Knock-out of the mouse RXRa gene was previously shown to result in a hypoplastic heart ventricular wall, histologically detectable in 12.5 dpc fetuses. We show here that a precocious differentiation can be detected as early as 8.5 dpc in ventricular cardiomyocytes of RXRα−/− mutants. This precocious differentiation, which is characterized by the presence of striated myofibrils, sarcoplasmic reticulum and intercalated disks, is found after 9.5 dpc in about 50% of RXRα−/− subepicardial myocytes. In contrast, wild-type subepicardial myocytes remain morphologically undifferentiated up to at least 16.5 dpc. A similar precocious differentiation was observed in 9.5 dpc subepicardial myocytes of several RXR0α−/− and RARα−/− mutants, as well as in vitamin A-deficient embryos. The proportion of differentiated subepicardial myocytes almost reached 100% in RXRα/RXRβ double null mutants, indicating a partial functional redundancy between RXRα and RXRβ. This differentiation defect was always paralleled by a decrease in the mitotic index. In addition, subepicardial myocytes of RXRα−/−, RXRα−/−1/RXRβ−/− or vitamin A deficient, but not of RXRβ−/− and RARα−/− embryos, were often flattened and more loosely connected to one another than those of WT embryos. Thus, retinoids are required at early stages of cardiac development to prevent differentiation, support cell proliferation and control the shape of ventricular myocytes, and both RXRs and RARs participate in the mediation of these functions.

Retinoic acid (RA), the major active metabolite of vitamin A, exerts its functions through two families of nuclear receptors. Retinoic acid receptors (RARot, P and y) are activated by all active forms of retinoic acid, whereas retinoid X receptors (RXRot, P and y) are activated specifically by 9-cis RA. RXRs form heterodimers with RARs, which are likely to be the actual functional units transducing the retinoid signal at the molecular level. In addition to their role in the action of RARs, RXRs may also function as homodimers, as well as within heterodimeric associations with a number of nuclear receptors, notably thyroid hormone receptors, vitamin D3 receptor, PPARs and several orphan receptors (for reviews, see Mangelsdorf and Evans, 1995; Chambon, 1996).

The respective roles of each RAR and RXR in the transduction of the developmental functions of retinoids has been addressed by engineering null mutations in each of these receptor genes (see Kastner et al., 1995 for a review). Studies of RAR and RXR single null mutants, RAR double mutants and RXR/RAR compound mutants have led to important conclusions. Firstly, as all the defects of the fetal vitamin A deficiency (VAD) syndrome were recapitulated in RAR single or double mutants, RARs are clearly required for the mediation of the known vitamin A functions during development (Mendelsohn et al., 1994; Lohnes et al., 1994; Luo et al., 1996; Ghyselinck et al., 1997). Secondly, RXRa null mutants exhibited some cardiac and ocular defects belonging to the fetal VAD syndrome (Sucov et al., 1994; Kastner et al., 1994), and furthermore a strong synergy was observed between the effects of the RXRα and RAR mutations, as RXRα/RAR compound mutants reproduced most of the defects occurring in RAR double mutants and VAD fetuses (Kastner et al., 1994, 1997). Therefore, RXRs are also involved in the mediation of the retinoid signal during development, most probably as heterodimeric partners for RARs.

All RXRα−/− mutants die in utero and, beside ocular defects, display a hypoplastic ventricular wall (Sucov et al., 1994; Kastner et al., 1994; Dyson et al., 1995). At the histological level, the severity of this cardiac defect is variable, which most probably accounts for the large time window during which death of RXRα−/− mutants occurs (11.5–17.5 days post-coitum (dpc); Kastner et al., 1994, and our unpublished results). The normal embryonic myocardium is subdivided into a peripheral compact zone and an inner trabecular zone. As one moves from the periphery towards the lumen of the ventricle, the rate of cell division decreases, whilst morphological features indicative of cardiomyocyte differentiation become more prominent (Rumyantsev, 1977). The rapid expansion of the compact layer permits an harmonious adjustment of the cardiac function to the body size increase. We have previously shown that, at 14.5 dpc, the RXRα−/− compact zone myocytes display aberrant differentiation features, in that some of these cells contained well organised striated myofibrils resembling those normally found in trabecular myocytes (Kastner et al., 1994). These observations suggested that the mutant myocytes of the compact zone failed to remain in their ‘normal’, relatively undifferentiated, state. To gain further insight into the timing of appearance of this precocious differentiation and to investigate the possibility that the variable severity of the RXRα−/− cardiac phenotype could be due to a partial functional compensation by RXRβ, we have now examined the ventricular myocytes at earlier stages of development, both in RXRα single, RXRβ single or RXRα/RXRβ double mutants. Furthermore, in order to correlate this differentiation defect to a possible function of vitamin A, we have also examined ventricular myocytes from VAD embryos and RAR mutant embryos.

Mice and embryos

All single mutant mice lines employed in this study have been described previously (Kastner et al., 1994, 1996; Lohnes et al., 1993; Lufkin et al., 1993; Ghyselinck et al., 1997). RXRα/RXRβ double null embryos were obtained from the mating of RXRα+/−/RXRβ+/− males with RXRα+/−/RXRβ+/− or RXRα+/−/RXRβ−/− females. Genotypings were usually performed on DNA extracted from yolk sac by Southern blotting, as described previously. The RXRα, RXRβ and RARα genotypes could also be determined by PCR (primers and PCR conditions available upon request).

Histology, electron microscopy and quantification of differentiated cells

Embryos and fetuses were fixed in 2.5% gluteraldehyde in 0.1 M sodium cacodylate buffer (pH 7.2) for 24 hours at 4°C, washed in 0.1 M cacodylate buffer for 30 minutes and post-fixed in 1% osmium tetroxide in 0.1 M cacodylate buffer for 1 hour at 4°C. Following stepwise dehydration with increasing concentrations of ethanol and embedding in Epon 812, semi-thin sections (7.5 pm thick) were stained with toluidine blue for light microscopy. Ultrathin sections were contrasted with uranyl acetate and lead citrate and observed with a Phillips 208 electron microscope. For all observations, care was taken to consider sections that correspond to comparable positions within the heart. Estimations of the percentage of differentiated cells were performed by scoring at least 100 cells. Scorings of different grids from the same embryo showed that the variations in the percentages of differentiated cells were within 10%.

Mitotic index

Mitotic indexes were estimated on semi-thin sections by scoring the number of mitotic events (metaphase and anaphase) in subepicardial cells. For each embryo, a minimum of 200 cells were analysed, by counting 2–3 sections, each containing about 80–100 subepicardial cells (to avoid counting the same cell twice, only one out of four consecutive sections was examined). Note that, for a given embryo, there was very little variation between the percentages of mitotic cells counted from each individual section (±1%). The error margin of the mitotic indexes obtained for each mutant is thus likely to be within 1%.

In situ hybridisation

The MLC2a probe was a 285 bp fragment recloned by RT-PCR, spanning nucleotides 128-413 (Kubalak et al., 1994). The MLC2v probe was a gift from K. Chien. Probe labeling, in situ hybridization and emulsion autoradiography were performed as described (Ruberte et al., 1990).

Differentiation features in wild-type, RXRα+/− and RXRα−/− ventricular myocytes from 8.5 dpc to 16.5 dpc

The differentiation state of cardiomyocytes was assessed at the ultrastructural level by the presence of structures involved in muscle contraction. Individual actin and myosin myofilaments first associate into bundles of myofilaments (F, Fig. 1c), which then form sarcomeres, the smallest contractile unit of cardiac myofibrils. Morphologically, a sarcomere is delimited by two successive Z-lines (Z, see Fig. 1b,d,e). At cellular junctions, myofibrils from two adjacent myocytes are connected by intercalated disks (ID, Fig. 1e). The sarcoplasmic reticulum (SR, Fig. 1e), a kind of smooth endoplasmic reticulum associated with myofibrils, is involved in the propagation of myofibril contraction by releasing Ca2+ ions (Krstic, 1984, and refs therein). These ultrastructural features were used to score the differentiation status of cardiac myocytes in wild type (WT) and RXR mutant embryos.

Fig. 1.

Ultrastructural features of ventricular myocytes of WT (a,c), RXRα+/ − (d) and RXRα−/− (b,e) embryos at 8.5 dpc. c and e are enlargements of the boxed areas in a and b, respectively. F, myofilaments; ID, intercalated disk; M, myocyte; N, nucleus; S, sarcomere; SR, sarcoplasmic reticulum; Z, Z-line. Magnification: 3800× (a,b); 24000× (c–e).

Fig. 1.

Ultrastructural features of ventricular myocytes of WT (a,c), RXRα+/ − (d) and RXRα−/− (b,e) embryos at 8.5 dpc. c and e are enlargements of the boxed areas in a and b, respectively. F, myofilaments; ID, intercalated disk; M, myocyte; N, nucleus; S, sarcomere; SR, sarcoplasmic reticulum; Z, Z-line. Magnification: 3800× (a,b); 24000× (c–e).

In WT embryos at 8.5 dpc (3–5 somites), the ventricular myocardium consists of 1–2 layers of myocytes, which usually exhibit a cuboidal shape (Figs 1a, 2a). Only a few of these myocytes contained bundles of myofilaments (F, Fig. 1c) and Z-lines were never seen. Note that the 8.5 dpc heart lacks an epicardium. At all subsequent stages from 9.5 dpc to 16.5 dpc, the vast majority of WT myocytes located immediately beneath the epicardium (the subepicardial myocytes, SMs) remained relatively undifferentiated; they displayed numerous short bundles of myofilaments scattered throughout their cytoplasm and occasional Z-lines (Table 1), but only rarely contained sarcomeres, intercalated disks or sarcoplasmic reticulum. Interestingly, the proportion of sarcomere-containing SMs in WT embryos appears to vary depending on the genetic background, being totally absent in WT pure 129/Sv embryos (derived from RXRα−/− intercrosses; Table 2), while present at a low frequency in WT embryos from a mixed 129/Sv/C57Bl/6 genetic background (derived from RARα+/− and RARβ+/− intercrosses; see below and Table 3). Note that there is a gradual increase in the degree of myocyte differentiation as one moves from the subepicardial cell layer to the trabecular region (Fig. 3, 16.5 dpc WT fetus). Given the constancy of this absence of differentiation in the subepicardial cell layer in WT embryos derived from RXRα+/− intercrosses at all stages examined (see Tables 1 and 2), we focused on this cell layer to assess the differentiation status of RXRa mutant ventricular myocytes.

Table 1.

Differentiation status of subepicardial, atrial and trabecular myocytes in RXRα mutants

Differentiation status of subepicardial, atrial and trabecular myocytes in RXRα mutants
Differentiation status of subepicardial, atrial and trabecular myocytes in RXRα mutants
Table 2.

Sacomeres and mitoses in subepicardial cells of 9.5 dpc RXR mutant embryos

Sacomeres and mitoses in subepicardial cells of 9.5 dpc RXR mutant embryos
Sacomeres and mitoses in subepicardial cells of 9.5 dpc RXR mutant embryos
Table 3.

Sacomeres and mitoses in vitamin A-deficient (VAD) and RAR mutant 9.5 dpc embryos

Sacomeres and mitoses in vitamin A-deficient (VAD) and RAR mutant 9.5 dpc embryos
Sacomeres and mitoses in vitamin A-deficient (VAD) and RAR mutant 9.5 dpc embryos

The RXRα+/− ventricular myocardium was indistinguishable from its WT counterparts at the histological level (not shown). However, the differentiation status of ventricular myocytes (at 8.5 dpc) or subepicardial myocytes (from 9.5 dpc onwards) clearly differed in mutant and WT embryos. At 8.5 dpc, most of the RXRα+/− myocytes already contained bundles of myofilaments, and isolated Z-lines and even sarcomeres were seen in some of them (Fig. 1d, Table 1). Note that at this stage the pro-portion of RXRα+/− cardiomyocytes containing Z-lines and/or sarcomeres varied greatly from embryo to embryo, ranging from 4% to 60%. Sarcomeres, intercalated disks and sarcoplasmic reticulum were visible in RXRα+/− SMs from 9.5 dpc onwards (Tables 1 and 2). However, more than two consecutive Z-lines were never observed in these myocytes.

At all stages examined, the RXRα−/− ventricular myocardium appeared thinner than its WT and RXRα−/− counterparts, and often formed a single layer of abnormal flattened and loosely associated cells [compare Fig. 2a,b (8.5 dpc) and c,e (9.5 dpc)]. At 8.5 dpc, bundles of myofilaments were identified in all ventricular myocytes. About 50% of these cells contained striated myofibrils, displaying on average 3–4 consecutive sarcomeres (Fig. 1b,e) as well as a sarcoplasmic reticulum, and these differentiated myocytes were connected by intercalated disks (Fig. 1e). All these features persisted in RXRα−/− SMs at subsequent developmental stages (Tables 1 and 2). Note that, as dying cells were seen very infrequently in both WT and mutant myocardia at all stages analysed, increased cell death is unlikely to be a major cause leading to the thinning of the ventricular wall in RXRα mutants.

Fig. 2.

Semi-thin sections through the ventricular wall of WT and RXR mutant embryos at 8.5 dpc (a,b) or 9.5 dpc (c-f). E, epicardium; EN, endocardium; M, myocyte; SM, subepicardial myocyte; T, trabeculae. Magnification: 370×.

Fig. 2.

Semi-thin sections through the ventricular wall of WT and RXR mutant embryos at 8.5 dpc (a,b) or 9.5 dpc (c-f). E, epicardium; EN, endocardium; M, myocyte; SM, subepicardial myocyte; T, trabeculae. Magnification: 370×.

Fig. 3.

Differentiation gradient of the myocytes through the myocardial wall of a WT fetus at 16.5 dpc. Brackets indicate sarcomeres, and the large arrow points towards the lumen of the ventricle. C, capillary, E, epicardium, N, nucleus, SM, subepicardial myocyte. Magnification: 5300×.

Fig. 3.

Differentiation gradient of the myocytes through the myocardial wall of a WT fetus at 16.5 dpc. Brackets indicate sarcomeres, and the large arrow points towards the lumen of the ventricle. C, capillary, E, epicardium, N, nucleus, SM, subepicardial myocyte. Magnification: 5300×.

Trabeculae were present in most 9.5 dpc WT and mutant ventricles, but they were, on average, thinner in RXRα−/− mutants (not shown). Both WT and mutant trabecular myocytes always displayed numerous sarcomeres and intercalated disks, and abundant sarcoplasmic reticulum (Table 1, and data not shown).

Early morphological differentiation features in RXRp mutant ventricular myocytes

To investigate whether the cardiac defects reflect specific functions of RXRoi, we analysed the hearts of six 9.5 dpc RXRβ−/− mutants. At the histological level, their myocardia appeared normal, with globular and tightly associated SMs (Fig. 2d; compare with Fig. 2c). In contrast, electron microscopic examination revealed the presence of well organised striated myofibrils, sarcoplasmic reticulum and intercalated disks in SMs of four out of six 9.5 dpc RXRβ−/− mutants examined (Fig. 4; Table 2). The remaining two 9.5 dpc RXRβ−/− mutants were affected to a lesser degree, as they displayed only isolated sarcomeres in a restricted population of SMs (embryos 13 and 16 in Table 2). Therefore, either RXRα or RXRβ deficiency can affect the differentiation status of subepicardial ventricular myocytes. However, there are some interesting differences between the RXRα and RXRβ mutant phenotypes: (1) the abnormal flattened aspect and the loose association between myocytes was seen only in RXRα null mutants (compare Fig. 2d with e); thus this alteration of cell shape can be dissociated from an abnormal presence of sarcomeres (note that the RXRβ mutant illustrated in Fig. 2d was as affected as RXRα null mutants with respect to the percentage of SMs containing sarcomeres); (2) individual RXRβ−/− embryos were much more variably affected than RXRβ−/− embryos, with phenotypes varying from very mild to as severe as that of RXRβ−/− embryos.

Fig. 4.

Ultrastructural aspect of 9.5 dpc WT and RXRβ−/− subepicardial myocytes. Note the presence of well developed myofibrils in the RXRP mutant and their absence in the WT myocyte. Open arrows point away from the myocardium. ID, intercalated disk; N, nucleus; S, sarcomere. Magnification: 5700×.

Fig. 4.

Ultrastructural aspect of 9.5 dpc WT and RXRβ−/− subepicardial myocytes. Note the presence of well developed myofibrils in the RXRP mutant and their absence in the WT myocyte. Open arrows point away from the myocardium. ID, intercalated disk; N, nucleus; S, sarcomere. Magnification: 5700×.

Increased proportion of differentiated subepicardial myocytes in RXRα/RXRβ double mutants

The incomplete penetrance, at the cellular level, of the precocious differentiation phenotype seen in either RXRα−/− or RXRβ−/− myocardia could be due to a partial functional redundancy between these receptors. We therefore examined RXRα−/−/RXRβ−/− double mutants. Many of the double mutant embryos were growth-retarded and displayed severe malformations at 9.5 dpc, and most of them died between 9.5 and 10.5 dpc (our unpublished results). The myocardial wall of these growth-retarded mutants was extremely thin and consisted almost exclusively of elongated and loosely connected cells (Fig. 2f). Nearly all these cells displayed sar-comeres, intercalated disks and sarcoplasmic reticulum (Table 2, and data not shown). Even though the thinness of the ventricular wall and the absence of trabeculae seen in these double mutants may be related to their developmental retardation, the elongated shapes and differentiated state of their myocytes has no equivalent in WT hearts at earlier developmental stages (compare Fig. 2f with Fig. 1a). Some RXRα−/−/RXRβ−/− mutants were developmentally less affected, of similar size as their littermates, and survived until 10.5 dpc. Morphologically, the myocardia of the double mutants belonging to this second category were similar to those of RXRα−/− mutants (data not shown). Moreover, the percentage of SMs exhibiting differentiated features in these less affected mutants (examined at 10.5 dpc) was also similar to that of RXRα−/− mutants (not shown). In any event, the severe abnormal cardiac phenotype seen in several RXRα−/−/RXRβ−/− double mutants clearly underscores the essential role of RXRs in preventing an early differentiation of ventricular myocytes, and also indicates that RXRα and RXRβ are partially redundant. It is also noteworthy that whenever a SM was differentiated, its degree of differentiation (as judged from myofibrillar organisation) was similar in single RXRα−/−, RXRβ−/− and RXRα−/−/RXRβ−/− mutants and never reached the differentiation degree seen in atrial and trabecular cells.

At 9.5 dpc, RXRα−/−/RXRβ+/ − embryos were externally indistinguishable from their RXRα−/− or WT littermates. Their ventricles were histologically indistinguishable from those of RXRα−/− mutants and the proportion of SMs exhibiting differentiated features was similar to that of RXRα−/− mutants (Table 2). Interestingly, one 9.5 dpc RXRα+/−/RXRβ−/− mutant displayed features that were essentially similar to those seen in RXRα+/− or mildly affected RXRβ−/− embryos, in that isolated sarcomeres were occasionally detected in approximately 30% of the cells (Table 2). Thus, the combination of RXRα het-erozygocity and RXRβ homozygocity does not result in an aggravation of the cardiac phenotype associated with these two genotypes (note that RXRα+/ −/RXRβ−/− mutants are viable).

Reduced proliferation of subepicardial cardiac myocytes in RXR mutants

Cell proliferation in 9.5 dpc SMs was evaluated by counting cells in metaphase and anaphase (Table 2). In WT myocardia, a high proportion (about 16%) of SMs were engaged into mitosis. The fraction of cells in mitosis was markedly reduced in RXRα−/−/RXRβ−/− mutants, as well as in RXRα−/− mutants. Interestingly, in the case of RXRβ−/− mutants, there was an apparent correlation between a reduced mitotic index and the presence of organised myofibrils (Table 2; compare embryos 13 and 16 with embryos 14 and 15). These observations suggest that a decrease in the rate of cell proliferation is associated with the precocious differentiation of myocardial cells in RXRα and β mutants.

Correct ventricular specification in RXR mutants

Morphological features of differentiation are seen in the cells localised at the posterior part of the heart tube, the presumptive atrium, in 5-to 7-somites embryos (our unpublished result). Thus, the question arises as to whether the precocious differentiation of RXR mutant ventricular cardiomyocytes could reflect an aberrant, atrial-like specification. In WT embryos, the myosin light chain 2v (MLC2v) exhibits a ventricular specificity from very early embryonic stages (e.g. 9.5 dpc, Kubalak et al., 1994). This gene was correctly expressed in the ventricles of 2 RXRα−/−/RXRβ−/− double mutants (compare Fig. 5d and f; data not shown), demonstrating that ventricular specification occurs correctly in these mutants. We also examined the expression of the MLC2a gene at 14.5 dpc in RXRα−/− mutants. This gene, which is first expressed in both atria and ventricles, was reported by Kubalak et al. (1994) to be specifically down-regulated in ventricles by 11.5 dpc. We found, however, that MLC2a expression, even though higher in the atria, still persisted in WT ventricles at 14.5 dpc. Moreover, we observed a consistently higher expression in the left atrium when compared to the right atrium (Fig. 5a). The differential expression of this gene can therefore be used to assess ventricular, as well as left versus right atrium, specifications. No difference in the pattern of expression of this gene could be observed between RXRα−/− mutant and WT hearts at 14.5 dpc (Fig. 5a,b), thus suggesting that the ventricular and atrial identities have been correctly specified in these mutants.

Fig. 5.

Expression of MLC2a and MLC2v transcripts in RXR mutant hearts. (a,b) Expression of MLC2a transcripts in WT and RXRα−/− hearts at 14.5 dpc. The colours indicate different intensities of signal, with blue, yellow and red corresponding respectively to low, intermediate and high levels of signal. Note the similarity of the expression pattern in WT and mutant fetuses, both with respect to the low ventricular expression and the higher expression in left versus right atrium. (c-f) expression of MLC2v at 9.5 dpc in WT and RXRα−/−/RXRβ−/− embryos. (c,e) Bright field exposures; (d,f) the corresponding dark fields exposures. A, atrium; C, conotruncus; LA, left atrium; LV, left ventricle; RA, right atrium; RV, right ventricle; V, ventricle. Magnification: 17× (a,b); 29× (c–f).

Fig. 5.

Expression of MLC2a and MLC2v transcripts in RXR mutant hearts. (a,b) Expression of MLC2a transcripts in WT and RXRα−/− hearts at 14.5 dpc. The colours indicate different intensities of signal, with blue, yellow and red corresponding respectively to low, intermediate and high levels of signal. Note the similarity of the expression pattern in WT and mutant fetuses, both with respect to the low ventricular expression and the higher expression in left versus right atrium. (c-f) expression of MLC2v at 9.5 dpc in WT and RXRα−/−/RXRβ−/− embryos. (c,e) Bright field exposures; (d,f) the corresponding dark fields exposures. A, atrium; C, conotruncus; LA, left atrium; LV, left ventricle; RA, right atrium; RV, right ventricle; V, ventricle. Magnification: 17× (a,b); 29× (c–f).

Vitamin A deficiency and RARa null mutation lead to an early differentiation of subepicardial myocytes

Fetuses from vitamin A deficient (VAD) rat dams exhibit a ‘spongy’ ventricular myocardium similar to that of RXRα−/− mutants (Wilson and Warkany, 1949). To determine whether the correspondence between the two phenotypes also holds for the early differentiation and proliferation defects, vitamin A-deficient embryos were generated by injection of antisense oligonucleotide for retinol binding protein into the yolk sac at 7.5 dpc, followed by a 48-hour in vitro culture (Bavik et al., 1996; control embryos were injected with the complementary sense oligonucleotide). Remarkably, the SMs of VAD embryos displayed the same advanced differentiation features of those of RXR mutant embryos (Fig. 6d and Table 3). In addition, the VAD SMs were often elongated and loosely connected to one another, again resembling those of RXRα−/− and RXRα−/−/RXRβ−/− mutants (Fig. 6b). The mitotic index of the VAD subepicardial cells (6%) was also lower than in control embryos (11%; see Table 3). These observations therefore strongly support the idea that RXRs are involved in the transduction of an early vitamin A function necessary to prevent a precocious differentiation of ventricular myocytes, to maintain a high rate of cell proliferation and to control myocyte shape and cohesiveness.

Fig. 6.

Effect of vitamin A deficiency (VAD) on the myocardial wall at 9.5 dpc. Semi-thin section through the myocardial wall (a,b) and ultrastructural aspect of subepicardial myocytes (c,d) of control and VAD embryos. E, epicardium; F, myofilament; MI, mitochondria; N, nucleus; S, sarcomere; SM, subepicardial myocyte; Z, Z-line. Magnification: 500× (a,b); 17000× (c–d).

Fig. 6.

Effect of vitamin A deficiency (VAD) on the myocardial wall at 9.5 dpc. Semi-thin section through the myocardial wall (a,b) and ultrastructural aspect of subepicardial myocytes (c,d) of control and VAD embryos. E, epicardium; F, myofilament; MI, mitochondria; N, nucleus; S, sarcomere; SM, subepicardial myocyte; Z, Z-line. Magnification: 500× (a,b); 17000× (c–d).

RXRs can mediate a vitamin A function either by exerting a 9-cis RA-dependent transactivation, or by providing a heterodimeric partnership to a RAR (or both). As we had previously observed a spongy myocardium in some RARα/RARγ double null mutants (Mendelsohn et al., 1994), RARs as well as RXRs could be involved in mediating the function of vitamin A in controlling the development of ventricular myocytes. We thus examined SMs of 9.5 dpc RARα−/−, RARβ−/− and RARγ−/− mutants, as well as control littermates.

Remarkably, all four RARα−/− mutants examined exhibited a high proportion of SMs containing sarcomeres (Table 3; note that almost all the SMs of mutant no. 11 were differentiated). In addition, the percentage of mitotic SMs was clearly reduced in all RARa mutants when compared to their heterozygote or WT littermates. However, abnormalities of cell shape and cohesiveness were not detected in these RARa mutants (data not shown).

About one third of RARβ null mutant SMs also exhibited differentiated features, but this proportion was only twofold higher than in their WT littermates (Table 3). The ultrastructure of RARγ−/− ventricular myocytes was normal (Table 3). Therefore, among RARs, RARα appears to play a major role in the control of the differentiation and proliferation of ventricular myocytes.

Early function of retinoids in the control of ventricular myocyte differentiation and proliferation

Vitamin A deficiency (Wilson and Warkany, 1949), inactivation of both RARα and RARγ (Mendelsohn et al., 1994) and inactivation of RXRα (Sucov et al., 1994; Kastner et al., 1994), were all previously shown to result in a hypoplastic ventricular wall. However, these studies were performed at late stages of heart development, and did not reveal the origin of this defect. The present study demonstrates that ventricular myocytes of RXR mutants (RXRα−/−, RXRβ−/− and RXRα−/−/RXRβ−/−), RAR mutants (RARα−/− and RARβ−/−) and VAD embryos all exhibit a premature differentiation of their ventricular myocytes, which can be visualized as early as 8.5 dpc, as well as a reduced rate of proliferation. In addition, RXRα−/−, RXRα−/−/RXRβ−/− and VAD ventricular myocytes appeared elongated instead of globular and loosely associated. Vitamin A is therefore required from early stages of cardiogenesis to prevent differentiation, maintain a high rate of cell proliferation, and control cell shape and cohesiveness of ventricular myocytes. Furthermore, both RXRs and RARs are involved in the mediation of these retinoid functions, most probably acting as heterodimers (Kastner et al., 1997).

Many cellular events are affected in the mutants (sarcomere, intercalated disk or sarcoplasmic reticulum assembly, cell cycle control, control of cell shape and control of cell-cell adhesion). As already pointed out, the precocious differentiation was always associated with a decreased mitotic index, suggesting that a common mechanism might regulate both differentiation and proliferation. On the other hand, abnormal differentiation and proliferation were not always accompanied by cell shape abnormalities, as this latter defect was seen only in mutants lacking RXRα (or both RXRα and RXRβ) and not in RXRβ or RARα single mutants, even though these mutants were often as affected as RXRα mutants in terms of proliferation and differentiation. Thus, the retinoid control of ventricular myocyte differentiation and proliferation on the one hand, and of cell shape and/or cell-cell association on the other one, might occur via distinct mechanisms.

To what extent do these early abnormalities of RXR and RAR single mutants lead to cardiac defects at later stages? Histologically, none of the RXRβ−/−, RARα−/− and RARβ−/− mutant fetuses analysed at 14.5 dpc or 18.5 dpc exhibited a spongy myocardium similar to that of RXRα−/− or VAD fetuses (our unpublished observations). Whether the reduced cellular proliferation observed in these mutants at 9.5 dpc is subsequently compensated for, or results in a decreased myocardial mass at later stages, cannot easily be assessed from examination of histological sections, because even normal hearts exhibit variations in the apparent thickness of their compact layer, due to variations in their contraction status at the time of fixation. In any event, these observations indicate that the differentiation and proliferation defects alone (which are at least as pronounced in RARα−/− as in RXRα−/− mutants) are not sufficient to lead to a RXRα−/−-like spongy myocardium. This phenotype may thus result, at least in part, from the abnormalities in cell shape and cohesivness. However, as a significant fraction of RXRβ and RARα null mutants die perinatally (Kastner et al., 1996; Lufkin et al., 1993), it will be of interest to investigate whether cardiac dysfunction may cause the death of some of these mutants.

Does the differentiation gradient across the myocardium involve a local production of RA in pericardial tissues?

The observation that the inhibitory effect of RA on differentiation is manifested only in the most peripheral layers of myocytes suggests that RA may be produced at the external periphery of the myocardium, possibly by the epicardium. This is supported by the presence of transcripts for retinaldehyde dehydrogenase 2 (an enzyme which is possibly involved in RA synthesis; Zhao et al., 1996), in the pericardium and their absence in the myocardium (Niederreither et al., 1997). In addition, the strong expression of a LacZ transgene driven by the RA-responsive RARβ2 promoter in the compact layer myocytes and its weaker expression in the trabeculae is also in agreement with this idea (Zimmer et al., 1994; our unpublished results). Together, these observations are consistent with the notion that a gradient of RA emanating from pericardial tissues might be invoved in establishing the gradient of myocyte differentiation accross the myocardium. Note however, that the epicardium is unlikely to correspond to that putative RA source since (1) the epicardium is not formed at 8.5 dpc (Viragh and Challice, 1981), a stage when RXRα−/− myocytes are already differentiated, and (2) the myocardial compact layer is apparently normal in mice embryos lacking integrin a4, in which the epicardium is absent.

That mutant ventricular myocytes express early differentiated features may not be too surprising, as WT myocytes located in the presumptive atrium and bulbus cordis are already morphologically differentiated at 8.5 dpc (data not shown). Thus, the absence of differentiation in 8.5-dpc WT ventricular myocytes, or in WT SMs at later developmental stages, might require a retinoid-induced inhibition of the molecular machinery triggering their differentiation. The RXR mutants show the physiological importance of such an inhibitory mechanism, as a delayed differentiation may favor growth and thus allow the production of an adequate mass of tissue. Could RA play a similar role in other systems that require a balance between proliferation and differentiation? The developing limb is a particularly interesting candidate, since RA may prevent limb blastema cells from being engaged into a chondrogenic program (Jiang et al., 1995; Paulsen and Solursh, 1988). Along the same vein, a gradient of RA is also thought to regulate proliferation and differentiation in the skin (see Darmon, 1991, for a review).

Do RXR mutant ventricular cells adopt an atrial phenotype?

Since WT atrial myocytes differentiate at early developmental stages, the precocious differentiation of RXR mutant ventricular myocytes might reflect an abnormal atrial specification, as suggested by Dyson et al. (1995); these authors reported, in 13.5 dpc RXRα−/− mutants, an abnormal persistence of ventricular expression of the MLC2a gene, which is highly expressed in atrial myocytes. Conversely, atrial/ventricular specifications may occur correctly, but the ventricular-specific differentiation program per se may be perturbed. Our data support this second possibility since: (1) the mutant ventricular cells never reached the degree of myofibril organisation displayed by atrial cells, and are therefore qualitatively different from atrial cells; (2) the ventricular-specific marker MLC2v is correctly expressed in ventricular myocytes of RXRα−/−/RXRβ−/− double mutants; (3) the differential expression of MLC2a between the atrium and ventricle is normal in our WT and RXRα mutant 14.5 dpc fetuses. The discrepancy between our observation and that of Dyson et al. (1995) could be due to the very weak ISH signals detected in their WT sample, which could have prevented the detection of a low level of ventricular expression. In fact, their RXRα−/− MLC2a expression pattern is very similar to ours, in that it clearly exhibits a weaker ventricular than atrial expression. Taken together, our observations do not support the possibility that the RXR mutant ventricles could have adopted an atrial identity, and lead to the conclusion that RXRs are acting on the realisation of a ventricular-specific differentiation program.

Specific or redundant functions for distinct retinoid receptors in the developing myocardium?

The present results indicate that at least four distinct retinoid receptors (RXRα, RXRβ, RARα and RARβ) are apparently involved in controling the development of the ventricular myocardium. This raises questions about functional specificity and redundancy of these receptors.

RXRα clearly has a key role in myocardial development, since among the single mutants analysed, RXRα−/− mutants most consistently reproduced the VAD phenotype. In particular, only RXRα, but not RXRβ, RARα or RARβ, seems to play a major role in determining cell shape and cell-cell adhesion properties of cardiomyocytes. In addition, in RXRβ and RARβ mutant SMs, the penetrance at the cellular level of the precocious differentiation phenotype was often weak. The crucial role of RXRα in cardiac development is also underscored by the presence of cardiac abnormalities in heterozygote mice: RXRα+/ − embryos display a milder form of the early differentiation phenotype, and adult RXRα+/ − mice exhibit an abnormal ventricular morphology (Gruber et al., 1996). Whether this prominence of RXRα reflects unique properties of this receptor, or merely results from a quantitative predominance, remains to be investigated.

The high percentage of myocytes displaying the abnormal differentiation features in RXRα−/−/RXRβ−/− mutants demonstrates that, with respect to preventing an early differentiation, RXRP and RXRa exert similar functions. A threshold level of RXRs in ventricular myocytes might be required for their normal differentiation-inhibitory function, and the increase in the proportion of affected cells in double mutants probably reflects this requirement. The incomplete expressivity, at the cellular level, of the precocious differentiation seen in RXRα and RXRβ single null mutants probably indicates stochastic variations, either in the levels of RXR protein present in given cells, or in the threshold level of RXR protein required for a correct function (or both).

Knock-out experiments have shown that the function of several gene products is critical for myocardial morphogenesis, including the transcription factors N-myc (Moens et al., 1993), TEF-1(Chen et al., 1994), Nkx2.5 (Lyons et al., 1995), WT-1 (Kreidberg et al., 1993), as well as the secreted factor neuregulin and its receptor erbB4 (Meyer and Birchmeier, 1995; Gassmann et al., 1995). It should be interesting to investigate whether absence of these genes similarly perturbs the differentiation programme of the myocytes and if the pathways affected in these mutants intersect with that of retinoic acid.

We thank K. Chien for the gift of the MLC2v plasmid, J.-L. Vonesch for computer processing of the MLC2a in situ hybridisation data, C. Bronn, B. Schubaur, B. Bondeau and L. Auvray for technical assistance, and the photography, illustration and secretarial staff for help in the preparation of this manuscript. This work was supported by funds from the Centre National de la Recherche Scientifique (CNRS), the Institut National de la Santé et de la Recherche Médicale (INSERM), the Collège de France, the Centre Hospitalier Universitaire Régional, the Association pour la Recherche sur la Cancer (ARC), the Human Frontier Science Program and Bristol-Myers Squibb. J.M.G. was a recipient of a fellowship from the EEC (TMR Program) and the Fondation pour la Recherche Médicale, respectively; S.W. was a recipient of a fellowship from the EMBO and the EEC (TMR Program), respectively.

Bavik
,
C.
,
Ward
,
S.
and
Chambón
,
P.
(
1996
).
Developmental abnormalities in cultured mouse embryos deprived of retinoic acid by inhibition of yolk-sac retinol binding protein synthesis
.
Proc. Nat. Acad. Sci. USA
93
,
3110
3114
.
Chambón
,
P.
(
1996
).
A decade of molecular biology of retinoic acid receptors. In The Retinoid Revolution
(ed.
W.J.
Whelan
).
FASEB J
.
10
,
940
954
.
Chen
,
Z.
,
Friedrich
,
G.A.
,
Soriano
,
P.
(
1994
).
Transcriptional enhancer factor 1 disruption by a retroviral gene trap leads to heart defects and embryonic lethality in mice
.
Genes Dev
8
,
2293
2301
.
Darmon
,
M.
(
1991
).
Retinoic acid in skin and epithelia
.
Sem. Dev Biol
.
2
,
219228
.
Dyson
,
E.
,
Sucov
,
H.M.
,
Kubalak
,
S.W.
,
Schmid-Schonbein
,
G.W.
,
DeLano
,
F.A.
,
Evans
,
R.M.
,
Ross
,
J.
, Jr
. and
Chien
,
K.R.
(
1995
).
Atrial-like phenotype is associated with embryonic ventricular failure in retinoid X receptor α−/− mice
.
Proc. Nat. Acad. Sci. USA
92
,
7386
7390
.
Gassmann
,
M.
,
Casagranda
,
F.
,
Orioli
,
D.
,
Simon
,
H.
,
Lai
,
C.
,
Klein
,
R.
and
Lemke
,
G.
(
1995
).
Abberant neural and cardiac development in mice lacking the ErbB4 neuregulin receptor
.
Nature
378
,
390
394
.
Ghyselinck
,
N.
,
Dupé
,
V.
,
Dierich
,
A.
,
Messaddeq
,
N.
,
Garnier
,
J.M.
,
Rochette-Egly
,
C.
,
Chambon
,
P.
and
Mark
,
M.
(
1997
).
Role of the retinoic acid receptor beta (RARp) during mouse development
.
Int J. Dev Biol
.
41
,
425
447
.
Gruber
,
P.J.
,
Kubalak
,
S.
,
Pexieder
,
T.
,
Sucov
,
H.
,
Evans
,
R.
and
Chien
,
K.
(
1996
).
RXRa deficiency confers genetic susceptibility for aortic sac, atrioventricular cushion and ventricular defects in mice
.
J. Clin. Invest
.
98
,
1332
1343
.
Jiang
,
H.
,
Soprano
,
D.R.
,
Li
,
S.W.
,
Soprano
,
K.J.
,
Penner
,
J.D.
,
Gyda
,
M.
3rd
,
Kochhar
,
D.M.
(
1995
).
Modulation of limb bud chondrogenesis by retinoic acid and retinoic acid receptors
.
Int. J. Dev. Biol
.
39
,
617
627
.
Kastner
,
P.
,
Grondona
,
J.
,
Mark
,
M.
,
Gansmuller
,
A.
,
LeMeur
,
M.
,
Décimo
,
D.
,
Vonesch
,
J.L.
,
Dollé
,
P.
and
Chambon
,
P.
(
1994
).
Genetic analysis of RXRa developmental function: convergence of RXR and RAR signalling pathways in heart and eye morphogenesis
.
Cell
78
,
987
1003
.
Kastner
,
P.
,
Mark
,
M.
and
Chambon
,
P.
(
1995
).
Nonsteroid nuclear receptors: what are genetic studies telling us about their role in real life?
Cell
83
,
859
869
.
Kastner
,
P.
,
Mark
,
M.
,
Leid
,
M.
,
Gansmuller
,
A.
,
Chin
,
W.
,
Grondona
,
J.M.
,
Décimo
,
D.
,
Krezel
,
W.
,
Dierich
,
A.
and
Chambon
,
P.
(
1996
).
Abnormal spermatogenesis in RXR| mutant mice
.
Genes Dev
.
10
,
80
92
.
Kastner
,
P.
,
Mark
,
M.
,
Ghyselinck
,
N.
,
Krezel
,
W.
,
Dupé
,
V.
,
Grondona
,
J. M.
and
Chambon
,
P.
(
1997
).
Genetic evidence that the retinoid signal is transduced by heterodimeric RXR/RAR functional units during mouse development
.
D evelopment
124
,
313
326
.
Kreidberg
,
J.A.
,
Sariola
,
H.
,
Loring
,
J.M.
,
Maeda
,
M.
,
Pelletier
,
J.
,
Housman
,
D.
and
Jaenisch
,
R.
(
1993
).
WT-1 is required for early kidney development
.
Cell
14
,
679
691
.
Krstic
,
R.V.
(
1984
).
Illustrated Encyclopedia of Human Histology
.
Springer Verlag
.
Berlin
.
Kubalak
,
S.W.
,
Miller-Hance
,
W.C.
,
O’Brien
,
T.X.
,
Dyson
,
E.
and
Chien
,
K. R.
(
1994
).
Chamber specification of atrial myosin light chain-2 expression precedes septation during murine cardiogenesis
.
J. Biol. Chem
.
269
,
16961
16970
.
Lohnes
,
D.
,
Kastner
,
P.
,
Dierich
,
A.
,
Mark
,
M.
,
LeMeur
,
M.
and
Chambon
,
P.
(
1993
).
Function of retinoic acid receptor y (RARy) in the mouse
.
Cell
73
,
643
658
.
Lohnes
,
D.
,
Mark
,
M.
,
Mendelsohn
,
C.
,
Dollé
,
P.
,
Dierich
,
A.
,
Gorry
,
P.
,
Gansmuller
,
A.
and
Chambon
,
P.
(
1994
).
Function of the retinoic acid receptors (RARs) during development. I. Craniofacial and skeletal abnormalities in RAR double mutants
.
Development
120
,
2723
2748
.
Lufkin
,
T.
,
Lohnes
,
D.
,
Mark
,
M.
,
Dierich
,
A.
,
Gorry
,
P.
,
Gaub
,
M.-P.
,
LeMeur
,
M.
and
Chambon
,
P.
(
1993
).
High postnatal lethality and testis degeneration in retinoic acid receptor a (RARa) mutant mice
.
Proc. Nat. Acad. Sci. USA
90
,
7225
7229
.
Luo
,
J
,
Sucov
,
H.
,
Bader
,
J.
,
Evans
,
R.
and
Giguère
,
V.
(
1996
).
Compound mutants for retinoic acid receptor (RAR) p and RARa 1 reveal developomental functions for multiple RARp isoforms
.
Mech. Dev
55
,
33
44
.
Lyons
,
I.
,
Parsons
,
L.M.
,
Hartley
,
L.
,
Li
,
R.
Andrews
,
J.E.
,
Robb
,
L.
,
Harvey
,
R.P.
(
1995
).
Myogenic and morphogenetic defects in the heart tubes of murine embryos lacking the homeobox gene Nkx2.5
.
Genes Dev
.
9
,
16541666
.
Mangelsdorf
,
D.J.
and
Evans
,
R.M.
(
1995
).
The RXR heterodimers and orphan receptors
.
Cell
83
,
841
850
.
Mendelsohn
,
C.
,
Lohnes
,
D.
,
Décimo
,
D.
,
Lufkin
,
T.
,
LeMeur
,
M.
,
Chambon
,
P.
and
Mark
,
M.
(
1994
).
Function of the retinoic acid receptors (RARs) during development. II. Multiple abnormalities at various stages of organogenesis in RAR double mutants
.
Development
120
,
2749
2771
.
Meyer
,
D.
and
Birchmeier
,
C.
(
1995
).
Multiple essential functions of neuregulin in development
.
Nature
378
,
386
389
.
Moens
,
C.B.
,
Stanton
,
B.R.
,
Parada
,
L.F.
and
Rossant
,
J.
(
1993
).
Defects in heart and lung development in compound heterozygotes for two different targeted mutations at the N-myc locus
.
Development
119
,
485
499
.
Niederreither
,
K.
,
McCaffery
,
P.
,
Drager
,
U.C.
,
Chambon
,
P.
and
Dollé
,
P.
(
1997
).
Restricted expression and retinoic acid-induced downregulation of the retinaldehyde dehydrogenase type 2 (RALDH-2) gene during mouse development
.
Mech. Dev
.
62
,
67
78
.
Paulsen
,
D.F.
and
Solursh
,
M.
(
1988
).
Microtiter micromass cultures of limbbud mesenchymal cells
.
In Vitro Cell Dev. Biol
.
24
,
138
147
.
Ruberte
,
E.
,
Dollé
,
P.
,
Krust
,
A.
,
Zelent
,
A.
,
Morriss-Kay
,
G.
and
Chambon
,
P.
(
1990
).
Specific spatial and temporal distribution of retinoic acid receptor gamma transcripts during mouse embryogenesis
.
Development
108
,
213222
.
Rumyantsev
,
P.P.
(
1977
).
Interrelations of the proliferation and differentiation processes during cardiac myogenesis and regeneration
.
Int. Rev. Cytol
.
91
,
187
273
.
Sucov
,
H.M.
,
Dyson
,
E.
,
Gumeringer
,
C.L.
,
Price
,
J.
,
Chien
,
K.R.
and
Evans
,
R.M.
(
1994
).
RXRa mutant mice establish a genetic basis for vitamin A signaling in heart morphogenesis
.
Genes Dev
.
8
,
1007
1018
.
Viragh
,
S.
and
Challice
,
C.
(
1981
).
The origin of the epicardium and the embryonic myocardial circulation in the mouse
.
Anat. Rec
.
201
,
157
168
.
Wilson
,
J.G.
and
Warkany
,
J.
(
1949
).
Aortic-arch and cardiac anomalies in the offspring of vitamin A deficient rats
.
Am. J. Anat
.
85
,
113
155
.
Yang
,
J.
,
Rayburn
,
H.
and
Hynes
,
R.
(
1995
).
Cell adhesion events mediated by a4 integrins are essential in placenta and cardiac development
.
Development
121
,
549
560
.
Zhao
,
D.
,
McCaffery
,
P.
,
Ivins
,
K.J.
,
Neve
,
R.L.
,
Hogan
,
P.
,
Chin
,
W.W.
and
Drager
,
U.C.
(
1996
).
Molecular identification of a major retinoic-acid synthesizing enzyme: a retinaldehyde-specific dehydrogenase
.
Eur. J. Biochem
.
240
,
15
22
.
Zimmer
,
A.
,
Zimmer
,
A.M.
and
Reynolds
,
K.
(
1994
).
Tissue specific expression of the retinoic acid receptor-| 2 regulation by short open reading frames in the 5’-noncoding region
.
J. Cell Biol
.
127
,
1111
1119
.