Induction of an epithelial-mesenchymal transition by an in vivo adheron-like complex

The embryonic vertebrate heart consists of two epithelia: the myocardium and endothelium, separated by the myocardial basement membrane (MBM). The myocardium has been shown to induce endothelial transformation into prevalvular mesenchyme in a temporally and site restricted manner. Previously, we hypothesized that the myocardial-endothelial interaction is mediated in vivo by aggregates of 30-nm particles in the MBM which can be removed by EDTA extraction. These MBM extracts contain fibronectin and other lower Mr proteins and can initiate an epithelial-mesenchymal transition in the AV (atrioventricular canal) endothelium of embryonic chick heart in collagen gel culture. These and other data suggested that the 30-nm multicomponent particles are similar, structurally and compositionally, to multimolecular complexes, termed adherons, secreted by L6 muscle cells in culture. The purpose of this study was to (1) test whether the removal of the 30-nm particles from MBM extracts of embryonic chick hearts would remove the in vitro biological activity and (2) determine if the fractionated MBM extracts can cause AV endothelial cells to follow the same differentiation pathway observed in vivo by monitoring immunohistochemically the cell surface expression of N-CAM. Results showed that centrifugation of extract at 100,000g for 1 hr produced a supernatant fraction that was unable to initiate mesenchyme formation from AV endothelium. However, the resuspended pellet fraction did initiate differentiation of endothelium into mesenchyme. Conditioned medium from L6 skeletal muscle cultures could not substitute for the EDTA extract of embryonic heart. Endothelial cells undergoing the transition to form mesenchyme, both in vivo and in vitro, showed a concomitant decrease in N-CAM staining. This suggested that the pellet-induced formation of migrating cells in the collagen gels is not the result a novel in vitro phenomenon.     

Myocardial specificity for initiating endothelial-mesenchymal cell transition in embryonic chick heart correlates with a particulate distribution of fibronectin

The early chick heart tube consists of myocardium and endothelium separated by a myocardially derived basement membrane (MBM). As development proceeds, the endothelium undergoes a transition into mesenchyme in a regionally specific manner; only the atrioventricular (AV) and outflow tract, but not the ventricular endothelium, is transformed into mesenchyme, the progenitor of heart septa and valves. Recent experiments have shown that an EDTA extract of MBM can initiate AV endothelium to form mesenchyme in an in vitro collagen gel culture system. Two-dimensional gel electrophoresis of AV region EDTA extracts showed potentially three isoelectric forms of fibronectin (Fn), while extracts from ventricle contained only two forms. The purpose of the present study was to further investigate the significance of these regional differences by testing of specific myocardial regions (AV vs ventricle) for their ability to induce endothelium to form mesenchyme in vitro, and to immunohistochemically determine if a regionally specific distribution of Fn exists in the MBM that can be correlated with previous electrophoretic data. Embryonic heart regions cultured on three-dimensional collagen gels showed that AV endothelium could only form mesenchyme if cocultured with AV myocardium. Coculture with ventricular myocardial explants did not initiate differentiation of AV endothelium. In contrast, ventricular endothelial cells did not form mesenchyme when cocultured with AV or ventricle myocardium. Immunohistochemical localization of Fn revealed three distinct morphological patterns of distribution in the AV-MBM, i.e., an intense lamina densa staining, diffuse staining in fibrils, and as particles. The Fn localized in particles (0.1 to 0.5 micron in diameter) appeared as a gradient of decreasing concentration extending from the myocardium toward the endothelium. In contrast, no particulate Fn staining was observed in the ventricular region. EDTA extraction selectively depleted the particulate form of Fn. Previous work has shown that this extract, which contains several lower Mr proteins in addition to Fn, is biologically active in initiating mesenchyme formation from AV endothelium in vitro. These results show that a regionally specific interaction of the myocardium with the endothelium is required to initiate the formation of prevalvular mesenchyme. This interaction may be mediated by a multicomponent complex involving Fn and other proteins which appear as a regionally distinct particulate only in areas of endothelial differentiation.     

Extracellular matrix from embryonic myocardium elicits an early morphogenetic event in cardiac endothelial differentiation

A critical step in early cardiac morphogenesis can be faithfully duplicated in culture using a hydrated collagen substratum, and thereby serves as a useful model system for studying the molecular mechanisms of cell differentiation. Results from previous work suggested that the myocardium in the atrioventricular canal (AV) region of the developing chick heart secretes extracellular proteins into its associated basement membrane, which may function to promote an epithelial-mesenchymal transition of endothelium to form prevalvular fibroblasts (E. L. Krug, R. B. Runyan, and R. R. Markwald, 1985, Dev. Biol. 112, 414-426; C. H. Mjaatvedt, R. C. Lepera, and R. R. Markwald, 1987, Dev. Biol., in press). In the present study we show that an EDTA-soluble extract of embryonic chick hearts can substitute for the presence of myocardium, the presumptive stimulator tissue, in initiating mesenchyme formation from AV endothelium in culture. Ventricular endothelium was unresponsive to this material in keeping with observed in situ behavior. AV endothelial cells did not survive beyond 4-5 days when cultured in the absence of either the EDTA-soluble heart extract, myocardial conditioned medium, or the myocardium itself. Antibody prepared against a particulate fraction of the EDTA-solubilized heart extract immunohistochemically localized this material to the myocardial basement membrane. In addition, conditioned medium from embryonic myocardial cultures effectively induced mesenchyme formation. Neither a variety of growth factors nor a sarcoma basement membrane preparation were effective in promoting mesenchyme formation indicating a selectivity of the responding embryonic AV endothelial cells to myocardial basement membrane. These observations reflect a truly inductive phenomenon as there was an absolute dependence on the presence of the stimulating substance/tissue and retention, in culture, of both the temporal and regional characteristics observed in situ. This is in contrast to the results of others investigating the cytodifferentiation of committed cells whose phenotypic expression can be either accelerated or diminished but not obligatorily regulated by a specific agent, thus making the interpretation of data difficult, if not irrelevant, to the study of differentiation. The results of this study provide direct experimental support for the hypothesis that extracellular matrix can indeed serve as a direct stimulator or "secondary inducer" of cytodifferentiation.     

Epithelial-mesenchymal cell transformation in the embryonic heart can be mediated, in part, by transforming growth factor beta

Progenitor cells of the valves and membranous septa of the vertebrate heart are formed by transformation of a specific population of endothelial cells into mesenchyme. Previous studies have shown that this epithelial-mesenchymal cell transformation is mediated by a signal produced by the myocardium of the atrioventricular (AV) canal and transferred across the extracellular matrix. Data are presented here that transforming growth factor beta (TGF beta 1 or TGF beta 2), in combination with an explant of ventricular myocardium, will produce an epithelial-mesenchymal transformation by cultured AV canal endothelial cells in vitro. Alone, neither component is capable of producing this effect. The factor provided by the ventricular explant cannot be substituted by either epidermal growth factor or basic fibroblast growth factor. Further experiments show that an antibody that blocks TGF beta activity is effective in preventing the epithelial-mesenchymal cell transformation normally produced by AV canal myocardium. Control antibodies are without effect. By immunological criteria, a member of the TGF beta family of molecules can be demonstrated in the chicken embryo and heart at the time overt valvular formation begins. Together, these data show that TGF beta 1 can produce mesenchymal cell formation in vitro and provide evidence that a member of the TGF beta family is present and plays a role in the process of epithelial-mesenchymal cell transformation in the embryonic heart.     

Bone morphogenetic protein-2 can mediate myocardial regulation of atrioventricular cushion mesenchymal cell formation in mice

Transformation of endocardial endothelial cells into invasive mesenchyme is a critical antecedent of cardiac cushion tissue formation. The message for bone morphogenetic protein (BMP)-2 is known to be expressed in myocardial cells in a manner consistent with the segmental pattern of cushion formation [Development 109(1990) 833]. In the present work, we localized BMP-2 protein in atrioventricular (AV) myocardium in mice at embryonic day (ED) 8.5 (12 somite stage) before the onset of AV mesenchymal cell formation at ED 9.5. BMP-2 protein expression was absent from ventricular myocardium throughout the stages examined. After cellularization of the AV cushion at ED 10.5, myocardial BMP-2 protein expression was diminished in AV myocardium, whereas cushion mesenchymal cells started expressing BMP protein. Expression of BMP-2 in cushion mesenchyme persisted during later stages of development, ED 13.5-16, during valuvulogenesis. Intense expression of BMP-2 persisted in the valve tissue in adult mice. Based on the expression pattern, we performed a series of experiments to test the hypothesis that BMP-2 mediates myocardial regulation of cardiac cushion tissue formation in mice. When BMP-2 protein was added to the 16-18 somite stage (ED 9.25) AV endocardial endothelium in culture, cushion mesenchymal cells were formed in the absence of AV myocardium, which invaded into collagen gels and expressed the mesenchymal marker, smooth muscle (SM) alpha-actin; whereas the endothelial marker, PECAM-1, was lost from the invaded cells. In contrast, when noggin, a specific antagonist to BMPs, was applied together with BMP-2 to the culture medium, AV endothelial cells remained as an epithelial monolayer with little expression of SM alpha-actin, and expression of PECAM-1 was retained in the endocardial cells. When noggin was added to AV endothelial cells cocultured with associated myocardium, it blocked endothelial transformation to mesenchyme. AV endothelium treated with BMP-2 expressed elevated levels of TGFbeta-2 in the absence of myocardium, as observed in the endothelium cocultured with myocardium. BMP-2-supported elevation of TGFbeta-2 expression in endocardial cells was abolished by noggin treatment. These data indicated that BMP signaling is required in and BMP-2 is sufficient for myocardial segmental regulation of AV endocardial cushion mesenchymal cell formation in mice.     

Initial expression of type I procollagen in chick cardiac mesenchyme is dependent upon myocardial stimulation

Formation of the atrioventricular (AV) mesenchyme is a critical step in early heart development. Endothelial cells are activated and transformed into a mesenchymal population that invades the cell-free myocardial basement membrane. This process can be duplicated in collagen gel culture, where it has been established that myocardium or its secretory products activate the endothelium. The purpose of the present study was to determine when these activated endothelial and/or mesenchymal cells start producing type I collagen in situ. These results were compared to those obtained from a culture model of mesenchyme formation. The production of type I collagen was monitored using a monoclonal antibody (M38) that recognizes the carboxy-terminal propeptide of human type I procollagen. The initial expression of the latter within activated AV endothelial and mesenchymal cells in ovo was 48 hr following activation. Prior to this time, only the myocardium was reactive with M38. AV explants of early hearts on collagen gels revealed staining of activated endothelial and mesenchymal cells with M38 after 48 hr in coculture with myocardial tissue. Explants that were prevented from activating (myocardium removed) never expressed the M38 antigen. Similarly, AV endothelial monolayers grown in the presence of myocardial conditioned medium activated and expressed type I collagen after 48 hr in culture, whereas those grown in standard medium did not. These results establish the initial expression of type I collagen within activated AV endothelium and mesenchyme. In addition, the data suggest that the expression of type I collagen within the AV mesenchyme may be dependent on extrinsic influences that induce the AV endothelium to transform into mesenchyme.     

Identification of four classes of cell surface antigens appearing at gastrulation in sea urchin embryos

An antiserum to isolated membranes of gastrula-stage embryos of the sea urchin Lytechinus variegatus was characterized by absorption and cell agglutination specificities. The antiserum was found to recognize four distinct classes of antigens on the embryonic cell surface: (1) an early embryonic class or “maternal” class present from the earliest stages of development, (2) an embryonic class of antigens which appeared on all cells beginning at gastrulation, (3) a class of antigens present on ectoderm cells, and (4) a class of antigens present on endoderm cells. All four classes of antigens were shown indirectly to be synthesized on embryonic mRNA since a hybrid embryo of the cross Tripneustes ♀ × Lytechinus ♂ expressed all four classes of Lytechinus-specific antigens beginning at gastrulation. Each class was Lytechinus specific in that hybrid cells were agglutinated if beyond the beginning of gastrulation, while normal Tripneustes ♀ × Tripneustes ♂ cells were not agglutinated.     

Spatially and temporally regulated expression of N-acetylglucosamine-6-O-sulfotransferase during mouse embryogenesis.

GlcNAc-6-O-sulfotransferase is involved in formation of 6-sulfo-N -acetyllactosamine-containing structures such as 6-sulfo sialyl Lewis x. We investigated the mode of expression of GlcNAc-6-O-sulfotransferase during postimplantation embryogenesis in the mouse by in situ hybridization. Sulfotransferase mRNA was not detected on embryonic day (E) 6.5, while on E7.5 it was detected in the mesoderm, ectoderm, and ectoplacental cone. On E10.5, the sulfotransferase signals were mainly observed in the nervous tissue. On E12.5 and 13.5, various tissues in the process of differentiation expressed this mRNA. Several epithelial and mesenchymal tissues undergoing epithelial-mesenchymal interactions strongly expressed the mRNA. For example, in the developing tooth strong sulfotransferase mRNA expression was found only in the condensing mesenchyme on E13.5. On E13.5 and 15.5, the sites showing intense expression of the sulfotransferase again became restricted. In the brain, sulfotransferase mRNA was frequently found as discrete signals in narrow regions. These results suggest that 6-sulfo-N-acetyllactosamine structures have important roles in development. On E13.5 and 15.5, G152 (6-sulfo sialyl Lewis x antigen) was expressed in the neocortex, and AG223 (6-sulfo Lewis x antigen) in the thalamus and neocortex where the sulfotransferase signal was detected. However, in other organs, expression of these antigens did not correlate with the sulfotransferase mRNA, implicating complex nature of regulation of expression of the fucosyl 6-sulfo antigens.     

Changes in the distribution of intermediate-filament types in Japanese quail embryos during morphogenesis

We examined the distribution of intermediate filaments in early quail embryos in order to determine whether these cytoskeletal proteins play a role in the epithelial-mesenchymal transitions that commonly occur during embryogenesis, e.g., the separation of neural-crest cells from the neural epithelium. The distribution of cytokeratins, vimentin, and desmin was examined in frozen sections of quail embryos at stages during which dramatic reorganizations of tissues take place. All embryonic tissues were found to contain either vimentin or cytokeratins, but the distribution of these cytoskeletal proteins was characteristic neither of the cellular organization (e.g., epithelium vs. mesenchyme) nor of the germ-layer derivation of the tissues. Cytokeratin monoclonal antibodies stained most embryonic epithelia (defined here as being sheet-like tissue with an underlying basement membrane), including epidermis and extraembryonic membranes derived in part from the ectoderm, splanchnopleure and kidney tubules derived from mesoderm, and endoderm. Cytokeratin antibodies did not stain some epithelia, including the neural tube, neural plate, and dermatome/myotome. Whereas the cytokeratin antibodies exclusively stained epithelia, the vimentin antibodies labeled both epithelial (the neural tube, dermatome/myotome, and somatic and splanchnic mesoderm) and mesenchymal tissues (the sclerotome and neural-crest cells), regardless of their germ-layer derivation. In early embryos, antibodies against desmin only stained the myotome and, in 4-day embryos, the heart and mesenchyme around the pharynx. As the distribution of intermediate-filament types did not reflect tissue organization or germ-layer derivation, we propose that the distribution of intermediate filaments in early avian embryos reflects the motile capacity of an embryonic cell and/or the presence of specialized cell junctions, i.e., desmosomes.     

Evidence that plasminogen activator is not involved in basement membrane penetration at avian gastrulation

We have used the method of radial caseinolysis in agarose to assay for the presence of plasminogen activator in the primitive streak cells of the early chick embryo. These cells are undergoing an epithelial-mesenchymal transformation in the process of the differentiation of mesoderm cells from the ectoderm. In the primitive streak, the epithelial cells, after penetrating the basement membrane, ingress and invade the underlying tissue space as fibroblast-like mesenchyme cells. In contrast to some other early embryonic invasionary and migratory events, we find no evidence for the participation of plasminogen activator in this process.     

Epithelial-mesenchymal interactions in the development of chick facial primordia and the target of retinoid action

The development of the chick face involves outgrowth of buds of tissue, accompanied by the differentiation of cartilage and bone in spatially defined patterns. To investigate the role of epithelial-mesenchymal interactions in facial morphogenesis, small fragments of facial tissue have been grafted to host chick wing buds to continue their development in isolation. Fragments of the frontonasal mass give rise to typical upper-beak-like structures: a long central rod of cartilage, the prenasal cartilage and an egg tooth. Meckel's cartilage, characteristic of the lower beak, develops from fragments of the mandible. Removal of the ectoderm prior to grafting leads to truncated development. In fragments of frontonasal mass mesenchyme only a small spur of cartilage differentiates and there is no outgrowth. The mandible is less affected; a rod of cartilage still forms but the amount of outgrowth is reduced. Retinoid treatment of chick embryos specifically affects the development of the upper beak and outgrowth and cartilage differentiation in the frontonasal mass are inhibited. The mandibles, however, are unaffected and develop normally. In order to investigate whether the epithelium or the mesenchyme of the frontonasal mass is the target of retinoid action, recombinations of retinoid-treated and untreated facial tissue have been grafted to host wing buds. Recombinations of retinoid-treated frontonasal mass ectoderm with untreated mesenchyme develop normally whereas recombinations of untreated ectoderm with retinoid-treated mesenchyme lead to truncations. The amount of outgrowth in fragments of mandibular tissue is slightly reduced when either the ectoderm or the mesenchyme has been treated with retinoids. These recombination experiments demonstrate that the mesenchyme of the frontonasal mass is the target of retinoid action. This suggests that retinoids interfere with the reciprocal epithelial-mesenchymal interactions necessary for outgrowth and normal upper beak development.     

Serum response factor, its cofactors, and epithelial-mesenchymal signaling in urinary bladder smooth muscle formation

Little is known about the mechanism of bladder smooth muscle differentiation. We hypothesize that epithelial-mesenchymal signaling induces the expression of smooth muscle proteins in bladder mesenchyme resulting in smooth muscle differentiation. We confirmed that smooth muscle differentiation in the mouse urinary bladder occurs first at gestational day 14 (E14) based upon immunohistochemical localization of smooth muscle alpha-actin (SMAA). To investigate murine bladder smooth muscle differentiation and epithlelial-mesenchymal signaling in the developing bladder, we analyzed gene expression profiles of intact embryonic murine bladders and separated epithelial and mesenchymal components at embryonic days E13, E14, E15, E16, and postnatal day 1 (P1). Using cDNA microarray, we identified regulators of vascular smooth muscle differentiation in bladder mesenchyme, including serum response factor (SRF) and its cofactors, ELK1 and SRF accessory protein (SAP)1, as well as two SRF-associated pathways, angiotension receptor II and transforming growth factor- beta2. Immunohistochemistry showed diffuse expression of SRF in the bladder at E12 with localization of expression to the peripheral mesenchyme at E13 and E14. Our results suggest that bladder smooth muscle differentiation may share a similar gene expression program as occurs during vascular smooth muscle differentiation. The unique structure of the urinary bladder makes it an ideal model for studies of smooth muscle differentiation and epithelial-mesenchymal signaling.     

Two FGF receptor genes are differentially expressed in epithelial and mesenchymal tissues during limb formation and organogenesis in the mouse.

Fibroblast growth factors (FGFs) can influence the growth and differentiation of cultured cells derived from neuroectoderm, ectoderm or mesenchyme. The FGFs interact with a family of at least four closely related receptor tyrosine kinases that are products of individual genes. To investigate the role of FGFs in the growth and differentiation of embryonic tissues and to determine whether the individual FGF receptor genes might have specific functions, we compared the localization of mRNA for two FGF receptor genes, FGFR1 (the flg gene product) and FGFR2 (the bek gene product), during limb formation and organogenesis in mouse embryos (E9.5-E16.5). Although the two genes were coexpressed in some tissues, the differential expression of FGFR1 and FGFR2 in most embryonic tissues was striking. FGFR1 was expressed diffusely in mesenchyme of limb buds, somites and organ rudiments. In contrast, FGFR2 was expressed predominantly in the epithelial cells of embryonic skin and of developing organs. The differential expression of FGFR1 and FGFR2 in mesenchyme and epithelium respectively, suggests the receptor genes are independently regulated and that they mediate different functions of FGFs during development.     

Micromere Differentiation in the Sea Urchin Embryo: Expression of Primary Mesenchyme Cell Specific Antigen during Development

The temporal and spatial expression of antigen specific for primary mesenchyme cell (PMC) lineage cells during early development of the sea urchins Hemicentrotus pulcherrimus and Stronglyocentrotus nudus was studied with a monoclonal antibody (P4). P4 was produced by a hybridoma cell line prepared by fusion of myeloma cells and spleen cells from a mouse immunized with cultured spicule-forming cells. Immunofluorescence studies demonstrated that P4 antibody reacted strongly with the surfaces of PMC's and spicule-forming cells of both species. Immunoblot analysis showed that P4 antibody reacted with several proteins including those of 140–kDa, 120–kDa, 53-kDa, 43–kDa, and 41–kDa in H. pulcherrimus and with those of 130–kDa, 110–kDa, 51–kDa, and 43–kDa in S. nudus . These proteins appeared sequentially after the hatching blastula stage. Tunicamycin inhibited the expressions of these P4 antigens as well as spicule formation. Two of the P4-reactive antigens, the 140–kDa and 43–kDa proteins, in H. pulcherrimus were synthesized de novo and shown to be identical to micromere differentiation specific proteins. These results suggest that P4 binds to specific molecules that are important in spicule formation in developing sea urchin embryos.     

Serogroups of the beer spoilage bacterium Megasphaera cerevisiae correlate with the molecular weight of the major EDTA-extractable surface protein

Megasphaera cerevisiae is a Gram-negative obligate anaerobe that causes turbidity and off-flavour and aroma in beer. Seven isolates of M. cerevisiae were obtained worldwide, and their extractable surface antigens were focused upon to determine if there is more than one serogroup of this bacterium. Sodium dodecyl sulphate polyacrylamide gel electrophoresis of ethylenediaminetetraacetic acid (EDTA) bacterial extracts revealed a predominant protein with apparent molecular weights of 46,000, 45,000, and 43,000 for three, two, and two isolates, respectively. When mouse anti-serum generated against any of the EDTA extracts was reacted with denatured bacterial proteins in immunoblots, all bacterial isolates exhibited extensive cross-reactivity involving three antigens, one being the major EDTA-extractable protein. In contrast, when the sera were tested for surface reactivity with intact bacteria, three cross-reactivity groups were observed, with the groups individually comprised of bacteria having the same size major EDTA-extractable surface protein. When BALB/c mice immunized with a bacterium from each of the three serogroups were used for monoclonal antibody (Mab) hybridoma production, bacterial surface-reactive Mabs were obtained whose reactivities parallel the three polyclonal antibody-defined serogroups. Through combining these surface-reactive Mabs, it will be possible to rapidly detect and identify beer contamination by M. cerevisiae belonging to any serogroup.     

An SEM analysis of the epithelial--mesenchymal interface in the mandible of the embryonic chick

In this paper the ultrastructural features of the epithelial-mesenchymal interface in mandibular processes of embryonic chicks have been examined using scanning electron microscopy. Mandibular epithelium is required for the mesenchyme to differentiate as osteoblasts and to deposit the membrane bones of the mandible. The surface morphology of the epithelium changes from the lateral to the medial face of the mandible from rounded cells, each with a central cilium to flattened cells with numerous microvilli. Treatment with trypsin and pancreatin was used to digest the basal lamina so as to separate epithelium from mesenchyme. This exposed a thick, fibrillar basement membrane (reticular lamina), which was thicker underlying the caudal epithelium than under the cephalad epithelium. Addition of collagenase to the trypsin/pancreatin solution degraded some of the basement lamella, especially that underlying epithelium on the caudal portion of each mandibular process. Selective degradation of basement lamella is postulated as one means of regulating inductive epithelial-mesenchymal interactions. EDTA was used to isolate basal laminae on mandibular mesenchyme. SEM was used to confirm the integrity of the basal lamina, its structure, and its association with overlying epithelial cells and underlying basement lamella.