Thyroxine-dependent modulations of the expression of the neural cell adhesion molecule N-CAM during Xenopus laevis metamorphosis

Numerous studies suggest that cell surface glycoproteins involved in the formation of cell-cell contacts play a critical role in determining cell patterning, movement and differentiation during embryonic development as well as in the control of the cellular events maintaining tissue integrity and repair throughout the life of the animal (Edelman, 1986a,b; Takeichi, 1988; Thiery, 1989). Using specific monoclonal antibodies that perturb the aggregation or the adhesion of cells in vitro, it has been possible to isolate and characterize a number of these molecules known as cell adhesion molecules or CAMs (Thiery et al. 1977; Gallin et al. 1983; Grumet and Edelman, 1984; Hatta et al. 1985; Volk and Geiger, 1986).

N-CAM is a cell surface glycoprotein that mediates Ca²⁺-independent aggregation of cells by a homophilic mechanism (Hoffman et al. 1982; Hoffman and Edelman, 1983). N-CAM is the prototype member of adhesion molecules belonging to the IgG superfamily (Edelman, 1987); it contains five Ig-like constant domains in the extracellular portion of all its polypeptide variants (Cunningham et al. 1987). In the chicken and in the mouse, the polypeptides differ in their cell-associated domains as a result of alternative splicing of RNA (Murray et al. 1986a,b;Owens et al. 1987; Hemperly et al. 1986a,b; Cunningham et al. 1987; Barbas etal. 1988; Santoni etal. 1989) transcribed from a single N-CAM gene (D’Eustacchio et al. 1985). The relative amount of each polypeptide varies in different tissues and at different epochs of development; a 180x10³Aï_r (K) form of the molecule (the Id polypeptide), for example, is synthesized exclusively in the nervous system and is first produced at relatively late stages of development while a 140K form (the sd polypeptide) is already present in the early blastula and then persists in many different tissues (Murray et al. 19866; Levi et al. 1987). These changes, most probably, reflect modulations of the capacity of N-CAM to interact with the cytoskeleton and/or with other molecules in the membrane, events that should ultimately affect the adhesive and morphogenetic function of the molecule.

During embryonic development, N-CAM is present transiently on derivatives of all three germ layers; its expression is strongly modulated in areas undergoing embryonic induction such as the neural plate and the placodes and in the somites (Thiery et al. 1982; Crossin et al. 1985; Levi et al. 1987). During later development, the expression of N-CAM becomes progressively restricted to neurons and astrocytes in the CNS and to cell bodies, unmyelinated axons and non-myelinating Schwann cells of the PNS (Chuong and Edelman, 1984; Daniloff et al. 1986a). In skeletal muscle, the molecule is transiently expressed before and during the formation of neuromuscular contacts and persists mainly in neuro-muscular junctions in older animals (Rieger et al. 1985; Covault and Sanes, 1986). In denervated muscles and in regenerating nerves, however, the expression of N-CAM can be reinduced (Daniloff et al. 19866; Covault and Sanes, 1985).

Despite the volume of descriptive information about the modulation of N-CAM expression at different stages of development, not much is yet known about the molecular signals and modes of regulation of the expression of the molecule in its various forms. So far, the only molecule known to influence N-CAM expression is the nerve growth factor NGF (Prentice et al. 1987; Doherty el al. 1988).

To establish a system to study other possible molecular controls of N-CAM expression, we have chosen to investigate the modulations of N-CAM expression occurring during Xenopus laevis metamorphosis, inasmuch as metamorphosis offers one of the most striking cases of morphogenesis and it occurs under direct hormonal control.

The phenomenon of metamorphosis in amphibia involves morphological, histological and physiological changes associated with the shift from aquatic to terrestrial life and from herbivorous to a carnivorous diet. Virtually all organs and tissues in the animal undergo profound remodeling. For example, the tail is resorbed; limbs develop; the nervous system becomes adapted to control limb movement with a de novo formation of the cerebellum, motor neurons and nerves, and profound changes in the ganglia; the digestive tract is redesigned by means of a complex process of simultaneous degeneration and regeneration; muscles are formed; bones and cartilages are reshaped and the skin restructured.

All these morphological changes take place in a few days under the direct control of thyroid hormones. That thyroxine, and no other systemic factor, is the essential stimulus for metamorphosis has been shown by treating various tissues in vitro with the hormone and observing their subsequent modification (For reviews see: Deuchar, 1975; Fox, 1981; White and Nicoll, 1981). One of the most interesting features of metamorphosis is reflected by the fact that several organs and tissues of the larva respond in such different ways to only one hormone. It has been shown that the nuclear receptor for thyroxine is the cellular equivalent of the oncogene v-erbA (Weinberger et al. 1986; Koening et al. 1988); it is conceivable that different variants of this molecule might activate totally different developmental programs leading either to histolysis or to further tissue differentiation (Miyajima et al. 1989).

We reasoned that, if N-CAM is involved in the control of morphogenetic events, its expression should be modulated by thyroid hormones at sites of tissue remodeling in metamorphosing animals. The identification of such modulation events might then provide a basis for the development of an in vitro model system in which to study the regulation of expression of a morphogenetic molecule. Such a system would be ideal for further experimental investigation inasmuch as the molecular mechanisms underlying the thyroxine-dependent control of gene expression are now beginning to be understood (Evans, 1988; Glass et al. 1988; Damm et al. 1989; Koenig et al. 1989).

Sexually mature Xenopus laevis were obtained from the Service d’Elevage de Xenopus of the Centre Nationale de la Recherche Scientifique (Montpellier). Animals at different stages through metamorphosis were purchased from Nasco (Fort Atkinson, Wisconsin). Animals were either used immediately upon delivery or maintained at 24°C and fed twice weekly with Nasco Xenopus brittle. Stages of development were determined according to Nieuwkoop and Faber (1967). Treatment of animals with thyroid hormones (TH) was achieved by rearing the animals in water containing 3x 10∼⁷M thyroxine (T₄) (Atkinson, 1981). The rearing solution was obtained by diluting a commercial aqueous solution of L-thyroxine (0.2 M, Roche) and was changed daily.

The preparation and the characterization of the polyclonal and monoclonal antibodies directed against Xenopus N-CAM used in this study has been previously described (Fraser et al. 1984; Levi et al. 1987). Characterization of NC-1 antibody and of its reactivity on early Xenopus nervous system has been previously described (Tucker et al. 1984, 1988).

Paraffin sections were prepared for staining using a previously published procedure (Levi etal. 1987; Gurdon etal. 1976) with some modifications required for the size of metamorphic animals. Whole animals or dissected organs were frozen in isopentane cooled in liquid nitrogen and immediately immersed in methanol at -80°C. The samples were then maintained in methanol at —80°C for periods ranging from one to four months depending on the size of the animal with weekly changes of cold methanol. The tissues were serially transferred to methanol equilibrated at -20°, 4°, and 20°C for at least one day at each step. The samples were then immersed twice in xylene until completely clarified for a total period not greater than 30 min and then transferred to a solution of 50% Paraplast (Monoject Scientific, St. Louis, MO) in xylene at 56°C for 20min in a vacuum oven, infiltrated three times with Paraplast at 56°C under vacuum for 45 min, and embedded in Paraplast. Sections 10 mm thick were cut using a microtome (Lemardeley, Paris, France), floated on distilled water at 45°C, collected on washed glass slides and dried on a heating plate at 45°C for at least one hour.

For immunofluorescent staining, the deparaffinized sections were incubated sequentially with the primary antibody (10/tgml^-1 in PBS, 5% foetal calf serum (FCS); overnight) and a rhodamine-conjugated goat anti-rabbit IgG secondary antibody (10 pg ml^-1 in PBS, 5% FCS. Nordic Immunology, Tilburg, The Netherlands) for 2h. For double labeling, the sections were treated with a mixture of the two primary antibodies (10 pg ml^-1 each in PBS, 5% FCS; overnight) followed by a mixture of rhodamine-conjugated goat antirabbit IgG and biotinylated goat anti-mouse secondary antibodies (10 pg ml^-1 each in PBS. 5% FCS; 2 h) and FITC-conjugated streptavidin 5 pg ml^- in PBS, 5% FCS; 30 min). The sections were observed with a Leitz epifluorescence microscope.

Tissues were dissected from animals at different stages of development, homogenized in 10 volumes of PBS containing 0.5% NP40, the protein concentration of the extract was measured, and SDS sample buffer was added to yield a final protein concentration of 2 mg ml^-1. Samples were resolved by SDS-PAGE (Laemmli, 1970) on 7% polyacrylamide gels followed by electrophoretic transfer of the proteins to nitro-cellulose paper (Towbin et al. 1979). The presence of N-CAM was revealed by incubation with 50 pg of antibodies followed by [^l25I]protein A and autoradiography. Autoradiograms were scanned and analyzed with an LKB Ultrascan XL computerized gel scanner.

Total RNA from organs was prepared by a modified guanidinium/cesium chloride centrifugation method (Mac Donald etal. 1987). Tissues were homogenized in ten volumes of 4 M guanidinium thiocyanate, 0.05 M sodium acetate, 2mM EDTA and IM /3-mercaptoethanol. Solid CsCl was further added to reach a concentration of 4.5 M. Homogenates were laid onto a cushion of 6.7M CsCl (p=1.82gml^-1), 0.05 sodium acetate, 1 mM EDTA. After an overnight centrifugation at 40 000 revs min^-1 at 20°C in a 70.1 Ti rotor (Beckman), RNA appeared as a band in the CsCl gradient and was recovered with a syringe and ethanol precipitated.

About 10 μg of RNA from liver and brain was heated at 60°C in 50% formamide in 4-morpholinopropane sulfonic acid buffer and then electrophoresed in a 1 % agarose gel containing 6% formaldehyde in the running buffer. RNAs were blotted onto nylon membrane (Hybond N, Amersham). Recombinant plasmid including a 1.8 kb insert corresponding to an EcoRI fragment of the 5’ end of Xenopus was a gift from Dr Kintner; the probe was labeled by random priming with α¹P-dCTP. Filters were hybridized with the random primed probe at 42°C in buffer containing 50 % formamide, 5 xSSPE, 2xDenhardt (0.2% polyvinylpyrrolidone/0.2% Ficoll/0.2% bovine serum albumin), 5% dextran sulfate, 0.1% SDS and 20 μg of salmon sperm DNA per ml. Blots were washed in O.lxSSC, 0.1 % SDS at 55°C and autoradiography performed on Kodak X-OMAT film for 1-3 days.

Using double labeling, we compared the patterns of staining of antibodies against Xenopus N-CAM to those of the monoclonal antibody NC-1. This monoclonal antibody recognizes a carbohydrate epitope that in Xenopus laeyis is present on most neurons and neural processes; in this species, it is therefore a useful marker of neural cells (Tucker et al. 1988).

In premetamorphic and prometamorphic animals, all neuronal cell bodies and processes of the CNS and the PNS and the neuroepithelial cells lining the ependymal canal were intensely stained by anti-N-CAM antibodies (Figs. IA, 2A). The NC-1 epitope was present in neural processes of the CNS and PNS (Figs. IB, 2B) and on cell bodies in the PNS but was only weakly expressed by CNS neural cell bodies and was not present in the neural epithelium (Fig. IB). Reactivity to both antibodies was limited to the cell surface. This pattern of staining persisted unchanged between stages 51 and 61 NF. During the metamorphic climax, between stages 62 and 66 NF, the staining of N-CAM, while persisting in the CNS, decreased dramatically and rapidly in the PNS, so that by stage 64, the dorsal root ganglia (DRG) (Fig. 1C) and peripheral nerves (Figs. IC, 2C) were virtually negative for the molecule, which persisted only in few unmyelinated fibers. In contrast, reactivity to NC-1 persisted unchanged on the cell bodies of both DRG (Fig. ID) and peripheral nerve (Figs. ID, 2D). Both the optic nerve (Fig. 2E,F) and the olfactory nerve (Fig. 8D) maintained a high level of N-CAM expression; embryologically and cytologically both these nerves constitute an integral part of the CNS.

Treatment of stage 56 animals with 3 × 10⁻⁷ M thyroxine for 4 days, although inducing profound morphological changes in the nervous system (e.g. the formation of the lateral motor columns) did not induce a rapid disappearence of N-CAM from the PNS. The disappearance of N-CAM was induced only by a much longer treatment (at least 10 days), suggesting that the down-regulation of N-CAM in the nervous system is not under the direct control of the hormone, but depends on a more complex cascade of events.

Throughout metamorphosis, N-CAM is expressed in the nervous system as three principal components of relative molecular mass 180, 140, and 120X10³. In premetamorphic stages, some of the molecule appears as a polydisperse region corresponding to the highly polysialylated E (embryonic) form that has been described in the early stages of development of all species examined so far (Hoffman et al. 1982; Friedlander et al. 1985; Levi et al. 1987). During metamorphosis and later development, the E form of N-CAM disappears and the relative amount of the molecule present in the nervous system diminishes gradually as normalized to total membrane protein (Fig. 5 lanes 5, 7, 8). Quantitative analysis of different autoradiograms showed that the relative amount of the three major polypeptides did not change considerably during development. Moreover, treatment of stage 56 tadpoles with thyroxine for 4 days does not produce any evident change in the forms of the molecule (Fig. 5 lane 5,6).

We have previously reported that adult Xenopus hepatocytes express high levels of a particular variant of N-CAM of relative molecular mass 160x10³. This molecule can be immunoprecipitated and recognized in immunoblots by polyclonal and monoclonal antibodies against Xenopus N-CAM as well as by cross-species polyclonal antibodies against chicken brain N-CAM (Levi et al. 1987). It is unusual to observe N-CAM in adult liver and other species examined so far do not show it.

Premetamorphic hepatocytes do not express N-CAM. The molecule first appears on parenchymal cells of livers at stage 59 and then persists throughout the life of the animal (Fig. 3). Liver N-CAM is present on the cell surface of hepatocytes and is accumulated in regions of cell-cell contact; the luminal aspect of the hepatocytes and other liver cell types such as the melanocytes are not stained by anti-N-CAM antibodies (Figs. 3, 4B).

Treatment of stage 52 animals with 3X10⁻⁷M thyroxine for periods as short as 48 h induced the appearance of N-CAM at the surface of hepatocytes (Fig. 4); shorter periods of treatment were not sufficient to induce the response. All hepatocytes responded simultanously to the hormone excluding the possibility of replacement of the population of parenchymal cells by newly produced cells. Hepatocytes were first responsive to thyroxine at stage 51-52; at earlier stages, N-CAM could not be induced in the liver even after prolonged hormonal treatment.

The induction of N-CAM in Xenopus liver could be confirmed by Western blot analysis. The major form of the molecule synthesized after hormonal treatment has a relative molecular mass of 160x10³ similar to that appearing during spontaneous metamorphosis and in adult animals; two minor components at 135 and 125K could also be observed in Fig. 5A (lanes 1-4).

In order to characterize differences between brain N-CAM and Xenopus liver N-CAM, we isolated mRNAs from adult Xenopus brain and liver and performed a Northern blot analysis using a 1.8 kb fragment corresponding to an EcoRI insert from the 5’ end of Xenopus N-CAM cDNA. Transcripts were present in both organs. In adult brain, in accord with what has been described (Kintner and Melton, 1987), two predominant transcripts at 9.6 and 4.2 kb and a minor component at 2.2 kb were observed (Fig. 5B lane 2). In the liver (Fig. 5B lane 1), a predominant transcript was present at 9.2 kb while other minor bands could be observed at 3.8, 3.3 and 2.2 kb. The two larger mRNAs of liver had a size unequivocally different from larger components found in the brain (Fig. 5B).

During prometamorphosis and metamorphosis, many new muscle bundles are formed while pre-existing muscle masses, such as the dorsal muscles, increase in volume due to the addition of new fibers. N-CAM is strongly expressed in all newly formed muscles from their first condensation up to their complete differentiation in contracting muscle bundles; at later stages N-CAM disappears rapidly (see also Kay et al. 1988). This type of modulation is particularly evident in the muscles of the developing limb (Figs. 6A, 2C) and in the ventral muscles surrounding the abdominal cavity (m. rectus abdominis and m. obliquus abdominis, Fig. 6B). During further development, N-CAM is expressed only in newly formed fibers that are added to differentiated muscle masses. This process can be seen in the dorsal muscles of the trunk, in the ventral muscles and in differentiated muscles of the limb (see Fig. 2C).

Treatment of animals with thyroid hormones accelerates greatly the myogenic process and the corresponding modulation in the expression of N-CAM, which is always transiently expressed in condensing and differentiating muscle masses.

In contrast to skeletal muscles, cardiac muscles express N-CAM from their first differentiation throughout metamorphosis. Both the walls and the trabeculae of the ventricle as well as the atrial walls are intensely stained by anti-N-CAM antibodies at all stages of development (Fig. 7A, B).

It has been reported that, during embryonic development, N-CAM is transiently expressed in certain epithelia particularly in areas where inductive processes are taking place (Crossin et al. 1985). During late development of Xenopus, two types of epithelia are strongly labeled by anti-N-CAM antibodies: the proliferating epithelium of the lens (Fig. 8A-C) and the sensory part of the olfactory epithelium (Fig. 8D). The staining of these tissues persists throughout metamorphosis. The staining of the lens epithelium is clearly polarized being limited to areas of cell-cell contact and being absent from the apical aspect of the cells (Fig. 8B,C). Lens fiber cells are not stained.

Other sites in the animal did not show major changes in N-CAM expression during metamorphosis.

In this study, we have examined the patterns of expression of a primary cell adhesion molecule, N-CAM, during metamorphosis of Xenopus laevis. As the morphological transitions that take place during metamorphosis occur under the direct control of thyroid hormones, we wished to detect any thyroxine-induced modulations in the expression of the molecule. Our major findings may be summarized as follows: (a) N-CAM is expressed by all the cell bodies, processes and neural epithelial cells of the CNS throughout metamorphosis. The optic and olfactory nerves, which belong embryologically and cytologically to the CNS, express high levels of the molecule throughout the life of the animal. In the CNS, N-CAM is present as three polypeptides of 180, 140 and 120K. The ratio of the three polypeptides does not change considerably during development, but the amount of N-CAM relative to total protein expressed in the CNS decreases gradually during the growth of the animal as seen by immunoblots. A polysialylated, polydisperse form of the molecule present in the premetamorphic brain diminishes during metamorphosis, (b) In the cell bodies and processes of the PNS, the expression of N-CAM is strongly repressed towards the end of metamorphosis, so that in postmetamorphic animals the peripheral ganglia and nerves are virtually N-CAM-negative. (c)

Premetamorphic liver does not express N-CAM. During metamorphosis a 160K polypeptide that is recognized by both polyclonal and monoclonal antibodies to N-CAM is strongly induced in the liver. It can first be observed at stage 59 and persists then throughout the life of the animal. Liver N-CAM is a cell membrane protein and is predominantly localized in regions of cell-cell contact. A cDNA probe of Xenopus brain N-CAM recognizes four transcripts (9.2, 3.8, 3.3 and 2.2 kb) in adult liver RNA, these transcripts are definitely different is size from those recognized in brain RNA (9.5, 4.2 and 2.2 kb), (d) Expression of N-CAM in the liver can be induced by a 48 h treatment of premetamorphic tadpoles with thyroxine, (e) In skeletal muscle, N-CAM is strongly expressed in fusing myoblasts and in extending myofibers, but in later stages of myogenesis N-CAM is no longer expressed. In contrast, cardiac muscle expresses N-CAM throughout the life of the animal, (f) Beside the liver, strong N-CAM expression is maintained in two other epithelia: the sensory olfactory epithelia and the epithelium of the lens. In both cases, N-CAM is present since the first formation of the epithelium and persists then unchanged through the life of the animal.

In Fig. 9, we have summarized the general types of modulation of N-CAM expression that are observed during metamorphosis; three different modes can be distinguished: (1) a sharp thyroxine-dependent induction of expression in the liver; (2) a sharp inhibition of expression in the PNS and in differentiated skeletal muscle and (3) a persistence of expression with a gradual reduction during the growth of the animal in the CNS, heart, lens and olfactory epithelia.

It has been proposed that N-CAM plays an important morphoregulatory role (Edelman, 1985, 1986a, b, 1988a,b). To fulfill this hypothesis, it is necessary that the levels of expression and molecular forms of the molecule change in sites where major changes in the morphology of the animal are taking place. Indeed, during embryonic development and in the course of nerve and muscle regeneration, the levels of N-CAM expression are strongly regulated at specific locations. For example, N-CAM is strongly modulated in sites of embryonic induction (Crossin et al. 1985; Levi et al. 1987), it increases in denervated muscle (Covault and Sanes, 1985; Daniloff et al. 1986b) and during nerve regeneration (Nieke and Schachner, 1985; Daniloff etal. 1986b).

So far, the molecular mechanisms by which the levels of N-CAM expression are modulated remain elusive. There are three published examples of factors affecting the expression of N-CAM: NGF on cultures of PC12 cells (Prentice et al. 1987; Doherty et al. 1988), retinoic acid in embryonal carcinoma cell lines (Husman et al. 1989) and laminin in cultures of N2A neuroblastoma cells (Pollerberg et al. 1985). In all these cases, however, the changes in N-CAM expression are associated either with cell differentiation into neurons or with changes in the morphology of the culture (e.g. neurite extension) and could be a consequence of these variations.

Amphibian metamorphosis provides a further ideal system to study the regulation of morphoregulatory molecules in vivo and possibly in vitro. During metamorphosis, the body plan of the animal is profoundly changed as the animal adapts to live in a completely new environment. The whole reshaping of the body plan takes place over a few days and is under the direct control of thyroid hormones; it depends extensively on processes of cell migration, adhesion, proliferation, differentiation and death which are directly affected by the presence of thyroid hormones (Deuchar, 1975; Frieden and Just, 1970; Fox, 1981; White and Nicoll, 1981).

We report here three different modes of regulation of N-CAM expression during Xenopus metamorphosis. The modulations that we observe in the CNS, the PNS and muscle could be ascribed to a decrease in the transcription rates of the N-CAM gene. Similar types of modulation are also observed during development of other species (Daniloff et al. 1986a; Rieger et al. 1985; Nieke and Schachner, 1985) and it could well be that the effect of thyroxine in these cases is only an acceleration of a preexisting regulatory mechanism of N-CAM. The diminution of N-CAM expression in the PNS of Xenopus is, however, much more abrupt and drastic than in chicken and in mouse; in less than a week at the end of the metamorphic climax (between stage 61 and 66), N-CAM expression is virtually abrogated in the PNS of Xenopus, while in other species a N-CAM decreases gradually during the postnatal development of the animal and persists at low levels in adult animals.

The mode of regulation of N-CAM expression that we observe in the liver is novel. In all species studied so far, the transcription of the single gene encoding the N-CAM molecule is strongly repressed in the parenchymal cells of the liver. In Xenopus liver, thyroid hormones induce the abrupt expression of a special form of N-CAM in all hepatocytes; the molecule is not expressed in the absence of the hormone. The fact that N-CAM expression can be induced in all hepatocytes already at early premetamorphic stages (stage 51-52) excludes the possibility that the responsive cells constitute a new cellular population replacing larval hepatocytes as is the case for cells competent for the estrogen-dependent vitellogenin synthesis (Kawahara etal. 1987, 1989).

Our observations can be interpreted in several ways. (1) It is possible that in Xenopus different N-CAM genes are under the control of different gene regulatory elements and those in the liver may be in part activated by thyroid hormones via the proto-oncogene c-erbA nuclear receptors. This activation would bypass the mechanism of repression normally acting in the liver in other species. Indeed, due to gene duplication, the Xenopus genome contains two N-CAM genes with slightly different sequence (Krieg, personal communication); nothing is known so far about their promotors. (2) The molecule expressed in the liver could represent the product of a completely different gene that would maintain, however, both at the protein and at the nucleic acid levels sufficient homology to N-CAM to be recognized by monoclonal antibodies and cDNA probes. (3) The expression of N-CAM in the liver might be the result of a more complex thyroxine-dependent mechanism implying the activation of a specific form of the molecule resulting from an organ-specific alternative splicing. In this case, the liver form of N-CAM could be induced only in an organ-specific context in which other forms of the molecule are repressed by liver factors.

The analysis of the sequence, gene organization and gene regulation of liver N-CAM (already started in our laboratory) should provide an answer to this aspect of the problem as the mechanisms of action at the transcriptional level of thyroid hormones are now beginning to be elucidated (Evans, 1988; Glass et al. 1988; Damm etal. 1989; Koenig et al. 1989).

A second question raised by our findings regards the function of liver N-CAM. Although the involvement of liver N-CAM in cell adhesion needs still to be proved directly in primary cultures of hepatocytes, its polarized distribution in areas of cell-cell contact in the membrane of hepatocytes and other epithelial cells of the liver is typical of molecules involved in intercellular adhesion. In all species described so far, the parenchymal cells of the liver express high levels of the Camdependent cell adhesion molecule L-CAM (E-cadherin or uvomorulin) (Gallin et al. 1983; Peyrieras et al. 1983; Takeichi, 1988). This molecule is also expressed in many other epithelia throughout development and in the adult (Thiery et al. 1984). In contrast to other species, it appears that the level of expression of L-CAM (E-cadherin) in Xenopus liver is relatively low compared to other epithelia (Levi et al. 1987; Choi and Gumbiner, 1989; Gumbiner, personal communication). These molecular differences might be at the origin of other cytological differences such as the massive invasion of melanocytes occurring in Xenopus liver.

These observations all suggest, therefore, that it is possible for the same organ in different species to utilize different adhesion systems to maintain its integrity. These types of changes suggest that regulatory genes controlling the expression of CAMs may act differently depending on the morphogenetic context (embryonic development, metamorphosis or regeneration) but nevertheless preserving the general body plan (Edelman, 19886).

We thank Dr Christopher Kinter for providing Xenopus N-CAM cDNA probes. This work was supported by the Centre National de la Recherche Scientifique and the Fondation pour la Recherche sur la Myopathie (contract AFM 89). G.L. was a recipient of a grant from the Associazione Italiana per la Promozione delle Ricerche Neurologiche, A.R.I.N.