The vertebrate embryonic body plan is constructed through the interaction of many developmentally regulated genes that supply cells with the essential positional and functional information they require to migrate to their appropriate destination and generate the proper structures. Some molecular cues involved in patterning the central nervous system, particularly in the hindbrain, are interpreted by the Hox homeobox genes. Retinoids can affect the expression of Hox genes in cells lines and embryonic tissues; the hindbrain and branchial region of the head are particularly sensitive to the teratogenic effects of retinoic acid. The presence of endogenous retinoic acid, together with the distribution of retinoid binding proteins and nuclear receptors in the developing embryo, strongly suggest that retinoic acid is a natural morphogen in vertebrate development. The molecular basis for the interaction between retinoic acid and the Hox genes has been aided in part by approaches using deletion analysis in transgenic mice carrying lacZ reporter constructs. Such studies have identified functional retinoic acid response elements within flanking sequences of some of the most 3' Hox genes, suggesting a direct interaction between the genes and retinoic acid. Furthermore, as demonstrated using transgenic mice carrying Hoxb-1/lacZ constructs, multiple retinoic acid response elements may cooperate with positive and negative regulatory enhancers to specify pattern formation in the vertebrate embryo. These types of studies strongly support the normal roles of retinoids in patterning vertebrate embryogenesis through the Hox genes.

HOX homeoproteins control cell identities during animal development by differentially regulating target genes. The homeoprotein encoded by the extradenticle (exd) gene can selectively modify HOX DNA binding, suggesting that it contributes to HOX specificity in vivo. HOX‐EXD interactions are in part mediated by a conserved stretch of amino acids termed the hexapeptide found in many HOX proteins. Here, we demonstrate that a 20 bp oligonucleotide from the 5′ region of the mouse Hoxb‐1 gene, a homolog of Drosophila labial (lab), is sufficient to direct an expression pattern in Drosophila that is very similar to endogenous lab. In vivo, this expression requires lab and exd and, in vitro, LAB requires EXD to bind this oligonucleotide. In contrast, LAB proteins with mutations in the hexapeptide bind DNA even in the absence of EXD. Moreover, a hexapeptide mutant of LAB has an increased ability to activate transcription in vivo. Partial proteolysis experiments suggest that EXD can induce a conformational change in LAB. These data are consistent with a mechanism whereby the LAB hexapeptide inhibits LAB function by inhibiting DNA binding and that an EXD‐induced conformational change in LAB relieves this inhibition, promoting highly specific interactions with biologically relevant binding sites.

In this study we have investigated the organization and regulation of the mouse Hox‐2.7 gene. There are several alternative transcripts some of which are conserved between mouse and humans. By Northern and in situ analysis we are able to identify at least three types of transcripts which are different in size and splicing pattern and have distinctly different boundaries of expression in the nervous system. One subset of the endogenous transcripts has a boundary of expression that corresponds to the adjacent Hox‐2.8 gene instead of Hox‐2.7. In another type of transcript there is an alternative reading frame which predicts a protein that has homology to an enzyme ATPase and suggests that a non‐homeobox containing gene may be located in the Hox‐2 cluster. A Hox‐2.7‐lacZ transgene is expressed in a similar pattern to the endogenous gene in that spatially‐restricted domains of expression are seen in the branchial arches, neural tube, paraxial mesoderm (somites), cranial ganglia, neural crest and gut. However, the anterior boundaries of transgene expression only correspond to the subset of Hox‐2.7 transcripts which map to the Hox‐2.8 boundary. The proximity of a Hox‐2.7 promoter to regions which regulate the adjacent Hox‐2.6 gene and the expression of transgenic and endogenous transcripts in a Hox‐2.8 pattern, suggest that regulatory elements may be shared by neighbouring genes to establish the complete expression pattern.

The mouse alpha-fetoprotein gene is activated in embryonic development in the visceral endoderm of the extraembryonic yolk sac and the fetal liver and gut. Transcription of the gene is subsequently repressed in the neonatal liver. To ask whether the DNA sequence elements required for tissue-specific activation are the same or different from those required for postnatal developmental regulation of the gene, modified copies of the alpha-fetoprotein gene were microinjected into fertilized mouse eggs. Those animals which developed to term and carried integrated copies of the modified gene were analyzed for expression. In approximately 50% of such animals, the introduced gene was active only in the three cell lineages which expressed the authentic alpha-fetoprotein gene. Furthermore, its expression was repressed in the neonatal liver. Thus, we conclude that the modified genes, which included either 7 or 14 kilobase pairs of 5'-flanking DNA, contained the DNA sequence information to direct both tissue-specific expression and developmental regulation. The observation that 50% of the mice which carried the modified gene did not express it in any tissue, combined with the fact that the level of expression was highly variable between expressing transgenic animals, suggested that the gene was susceptible to its site of integration in the mouse genome.

We have compared the ways in which vertebrate Hox genes are used in the patterning of three distinct embryonic contexts, the branchial region, the somites, and the limb. We have identified common features of the three systems, but have suggested on the basis of their differences (in both embryological properties and use of Hox genes) that it is better to consider each as an independent system for regional specification. Nevertheless, there are sufficient common features to expect that exploitation of the distinct experimental advantages of each system will provide important insights to the mode of operation of the others.