJ, Garcia JM, Ordonez-Moran P, Pena C, Peiro S, Puig I, Rodriguez R, de la Fuente R, Bernad A, Pollan M, Bonilla F, Gamallo C, de Herreros AG and Munoz A . (2004). Nat. Med., 10, 917–919.

  54. Park BK, Zeng X and Glazer RI . (2001). Cancer Res., 61, 7647–7653.

  55. Perkins J, St John J and Ahmed A . (2002). Mol. Med., 8, 847–856.

  56. Piek E, Moustakas A, Kurisaki A, Heldin CH and ten Dijke P . (1999). J. Cell Sci., 112, 4557–4568.

  57. Ponzetto C, Bardelli A, Zhen Z, Maina F, dalla Zonca P, Giordano S, Graziani A, Panayotou G and Comoglio PM . (1994). Cell, 77, 261–271.

  58. Qiu Q, Yang M, Tsang BK and Gruslin A . (2004). Reproduction, 128, 355–363.

  59. Quilliam LA, Castro AF, Rogers-Graham KS, Martin CB, Der CJ and Bi C . (1999). J. Biol. Chem., 274, 23850–23857.

  60. Radisky DC, Levy DD, Littlepage LE, Liu H, Nelson CM, Fata JE, Leake D, Godden EL, Albertson DG, Nieto MA, Werb Z and Bissell MJ . (2005). Nature, 436, 123–127.

  61. Ringel MD, Hayre N, Saito J, Saunier B, Schuppert F, Burch H, Bernet V, Burman KD, Kohn LD and Saji M . (2001). Cancer Res., 61, 6105–6111.

  62. Rodrigo I, Cato AC and Cano A . (1999). Exp. Cell Res., 248, 358–371.

  63. Salomon DS, Bianco C, Ebert AD, Khan NI, De Santis M, Normanno N, Wechselberger C, Seno M, Williams K, Sanicola M, Foley S, Gullick WJ and Persico G . (2000). Endocr. Relat. Cancer, 7, 199–226.

  64. Samuels Y, Diaz Jr LA, Schmidt-Kittler O, Cummins JM, Delong L, Cheong I, Rago C, Huso DL, Lengauer C, Kinzler KW, Vogelstein B and Velculescu VE . (2005). Cancer Cell, 7, 561–573.

  65. Samuels Y and Velculescu VE . (2004). Cell Cycle, 3, 1221–1224.

  66. Samuels Y, Wang Z, Bardelli A, Silliman N, Ptak J, Szabo S, Yan H, Gazdar A, Powell SM, Riggins GJ, Willson JK, Markowitz S, Kinzler KW, Vogelstein B and Velculescu VE . (2004). Science, 304, 554.

  67. Shepherd T and Hassell JA . (2001). J. Mammary Gland Biol. Neoplasia, 6, 129–140.

  68. Soo K, O'Rourke MP, Khoo PL, Steiner KA, Wong N, Behringer RR and Tam PP . (2002). Dev. Biol., 247, 251–270.

  69. Staal SP . (1987). Proc. Natl. Acad. Sci. USA, 84, 5034–5037.

  70. Strizzi L, Bianco C, Normanno N, Seno M, Wechselberger C, Wallace-Jones B, Khan NI, Hirota M, Sun Y, Sanicola M and Salomon DS . (2004). J. Cell. Physiol., 201, 266–276.

  71. Sun M, Wang G, Paciga JE, Feldman RI, Yuan ZQ, Ma XL, Shelley SA, Jove R, Tsichlis PN, Nicosia SV and Cheng JQ . (2001). Am. J. Pathol., 159, 431–437.

  72. Suriano G, Oliveira C, Ferreira P, Machado JC, Bordin MC, De Wever O, Bruyneel EA, Moguilevsky N, Grehan N, Porter TR, Richards FM, Hruban RH, Roviello F, Huntsman D, Mareel M, Carneiro F, Caldas C and Seruca R . (2003). Hum. Mol. Genet., 12, 575–582.

  73. Taki M, Kamata N, Yokoyama K, Fujimoto R, Tsutsumi S and Nagayama M . (2003). Cancer Sci., 94, 593–597.

  74. Tan C, Costello P, Sanghera J, Dominguez D, Baulida J, de Herreros AG and Dedhar S . (2001). Oncogene, 20, 133–140.

  75. Testa JR and Bellacosa A . (2001). Proc. Natl. Acad. Sci. USA, 98, 10983–10985.

  76. Testa JR and Tsichlis PN . (2005). Oncogene Rev., 24, 7391–7393.

  77. Thiery JP . (2002). Nat. Rev. Cancer, 2, 442–454.

  78. Thiery JP . (2003). Curr. Opin. Genet. Dev., 13, 365–371.

  79. Thiery JP and Morgan M . (2004). Nat. Med., 10, 777–778.

  80. Thisse B, el Messal M and Perrin-Schmitt F . (1987). Nucleic Acids Res., 15, 3439–3453.

  81. Tian Q, Feetham MC, Tao WA, He XC, Li L, Aebersold R and Hood L . (2004). Proc. Natl. Acad. Sci. USA, 101, 15370–15375.

  82. Timmerman LA, Grego-Bessa J, Raya A, Bertran E, Perez-Pomares JM, Diez J, Aranda S, Palomo S, McCormick F, Izpisua-Belmonte JC and de la Pompa JL . (2004). Genes Dev., 18, 99–115.

  83. Valles AM, Boyer B, Badet J, Tucker GC, Barritault D and Thiery JP . (1990). Proc. Natl. Acad. Sci. USA, 87, 1124–1128.

  84. Van de Putte T, Maruhashi M, Francis A, Nelles L, Kondoh H, Huylebroeck D and Higashi Y . (2003). Am. J. Hum. Genet., 72, 465–470.

  85. Veeraraghavalu K, Subbaiah VK, Srivastava S, Chakrabarti O, Syal R and Krishna S . (2005). J. Virol., 79, 7889–7898.

  86. Vestweber D, Gossler A, Boller K and Kemler R . (1987). Dev. Biol., 124, 451–456.

  87. Vestweber D and Kemler R . (1984). Exp. Cell Res., 152, 169–178.

  88. Vleminckx K, Vakaet Jr L, Mareel M, Fiers W and van Roy F . (1991). Cell, 66, 107–119.

  89. Wakamatsu N, Yamada Y, Yamada K, Ono T, Nomura N, Taniguchi H, Kitoh H, Mutoh N, Yamanaka T, Mushiake K, Kato K, Sonta S and Nagaya M . (2001). Nat. Genet., 27, 369–370.

  90. Ward KR, Zhang KX, Somasiri AM, Roskelley CD and Schrader JW . (2004). Oncogene, 23, 1187–1196.

  91. Watanabe K and Yamaguchi Y . (1996). J. Biol. Chem., 271, 22945–22948.

  92. Weinstein IB . (2002). Science, 297, 63–64.

  93. Weston CR and Davis RJ . (2001). Science, 292, 2439–2440.

  94. Wu C, Keightley SY, Leung-Hagesteijn C, Radeva G, Coppolino M, Goicoechea S, McDonald JA and Dedhar S . (1998). J. Biol. Chem., 273, 528–536.

  95. Wu C . (1999). J. Cell Sci., 112 (Part 24), 4485–4489.

  96. Yamada K, Yamada Y, Nomura N, Miura K, Wakako R, Hayakawa C, Matsumoto A, Kumagai T, Yoshimura I, Miyazaki S, Kato K, Sonta S, Ono H, Yamanaka T, Nagaya M and Wakamatsu N . (2001). Am. J. Hum. Genet., 69, 1178–1185.

  97. Yang J, Mani SA, Donaher JL, Ramaswamy S, Itzykson RA, Come C, Savagner P, Gitelman I, Richardson A and Weinberg RA . (2004). Cell, 117, 927–939.

  98. Yang Z, Rayala S, Nguyen D, Vadlamudi RK, Chen S and Kumar R . (2005). Cancer Res., 65, 3179–3184.

  99. Yook JI, Li XY, Ota I, Fearon ER and Weiss SJ . (2005). J. Biol. Chem., 280, 11740–11748.

  100. Yuan ZQ, Sun M, Feldman RI, Wang G, Ma X, Jiang C, Coppola D, Nicosia SV and Cheng JQ . (2000). Oncogene, 19, 2324–2330.

  101. Zoltan-Jones A, Huang L, Ghatak S and Toole BP . (2003). J. Biol. Chem., 278, 45801–45810.

Download references


This work was supported by grants from the Ligue Nationale Contre le Cancer (Equipe labellisée) and Cancéropole, NIH Grants CA105008, CA06927, and an appropriation from the Commonwealth of Pennsylvania to the Fox Chase Cancer Center. We would like to thank Drs Brigitte Boyer, John Burch, Jonathan Chernoff and Beatrice Mintz for helpful discussion and critical reading of the manuscript; all the members of their laboratories for helpful discussions; and Rose Sonlin for expert secretarial assistance.

Author information



Corresponding authors

Correspondence to Lionel Larue or Alfonso Bellacosa.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Larue, L., Bellacosa, A. Epithelial–mesenchymal transition in development and cancer: role of phosphatidylinositol 3′ kinase/AKT pathways. Oncogene 24, 7443–7454 (2005).

Download citation


  • epithelial–mesenchymal transition
  • PI3K
  • AKT
  • E-cadherin
  • invasion
  • metastasis

Further reading


Quick links