Differentiation (1992) 51 :9-20 Ontogeny, Neoplasia and Differentiation Therapy

0 Springer-Verlag 1992

Distribution pattern of connexin 43, a gap junctional protein, during the differentiation of mouse heart myocytes Catherine Fromaget, Abdelhakim El Aoumari, and Daniel Gros Laboratoire de Biologie de la Differenciation Cellulaire, UA CNRS 179, Facult6 des Sciences de Luminy, Universite d’Aix-Marseille 2, F-13288 Marseille-Ctdex, France Accepted in revised form April 29, 1992

Abstract. In the cardiac muscle, the electrical coupling of myocytes by means of gap (or communicating) junctions, allows the action potentials to be propagated. Connexin 43 (CX 43) is the major constitutive protein of the gap junctions in the mammalian myocardium. In this organ, the abundance of CX 43 and of its messenger, as well as the spatial expression of this protein, are developmentally regulated. These findings are complemented by the results presented in this article, which deals with the distribution of CX 43 in the ventricular myocytes of mouse heart during differentiation, between the 11 days post coitum embryo stage and adulthood. By immunoelectron microscopy experiments on ultrathin sections of cardiac ventricular tissue of oneweek-old mouse, we have provided confirmation that the anti-CX 43 antibodies used here specifically recognized the gap junctions. Double labeling immunofluorescence experiments have been undertaken to localize, within the same cells, either CX 43 and desmin, or CX 43 and Con A or WGA receptor sites. From the earliest stage investigated (1 1 days post coituni) onwards, expression of CX 43 is always associated with desmin-positive cells, that is, with the myocytes. Up to birth, there is in the ventricular wall a gradient of expression of CX 43 which is superimposable on a gradient of expression of desmin. Immunoreactivity to anti-CX 43 and anti-desmin antibodies is high in the sub-endocardial trabeculae and low (or even undetectable for CX 43, in the early stages) in the sub-epicardial cell layers. In the embryonic stages, the expression sites of CX 43 are visible in the form of small dots, whose abundance increases as development proceeds. During these stages, the immunoreactive sites are distributed in a relatively homogeneous pattern throughout the membrane of the myocytes. One week after birth, the CX 43 expression is restricted to the two ends of the myocytes (where the intercalated discs develop), and the adjacent lateral regions. This polarization of CX 43 is more pronounced at the two and three weeks post natal stages Correspondence to: D. Gros

and in the fully differentiated ventricular myocytes (adult stage) CX 43 is only present in the intercalated discs.

Introduction The gap junctions are structures composed of two closely apposed membranes containing tightly packed cell-tocell channels, called junctional channels, which are responsible for intercellular metabolic cooperation and electrical coupling [ l , 421. Important biological functions are attributed to these junctional channels, including the regulation of embryonic development [25] and cell growth [48], the homeostasis of tissular compartments (see for example [50]) and the synchronisation of myocardial cell beating [61]. The junctional channels are composed of transmembrane proteins which belong to the connexin family [6, 721. In mammals, ten members of this family have been cloned and sequenced. These are: connexin 46 (CX 46) [59]; connexin 45 (CX 45) [32]; connexin 43 (CX 43; Mr 43-47,000) [4, 5, 17, 401; connexin 40 (CX 40) [26]; connexin 37 (CX 37) [72]; connexin 33 (CX 33) [26]; connexin 32 (CX 32; Mr 27-28,000 [36, 581; connexin 31.1 (CX 31.1) [26]; connexin 31 (CX 31) [29] and connexin 26 (CX26; Mr 21-22,000) [76]. Homologs of some of these connexins have been cloned and sequenced from chicken [3, 511 and Xenopus [21] sources. Transcripts of CX 46, CX 43, CX 40 and CX 37 genes have been detected in rat and mouse heart (5, 26, 59, 731 but the abundance of CX 46, CX 40 and CX 37 gene transcripts, in this organ, is low compared to that of CX 43 gene transcript [26, 59, 731. So far, only the expression of the CX 43 protein has been demonstrated in rat and mouse heart. Although this junctional protein has been identified in a variety of other tissues [7, 13, 151 it is in heart that its expression is best documented.

10

CX 43 is a major component of gap junctions isolated from cardiac muscle [ 131. More specifically, its association with myocyte gap junctions has been demonstrated by immunofluorescence and electron microscopy imniunogold labeling by Beyer et al. [7], Yancey et al. [75], El Aoumari et al. [15, 161 and Laird and Revel [38]. In adult rat heart, Micevych and Abelson [49] have shown that antisense CX 43 riboprobe hybridized with myocytes. No hybridization signal with this probe was ever observed to be associated with cardiac fibroblasts. These results are in agreement with those of Rook et al. [60, 621 who demonstrated that cardiac fibroblast junctional channels have different electrophysiological and immunological properties from those of myocyte junctional channels that contain CX 43. CX 43 expressed in myocytes has multiple phosphorylation sites and its half-like in cultured neonatal rat myocytes was established to be 1-2 h [39, 411. In these same cells, Rook et al. [61] have shown that a rapid initial formation of gap junctions occurs during the 2-20 min which preceed the synchronization of contractions, probably by assembly of functional channels from precursors already present in the plasma membrane. During rat and mouse heart organogenesis, the expression of CX 43 is regulated [18, 191. The abundance of both mRNA and protein increases from the early developmental stages to the first week after birth, then decreases until the adult stage. A regulation phenomenon similar to that described above has also been shown for CX 45 mRNA during chicken heart development [3]. mRNA for CX 45 is detected in chicken heart as early as the sixth day of embryonic life; its abundance falls to about 10% at the sixteenth day, then stays low throughout adulthood. During the same period of organogenesis, the abundance of mRNAs for CX 43 and CX 42 remains relatively stable and variations are little more than twofold. As well as the expression level, the distribution of CX 43 in the different cardiac tissues is regulated during mammal heart organogenesis. Using an immunohistochemical technique, Van Kempen et al. [70] have shown that CX 43 is first detected, at the thirteenth day of gestation, in the free anterior and lateral walls of the atria and in the ventricular trabeculae of the rat heart. As development proceeds, CX 43 is progressively expressed in all the other regions of neonatal heart but one: the His bundle and the apical part of the bundle branches. The same distribution pattern of CX 43, which persists in adult rat heart, was also found in mouse and human heart [55, 701. At cell level, the organization of the gap junctions in the intercalated discs of adult myocytes, their formation and ultrastructural modifications during heart organogenesis have been extensively investigated by electron microscopy [23, 30,45, 57, 661. This technique however only allows the sampling of small portions of the plasma membrane and cannot provide a comprehensive overview of the distribution of gap junctions over the cell surface. The characterization of anti-CX 43 antibodies and their use on sections, by immunofluorescence, means that the problem of limited sampling can now be avoided, and consequently it is possible to carry

out, for example, a detailed analysis of connections between the differentiated myocytes [43] or of the distribution of the entire population of gap junctions within the intercalated discs [22]. We have recently characterized several site-directed antipeptide antibodies to rat CX 43 and shown that they cross-react with CX 43 from various mammalian species [13, 15, 161. In this paper we have used the antipeptide antibodies directed to residues 314-322 of CX 43 to analyse the distribution of gap junctions during the differentiation of mouse heart ventricular myocytes. On the basis of previous studies [33, 37. 56, 641 we have used desmin as a differentiation marker for identifying the myocytes, especially at the early stages of organogenesis of the myocardium.

Methods Biologicul muterid. Swiss mice were mated overnight and the fertilized females were selected the next morning on the basis of the presence of a vaginal plug [63]. The development stages investigated were the following: 11, 14 and 16 days post coitum (dpc), newborn (i.e. 20 days post coitum), 1, 2 and 3 weeks post partum (wpp) and the adult stage (2-month-old mice).

Immunofluorescence. Whole embryos (for the 11 and 14 dpc stages) or ventricles (for the others stages) were embedded in Tissue-Tek OCT compound (Miles, Elkhart, Ind., USA) and frozen by immersion in liquid nitrogen or in precooled isopentane (-20" C). Cryostat sections of 3-4 pm were mounted on gelatinized slides. The slides were stored at -20" C until use. Double labeling irnmunofluorescence experiments for the detection of both CX 43 and desmin were carried out as follows. Tissue sections were incubated for 30 min with phosphate-buffered saline (PBS) containing 1YObovine serum albumin (BSA). After overnight incubation at 4" C with a 3 pg/ml solution of affinity-purified rabbit antibodies to residues 314322 of rat CX 43 in PBS-BSA [15], the slides were washed with PBS and incubated for one hour at 37" C with a 1 :5 dilution in PBS-BSA of anti-desmin mouse monoclonal (clone DE-U-10, characterized by Debus et al. [I I] and commercialized by Sigma-Chimie SA, La Verpilliere, France). The slides were then washed and incubated for one hour at room temperature with PBS-BSA containing both tetraethyl rhodamine isothiocyanate (TR1TC)-conjugated sheep anti-rabbit IgGs diluted to 1 : 500 (Immunotech, Marseilles, France) and fluorescein isothiocyanate (F1TC)-conjugated goat anti-mouse IgCs diluted to 1 :2000 (Biosys SA, Compiegne, France). The slides were washed and mounted with Citifluor (Citifluor, London, UK). Double labeling inimunofluorescence experiments for the immunodetection of CX 43 and the labeling of cell surfaces with lectins (concanavalin A, ConA, or wheat germ agglutinin, WGA) [24] were carried out as follows. After overnight incubation at 4" C with the anti-CX 43 antibodies as described above, the sections were washed with PBS and incubated for one hour at room temperature with a 1 : 500 dilution in PBS-BSA of TRITC-conjugated sheep anti-rabbit IgGs. The slides were washed with PBS and incubated for 30 min at room temperature with either FITCconjugated ConA or WGA (Sigma Chimie SA, La Verpilliere, France) diluted at 100 pg/ml or 250 pg/ml in PBS, respectively. Slides were washed and mounted as described above. Control experiments were carried out using the secondary antibodies only or by replacing the anti-CX 43 primary antibodies with a preimmune serum affinity-purified fraction (see [ I 3 and 151 for the preparation of the preirnmune sera fractions). The sections were examined with a Zeiss 111 light microscope equipped for epifluorescence with appropriate filters to distinguish FITC

11 emission from TRITC emission. Tri-X Pan ASA 400 film was used for the photographs. Immunogold labeling. Tissue and sections were prepared as described by Berryman and Rodewald [2]. Small samples of ventricular tissue from one week-old newborn mouse were fixed for one hour at room temperature in a periodate-lysine-paraformaldehyde solution [46]. The samples were washed overnight at 4" C in 0.1 M phosphate buffer (pH 7.4) containing 0.5 m M calcium chloride, then for one hour at 4" C in 0.1 M phosphate buffer containing 50 mM ammonium chloride and 0.5 m M calcium chloride. To remove phosphate ions, the samples were rinsed with cold 0.1 M maleate buffer at pH 5.5 (4 times 15 rnin). They were then postfixed for 2 h at 4" C with 2% uranyl acetate in maleate buffer (pH 5.5). After rinsing with several changes of cold maleate buffer the samples were dehydrated for 45 rnin at 4" C in 50% acetone followed by sequential 45 min incubations at -20" C in 70% and 90% acetone. Embedding in LR Gold resin was performed according to the manufacturer's instructions (London Resin Company, Hants, UK). Ultrathin sections, mounted in 400-mesh gold grids, were treated for 5 rnin with TRIS-buffered saline (TBS): 5 0 mM TRIS, 150 m M sodium chloride, pH 7.4. Sections were then saturated for 10min with 0.5% ovalbumin (grade 111, Sigma Chimie SA, La Verpillikre, France) in TBS, then incubated for 3-4 h with rabbit antipeptide antibodies to residues 314322 of rat CX 43 (151 diluted (10 pg/ml) in TBS containing 0.1% BSA. Grids were washed by immersion for 3 rnin in TBS-BSA then by transfer through five successive drops of TBS-BSA (3 min per drop). Incubation with goat-antirabbit antibodies coupled to 10 nm gold particles (Janssen Pharmaceutical, Beerse, Belgium) diluted in TBS was carried out for 1 h. Grids were transferred through five drops of TBS (3 rnin per drop) and then rinsed with double-distilled water. Antibody complexes were stabilized with 2% aqueous glutaraldehyde for 5 min. After rinsing with water, sections were fixed for 15 min with aqueous osmium tetroxide and counterstained with uranyl acetate and lead citrate. Sections were examined with a Hitachi H 600 electron microscope. Control experiments were carried out using the secondary gold-labeled antibodies only or by replacing the primary antibodies with a preimmune serum affinity-purified fraction (see the previous paragraph). Freezezfracturing. Ventricles were dissected from hearts of oneweek-old mice. Small samples were fixed for 30 min at 4" C in a 5% glutaraldehyde solution in 0.1 M cacodylate buffer (pH 7.4) then immersed for 20 min sequentially in 10, 20 and 30% glycerol. The samples were mounted on gold discs and frozen in Freon 22 cooled with liquid nitrogen. Freeze-fracturing at - 100" C and platinum carbon shadowing were carried out using a Bakers freezeetch apparatus. Cleaned replicas were examined with a Hitachi H 600 electron microscope. Density of inzrnunoreactive sites to unti-CX 43 antibodies. The density of immunoreactive sites to anti-CX 43 was determined for the following development stages: 11, 14, 16 days post coitum and newborn. The number of sites was counted on immunofluorescent photographs ( x 800) representative of CX 43 positive ventricular regions. The surface areas taken into consideration for counting the sites ranged from 123 x lo3 pm2 (1 1 days post coitum) to 228 x lo3 Fm2 (newborn). The variance analysis of the results was carried out using the programme ANOVA.

Results The stages of development examined in this work (1 1, 14 and 14 days post coitum, newborn, 1, 2 and 3 weeks post partum and the adult stage) are the same as those which were studied previously to estimate the abundance of CX 43 and its mRNA during heart ontogenesis [19].

Fig. 1. Ventricular myocardium of one-week-old newborn mouse. Freeze-cleave replicas reveal in this tissue numerous gap junctions (urrows in a) scattered across the fracture faces. On ultrathin sections incubated with anti-conrexin CX 43 antibodies and gold labeled secondary antibodies (as described in Methods) the gap junctions, identified by their narrow extracellular space, are clearly visible festooned with colloidal gold particles (b and c). The nonjunctional membranes (indicated by urrowheud.7 in c) are free of labeling. No labeling was seen in the control experiment sections (see Methods). Cyt, cytoplasm; my, myofilaments; P, P facture face. Burs, 0.5 pm

The specificity of antipeptide antibodies used in this study was previously demonstrated and will be recalled later, in the Discussion. However, to confirm that in the biological material investigated these antibodies recognize specifically the CX 43-containing gap junctions, immunoelectron microscopy was performed on ultrathin sections of ventricle of one-week-old mouse. At this stage of development, immunoreactivity to anti-CX 43 antibodies is relatively abundant i n the ventricular myocardium (see Fig. 6) and freeze-cleave replicas of the tissue reveal numerous gap junctions of various sizes scattered across the fracture faces (Fig. la ). The results of immunocytochemistry experiments, illustrated in Fig. 1 b and Ic, show gold particles associated with gap junc-

12

Fig. 2. Cross-sections in the ventricle of 11 days post coitum mouse embryo myocardium. Double immunofluorescence experiments. a and b show a crosssection observed on the fluorescein isothiocyanate (FITC) channel for desmin detection (a) and on the tetra methyl rhodamine (TRITC) channel for CX 43 detection (b). c and d show another cross-section observed, at higher magnification, on the FITC channel (desmin detection, c) and on the TRITC channel (CX 43 detection. d). e is a phase contrast photograph of the section shown in c and d. Note, in a and c, the desniin expression gradient, on which the CX 43 expression gradient is superimposed (b and d). The striations characteristic of differentiated myofibrils are visible in the cells of the trabeculae (c). CX 43 is not detected in the cells that are desmin-negative (arrows in c, d and e). Tra, sub-endocardial trabeculae; Sep, subepicardial layers; Vc, ventricular cavity. Bars, 50 pm in a and b; 25 pm in c, d and e

tions, identified by their narrow extracellular space, whereas non-junctional membranes are free of labeling (arrowheads in Fig. lc). No labeling was seen in the control experiment sections treated as described in Methods (not shown). At 11 days post coitum (dpc), a gradient of expression of desmin is clearly visible in ventricle cross-sections (Fig. 2a). The most intense immunoreactivity to antidesmin antibodies is observed in the sub-endocardial trabeculae; it is much weaker in the sub-epicardial cell layers. The striations characteristic of differentiated myofibrils are seen only in the trabeculae (Fig. 2c). Not all the cells (sub-endocardial or sub-epicardial), however, express desmin (compare Figs. 2c and 2e). A gradient of expression of CX 43 is superimposed on the

desmin gradient (compare Figs. 2 a and 2 b ; 2c and 2d): immunoreactivity to anti-CX 43 antibodies is associated with the sub-endocardial trabeculae; it is not detected in the sub-epicardial layers. CX 43 was never observed to be associated with desmin-negative cells (compare Figs. 2c, 2d and 2e). The gradients of expression of desmin and CX 43 persist until birth, at which stage they disappear (compare Figs. 4 a and 4b, for the 16 dpc stage, with Figs. 5 a and 5b, for the newborn stage). Up to this stage, the desmin-negative cells are present in the ventricle wall (Figs. 4c and 4d for the 16 dpc stage) but their occurrence decreases a lot between 11 dpc and the newborn stage. At the early stages of development and at the newborn stage, the distribution of CX 43 is relatively homo-

13

Fig. 3. Longitudinal sections in the ventricle sub-endocardial layers of 14 days postcoitum mouse embryo myocardium. Double immunofluorescence experiments. a and b show a section observed on the FITC channel for desmin detection (a) and on the TRITC channel for CX 43 detection (b). c and d show another section observed, at higher magnification, on the TRITC channel (CX 43 detection, c) and on the FITC channel (detection of WGA receptor sites, d). The abundance of the striations, in a, due to the immunoreactivity of the Z lines, is evidence of myocyte differentiation. Note, in b, the relatively homogeneous distribution of CX 43 in the sub-endocardial layers. Arrowheads, in c and d, show that the anti-CX 43 antibody reactive sites are localized at the periphery of the cells. Note that CX 43 is not detected around all the cells (arrou~sin c and d). Burs, 50 pm in a and b; 25 pm in c and d

geneous in the sub-endocardial trabeculae (11, 14 and 16 dpc stages) and in the ventricle wall (newborn) (Figs. 3a and 3 b; 4a and 4b; 5a and 5 b). The immunoreactive sites to anti-CX 43 antibodies are always associated with desmin-positive cells (Figs. 4c, 4d and 4e; 5a and 5b); their abundance increases as development proceeds (compare Figs. 2 b, 3 b, 4 b and 5 b) (Table 1). Comparison of observations on the TRITC channel (anti-CX 43 antibody) and on the FITC channel (Con A and WGA) of the Same specimen, shows that the anti-^^ 43 antibody labeling coincides with that of the WGA or Con A receptor sites which mark the outer boundaries

Table 1. Stages

Density (sites/1000 pm2 & SD)

Days post coitum Days post coitunl D~~~post coitum Newborn

20+0.9 13.5k2.5 14.1 i 0 . 3 5 25.4 - 1.9

+

Density of connexin (cX)43 immunoreactive sites. The variance analysis shows that the differences of density between the stages are significant except between stages 14 and 16 days post coitum SD, standard deviation

14

Fig. 4. Cross-section in the ventricle of 16 days post coitum mouse embryo myocardium. Double immunofluorescence experiment. a and b show a cross-section observed on the FITC channel for desmin detection (a) and on the TRITC channel for CX 43 detection (b). c (phase contrast), d (desmin detection) and e (CX 43 detection) correspond to an enlarged part of the section shown in a and b. Note, in a and b, the expression gradients of desmin and CX 43, respectively. A r r o w in c, d and e indicate groups of desmin-negative cells that do not express CX 43. Tru. sub-endocardial trabeculae; Sep, external subepicardial layers; Vc, ventricular cavity. Bars, 50 pm in a and b ; 25 bm in c, d and e

of the cells (Figs. 3c and 3d). In contrast to the results obtained on cultured myocytes [62], no intracellular labeling was observed either at the embryo stages or later, during the post natal stages. After birth, the diameter and length of the cells increase [27, 281. The differentiation of the myocytes is marked by the appearance of intercalated discs which are detectable owing to their strong inimunoreactivity to anti-desmin antibodies (Figs. 6c and 7a). At the oneweek post natal stage, the immunoreactivity of CX 43

is much more intense than during the preceding stages (compare Fig. 6 a with Figs. 4e and 3c). The labeling still appears, however, in the form of small dots distributed in a relatively homogeneous fashion between the myocytes (Figs. 6 a and 6b). Strong labeling, in the form of short lines, is visible at the level of the few differentiated intercalated discs (Figs. 6 a and 6b). The progressive polarization which characterizes the distribution of CX 43 after the one-week post partum (1 wpp) stage, is illustrated in the set of Figs. 6d (2 wpp),

15

Fig. 5. Cross-section in the ventricle wall of newborn (20 days post coitum) mouse embryo myocardium. Double immunofluorescence experiment. a and b show a cross-section observed on the FITC channel for desmin detection (a) and on the TRITC channel for CX 43 detection (b). Note the absence of expression gradients of desmin and CX 43 (compare with the Figs. 4a and 4b). Vc, ventricular cavity. Burs, 50 bm

7b and 7c (3 wpp) and 8 (adult stage). At the 2 and 3 wpp stages, CX 43 is detected in all the intercalated discs (which are now differentiated at the two ends of all the myocytes), but also, quite frequently, in the lateral plasma membrane (Figs. 7c and 7d). At the adult stage, when the myocytes are fully differentiated, the localisation of CX 43 is exclusively limited to the intercalated discs (Fig. 8). No fluorescent labeling was observed, at any of the developmental stages, in the sections of control experiments (not shown).

Discussion The antibodies used in this study are directed against the peptide SAEQNRMGQ [15]. This sequence is conserved in its entirety in the CX 43 of rat [5], mouse [4, 541, chicken [51], Xenopus [21] and in human [17]. These antibodies are specific for CX 43. Proof of this specificity has been provided previously by immunoblotting experiments and immunogold labeling by electron microscopy [ 15, 201. These immunogold labeling experiments were carried out in vitro, on isolated rat heart gap junctions [15], and in situ, on sections of adult rat myocardium [ 151 and cultured mouse astrocytes [20]. These same antibodies were also used to immunopurify CX 43 from rat brain [14]. We have also shown by immunoblotting that these antibodies recognize only CX 43 in total extracts prepared from myocardium of mouse embryos and newborns as well as adult mouse [19]. The results of immunoelectron microscopy experiments carried out on ultrathin sections of ventricle of one-weekold newborn mouse (Fig. 1) confirm the specificity of these antibodies which are therefore quite suitable for analyzing by immunofluorescence the distribution of CX

43 and gap junctions during the differentiation of mouse ventricle myocytes. This investigation has raised four main points, which are discussed in the following paragraphs: 1) CX 43 is always associated with desmin-positive cells, that is, with differentiating or differentiated myocytes. However, in the early stages of cardiac ontogenesis, CX 43 is undetectable in cells which express desmin only weakly. We have detected CX43 in the sub-endocardial trabeculae from the 11 days post coitum stage onwards. Schaart et al. [64] have shown that in early postimplantation mouse embryos, desmin is first detected at 8.25 days in the ectoderm. Later, at 8.5 days post coitum, desmin is found exclusively in the heart rudiment. It is at this stage, in mouse, that the first beats occur [63, 681. The first striations of desmin, which are indicative of the appearance of the Z lines of myofibrils, are visible in the myocardium from 9.5 days post coitum [9]. Taking desmin as an early marker of differentiation, we have been able to demonstrate that large mesenchymous cells, which do not express CX 43, persist at least until birth in the ventricle wall. Therefore, the junctional channels, made of CX 43, would appear to exclusively link up the myocytes (desmin-positive cells), which are the only ventricular cells that contract in response to the passage of action potentials. 2) At the embryonic stages, there is a gradient of CX 43 expression in the ventricle wall which is superimposable on a less distinct gradient of desmin expression. The intensity of the immunoreactivity to the anti-CX 43 antibody is at its highest in the sub-endocardial trabeculae, and is at a minimum (or even undetectable, in the early stages) in the sub-epicardial cell layers. This CX 43 spatial distribution pattern has also been seen in the ventricular tissue of rat embryo myocardium [70].

16

Fig. 6. Longitudinal sections in the ventricle wall of one-week (1 wpp, a and b) and two-weekold (2 wpp, c and d) mouse myocardium. Double immunofluorescence experiments. a and b show a section, in the ventricle wall of a one-week-old mouse, observed on the TRITC channel for CX 43 detection (a) and on the FITC channel for Con A receptor sites detection (b). At the one week post natal stage (a and b) CX 43 labeling still appears mainly punctate, as for the earlier stages, and the immunoreactive sites are distributed in a relatively homogeneous fashion between the myocytes. This distribution pattern is visible in the cells indicated by asterisks in a and b. CX 43 immunoreactivity accumulates, however, at the level of the few rare intercalated discs that are differentiated at this stage (arrou~s in a and b). c and d show a section in the ventricular wall of a two-week-old mouse, observed on the FITC channel for desmin detection (c) and on the TRITC channel for CX 43 detection (d). At the two-weeks post natal stage (c and d), the intercalated discs are differentiated at both ends of a11 the myocytes (arrowheads in c) and CX 43 is localized both in the discs (arrowheads in d) and in the lateral membranes. Burs, 25 pm in a and b; SO pm in c and d

From a physiological point of view, this distribution pattern implies a slowing down (or even arrest) of conduction from the sub-endocardial trabeculae towards the sub-epicardial layers. There is however no experimental data to confirm this hypothesis. 3) Up to the one-week post partuni stage, the CX 43 expression sites appear as small dots, each probably corresponding to a gap junction. In later stages, the accumulation of junctions in the intercalated discs makes it difficult to distinguish the immunoreactive sites individually. Comparison of observations (Table 1) made at different embryonic development stages (11, 14, 16 days post coitum and newborn) shows that the abundance (or density) of the immunoreactive sites, and

therefore of the gap junctions, increases as development proceeds. These results are consistent with those obtained previously either by morphometric analysis of the gap junctions by electron microscopy, or by semi-quantitative estimation of CX 43 [19,23, 671. The physiological implications of this have been discussed elsewhere by Fromaget et al. [19]. 4) The differentiation of the myocytes, which is essentially the differentiation of the inyofibrils and the intercalated discs, is accompanied by the progressive cellular polarization of CX 43 expression. In the early stages of myocardium ontogenesis, CX 43 is distributed in a relatively homogeneous pattern in the myocyte membrane. During differentiation, the myocytes lengthen,

17

Fig. 7. Longitudinal sections in the ventricle wall of three-weekold mouse myocardium. Double immunofluorescence experiments. a and b show a section observed on the FITC channel (detection of desmin, a) and on the TRITC channel (detection of CX 43, b). c and d show another section observed on the TRITC channel (detection of CX 43, c) and on the FITC channel (detection of Con A receptor sites, d). The polarization of CX 43 that started during the preceding stage (two weeks post partum) is now more pronounced. CX 43 is localized in all the intercalated discs (arrowheads in a and b), but in certain myocytes, CX 43 is still present in the lateral membranes. Asterisks in c and d indicate a myocyte in which CX 43 is localized both in the intercalated discs and in the lateral plasma membrane (arrowheads). Stars in c and d indicate a myocyte in which localisation of CX 43 is restricted to the intercalated discs alone. Ears, 50 pm

and concomitantly, CX 43 expression is progressively restricted to the opposite ends of the cells and in the adjacent lateral regions. At the end of differentiation, CX 43 only persists in the intercalated discs. This highly polarized CX 43 distribution pattern is characteristic of adult ventricular myocytes. In contrast, the occurrence of CX 43 is frequent in the lateral membranes of the atrial myocytes [22]. Observations dealing with polarization demonstrate clearly the dynamics of the distribution of CX 43, and of the gap junctions of which it is the constituent, during myocytes differentiation. Studies carried out previously by electron microscopy, either on ultrathin sections [28], or by freeze-fracture [23, 671, failed to bring this phe-

nomenon to light. With the techniques used, it was possible to demonstrate the increase in the average size of the gap junctions during myocardium development, but they were not particularly well suited to investigation of their distribution patterns. Gap junctions can only form between two cells in areas where the extracellular space is sufficiently narrow to allow recognition of the connexins in the adjacent membranes. A priori, these areas can only correspond to or be close to cell-cell contacts that mediate the appropriate molecules, that is to say, cell-adhesion molecules (CAM). Several authors [31, 44, 47, 521 have shown that the transfection of communication-deficient cell lines with cDNAs encoding for Ca' +-dependent CAMS,

18

the intercalated discs [65]. These adhesion molecules might well play a similar role in the formation of gap junctions to that of the N- and E-cadherins. In the myocytes, the fasciae adherens and the desmosomes, adhesive junctions where the N-cadherin and the desmogleins respectively are expressed, are the anchorage sites of two elements of the cytoskeleton: actin (for the fasciae adherens) and desmin (for the desmosomes). It would thus appear that the formation of the cytoskeleton probably determines, through the intermediary of the Ca' +-dependent CAMs, the cell distribution of the gap junctions. The occurrence in the ventricle wall of the embryonic myocardium of a CX 43 expression gradient, superimposed on the desmin expression gradient, would seem to suggest a close interaction between the cytoskeleton and the communicating junctions. N-CAM is a Ca++-independentcell adhesion molecule. Its expression might modulate, as Ca' +-dependent CAMs do, the cellular interactions during heart organogenesis [74]. It has been shown that, in vitro, chronic treatment of chick neurectoderm with antibody Fab fragments against N-CAM interferes with the development of junctional communication [34], which suggests that N-CAM mediated adhesion may promote the formation of cell-to-cell channels. However, the reverse experiment to confirm the involvement of N-CAM expression in inducing junctional communication, has yet to be performed.

Fig. 8. Longitudinal section in the ventricle wall of adult mouse myocardium incubated with anti-CX 43 antibodies and TRITClabeled secondary antibodies. Localisation of CX 43 is restricted to the intercalated discs alone (compare with the Figs. 6a, 6d and 7c). Bur, 50 pm

Acknowledgements. We wish to thank M. Berthoumieux and J.P. Chauvin for their expert technical assistance as well as Dr J.P. Durbec (Centre d'Ockanologie, Luminy) for his help with the statistical analyses. This study was supported by INSERM (grant 88.50.09), the Fondation pour la Recherche Mtdicale and EEC (grant SC1-CT91-0644).

References such as chick L-CAM (corresponding to murine E-cadherin or uvomorulin), mouse N-cadherin (corresponding to chick A-CAM) or mouse E-cadherin, induced the formation of gap junctions (visible by electron microscopy or by immunofluorescence) and re-established junctional communication. In one case, at least (S 180 mouse sarcoma cells), transfection caused an alteration in the shape of cells [47]. In others words, the expression of C a + + dependent CAMs would appear to be a prerequisite for the assembly of connexin molecules [52] and the establishment of functional gap junctions. In this context, it should be pointed out that N-cadherin (or A-CAM) is expressed from the onset of cardiogenesis by the presumptive myocardial cells [12], and it is predominantly associated with the cell surfaces involved in cell-cell contact. The expression of this cadherin in the cardiac muscle is permanent [69] and in the differentiated myocytes it is exclusively localized in the fasciae adherens, which, like the gap junctions, are situated within the intercalated discs [71]. The desmogleins, which are the transmembrane constituents of the desmosomes, are members of the cadherin family of cell adhesion molecules [lo, 35, 531. In the differentiated myocytes, the desmogleins (and the desmosomes) are localized, like the N-cadherin, in

1. Bennett MVL, Barrio LC, Bargiello TA, Spray DC, Hertzberg E, Saez JC (1991) Gapjunctions: new tools, new answers, new questions. Neuron 6 : 305-320 2. Berryman MA, Rodewal R D (1990) An enhanced method for post-embedding immunocytochemical staining which preserves cell membranes. J Histochem Cytochem 38: 149-170 3. Beyer EC (1990) Molecular cloning and developmental expression of two chick embryo gap junction proteins. J Biol Chem 265: 14439-14443 4. Beyer EC, Steinberg TH (1991) Evidence the gap junction protein connexin-43 is ATP induced pore of mouse macrophage. J Biol Chem 266: 7971-7974 5. Beyer EC, Paul DL, Goodenough DA (1987) Connexin 43, a protein from rat heart homologous to gap junction protein from liver. J Cell Biol 195:2621-2629 6 . Beyer EC, Goodenough DA, Paul DL (1988) The connexins, a family of related gap junction proteins. Modern Cell Biol 71175-176 7. Beyer EC, Kistler J, Paul DL, Goodenough DA (1989) Antisera directed against connexin 43 peptides react with a 43-kD protein localized to gap junctions in myocardium and other tissues. J Cell Biol 108: 595--605 8. Beyer EC, Paul DL, Goodenough DA (1990) Connexin family of gap junction protein. J Membr Biol 116: 187-194 9. Bignami A, Dahl D (1984) Early appearance of desmin, the muscle-type intermediate filament protein in the rat embryo. J Histochem Cytochem 32:473-476

19 10. Collins JE, Legan KP, Kenny TP, MacGarvie J, Holton JL, Garrod DR ( I 991) Cloning and sequence analysis of desmosoma1 glycoproteins 2 and 3 (desmocollins) : cadherin-like desmosoma1 adhesion molecules with heterogeneous cytoplasmic domains. J Cell Biol 113:384-391 12. Debus E, Weber K, Osborn M (1983) Monoclonal antibodies to desmin, the muscle-specific intermediate filament protein. EMBO J 2:2305-2312 12. Duband JL, Volberg T, Sabanay I, Thitry JP, Geiger B (1988) Spatial and temporal distribution of the adherens-junction associated adhesion molecule A-CAM during avian morphogenesis. Development 103:325-344 13. Dupont E, El Aoumari A, Fromaget C, Briand JP, Gros D (1988) Immunological characterization of rat cardiac gap junctions. Presence of common antigenic determinant in heart of other vertebrate species and in various organs. J Membr Biol 104: 1 19--128 14. Dupont E, El Aoumari A, Fromaget C, Briand JP, Gros D (1991) Affinity purification of a rat brain junctional protein, connexin 43. Eur J Biochem 200:263-270 15. El Aoumari A, Fromaget C, Dupont E, Reggio H, Durbec P, Briand JP, Boller K, Kreitman B, Gros D (1990) Conservation of a cytoplasmic carboxyl-terminal domain of connexin 43, a gap junctional protein, in mammalian heart and brain. J Membr Biol 115 :229-240 16. El Aoumari A, Dupont E, Fromaget C, Jarry T, Briand JP, Kreitman B, Gros D (1991) Immunolocalization of an extracellular domain of connexin 43 in rat heart gap junctions. Eur J Cell Biol 56: 391-400 17. Fishman GI, Spray DC, Leinwand LA (1990) Molecular characterization and functional expression of the human cardiac gap junction channel. J Cell Biol 11 1 : 589-598 18. Fishman GI, Hertzberg EL, Spray DC, Leinwand LA (1991) Expression of connexin 43 in the developing rat heart. Circ Res 68 :782-787 19. Fromaget C, El Aoumari A, Dupont E, Briand JP, Gros D (1990) Changes in the expression of connexin 43, a cardiac gap junctional protein, during mouse heart development. J Mol Cell Cardiol 22: 1245-1258 20. Giaume C, Fromaget C, El Aoumari A, Cordier J, Glowinski J , Gros D (1991) Gap junctions in cultured astrocytes: singlechannel currents and characterization of channel-forming protein. Neuron 6: 133-143 21. Gimlich RL, Kumar NM, Gilula NM (1990) Differential regulation of the levels of three gap junction mRNAs in Xenopus embryos. J Cell Biol 110:597-605 22. Gourdie RG, Green CR, Severs NJ (1991) Gap junction distribution in adult mammalian myocardium revealed by an antipeptide antibody and laser scanning confocal microscopy. J Cell Sci 99:41-55 23. Gros D, Lee I, Challice CE (1982) Formation and growth of myocardial gap junctions: in vivo and in vitro studies. In: Bouman LN, Jongsma HJ (eds) Cardiac rate and rhythm. Physiological, morphological and developmental aspects. Martinus Nijhoff Publisher, The Hague, Netherlands, pp 243-264 24. Gros D, Bruce B, Kieda C, Delmotte F, Monsigny M, Schrkvel J (1983) Evolution of the surface of mouse myocardial cells during ontogenesis. 1 Changes of fluorescent lectin-binding sites. J Mol Cell Cardiol 15 :93-104 25. Guthrie SC, Gilula NB (1989) Gap junctional communication and development. Trends Neurosci 12: 12-16 26. Haefliger JA, Bruzzone R, Jenkins NA, Gilbert DJ, Copeland NG, Paul DL (1992) Four novel members of the connexin family of gap junction proteins. Molecular cloning, expression and chromosome mapping. J Biol Chem 267 :2057-2064 27. Hirakow R (1974) Myocardial cell growth and T-tubule formation during postnatal development of mouse. Acta Anat Nippoll 49 :7 4 7 5 28. Hirakow R, Gotoh T (1975) A quantitative ultrastructural study on the developing rat heart. In: Lieberman M, Sano

T (eds) Development and physiological correlates of cardiac muscle. Raven Press, New York, pp 37-49 29. Hoh JH, John SA, Revel JP (1991) Molecular cloning and characterization of a new member of the gap junction gene family, connexin 31. J Biol Chem 266:6524-6531 30. Hoyt RH, Cohen ML, Saffitz JE (1989) Distribution and threedimensional structure of intercellular junctions in canine myocardium. Circ Res 6: 563-574 31. Jongen WMF, Fitzgerald DJ, Asamoto M, Piccoli C, Slaga TJ, Gros D, Takeichi M, Yamasaki H (1991) Regulation of connexin 43-mediated gap junctional intercellular communication by C a + + in mouse epidermal cells is controlled by Ecadherin. J Cell Biol 114: 545-555 32. Kanter HL, Saffitz JE, Beyer EC (1992) Cardiac myocytes express multiple gap junction proteins. Circ Res 70:438-444 33. Kartenbeck J, Franke WW, Moser KG, Stoffels V (1983) Specific attachment of desmin filaments to desmosomal plaques in cardiac myocytes. EMBO J 2: 735-742 34. Keane RW, Mehta PP, Rose B, Honig LS, Loewenstein WR, Rutishauser V (1988) Neural differentiation, N-CAM-mediated adhesion and gap junctional communication in neuroectoderm. A study in vitro. J Cell Biol 106: 1307-1319 35. Koch PJ, Walsh MJ, Schmelz M, Goldschmidt MD, Zimbelmann R, Franke WW (1990) Identification of desmoglein, a constitutive desmosomal glycoprotein, as a member of the cadherin family of cell adhesion molecules. Eur J Cell Biol53 : 1-12 36. Kumar NM, Gilula NB (1986) Cloning and characterization of human and rat liver cDNAs coding for a gap junction protein. J Cell Biol 103: 767-776 37. Kuruc N, Franke WW (1988) Transient coexpression of desmin and cytokeratins 8 and 38 in developing myocardial cells of some vertebrates species. Differentiation 38 : 177-193 38. Laird DW, Revel JP (1990) Biochemical and immunological analysis of the arrangement of connexin 43 in rat heart gap junction membrane. J Cell Sci 97: 109-1 17 39. Laird DW, Puranam KL, Revel JP (1991) Turnover and phosphorylation dynamics of connexin 43 gap junction protein in cultured cardiac myocytes. Biochem J 273 :67-72 40. Lash JA, Crister ES, Pressler ML (1990) Cloning of a gap junctional protein from vascular smooth muscle and expression in two cell mouse embryos. J Biol Chem 265:13113-13117 41. Lau AF, Hatchpigott V, Crow DS (1991) Evidence that heart connexin 43 is a phosphoprotein. J Mol Cell Cardiol 23:659667 42. Loewenstein WR (1981) Junctional intercellular communication: the cell to cell membrane channel. Physiol Rev 61 :829-91 3 43. Luke RA, Beyer EC, Hoyt RH, Saffitz JE (1989) The quantitative analysis of intercellular connections by immunohistochemistry of the cardiac gap junction protein connexin 43. Circ Res 65:1450-1457 44. Matsuzaki F, Mege RM, Jaffe SH, Friedlander DR, Gallin WJ, Goldberg JI, Cunningham BA, Edelman GM (1990) cDNAs of cell adhesion molecules of different specificity induce changes in cell shape and border formation in cultured S 180 cells. J Cell Biol 110:1239-1252 45. Mazet F (1977) Freeze fracture studies of gap junctions in the developing and adult amphibian cardiac muscle. Dev Biol 601139-152 46. McLean IW, Nakane PK (1974) Periodate-lysine-paraformaldehyde fixative. A new fixative for immuno-electron microscopy. J Histochem Cytochem 22 : 1077-1 083 47. Mege RM, Matsuzaki F, Gallin WJ, Goldberg JI, Cunningham BA, Edelman G M (1988) Construction of epitheloid sheet by transfection of mouse sarcoma cells with cDNAs for chicken cell adhesion molecules. Proc Natl Acad Sci USA 85 :7274-7278 48. Mehta P, Bertram J, Loewenstein WR (1986) Growth inhibition of transformed cells correlates with junctional communication with normal cells. Cell 44: 187-196 49. Micevych PE, Abelson L (1991) Distribution ofmRNAs coding for liver and heart gap junction proteins in the rat central nervous system. J Comp Neurol305: 96-1 18

20 50. Mugnani E (1986) Cell functions of astrocytes, ependyma and related cells in the mammalian central nervous system with emphasis on the hypothesis of a generalized functional syncytium of supporting cell. In: Fedoroff S, Vernadakis A (eds) Astocytes, Vol. 1. Academic Press, New York, pp 329-371 51. Musil LS, Beyer EC, Goodenough DA (1990) Expression of the gap junction protein connexin 43 in embryonic chick lens: molecular cloning, ultrastructural localization and post-translational phosphorylation. J Membr Biol 116: 163-175 52. Musil LS, Cunningham BA, Edelman GM, Goodenough DA (1990) Differential phosphorylation of the gap junction protein connexin 43 in junctional communication-competent and deficient cell lines. J Cell Biol 111 :2077-2088 53. Nilles LA, Parry DAD, Powers EE, Angst BD, Wagner RW, Green KJ (1991) Structural analysis and expression of human desmoglein: a cadherin-like component of the desmosome. J Cell Sci 99 :809-821 54. Nishi M, Kumar NM, Gilula NB (1991) Developmental regulation of gap junction gene expression during mouse embryonic development. Dev Biol 146: 117-130 55. Ooesthoek PW, Gros D, Mijinders TAM, Wessels A, Vermeulen JLM, Moorman AFM, Lamers WH (1991) Distribution of gap junctions in the human neonatal heart. Cell Biol Int Reports 14 (Abs Suppl):22 56. Osinska HE, Lemanski LF (1 989) Immunofluorescent localization of desmin and vimentin in developing cardiac muscle of Syrian hamster. Anat Rec 223 :406-41 3 57. Page E, Manjunath CK (1986) Communicating junctions between cardiac cells. In: Fozzard HA, Haber E, Katz AM, Morgan HE (eds) The heart and cardiovascular system. Raven Press, New York, pp 573-600 58. Paul DL (1986) Molecular cloning of cDNA for rat liver gap junction protein. J Cell Biol 103: 123-1 34 59. Paul DL, Ebihara L, Takemoto LJ, Swenson KI, Goodenough DA (1991) Connexin 46, a novel lens gap junction protein, induces voltage-gated currents in non-junctional plasma membrane of Xenopus oocytes. J Cell Biol 115 : 1077-1089 60. Rook MB, Jongsma HJ, Jonge B de (1989) Single channel currents of homo- and heterologous gap junctions between cardiac fibroblasts and myocytes. Pflugers Arch 414: 95-98 61. Rook MB, Jonge B de, Jongsma HJ, Masson-Pkvet MA (1990) Gap junction formation and functional interaction between neonatal rat cardiocytes in culture: A correlative physiological and ultrastructural study. J Membr Biol 118:179-192 62. Rook MB, Van Ginneken ACG, Jonge B de, El Aoumari A, Gros D, Jongsma HJ (1992) Differences in gap junction chan-

nels between cardiac myocytes, fibroblasts and heterologous pairs. Am J Physiol (in press) 63. Rugh R (1990) The Mouse. Its reproduction and development. Oxford University Press, Oxford 64. Shaart G, Viebahn C, Langmann W, Ramaekers F (1989) Desmin and titin expression in early post implantation mouse embryo. Development 107: 585-596 65. Schmelz M, Duden R, Cowin P, Franke WW (1986) A constilutive transmembrane glycoprotein of Mr 165,000 (desmoglein) in epidermal and non-epidermal desmosomes. 11. Immunolocalization and microinjection studies. Eur J Cell Biol42: 184199 66. Severs NJ (1990) Review. The cardiac gap junction and intercalated disk. Int J Cardiol26: 137-173 67. Shibata Y, Nakata K , Page E (1980) Ultrastructural changes during development of gap junctions in rabbit left ventricular myocardial cells. J Ultrastruct Res 71 :258-271 68. Sissman N (1970) Developmental landmarks in cardiac morphogenesis comparative chronology. Am J Cardiol 25 : 141-148 69. Takeichi M (1988) The cadherins: cell - cell adhesion molecules controlling animal morphogenesis. Development 102 :639-655 70. Van Kempen MJA, Fromaget C, Gros D, Moorman AFM, Lamers WH (1991) Spatial distribution of connexin 43, the major cardiac gap junction protein, in the developing and adult rat heart. Circ Res 68:1638-1651 71. Volk T, Geiger B (1985) A-CAM: a 135-kD receptor of intercelMar adherens junctions. I. Immunoelectron microscopy localization and biochemical studies. J Cell Biol 103: 1441-1450 72. Willecke K, Henneman H, Dahl E, Jungbluth S, Heynkes R (1991) The diversity of connexin genes encoding gap junctional proteins. Eur J Cell Biol 56: 1-7 73. Willecke K, Heynkes R, Dahl E, Stutemkemper R, Henneman H, Jungbluth S, Suchyna T, Nicholson BJ (1991) Mouse connexin 37: cloning and functional expression of gap junction gene highly expressed in lung. J Cell Biol 114: 1049-1057 74. Wharton J, Gordon L, Walsh FS, Flanigan TP, Moore SE, Polak JM (1989) Neural cell adhesion molecule (N-CAM) expression during cardiac development in the rat. Brain Res 4831170-176 75. Yancey BJ, John SA, La1 R, Austin BJ, Revel JP (1989) The 43-kD polypeptide of heart gap junctions. Immunolocalization, topology and functional domains. J Cell Biol 108:2241-2254 76. Zhang JT, Nicholson BJ (1989) Sequence and tissue distribution of a second protcin of hepatic gap junction, CX 26, as deduced from its cDNA. J Cell Biol 109:3391-3401