SEARCH

SEARCH BY CITATION

Keywords:

  • epithelial–mesenchymal transition;
  • embryo;
  • Metazoa;
  • mesoderm;
  • evolution

Abstract

  1. Top of page
  2. Abstract
  3. Making of a Pluricellular Organism
  4. Epithelium and Mesenchyme in Embryonic Development
  5. Epithelial–Mesenchymal Transitions in Development
  6. Vertebrate Development Shows a Pattern of Recurrent EMTs
  7. Molecular Regulation of EMTs
  8. Morphogenetic Role of EMTs, and Phylogenetic Implications
  9. EMTs in Adult Organisms
  10. CONCLUSIONS AND PERSPECTIVES
  11. Acknowledgements
  12. LITERATURE CITED

Epithelial–mesenchymal transitions (EMTs) are well known processes in which new mesenchyme is locally generated from epithelia. During the development of the vertebrate embryo, EMTs are a source of mesenchyme in diverse places and stages through embryonic morphogenesis, especially in mesodermal domains. In the present work we consider the embryo as a two-state system in which epithelium and mesenchyme represent the stable and unstable states, respectively. We think that a pattern of recurrent oscillations between the plasticity and exploratory behaviour of the mesenchyme and the stability of the epithelia can be recognized in the embryogenesis of vertebrates and, probably, in most tripoblastic Metazoans. Mesoderm, in particular, might be regarded as a cell layer able to oscillate between epithelial and mesenchymal states. The cellular and molecular mechanisms that enable these recurrent oscillations between stable (epithelial) and unstable (mesenchymal) states during embryogenesis provide the mesoderm with a large plasticity, an extended potential for innovation, and a better control of the three-dimensional (3D) body organization. In this scenario, it is conceivable that the origin of the mesoderm itself might be related to ancestral mechanisms regulating cell adhesion and detachment. We conclude that EMTs played a key role in the evolution of Metazoans, and are involved in the pathological and reparative processes of adult organisms. Anat Rec 268:343–351, 2002. © 2002 Wiley-Liss, Inc.


Making of a Pluricellular Organism

  1. Top of page
  2. Abstract
  3. Making of a Pluricellular Organism
  4. Epithelium and Mesenchyme in Embryonic Development
  5. Epithelial–Mesenchymal Transitions in Development
  6. Vertebrate Development Shows a Pattern of Recurrent EMTs
  7. Molecular Regulation of EMTs
  8. Morphogenetic Role of EMTs, and Phylogenetic Implications
  9. EMTs in Adult Organisms
  10. CONCLUSIONS AND PERSPECTIVES
  11. Acknowledgements
  12. LITERATURE CITED

Primitive forms of life arose in our planet as unicellular prokaryotic organisms with a basic design that subsequently became more complex through eukaryotic advances such as intracellular membrane septation and gain of organelles through endosymbiotic events (Margulis, 1970, 1975). Acquisition of pluricellularity allowed for large increases of organismic size, as well as for the cellular division of functions characteristic of cells from plants, fungi, and animals. Progressive accumulation of innovations at the cellular level also allowed for a wide diversification of the primitive pluricellular design.

Pluricellularity required 1) the development of systems of intercellular adhesion (basically junctional complexes mediated by cadherins and cadherin-associated proteins, such as catenins); 2) synthesis and secretion of insoluble substances to the intercellular medium—mainly collagens, adhesive glycoproteins, and proteoglycans; 3) development of systems of intercellular communication (receptor-mediated and gap junctions); and 4) development of systems of attachment to the extracellular medium, which are also involved in signal transduction (integrin receptors). The first immediate consequence of the evolutive outlining of pluricellularity was the formation of the primordial epithelial structures. As a matter of fact, the association of single, isolated cells to form a pluricellular organism is considered the first step in the emergence of true tissues (Valentine, 1978), and the differentiation of specialized intercellular junctions giving rise to the earliest epithelial structures is thought to be a main synapomorphy of the Metazoa (Nielsen, 1995). Primitive pluricellular organisms probably had a limited cell-to-cell communication and homeostatic regulation. These two essential properties developed through time, reaching a structural level similar to that displayed by Poriferans (sponges), the most primitive Metazoans (Brusca and Brusca, 1990). Sponges present septate junctions between epithelial cells that isolate from the outer environment the cavities in which minerals precipitate and the spicules are formed (Ledger, 1975).

The development of the pluricellular organisms involves changes not only in the number of cells, but also in the cell organization that determines the shape and proportions of the organism. These changes can be achieved by differential growth due to localized cell proliferation, cell hypertrophy, cell migration, or cell death. Differential growth due to cell proliferation and cell hypertrophy is a common phenomenon in plants, in which the spatial relationships between cells during development are more conservative than in Metazoans. Only in animal species, during the process of morphogenesis, are cells able to migrate and relocate themselves in embryonic regions where their presence is required in order to signal and/or to respond to local signals (Bard, 1990). It is important to emphasize that a main difference between the development of plants/fungi and Metazoans is the existence in the latter of mesenchymal cells, which are fibroblastoid in shape, equipped with motor proteins and a proteolytic arsenal, and able to migrate throughout the extracellular matrix (ECM). The lack of a cell wall is indeed a prerequisite for the acquisition of these properties.

Epithelium and Mesenchyme in Embryonic Development

  1. Top of page
  2. Abstract
  3. Making of a Pluricellular Organism
  4. Epithelium and Mesenchyme in Embryonic Development
  5. Epithelial–Mesenchymal Transitions in Development
  6. Vertebrate Development Shows a Pattern of Recurrent EMTs
  7. Molecular Regulation of EMTs
  8. Morphogenetic Role of EMTs, and Phylogenetic Implications
  9. EMTs in Adult Organisms
  10. CONCLUSIONS AND PERSPECTIVES
  11. Acknowledgements
  12. LITERATURE CITED

The mesenchyme and epithelium are thus the basic cell types that constitute the metazoan embryo, and they have been present since the inception of the Metazoans. It is currently accepted that the epithelia precede (both evolutionarily and ontogenetically) the mesenchyme, and that the mesenchyme was originally an epithelial derivative (Hay, 1968).

Mesenchymal and epithelial cell types are different in their intracellular organization and their relationships with the ECM and adjacent cells. Epithelial cells are usually polarized (with a baso-apical polarity) and they adhere to their neighbours to form layers, with the apical surface facing a cavity and the basal surface attached to a specialized ECM structure called the basal lamina (Bard, 1990). Epithelial structures can be found throughout the developing embryo, but most of them only partially satisfy the epithelial description, since the epithelia progressively acquire their specific features as development proceeds. Therefore, intermediate cases of epithelia can be reported in which some epithelial features are not present. Epiblast presents cells that are tightly joined by union complexes, but they constitute a pseudostratified epithelium that cannot be considered exactly to be polarized (Viebahn et al., 1995). Other example is provided by the so-called somitic epithelium, which has been reported to have tight junctions (Cheney and Lash, 1984) but is not an efficient barrier to the flow of diverse ions (Lobo et al., 1999). As a general rule, we consider embryonic cell layers to be epithelia when they meet two criteria: 1) they constitute physical barriers between two different extracellular environments, with or without the presence of a basal lamina; and 2) they are static structures, i.e., when they migrate they move as a whole structure, while keeping a constant close connection between cells.

On the other hand, mesenchymal cells are characterized by their migratory abilities or, at least, by their ability to interact with the ECM in a three-dimensional (3D) environment. They only establish transient contacts with other cells, and their polarization (if it exists) is determined by the direction of their movement. Mesenchymal cells receive changing information from their environment, the ECM, and secrete substances that modify it, sometimes in an irreversible way. Through normal development, most embryonic mesenchymal cells reach their final destination to differentiate and, eventually, regain the epithelial phenotype (Hay, 1996).

Therefore, it seems reasonable to consider epithelia as developmentally stable structures with a basic organization that enables accurate control of their cells. Epithelial cells are coupled through junctional structures of diverse complexity (Farquhar and Palade, 1963; Staehelin and Hull, 1978) that communicate cells between them. Considering that epithelial cells partially share information at the molecular level, an extreme theoretical point of view would lead us to describe the epithelium as an “informative syncytium” due to equivalence between cells included in the epithelium. From a mathematical point of view, information (H) is a function of the associated probability to find a “type” inside a diverse population (H = −Σpi log pi) (Shannon and Weaver, 1963). In this model, equality between cells reduces the amount of information, i.e., if there is only “one type” of cell, then pi = 1, log pi = 0 and H = 0. Information reaches its maximum in heterogeneous populations when pi1 = pi2 = pi3 = … = pin. Mesenchyme would be also an extreme theoretical example of this second informative type. Mesenchymal cells represent a rather heterogeneous population in which motile cells receive changing information from their environment, the ECM. Therefore, the new-found conditions can lead to differentiation and subsequent morphogenesis of mesenchymal cells. Of course, this model accounts only for defined regions of an epithelium, because cells belonging to the same epithelium can be instructed by quite different underlying tissues which can, in turn, elicit specific responses in the epithelial population, breaking its informative homogeneity. The transformation of the epithelial cells into migratory mesenchymal cells is one such type of specific response.

Epithelial–Mesenchymal Transitions in Development

  1. Top of page
  2. Abstract
  3. Making of a Pluricellular Organism
  4. Epithelium and Mesenchyme in Embryonic Development
  5. Epithelial–Mesenchymal Transitions in Development
  6. Vertebrate Development Shows a Pattern of Recurrent EMTs
  7. Molecular Regulation of EMTs
  8. Morphogenetic Role of EMTs, and Phylogenetic Implications
  9. EMTs in Adult Organisms
  10. CONCLUSIONS AND PERSPECTIVES
  11. Acknowledgements
  12. LITERATURE CITED

The segmentation of the vertebrate zygote leads to the establishment of an epithelial layer that frequently adopts the shape of a sphere or a disc surrounding a cavity (the blastocoel). Thus, epithelium is the earliest type of embryonic cell organization. It can be inferred from this that all embryonic mesenchymal cells have their origin in epithelial cells (Hay, 1968). The main morphogenetic event accounting for mesenchymal cell formation from epithelial cells is called the epithelial-to-mesenchymal transition or transformation (EMT) (reviewed in Hay, 1996).

EMTs constitute a fundamental kind of cellular process, and they are currently receiving increasing attention from developmental biologists. The EMT involves a phenotypical shift from the epithelial to the mesenchymal phenotype, frequently in response to an extrinsic signaling system. In order to accomplish such a radical change, the transforming epithelial cells have to reorganize their cytoskeleton, detach from the epithelial lining, degrade the basal lamina, and invade the ECM. Morphological evidence concerning cellular features of pretransforming epithelia include cell hypertrophy, loss of cell adhesion, cell overriding, and production of basal cytoplasmic projections (Markwald et al., 1975, 1977). Many transforming epithelia also change their intermediate filaments from cytokeratin to vimentin (Greenburg and Hay, 1989; Hay, 1990; Pérez-Pomares et al., 1997), a cytoskeletal shift that seems to be mandatory to start the transformation (Greenburg and Hay 1989; Hay, 1990).

Mesenchymal-to-epithelial transition (MET), i.e., the reacquisition of the epithelial phenotype by mesenchymal cells, also occurs during development. This process underlies the morphogenesis of some organs, such as the kidney. In this case, the metanephrogenic mesenchyme is induced by the ureteric bud to aggregate, express cell adhesion molecules (such as E-cadherin), and secrete the components of the basal membrane (Davies, 1996). Through embryogenesis, mesenchymal cells that do not regain an epithelial state through MET usually differentiate and give rise to muscle, bone, nerve, or connective tissue.

After an event of EMT, differentiation of the newly produced mesenchyme follows. The highest diversity in cell lineage differentiation, i.e., the largest developmental potential, occurs in the mesenchyme derived from the chronologically earliest EMT (the formation of mesoderm). This developmental potential progressively decreases in later EMTs (see below). Epithelial cell attachment and coupling can be considered as homogenizing factors limiting the developmental and evolutionary potential of epithelial cells. It is tempting to suggest that these factors have been balanced through evolution with a gain of developmental potentials in epithelium-derived mesenchymal cells. The transient instability generated by the EMT might be regarded as necessary in the acquisition of potential to explore new microenvironments, as well as for the exposure to new signaling molecules in an instance of what has been called exploratory behaviour (Gerhart and Kirschner, 1997). These mechanisms enable much greater complexity of responses than can be accommodated by conventional contingency (i.e., cell response) mechanisms alone.

Epithelium appears to be the stable state of cell organization in Metazoans, but it remains to be determined how this stability is controlled. Epithelia are able to maintain their structure with remarkable accuracy in spite of the challenges posed by cell turnover, external aggressions, etc. This observation leads to the concept of the “morphostat,” i.e., morphogen-like controller molecules responsible for maintaining normal epithelial tissue microarchitecture (Potter, 2001).

Vertebrate Development Shows a Pattern of Recurrent EMTs

  1. Top of page
  2. Abstract
  3. Making of a Pluricellular Organism
  4. Epithelium and Mesenchyme in Embryonic Development
  5. Epithelial–Mesenchymal Transitions in Development
  6. Vertebrate Development Shows a Pattern of Recurrent EMTs
  7. Molecular Regulation of EMTs
  8. Morphogenetic Role of EMTs, and Phylogenetic Implications
  9. EMTs in Adult Organisms
  10. CONCLUSIONS AND PERSPECTIVES
  11. Acknowledgements
  12. LITERATURE CITED

The primitive epiblast represents the earliest epithelial state in vertebrate development. In the amniote embryo, migratory cells ingressing from the epiblast give rise to the mesodermal mesenchyme (with the exception of the prechordal mesenchyme (Viebahn, 1995; Viebahn et al., 1995) and the endodermal progenitors (Gilbert, 1995). This process is called the “primary EMT,” and the original mesodermal mesenchyme is the “primary mesenchyme.” This term has been used for the earliest mesenchyme that appears in the echinoderm embryo shortly after gastrulation, in order to distinguish it from the secondary mesenchyme, which originates from the top of the archenteron and gives rise to the mesodermal organs (Fink and McClay, 1985). In the present work we use the term “primary mesenchyme” in a broader sense, to include the first wave of mesenchymal cells that populate the blastocoel of vertebrates.

Immediately after formation of the primary mesenchyme, most mesodermal and all endodermal cells reorganize in primitive epithelia. These secondary epithelia include the paraxial (somitic) and lateral plate mesoderm, i.e., somatopleura and splanchnopleura (Hay, 1990). Although some of those epithelia do not display all the cellular features that usually define an adult epithelium, they meet the minimum criteria to be considered epithelial states, as discussed above.

Secondary epithelia give rise to secondary mesenchymal cells in a set of EMTs. Sclerotomic cells, dermal cells, and limb bud myoblasts detach from the originally epithelial somite (Ordahl, 1993). The precardiac mesoderm is another structure organized as an epithelium (Peng et al., 1990), which is continuous with the splanchnopleura; precardiac epithelium undergoes an EMT that generates the endocardial progenitors (Sugi and Markwald; 1996, Mjaatvedt et al., 1999; Lough and Sugi, 2000). Intermediate mesoderm originates the nephrogenic mesenchyme, which coalesces in tertiary epithelial structures (e.g., nephrones, nephric ducts, and Müllerian ducts, see Gilbert, 1995).

A portion of the lateral mesodermal epithelia (somatopleural and splanchnopleural) remain as epithelia (visceral and parietal coelomic mesothelia, including the epicardial and pericardial mesothelia). These coelomic mesothelia give rise to another set of secondary mesenchyme, a fact that was well known by earlier anatomists (Gruenwald, 1942), but which has frequently been neglected in the recent literature. For example, the cells that form the adrenal cortex arise from the peritoneal epithelium of the mesonephric ridges (Bellairs and Osmond, 1998). Sertoli cells of the testicle are also mesothelial derivatives (Bellairs and Osmond, 1998; Karl and Capel, 1998). Mesothelium-derived mesenchyme differentiate into a number of cell types, including fibroblasts and smooth muscle cells (Dettman et al., 1998; Pérez-Pomares et al., 1999). Some authors have proposed a broader potential of the visceral mesothelium-derived cells, which would include endothelial and hemopoietic progenitors (Olah et al., 1988; Muñoz-Chápuli et al., 1999, 2001). Clearly, the endothelium of the liver sinusoids and the coronary vessels differentiates from mesothelium-derived mesenchymal cells (LeDouarin, 1975; Moore et al., 1998; Muñoz-Chápuli et al., 2001) (Pérez-Pomares et al., unpublished results).

A case of tertiary EMT, involving the endocardial endothelium, is well known: atrioventricular and conal endocardium transforms into valvuloseptal mesenchyme, originating the fibrocytic component of endocardial cushions, i.e, the primordia of the heart valves (Markwald et al., 1975, 1977, 1996). The vascular endothelium could also be a source of smooth muscle cells in the developing embryo (DeRuiter et al., 1997; Arciniegas et al., 2000). The formation of Müllerian duct-derived cells constitutes another case of tertiary mesenchyme (Trelstad et al., 1982).

It can be concluded from the preceding descriptions that most EMTs occur in the mesodermal lineage. An exception to this rule is the ectodermal EMT, which gives origin to the neural crest in vertebrates. The formation of the neural crest from the epithelial neural plate gives rise to a highly migratory population of mesenchyme that invades all the vertebrate body and differentiates in a number of cell types. However, some authors consider the neural crest as an “ectomesoderm” (Nielsen, 1995), reinforcing the view that EMT are developmental events linked to the mesodermal populations. As far as we know, no EMT has been reported for normal endodermal cells in vivo, although liver parenchymal cells, for example, can accomplish mesenchymal transformation in vitro (Haynes et al., 1988). Examples of successive EMTs in embryonic epithelia are illustrated in Figure 1.

thumbnail image

Figure 1. Semithin sections of chick embryos showing different transitions between epithelium and mesenchyme. A: Stage H/H6. Origin of the mesoderm. The epithelial epiblast (EB) transforms into a migratory mesenchyme (M) at the primitive streak (S). This primary mesenchyme migrates between the epiblast and the endoderm (EN) (×220). B: Stage H/H7+. MET of the primary mesenchyme. Mesenchymal cells (M) aggregate to form two epithelia (the somatopleure (SO) and splanchnopleure (SP)), which line the coelom (C). At this level, and probably induced by the endoderm (EN), the splanchnopleural epithelium gives rise to secondary mesenchymal cells that will constitute the endocardial precursors (arrowhead). EB: epiblast, NC: notochord, NP: neural plate (×220). C: Stage H/H18. EMT of the splanchnopleural mesothelium. Mesenchymal cells (arrow) migrate from the splanchnopleural mesothelium (SP) to the splanchnopleural mesoderm (SM). Note the difference between this transforming mesothelium and the somatopleural mesothelium (SO), which is quiescent at this level and developmental stage. G, gut; H, heart. (×336). D: Stage H/H18. Late EMTs in the developing heart. Mesenchymal cells originate from the endocardium (EC) and the epicardium (EP) and populate the endocardial cushion mesenchyme (CM) and subepicardial mesenchyme (SEM). This transition is probably induced by signal(s) locally produced by the myocardium (MC) (×270).

Download figure to PowerPoint

Some cell lineages undergo three consecutive EMTs, as shown in Table 1. However, the mesenchyme derived from a given EMT seems to show a lesser differentiation potential than that originated in the previous EMT, as mentioned above. This fact is summarized in the model depicted in Figure 2.

Table 1. The starting points of three epithelial-mesenchymal transitions (EMT), occurring in ontogenetically related tissues of the avian embryo*
TissuesEMT starting point
HoursH/H stages
  • *

    The second and third EMT start at a developmental stage which doubles the stage of the preceding EMT, if we estimate the time of development either by hours of incubation or by the Hamburger and Hamilton (1951) stages.

EMT-I. Epiblast to mesoderm133+/4
EMT-II. Splanchnopleural epithelium to mesenchyme (including endocardial progenitors)23–267/7+
EMT-III. Endocardium to valvuloseptal mesenchyme50–5615/16
thumbnail image

Figure 2. EMTs involve some degree of dedifferentiation, which increases the potentiality of the mesenchymal cells that originate from the epithelium. The differentiation potential of the mesodermal cells throughout development decreases, whereas the time period for the initiation of the new EMT increases (time lapse approximately doubles each time in some cases, as shown in Table 1). This process can be represented as a pattern consisting of peaks and valleys, as shown in the diagram.

Download figure to PowerPoint

We can recognize a similar pattern of recurrent EMTs in nonvertebrate metazoans. Primary and secondary mesenchyme in echinoderms are good examples of a phenotypical shift from epithelial-like cells lined by a basal lamina, to migratory cells that display extremely fine filopodia (probably with an exploratory function) and are capable of responding to positional information (Gilbert, 1995). The early mesenchyme of most invertebrates frequently reorganize in epithelial structures, such as the metanephridia/renopericardial canal and gonadal structures in molluscs (Raven, 1966). In some arthropods, muscles are formed from the disaggregate coelomic epithelium (Anderson, 1969), which is linked to the importance of the coelom as a basic structure in Metazoan evolution. Finally, it is important to note that coelomocytes originate from the dissociation of cells from the coelomic mesothelium (Mano, 1964; Jamieson, 1992; Mukai et al., 1997) in a process that could be considered a true EMT. Coelomocytes differentiate into specialized lineages that acquire functions similar to those of the vertebrate blood cells, i.e., defense against microbial aggression, and transport of respiratory pigments.

We conclude that the origin of the mesoderm, which was probably related to the earliest of EMTs, provided the organisms with a population of cells that organize themselves in epithelia that are able to undergo recurrent cycles of stability and instability, increasing the potential to generate differentiative complexity in each cycle. In this context, the claim made by some zoologists that the neural crest develops as a type of mesoderm (Nielsen 1995) appears to be justified.

Molecular Regulation of EMTs

  1. Top of page
  2. Abstract
  3. Making of a Pluricellular Organism
  4. Epithelium and Mesenchyme in Embryonic Development
  5. Epithelial–Mesenchymal Transitions in Development
  6. Vertebrate Development Shows a Pattern of Recurrent EMTs
  7. Molecular Regulation of EMTs
  8. Morphogenetic Role of EMTs, and Phylogenetic Implications
  9. EMTs in Adult Organisms
  10. CONCLUSIONS AND PERSPECTIVES
  11. Acknowledgements
  12. LITERATURE CITED

It can be argued that the different EMTs are related only by a superficial morphological resemblance, lacking any other kind of significant relationship. However, new developments in our understanding of the molecular mechanisms that regulate different EMTs reveal striking similarities. For example, the transcription factor Slug is expressed during, and it is probably essential for, the EMTs that give origin to the mesoderm (i.e., gastrulation) and the neural crest of the avian embryo (Nieto et al., 1994). Slug is also involved in the differentiation of the endocardial cushion and subepicardial mesenchyme, two late processes of EMT (Romano and Runyan, 1999, 2000; Carmona et al., 2000). Slug belongs to the Snail family of zinc-finger transcription factors, which are essential in the development of the mesoderm in Drosophila (Alberga et al., 1991). It is probable that Snail plays the same role as Slug in mammalian development (Sefton et al., 1998). Some of the functions of Snail are probably mediated through down-regulation of E-cadherin (Cano et al., 2000; Batlle et al., 2000) thus dissociating intercellular junctions and increasing the pool of cytoplasmic β-catenin, a key component of the Wnt signaling pathway. In other systems of EMT, where the epithelium lacks E-cadherin, other cadherins might play similar roles. Such is the case for the cardiac valve primordia, wherein the endocardium expresses VE-cadherin.

Slug expression in neural crest cells has been reported to be modulated by bone morphogenetic proteins (Sela-Donenfeld and Kalcheim, 1999), which are also involved in the transformation of the precardiac epithelium into mesenchymal endocardial progenitors (Lough and Sugi, 2000). Bone morphogenetic proteins, together with the closely related transforming growth factor (TGF)-β, appear to play a very important inductive role in gastrulation (Beppu et al., 2000), the transformation of endocardial cushions (Nakajima et al., 2000; Romano and Runyan, 2000), and the disappearance of the palatal medial edge epithelium (Kaartinen et al., 1997; Cui and Shuler, 2000). The specific role of the molecules of the TGF-β superfamily in the regulation of the EMT process has been much discussed, but has yet to be defined.

Finally, it is important to emphasize the role played by a number of tyrosine-kinase receptor binding growth factors, such as fibroblast growth factors (FGFs) and hepatocyte growth factors (HGFs). These growth factors affect EMT in many experimental systems in vitro and are probably involved directly in the EMTs that occur in the development of the dermomyotome, heart, and kidney. Tyrosine-kinase membrane receptors are present throughout the animal kingdom, and have been described even in sponges. It has been proposed that receptor tyrosine kinases are molecular synapomorphies of Metazoans arising from ancestral serine/threonine kinases (Muller et al., 1999).

We think that the existence of common regulatory genes acting in diverse examples of EMT and in different animal phyla supports the notion of an ancestral mechanism common to all the EMTs, whose origin may to be related to the origin of the mesodermal cell layer (see below). EMT-inducing molecules have to overcome normal epithelial stability. The evolutionary conservation of a common triggering system for EMT is probably linked to the necessity of a tight control of the EMT process to avoid undesirable generation of migratory and undifferentiated cells. The clinical relevance of the control mechanisms for EMT in the adult organism is discussed below.

Morphogenetic Role of EMTs, and Phylogenetic Implications

  1. Top of page
  2. Abstract
  3. Making of a Pluricellular Organism
  4. Epithelium and Mesenchyme in Embryonic Development
  5. Epithelial–Mesenchymal Transitions in Development
  6. Vertebrate Development Shows a Pattern of Recurrent EMTs
  7. Molecular Regulation of EMTs
  8. Morphogenetic Role of EMTs, and Phylogenetic Implications
  9. EMTs in Adult Organisms
  10. CONCLUSIONS AND PERSPECTIVES
  11. Acknowledgements
  12. LITERATURE CITED

After the formation of the blastocoel and the successive processes of gastrulation and coelomogenesis, the early metazoan organism had to reshape its internal cavities and develop specialized organs. This complex event involved the three blastodermal layers (the ectoderm, endoderm, and mesoderm).

It is important to note that endodermal organs usually increase volume and surface through fractal-like growth strategies (lungs, liver, folds of the gut epithelium, etc.) in which the epithelium migrates as a whole. Ectodermal derivatives other than the neural crest can also show fractal-like growth (mammary epithelium) or formation of tubular or hollow vesicular structures. Differences between the growth strategies of the blastodermal layers may have been essential in the diversification of Metazoans.

A simplification of the complex embryonic process that leads to the acquisition of 3D organization in the embryo would involve two basic morphogenetic mechanisms: 1) tubule formation in ectodermal and endodermal epithelia (neural tube, gut formation, and outgrowth of visceral primordia, such as lungs or liver), and 2) mesodermal EMTs. In most cases, an epithelium acts as a source of morphogenetic signals that control the EMTs of other adjacent epithelia, with a subsequent recruitment of the newly generated mesenchyme and the induction of its differentiation. Examples are the precardiac epithelium (Sugi and Markwald, 1996; Lough and Sugi, 2000), the developing liver (LeDouarin, 1975) and lungs (Weaver et al., 1999), and the growing blood vessels (for review, see Carmeliet, 2001). A model of this mechanism is illustrated in Figure 3.

thumbnail image

Figure 3. The characteristic interaction that develops between some adjacent epithelia in vertebrate embryogenesis. Epithelium I induces an EMT process in epithelium II (black arrows) through the secretion of inducers (shown as purple dots) (see text for details). The epithelium II-derived mesenchymal population (green) is recruited by epithelium I (green-to-blue-graded arrows) and differentiates (blue cells) according to the molecular information arising from the inducing tissue (red dots). Endoderm, which originates tubular structures with a complex branching pattern, would be a classic example of type I epithelia. Mesothelia (mesodermal epithelia) are examples of type II epithelia.

Download figure to PowerPoint

It is tempting to speculate that in absence of any EMT mechanisms, animal evolution would have emphasized the development of 3D epithelial structures through fractal-like growth, generating a variety of sac-like structures and complex networks of cavities, with a limited number of cell types. The resulting organisms would probably look like the Cnidarians. Instead, with the addition of the mesoderm, a cell layer is formed that is able to oscillate between epithelial and mesenchymal states. It is conceivable that the organogenetic potential of the mesoderm can be linked to this unique ability, which would account for the diverse morphological characteristics of the tripoblastic metazoans, as proposed in Figure 4. At this point it should be noted that EMT events are placed at the origin of important animal taxa, constituting true synapomorphies (mesoderm for bilateria, neural crest for vertebrates, and trophoblast for placental mammals).

thumbnail image

Figure 4. An illustration of various examples of pluricellular organization. The final shape and complexity of the structure is a consequence of the mechanisms used by the tissues to grow and develop (mechanisms for each case are indicated in red). Double-headed green arrows indicate the places of epithelial–mesenchyme interactions and subsequent EMTs. 1 = blastocoel, 2 = archenteron (primitive gastrovascular/endodermal cavity), 3 = primary mesoderm (blue mesenchymal cells), and 4 = coelom.

Download figure to PowerPoint

When did EMTs appear in the phylogeny? We could hypothesize that the origin of the EMT is linked to the origin of the mesoderm, thus arising in the earliest triplobastic metazoans. It is also possible that the mesoderm used a much older cellular mechanism, which perhaps originated before the Metazoans.

Dictyostelium discoideum is an unicellular organism belonging to the group of myxoamoebas. When subjected to starvation, Dictyostelium forms pluricellular aggregates of about 105 cells, which move as an “organism” that gives off specialized cells and spores. The transition from the unicellular to the pluricellular phase is regulated by a protein, Aardvark, which is expressed in the intercellular junctions that appear during cell aggregation. Interestingly, Aardvark is homologous to β-catenin, a protein that is associated to junctional cadherins, and can also act as a transcription factor when associated to T-cell factor (TCF) (Grimson et al., 2000). Like β-catenin, Aardvark can translocate to the nucleus and act as a transcription factor when it is phosphorylated by GSKA, the Dictyostelium homologous of glyceraldehyde-synthase kinase-3, a negative regulator of β-catenin in Metazoans. A β-catenin homolog has also been found in Hydra, and the origin of this protein has been related to the origin of the epithelial organization in Metazoans (Hobmayer et al., 1996).

We can imagine that the transition between epithelial and mesenchymal states, which generates, for example, the mesogleal amoebocytes of Cnidarians, could be regulated by a similar molecular system. Thus, the origin of the mesoderm could be related to the ability (acquired by embryonic mesenchymal cells) to reverse this system in order to regain a transitional epithelial state which subsequently can give off new mesenchyme.

EMTs in Adult Organisms

  1. Top of page
  2. Abstract
  3. Making of a Pluricellular Organism
  4. Epithelium and Mesenchyme in Embryonic Development
  5. Epithelial–Mesenchymal Transitions in Development
  6. Vertebrate Development Shows a Pattern of Recurrent EMTs
  7. Molecular Regulation of EMTs
  8. Morphogenetic Role of EMTs, and Phylogenetic Implications
  9. EMTs in Adult Organisms
  10. CONCLUSIONS AND PERSPECTIVES
  11. Acknowledgements
  12. LITERATURE CITED

The switch that regulates the transition from epithelial to mesenchymal phenotype must be strictly controlled, especially when embryonic development has been completed, in order to avoid chaotic cell behaviour in the adult organism. It is well known that loss of epithelial morphology and the acquisition of mesenchymal characteristics are involved in carcinoma progression, and correlate with metastatic potential (Birchmeier et al., 1996). Some features of the invasive behaviour of tumoral cells can be explained by the down-regulation of the molecular mechanisms that maintain the epithelial state after the transient instabilities required during the embryonic development. In this context, it is interesting to consider, from a clinical point of view, the stabilizing mechanisms that regulate the shift from mesenchymal to epithelial phenotype and may act to counterbalance the invasive properties of the tumoral cells. The concept of “morphostats” (Potter, 2001) as hypothetical morphogens that keep cells at the epithelial state may be relevant in this context (see above). An example of a stabilizing mechanism is the impairment of the capacity to form metastases in breast cancer cells after transfection with E-cadherin cDNA (Mbalaviele et al., 1996), a key molecule in the MET. An opposite and striking example comes from studies on the expression of genes in the Snail family. When normal epithelial cells are forced to express Snail, they adopt a fibroblastoid phenotype and acquire tumorigenic and invasive properties. Expression of Snail in mouse and human carcinoma cell lines and tumours shows a strong correlation with their invasive and metastatic properties (Cano et al., 2000; Batlle et al., 2000). As we have indicated, the Snail/Slug genes are candidates to be the main regulators of different embryonic EMTs, and they are essential in the differentiation of the mesoderm. The relationship between the invasiveness of tumoral cells and the expression of these genes also seems to support our point of view.

If metastasis is linked to the activation of embryonic mechanisms of the EMT, it should be possible to explain why cancer occurs only in Metazoans. Plants sometimes show uncontrolled cell growth that forms tumour-like masses, but the lack of a metastatic process prevents these pathologies from becoming lethal (Gaspar, 1998). Thus, cancer-like pathology in plants has different features from that in animals. Our proposal that the process of EMT arose in the Metazoans accounts for this observation.

On the other hand, a certain degree of controlled destabilization of the tissues can be useful for reparative or regenerative processes. Angiogenesis, the formation of new vessels from preexisting ones, involves some steps in common with embryonic EMT and cancer. For example, activated endothelial cells secrete proteolytic enzymes that degrade the basement lamina and ECM, and then proliferate and migrate into adjacent tissues where they regain the quiescent (stable) endothelial phenotype. It has recently been reported that angiopoietin-2 plays a key role in the destabilization of the vessels, which is a prerequisite for angiogenesis (reviewed in Yancopoulos et al., 2000). It is uncertain whether some steps of the angiogenic process are regulated by the same molecular systems that control the embryonic EMTs, but it seems significant that many secreted molecules involved in the EMTs (e.g., the basic FGF-2 and the HGF/scatter factor, Rosen et al., 1997) are also potent angiogenic factors. It is hoped that future findings of other common molecular regulatory mechanisms in the EMT and pathological/reparative processes will support our model.

CONCLUSIONS AND PERSPECTIVES

  1. Top of page
  2. Abstract
  3. Making of a Pluricellular Organism
  4. Epithelium and Mesenchyme in Embryonic Development
  5. Epithelial–Mesenchymal Transitions in Development
  6. Vertebrate Development Shows a Pattern of Recurrent EMTs
  7. Molecular Regulation of EMTs
  8. Morphogenetic Role of EMTs, and Phylogenetic Implications
  9. EMTs in Adult Organisms
  10. CONCLUSIONS AND PERSPECTIVES
  11. Acknowledgements
  12. LITERATURE CITED

The ectoderm and endoderm are the primordial epithelia that constitute the original Metazoan body. Their growth strategies, which basically include differential growth and folding, are quite similar to those found in fungi and plants. The mesoderm may have arisen by means of ancestral mechanisms regulating cell adhesion and detachment. The ability to go through cycles of epithelial and mesenchymal states provided the mesoderm with a larger plasticity, more potential for innovation, and a better control of the 3D body organization. Thus, mesodermal cells were able to diversify and contribute to the development of somatic (skeleton, muscle, and connective tissue) and visceral (circulation, blood, excretion, and reproduction) systems. Major differences between the development and organization of fungi, plants, and animals may to be related to the inception of EMT mechanisms. On the other hand, the recognition of EMT features in pathological processes, such as tumoral metastasis, can provide us with new approaches to understanding and treating cancer.

LITERATURE CITED

  1. Top of page
  2. Abstract
  3. Making of a Pluricellular Organism
  4. Epithelium and Mesenchyme in Embryonic Development
  5. Epithelial–Mesenchymal Transitions in Development
  6. Vertebrate Development Shows a Pattern of Recurrent EMTs
  7. Molecular Regulation of EMTs
  8. Morphogenetic Role of EMTs, and Phylogenetic Implications
  9. EMTs in Adult Organisms
  10. CONCLUSIONS AND PERSPECTIVES
  11. Acknowledgements
  12. LITERATURE CITED
  • Alberga A, Boulay JL, Kempe E, Dennefeld C, Haenlin M. 1991. The Snail gene required for mesoderm formation in Drosophila is expressed dynamically in derivatives of all three germ layers. Development 111: 983992.
  • Anderson DT. 1969. On the embryology of the cirripede crustaceans Tetraclita rosea (Krauss), Tetraclita purpurascens (Wood), Chthamalus antennatus (Darwin) and Chamaesipho columna (Spengler) and some considerations of crustaceans phylogenetic relationships. Phil Trans R Soc B 256: 183235.
  • Arciniegas E, Graterol A, Sanchez F. 2000. Characterization of nonmuscle cells present in the intima of normal adult bovine pulmonary artery. Anat Rec 258: 262268.
  • Bard J. 1990. Morphogenesis. The cellular and molecular processes of developmental anatomy. Cambridge: Cambridge University Press. 303 p.
  • Batlle E, Sancho E, Franci C, Dominguez D, Monfar M, Baulida J, Garcia de Herreros A. 2000. The transcription factor snail is a repressor of E-cadherin gene expression in epithelial tumour cells. Nat Cell Biol 2: 8489.
  • Bellairs R, Osmond M. 1998 The atlas of chick development. San Diego: Academic Press. 323 p.
  • Beppu H, Kawabata M, Hamamoto T, Chytil A, Minowa O, Noda T, Miyazono K. 2000. BMP type II receptor is required for gastrulation and early development of mouse embryos. Dev Biol 221: 249258.
  • Birchmeier C, Birchmeier W, Brand-Saberi B. 1996. Epithelial–mesenchymal transitions in cancer progression. Acta Anat 156: 217226.
  • Brusca RC, Brusca GJ. 1990. Invertebrates. Sunderland: Sinauer. 922 p.
  • Cano A, Pérez-Moreno MA, Rodrigo I, Locascio A, Blanco MJ, Del Barrio MG, Portillo F, Nieto MA. 2000. The transcription factor Snail controls epithelial–mesenchymal transitions by repressing E-cadherin expression. Nat Cell Biol 2: 7683.
  • Carmeliet P. 2001. Cardiovascular biology. Creating unique blood vessels. Nature 412: 868869.
  • Carmona R, González-Iriarte M, Macías D, Pérez-Pomares JM, García-Garrido L, MuÑoz-Chápuli R. 2000. Immunolocalization of the transcription factor Slug in the developing avian heart. Anat Embryol 201: 103109.
  • Cheney CM, Lash JW. 1984. An increase in cell-cell adhesion in the chick segmental plate results in a meristic pattern. J Embryol Exp Morphol 79: 110.
  • Cui XM, Shuler CF. 2000. The TGF-beta type III receptor is localized to the medial edge epithelium during palatal fusion. Int J Dev Biol 44: 397402.
  • Davies JA. 1996. Mesenchyme to epithelium transition during development of the mammalian kidney tubule. Acta Anat 156: 187201.
  • DeRuiter MC, Poelmann RE, Van Munsteren JC, Mironov V, Markwald RR, Gittenberger-de Groot AC. 1997. Embryonic endothelial cells transdifferentiate into mesenchymal cells expressing smooth muscle actins in vivo and in vitro. Circ Res 80: 444451.
  • Dettman RW, Denetclaw W, Ordahl CP, Bristow J. 1998. Common epicardial origin of coronary vascular smooth muscle, perivascular fibroblasts, and intermyocardial fibroblasts in the avian heart. Dev Biol 193: 169181.
  • Farquhar MG, Palade GE. 1963. Junctional complexes in various epithelia. J Cell Biol 17: 375412.
  • Fink RD, McClay DR. 1985. Three cell recognition changes accompany the ingression of sea urchin primary mesenchyme cells. Dev Biol 107: 6674.
  • Gaspar T. 1998. Plants can get cancer. Plant Physiol Biochem 36: 203204.
  • Gerhart J, Kirschner M. 1997. Cells, embryos and evolution. Malden: Blackwell. 642 p.
  • Gilbert SF. 1995. Developmental biology. Sunderland, MA: Sinauer Associates. 894 p.
  • Greenburg G, Hay ED. 1989. Epithelia suspended in collagen gels can lose polarity and express characteristics of migrating mesenchymal cells. J Cell Biol 95: 333339.
  • Grimson MJ, Coates JC, Reynolds JP, Shipman M, Blanton RL, Harwood AJ. 2000. Adherens junctions and beta-catenin-mediated cell signalling in a non-Metazoan organism. Nature 408: 727731.
  • Gruenwald P. 1942. Common traits in development and structure of organs originating from the coelomic wall. J Morphol 70: 353387.
  • Hamburger V, Hamilton HL. 1951. A series of normal stages in the development of the chick embryo. J Morphol 88: 4992.
  • Hay ED. 1968. Organization and fine structure of epithelium and mesenchyme in the developing chick embryo. In: FleischmajerR, BillinghamRE, editors. Epithelial–mesenchymal interactions. Baltimore: Williams and Wilkins. p 3155.
  • Hay ED. 1990. Epithelial–mesenchymal transitions. Semin Dev Biol 1: 347–356.
  • Hay ED. 1996. An overview of epithelio–mesenchymal transformation. Acta Anat 154: 820.
  • Haynes BF, Martin ME, Kay HH, Kurtzberg J. 1988. Early events in human T cell ontogeny. Phenotypic characterization and immunohistologic localization of T cell precursors in early human fetal tissues. J Exp Med 168: 10611080.
  • Hobmayer E, Hatta M, Fischer R, Fujisawa T, Holstein TW, Sugiyama T. 1996. Identification of a Hydra homolog of the beta-catenin/plakoglobin/armadillo gene family. Gene 172: 155159.
  • Jamieson BGM. 1992. Oligochaeta. In: HarrisonFW, editor. Microscopic anatomy of invertebrates. Vol. VII. New York: Wiley-Liss. p 217322.
  • Kaartinen V, Cui XM, Heisterkamp N, Groffen J, Shuler CF. 1997. Transforming growth factor-beta3 regulates transdifferentiation of medial edge epithelium during palatal fusion and associated degradation of the basement membrane. Dev Dyn 209: 255260.
  • Karl J, Capel B. 1998. Sertoli cells of the mouse testis originate from the coelomic epithelium. Dev Biol 203: 323333.
  • Ledger PW. 1975. Septate junctions in the calcareous sponge Sycon ciliatum. Tissue Cell 7: 1318.
  • LeDouarin N. 1975. An experimental analysis of liver development. Med Biol 53: 427455.
  • Lobo MV, Santander RG, Cuadrado GM, Alonso FJ. 1999. Cytochemical localization of calcium in prefusion myoblasts from the chick embryo myotome. Histochem J 31: 347355.
  • Lough J, Sugi Y. 2000. Endoderm and heart development. Dev Dyn 217: 327342.
  • Mano R. 1964. The coelomic corpuscles and their origin in the freshwater bryozoan, Lophopodella carteri. Sci Rep Tokyo Kyoiku Dai 11: 211235.
  • Margulis L. 1970. Origin of eukariotic cells. New Haven: Yale University Press.
  • Margulis L. 1975. Symbiotic theory of the origin of eukaryotic organelles; criteria for proof. Symp Soc Exp Biol 29: 2138.
  • Markwald RR, Fitzharris TP, Smith WN. 1975. Structural analysis of endocardial cytodifferentiation. Dev Biol 42: 160180.
  • Markwald RR, Fitzharris TP, Manasek FJ. 1977. Structural development of endocardial cushions. Am J Anat 148: 85120.
  • Markwald RR, Eisenberg C, Eisenberg L, Trusk T, Sugi Y. 1996. Epithelial–mesenchymal transformations in early avian heart development. Acta Anat 156: 173186.
  • Mbalaviele G, Dunstan CR, Sasaki A, Williams PJ, Mundy GR, Yoneda T. 1996. E-cadherin expression in human breast cancer cells suppresses the development of osteolytic bone metastases in an experimental metastasis model. Cancer Res 56: 40634070.
  • Mjaatvedt CH, Yamamura H, Wessels A, Ramsdell A, Turner D, Markwald RR. 1999. Mechanisms of segmentation, septation, and remodelling of the tubular heart: endocardial cushion fate and cardiac looping. In: RosenthalN, HarveyR, editors. Heart development. New York: Academic Press. p 159177.
  • Moore AW, Schedl A, McInnes L, Doyle M, Hecksher-Sorensen J, Hastie ND. 1998. YAC transgenic analysis reveals Wilms' tumour 1 gene activity in the proliferating coelomic epithelium, developing diaphragm and limb. Mech Dev 79: 169184.
  • Mukai H, Terakado K, Reed CG. 1997. Bryozoa. In: HarrisonFW, editor. Microscopic anatomy of invertebrates. Vol. XIII. New York: Wiley-Liss. p 45206.
  • Muller WE, Kruse M, Blumbach B, Skorokhod A, Muller IM. 1999. Gene structure and function of tyrosine kinases in the marine sponge Geodia cydonium: autapomorphic characters in Metazoa. Gene 238: 179193.
  • Muñoz-Chápuli R, Pérez-Pomares JM, Macías D, García-Garrido L, Carmona R, González M. 1999. Differentiation of hemangioblasts from embryonic mesothelial cells? A model on the origin of the vertebrate cardiovascular system. Differentiation 64: 133141.
  • Muñoz-Chápuli R, Carmona R, González-Iriarte M, Pérez-Pomares JM, Macías D, Atencia G, Aranda MJ. 2001. Origin of endothelial cells from mesothelial-derived mesenchymal cells in the liver of avian embryos. Int J Dev Biol 45: S155S156.
  • Nakajima Y, Yamagishi, T, Hokari S, Nakamura H. 2000. Mechanisms involved in valvuloseptal endocardial cushion formation in early cardiogenesis: roles of transforming growth factor (TGF)-beta and bone morphogenetic protein (BMP). Anat Rec 258: 119127.
  • Nielsen C. 1995. Animal evolution. Interrelationships of the living phyla. Oxford: Oxford University Press. 467 p.
  • Nieto MA, Sargent MG, Wilkinson DG, Cooke J. 1994. Control of cell behavior by Slug, a zinc finger gene. Science 264: 835839.
  • Olah I, Medgyes J, Glick B. 1988. Origin of aortic cell clusters in the chicken embryo. Anat Rec 222: 6068.
  • Ordahl CP. 1993. Myogenic lineages within the developing somite. In: BernfieldM, editor. Molecular basis of morphogenesis. New York: Wiley-Liss.
  • Peng I, Dennis JE, Rodríguez-Boulan E, Fischman D. 1990. Polarized release of enveloped viruses in the embryonic chicken heart: demonstration of epithelial polarity in the presumptive myocardium. Dev Biol 141: 164172.
  • Pérez-Pomares JM, Macías D, García-Garrido L, Muñoz-Chápuli, R. 1997. Contribution of the primitive epicardium to the subepicardial mesenchyme in hamster and chick embryos. Dev Dyn 210: 96105.
  • Pérez-Pomares JM, Macías D, García-Garrido L, Muñoz-Chápuli R. 1999. Immunohistochemical evidence for a mesothelial contribution to the ventral wall of the avian aorta. Histochem J 31: 771779.
  • Potter JD. 2001. Morphostats: a missing concept in cancer biology. Cancer Epidemiol Biomark Prevent 10: 161170.
  • Raven CP. 1966. Morphogenesis: the analysis of molluscan development, 2nd ed. Oxford: Pergamon Press. 365 p.
  • Romano LA, Runyan RB. 1999. Slug is a mediator of epithelial-mesenchymal cell transformation in the developing chick heart. Dev Biol 212: 243254.
  • Romano LA, Runyan RB. 2000. Slug is an essential target of TGFbeta2 signaling in the developing chicken heart. Dev Biol 223: 91102.
  • Rosen EM, Lamszus K, Laterra J, Polverini PJ, Rubin JS, Goldberg ID. 1997. HGF/SF in angiogenesis. Ciba Found Symp 212: 215229.
  • Sefton M, Sánchez S, Nieto MA. 1998. Conserved and divergent roles for members of the Snail family of transcription factors in the chick and mouse embryo. Development 125: 31113121.
  • Sela-Donenfeld D, Kalcheim C. 1999. Regulation of the onset of neural crest migration by coordinated activity of BMP4 and Noggin in the dorsal neural tube. Development 126: 47494762.
  • Shannon CE, Weaver W. 1963. The mathematical theory of communication. Urbana: University of Illinois Press.
  • Staehelin LA, Hull BE. 1978. Junctions between living cells. Sci Am 238: 141152.
  • Sugi Y, Markwald RR. 1996. Formation and early morphogenesis of endothelial precursors cells and the role of endoderm. Dev Biol 175: 6683.
  • Trelstad RL, Hayashi A, Hayashi K, Donahoe PK. 1982. The epithelial-mesenchymal interface of the male rat Müllerian duct: loss of basement membrane integrity and ductal regression. Dev Biol 131: 455474.
  • Valentine JW. 1978. The evolution of multicellular plants and animals. Sci Am 239: 140158.
  • Viebahn C. 1995. Epithelio-mesenchymal transformation during formation of the mesoderm in the mammalian embryo. Acta Anat 154: 7997.
  • Viebahn C, Mayer B, Hrabé de Angelis M. 1995. Signs of the principle body axes prior to primitive streak formation in the rabbit embryo. Anat Embryol 192: 159169.
  • Weaver M, Yingling JM, Dunn NR, Bellusci S, Hogan BL. 1999. BMP signaling regulates proximal-distal differentiation of endoderm in mouse lung development. Development 126: 40054015.
  • Yancopoulos GD, Davis S, Gale NW, Rudge JS, Wiegand SJ, Holash J. 2000. Vascular-specific growth factors and blood vessel formation. Nature 407: 242248.