Two families of nuclear receptors for retinoic acid (RA) have been characterized. Members of the RAR family (types alpha, beta and gamma and their isoforms alpha 1, alpha 2, beta 1 to beta 4, and gamma 1 and gamma 2) are activated by most physiologically occurring retinoids (all-trans RA, 9-cis RA, 4oxo RA and 3,4 dihyroRA). In contrast, members of the RXR family (types alpha, beta and gamma and their isoforms) are activated by 9cis-RA only. In addition to the multiplicity of receptors, the complexity of retinoid signalling is further increased by the fact that, at least in vitro, RARs bind to their cognate response elements as heterodimers with RXRs. Moreover, RXRs can also bind, in vitro, to some DNA elements as homodimers, and are heterodimeric partners for other nuclear receptors, including TRs, VDR, PPARs and a number of orphan nuclear receptors. To evaluate the functions of the different RARs and RXRs types and isoforms, we have generated null mutant mice by targeted gene disruption in ES cells. As to the functions of RARs, we found that RAR alpha 1 and RAR gamma 2 null mutant mice are apparently normal. Mice deficient in RAR alpha or RAR gamma (i.e., all alpha or gamma isoforms disrupted) show aspects of the post-natal vitamin A deficiency (VAD) syndrome which can be cured or prevented by RA, including post-natal lethality, poor weight gain and male sterility. RAR beta 2 (and RAR beta) null mutants display a retrolenticular membrane which represents the most frequent defect of the fetal VAD syndrome. That these abnormalities were restricted to a small subset of the tissues normally expressing these receptors suggested that some degree of functional redundancy should exist in the RAR family. To test this hypothesis we then generated RAR double null mutants. RAR alpha beta, RAR alpha gamma and RAR beta gamma compound mutants exhibit all the malformations of the fetal VAD syndrome, thus demonstrating that RA is the vitamin A derivative which plays a crucial role at many different stages and in different structures during organogenesis. Interestingly, almost all the structures derived from mesenchymal neural crests cells (NCC) are affected in RAR compound mutants. As to the functions of RXRs, RXR gamma null mutants are viable, fertile and morphologically normal. In contrast, RXR alpha null fetuses display a thin ventricular wall and die in utero from cardiac failure. A myocardial hypoplasia has also been observed in some RAR compound mutants as well as in VAD fetuses. Thus, RXR alpha seems to act as an inhibitor of ventricular cardiocyte differentiation and/or as a positive regulator of their proliferation, and these functions might involve heterodimerization with RARs and activation by RA. RXR beta null mutants are viable but the males are sterile, most probably because of an abnormal lipid metabolism in the Sertoli cells. New abnormalities, absent in RXR alpha mutants, are generated in RXR alpha/RAR (alpha, beta or gamma) compound mutants. All these abnormalities are also seen in RAR double mutants as well as in VAD fetuses. In contrast, such manifestations of synergism are not observed between the RXR beta or RXR gamma and the RAR (alpha, beta or gamma) null mutations. These data strongly support the conclusion that RXR alpha/RAR heterodimers represent the main functional units of the RA signalling pathway during embryonic development. Moreover, since RXR gamma-/-/RXR beta-/-/RXR alpha +/-mutants are viable, a single allele of RXR alpha can perform most of the developmental RXR functions.