Transforming growth factor beta (TGFbeta) signalling in palatal growth, apoptosis and epithelial mesenchymal transformation (EMT).

Formation of the medial edge epithelial (MEE) seam by fusing the palatal shelves is a crucial step of palate development. The opposing shelves adhere to each other at first by adherens junctions, then by desmosomes in the MEE. The MEE seam disappears by epithelial mesenchymal transformation (EMT), which creates confluence of connective tissue across the palate. Cleft palate has a mutifactorial etiology that often includes failure of adherence of apposing individual palatal shelves and/or EMT of the MEE. In this review, we first discuss TGFbeta biology, including functions of TGFbeta isoforms, receptors, down stream transcription factors, endosomes, and signalling pathways. Different isoforms of the TGFbeta family play important roles in regulating various aspects of palate development. TGFbeta1 and TGFbeta2 are involved in growth, but it is TGFbeta3 that regulates MEE transformation to mesenchyme to bring about palatal confluence. Its absence results in cleft palate. Understanding of TGFbeta family signalling is thus essential for development of therapeutic strategies. Because TGFbeta3 and its downstream target, LEF1, play the major role in epithelial transformation, it is important to identify the signalling pathways they use for palatal EMT. Here, we will discuss in detail the mechanisms of palatal seam disappearance in response to TGFbeta3 signalling, including the roles, if any, of growth and apoptosis, as well as EMT in successful MEE adherence and seam formation. We also review recent evidence that TGFbeta3 uses Smad2 and 4 during palatal EMT, rather than beta-Catenin, to activate LEF1. TGFbeta1 has been reported to use non-Smad signalling using RhoA or MAPKinases in vitro, but these are not involved in activation of palatal EMT in situ. A major aim of this review is to document the genetic mechanisms that TGFbeta uses to bring about palatal EMT and to compare these with EMT mechanisms used elsewhere.

Laser capture microdissection (LCM) for analysis of gene expression in specific tissues during embryonic epithelial-mesenchymal transformation.

The analysis of gene expression in developing organs is a valuable tool for the assessment of genetic fingerprints during the various stages of tissue differentiation and epithelial-mesenchymal transformation (EMT). However, the variety of differentiating cells and the close association of epithelial and mesenchymal cells makes it difficult to extract protein and mRNA from specific cells and tissue and, thus, to assign expressed genes to specific cell populations. We report here the analysis of LEF1 mRNA in epithelial and mesenchymal cells isolated by LCM from different stages of EMT during development of the mouse palate and describe our techniques in detail. By applying a laser capture microdissection (LCM) technique and real-time polymerase chain reaction, we were able to determine mRNA levels that accurately reflect changes in gene expression in specific cells. The sensitivity of the technique is remarkable. Indeed, the mRNAs can be detected for many proteins too low in abundance to stain with antibodies. These techniques will enable embryologists to collect homogeneous groups of cells from heterogeneous populations in developing organs, which otherwise would not be available for gene analysis.

The mechanism of palatal clefting in the Col11a1 mutant mouse.

The occurrence of cleft palate in mutant mice offers an opportunity to understand the possible role of specific genes in palatogenesis. Here, cleft palate in mice carrying the chondrodysplasia (cho) defect, which consists of an autosomal-recessive mutation in the collagen gene Col11a1, was investigated. The proposed cause of cleft palate in cho homozygous mice is failure of the palatal shelves to adhere and make contact due to mandibular growth abnormalities. Another cause of cleft palate that has recently been demonstrated in other animal models is failure of the midline epithelial seam forming between the shelves to undergo epithelial-mesenchymal transformation (EMT). The present strategy to test the likelihood of this second possibility was to culture the unfused cho/cho palatal shelves at different stages of development to see if they were capable of adhering and undergoing EMT in vitro. By using carboxydichlorofluorescein succinimidyl ester to trace the fate of the medial-edge epithelium (MEE), it was shown that cho/cho palates have full potential for MEE adherence and EMT up to embryonic day 17.5/18.5, when epithelia keratinize before birth, preventing the adherence of both the normal and homozygous palatal shelves. Thus, the major effect of the mutant collagen gene on the palate is likely to be via mandibular growth disruption. The possibility that unfused palatal shelves in other clinical syndromes can adhere and undergo EMT if brought into contact at appropriate times before birth has important therapeutic implications.

New evidence that nuclear import of endogenous beta-catenin is LEF-1 dependent, while LEF-1 independent import of exogenous beta-catenin leads to nuclear abnormalities.

The once accepted idea that LEF-1 transports beta-catenin into nuclei has recently been challenged by experiments using exogenous beta-catenin. Here, we investigated the effects of beta-catenin and LEF-1 on nuclear import of beta-catenin using different combinations of exogenous and endogenous molecules over longer lengths of time than previously studied. Nuclear beta-catenin is not detectable in corneal fibroblasts and epithelia or NIH 3T3 and MDCK cells. In LEF-1 transfections, we show that the B-box of LEF-1 is required to move cytoplasmic endogenous beta-catenin into the nuclei of such cells, proving that LEF-1 does transport endogenous beta-catenin into nuclei. Moreover, transfection of uveal melanoma cells with B-box deficient LEF-1 inhibits nuclear import of beta-catenin by endogenous LEF-1. However, the movement of overexpressed exogenous beta-catenin into nuclei is unaffected by the presence or absence of LEF-1 and forms abnormal nuclear aggregates that are a prelude to subsequent apoptosis. We conclude that nuclear transport of exogenous beta-catenin independently of LEF-1 has questionable physiological significance.

Epithelial-mesenchymal transformation is the mechanism for fusion of the craniofacial primordia involved in morphogenesis of the chicken lip.

We have previously demonstrated that epithelial-mesenchymal transformation (EMT) brings about TGF beta 3-induced confluence of craniofacial primordia that derive from the maxillary processes and give rise to the avian palate. The upper lip of the chick embryo forms by confluence of primordia also derived from the maxillary processes, but in this case, they fuse with the intermaxillary segment of the nasofrontal process. Here, we ask whether the bilateral epithelial seams formed when these primordia contact each other in vivo are removed by apoptosis (as formerly was believed to occur in developing palate) or by EMT. We found that, as is the case in the palate, the periderm of the two-layered embryonic epithelium begins to slough shortly before these primordia fuse, bringing the basal epithelial cells into close contact. We show by TUNEL staining and confirm by TEM that apoptosis occurs only in periderm. TEM reveals that basal epithelial cells contacting each other to form the midline seam produce numerous desmosomes with each other. Then, basement membrane begins to disappear, numerous filopodia extend from the basal surfaces of epithelial cells, the space between them enlarges, and the seam breaks apart, leaving mesenchymal cells in its wake. Transformation of the carboxyfluorescein (CCFSE)-labeled epithelial seam is demonstrated in vivo by detection of CCFSE bodies in mesenchymal cells that replace it. This demonstration of EMT in avian lip development lays important groundwork for understanding the causes of human cleft lip and analyzing the mechanism of action of growth factors, such as SHH and BMPs, that have been shown (J. A. Helms et al., 1997, Dev. Biol. 187, 25-35) to be involved in avian lip confluence.

Overexpression of beta-catenin induces apoptosis independent of its transactivation function with LEF-1 or the involvement of major G1 cell cycle regulators.

beta-Catenin promotes epithelial architecture by forming cell surface complexes with E-cadherin and also interacts with TCF/LEF-1 in the nucleus to control gene expression. By DNA transfection, we overexpressed beta-catenin and/or LEF-1 in NIH 3T3 fibroblasts, corneal fibroblasts, corneal epithelia, uveal melanoma cells, and several carcinoma cell lines. In all cases (with or without LEF-1), the abundant exogenous beta-catenin localizes to the nucleus and forms distinct nuclear aggregates that are not associated with DNA. Surprisingly, we found that with time (5-8 d after transfection) cells overexpressing beta-catenin all undergo apoptosis. LEF-1 does not need to be present. Moreover, LEF-1 overexpression in the absence of exogenous beta-catenin does not induce apoptosis, even though some endogenous beta-catenin moves with the exogenous LEF-1 into the nucleus. TOPFLASH/FOPFLASH reporter assays showed that full-length beta-catenin is able to induce LEF-1-dependent transactivation, whereas Arm beta-catenin totally abolishes the transactivating function. However, Arm beta-catenin, containing deletions of known LEF-1-transactivating domains, has the same apoptotic effects as full-length beta-catenin. Overexpressed beta-catenin also induces apoptosis in cells transfected with nuclear localization signal-deleted LEF-1 that localizes only in the cytoplasm. Thus, the apoptotic effects of overexpressed exogenous beta-catenin do not rely on its transactivating function with nuclear LEF-1. Overexpressed delta-catenin, containing 10 Arm repeats, induces only minor apoptosis, suggesting that the major apoptotic effect may be due to domains specific to beta-catenin as well as to Arm repeats. The absence of p53, Rb, cyclin D1, or E2F1 does not affect the apoptotic effect of overexpressed beta-catenin, but Bcl-x(L) reduces it. We hypothesize that in vivo apoptosis of cells overexpressing beta-catenin might be a physiological mechanism to eliminate them from the population.

Tissue-specific expression of beta-catenin in normal mesenchyme and uveal melanomas and its effect on invasiveness.

This paper is the first in a series aimed at understanding the role of beta-catenin in epithelial-mesenchymal transformation (EMT) and acquisition of mesenchymal invasive motility. Here, we compare the expression of this and related molecules in the two major tissue phenotypes, epithelial and mesenchymal, the latter including normal avian and mammalian fibroblasts and malignant human uveal melanoma cells. Previously, it was proposed that src initiates EMT by tyrosine phosphorylation of the cadherin/catenin complex resulting in a negative effect on epithelial gene expression. On the contrary, we found that although beta-catenin becomes diffuse in the cytoplasm during embryonic EMT, the cytoplasmic beta-catenin of the embryonic and adult mesenchymal cells we examined is not tyrosine phosphorylated. Pervanadate experiments indicate that cytoplasmic PTPases maintain this dephosphorylation. GSK-3beta is present, but little or no APC occurs in normal and neoplastic mesenchymal cells. The function of the nonphosphorylated cytoplasmic beta-catenin in mesenchyme may be related to invasive motility. Indeed, in order to invade extracellular matrix, transitional (Mel 252) melanoma cells transform from an epithelial to a mesenchymal phenotype with increased cytoplasmic beta-catenin. Moreover, antisense beta-catenin and plakoglobin ODNs inhibit Mel 252 and corneal fibroblast invasion of collagen. All fibroblastic, transitional, and spindle melanoma cells contain nuclear as well as cytoplasmic beta-catenin, but they are not significantly more invasive than normal fibroblasts that contain only cytoplasmic beta-catenin.

Simultaneous loss of expression of syndecan-1 and E-cadherin in the embryonic palate during epithelial-mesenchymal transformation.

Epithelial-mesenchymal transformation (EMT) is the key mechanism for fusion and confluence of the rodent palate. During this process, medial edge epithelia (MEE) form a midline seam that subsequently transforms to mesenchymal cells. We studied syndecan-1 and E-cadherin, two molecules which have been shown to promote the epithelial phenotype, to determine their fate during palatal EMT. We found that both syndecan-1 and E-cadherin are expressed on basolateral surfaces of the MEE at day 14. Twelve hours later, when a midline seam has formed, syndecan-1 and E-cadherin are still present on its basal and lateral epithelial surfaces and they persist after the seam breaks up into epithelial islands. Then, expression of both molecules is lost simultaneously and abruptly when EMT occurs. On the contrary, previous in vitro studies of cell lines transfected with antisense cDNAs suggested that loss of syndecan-1 would lead to loss of E-cadherin or vice versa. We conclude that in vivo, synthesis of both E-cadherin and syndecan-1 is downregulated synchronously by the initiation of EMT, leading to an effective and correctly timed conversion of the epithelial cells to mesenchyme.

TGFbeta3 promotes transformation of chicken palate medial edge epithelium to mesenchyme in vitro.

Epithelial-mesenchymal transformation plays an important role in the disappearance of the midline line epithelial seam in rodent palate, leading to confluence of the palate. The aim of this study was to test the potential of the naturally cleft chicken palate to become confluent under the influence of growth factors, such as TGFbeta3, which are known to promote epithelial-mesenchymal transformation. After labeling medial edge epithelia with carboxyfluorescein, palatal shelves (E8-9) with or without beak were dissected and cultured on agar gels. TGFbeta1, TGFbeta2 or TGFbeta3 was added to the chemically defined medium. By 24 hours in culture, medial edge epithelia form adherent midline seams in all paired groups without intact beaks. After 72 hours, seams in the TGFbeta3 groups disappear and palates become confluent due to epithelial-mesenchymal transformation, while seams remain mainly epithelial in control, TGFbeta1 and TGFbeta2 groups. Epithelium-derived mesenchymal cells are identified by carboxyfluorescein fluorescence with confocal microscopy and by membrane-bound carboxyfluorescein isolation bodies with electron microscopy. Labeled fibroblasts completely replace the labeled epithelia of origin in TGFbeta3-treated palates without beaks. Single palates are unable to undergo transformation, and paired palatal shelves with intact beaks do not adhere or undergo transformation, even when treated with TGFbeta3. Thus, physical contact of medial edge epithelia and formation of the midline seam are necessary for epithelial-mesenchymal transformation to be triggered. We conclude that there may be no fundamental difference in developmental potential of the medial edge epithelium for transformation to mesenchyme among reptiles, birds and mammals. The bird differs from other amniotes in having developed a beak and associated craniofacial structures that seemingly keep palatal processes separated in vivo. Even control medial edge epithelia partly transform to mesenchyme if placed in close contact. However, exogenous TGFbeta3 is required to achieve complete confluence of the chicken palate.

Effects of electroporation on the tubulin cytoskeleton and directed migration of corneal fibroblasts cultured within collagen matrices.

Electroporation provides a useful method for loading fibroblasts with fluorescent probes for the cytoskeleton, but the possible deleterious effects of this loading technique on cell motility are unknown. We have used conventional and confocal microscopy of living cells and immunohistochemistry to examine the migration and cytoskeleton of chick embryo corneal fibroblasts electroporated while cultured within collagen gels. Fibroblasts cultured in collagen (1 mg/ml) are successfully electroloaded (0.5-1.0 kVcm-1/960 microF in DMEM/F12/20 mM Hepes, pH 7.2) with dextran (4-150 kDa) and immunoglobulin, but subsequently display uncoordinated pseudopodia and hence are unable to migrate effectively in any one direction. The lack of directed movement is due to depolymerization of microtubules and/or a perinuclear collapse of vimentin filaments, seemingly caused by millimolar levels of Ca2+ ions derived from culture medium following electroporation. Fibroblasts loaded in a buffer which resembles intracellular fluid (< or = 10 microM Ca2+) maintain their cytoskeleton and continue to migrate, when returned to culture medium within 10 min. Using this novel approach, we have loaded fibroblasts migrating through extracellular matrix (ECM) with rhodamine phalloidin and monitored the behavior of the labeled actin cortex by confocal microscopy. During migration phalloidin-actin accumulates near the base of pseudopodia and at the rear of the cell where it is subsequently left behind. We conclude that electroporation is a valuable technique for loading fibroblasts to study migration within ECM, provided that the conditions used support stability of the tubulin cytoskeleton.

E-cadherin transforms embryonic corneal fibroblasts to stratified epithelium with desmosomes.

Important and precisely regulated transitions in tissue phenotype from epithelium to mesenchyme and from mesenchyme to epithelium occur in the developing embryo. The gene for E-cadherin has been shown to cause fibroblastic cell lines to become epithelioid in culture. We asked whether or not the activities of the E-cadherin gene could cause a definitive embryonic mesenchyme to transdifferentiate into an epithelial phenotype. Primary corneal fibroblasts from 6- to 7-day-old chick embryos were contransfected by impact loading with plasmids containing E-cadherin and Neo genes and selected in G418. The fibroblasts expressing E-cadherin aggregate, localize E-cadherin to lateral surfaces, and form stratified epithelia that develop zonulae occludentes and adherentes, connexin 43, cytokeratin, desmoplakin, and desmosomes. Vimentin intermediate filaments persist and no basement membranes appear, even though the cells synthesize laminin and type IV collagen. Our study is the first to demonstrate the ability of E-cadherin to induce fibroblasts to form desmosomes and stratified epithelia. The primary embryonic fibroblasts apparently have more developmental potential to transdifferentiate into epithelia than do the fibroblastic cell lines previously studied. We conclude that E-cadherin is likely to play an important role in transformation of mesenchyme to epithelium in the embryo.

Transformations between epithelium and mesenchyme: normal, pathological, and experimentally induced.

In this review, we define the two major tissue types, epithelium and mesenchyme, and we describe the transformations (transdifferentiations) of epithelium to mesenchyme (EMT) and mesenchyme to epithelium (MET) that occur during embryonic development. The differentiation of the metanephric blastema provides a striking example of MET. Differentiation of metanephric epithelium is promoted by matrix molecules and receptors (nidogen, laminins, alpha 6 integrins), hepatic growth factor/scatter factor, and products of the genes wnt-1, wnt-4, and Pax-2. Transformation of MDCK epithelium to mesenchyme-like cells is promoted in vitro by antibodies to E-cadherin, products of v-src, v-ras, and v-mos, and by manipulation of the epithelium on collagen gels. Suspension in collagen gel, transforming growth factors, and c-fos have also been shown to promote EMT in epithelia. We present studies from our laboratory showing that alpha 5 beta 1 integrin has a role in the EMT of lens epithelium that is brought about by suspension in collagen gel. Our laboratory has also shown that transfection with the E-cadherin gene induces embryonic corneal fibroblasts to undergo MET and that this MET is enhanced by interaction of the differentiating epithelium with living fibroblasts. This review calls attention to the roles that EMT and MET might have in kidney pathologies and urges further study of the involvement of these phenomena in renal development, renal injury, and renal malignancy.

An overview of epithelio-mesenchymal transformation.

Epithelium is the tissue phenotype of early embryos and primitive adults of the chordate phylum. A second tissue type, however, is produced by epithelial-mesenchymal transformation (EMT) in higher chordates, such as vertebrata. Mesenchymal cells have the ability, which true epithelia do not, to invade and migrate through the extracellular matrix (ECM) to create dramatic cell transpositions. The first-formed or primary mesenchymal cells in amniote vertebrates migrate from the primitive streak to differentiate into the mesodermal and endodermal epithelia. Definitive mesenchyme with connective tissue and muscle potentials arises from the epithelial mesoderm at about the same time as the neural crest mesenchyme forms from the ectoderm. Later on in embryogenesis. EMT is used to remodel unwanted epithelia, such as that of the palate medial edges. We discuss the mechanisms by which epithelial cells transform into mesenchyme and vice versa. On the one hand, cells activate putative mesenchymal master genes, turn off epithelial genes, and acquire motility machinery that allows them to interact in 3 dimensions (3D) with ECM via actin cortex while sliding their endoplasm into their new front ends. On the other hand, primary mesenchymal cells can reactivate epithelial regulatory genes, such as E-cadherin, turn off the motility machinery for invading ECM, and reexpress apical-basal polarity. We review the genes, such as FSP1, src, ras, and fos, that are activated in cells transforming to mesenchyme and the genes their neighbors activate to induce EMT, such as those for TGF beta, NT-3, and sonic hedgehog. Suspension in 3D collagen gels can induce adult epithelium to undergo EMT; alpha 5 beta 1 integrin is activated on surfaces in contact with collagen, including apical surfaces that do not normally express integrins. In vivo, it is possible that pathological manipulations of a cell's environment likewise induce EMT. Of the examples we give, the creation of invasive metastatic carcinoma cells by EMT is the most fearful. Interestingly, transfection of either metastatic cells or normal embryonic fibroblasts with the E-cadherin gene converts them to the epithelial phenotype. It may be possible in the future to manipulate the tissue phenotype of diseased cells to the advantage of the animal.

Expression of beta 1 integrins changes during transformation of avian lens epithelium to mesenchyme in collagen gels.

Remarkably, a number of definitive epithelia, such as that of the anterior lens, give rise when suspended within 3D gels of type I collagen, to elongate, bipolar shaped cells that exhibit the ultrastructure, polarity, and migratory ability of mesenchymal cells. They begin producing type I collagen and stop producing crystallins, type IV collagen, and laminin. Here, we investigated changes in beta 1 integrins and their extracellular matrix (ECM) ligands during this transdifferentiation. The former free surface of the lens epithelium that is now in contact with collagen begins within a day to stain intensely for beta 1 and it is this surface rather than the surface facing the basement membrane that gives rise to mesenchymal cells. Immunoprecipitation experiments reveal a large increase in the beta 1 integrin subunit on mesenchymal cells as compared to the epithelium of origin. The alpha 5 integrin subunit, which is barely detectable in the lens, increases in the mesenchymal cells and alpha 3 continues to be expressed at about the same level as in the epithelium. alpha 6, the epithelial integrin subunit, and laminin, its ECM ligand, are not detected immunohistochemically or biochemically in the mesenchyme. Rather, the mesenchymal cells secrete abundant fibronectin, the major ECM ligand for alpha 5 beta 1. RGD peptides do not inhibit the transformation but antibodies to beta 1 do perturb the emigration of mesenchymal cells from the lens apical surface. We conclude that the beta 1 integrins newly expressed on the apical epithelial surface interact with the surrounding 3D collagen gel to help bring about this unusual epithelial-mesenchymal transition.

Cytoplasmic loading of dyes, protein and plasmid DNA using an impact-mediated procedure.

We describe a method and apparatus designed to rapidly and reproducibly produce transient, survivable plasma membrane disruptions--"wounds"--in order to gain access to the cytoplasm of eukaryotic cells growing in culture. Compressed gas is used to propel glass beads, dispersed as a uniform aerosol, at adherent cells growing on a culture substratum. The impact of beads with the cells creates plasma membrane wounds. Macromolecules, such as dyes, proteins and plasmid DNAs, diffuse from the extracellular environment directly into the cytoplasmic compartment of the cell through these wounds. Resealing of the plasma membrane, necessary for cell survival, traps macromolecules within the cytoplasm of the cell.

Extracellular matrix alters epithelial differentiation.

Extracellular matrix (ECM) induces and maintains the differentiation of epithelial cells, not by totally altering their state of differentiation, but by activating overt differentiation. Recent studies of cultured mammary cells provide an elegant molecular analysis of this kind of progressive cell differentiation. Other studies show that ECM can not only activate and enhance a differentiated state, but can even alter it in bringing about transformation of epithelium to mesenchyme.

Retinoic acid inhibits formation of mesenchyme from lens epithelium in collagen gels.

PURPOSE: To examine the possibility that retinoic acid (RA), a stabilizer of the epithelial phenotype, would inhibit formation of mesenchymal cells from avian lens epithelium in three-dimensional collagen. METHODS: Lens epithelia from 11-day-old chick embryos were cultured for 6 days in collagen gels in the presence of RA. The number of mesenchymal cells emigrating into the gels was quantitatively compared with control cultures to which RA was not added. RESULTS: It was found that few fibroblast-like cells form at the highest dose used (10(-5) M RA) and outgrowth approaches control levels at lower doses of RA. The mesenchymal cells that form after RA treatment are not ultrastructurally different from those of controls. Many have well-developed rough endoplasmic reticulum and undoubtedly produce the collagen fibrils that accumulate around the cells. Others, although spindle-shaped, still exhibit lenslike cytoplasm. New basement membrane is deposited on the former free surface of RA-treated lens epithelium, but is not present at the former free surface of control epithelium. CONCLUSIONS: It is possible that RA inhibition of lens transformation to fibroblast-like cells is at least partly due to the ability of RA to stimulate production of basement membrane components by epithelia. More studies of RA action on epithelial-mesenchymal transformation in collagen gels may reveal additional mechanisms. It is also suggested that mesenchymal genes similar to those activated in lens epithelium by suspension in collagen may turn on in pathologic transformations (ie, in anterior capsular cataract, fibroblast-like cells arise from lens epithelium.