Prx2 suggest roles in development of the heart and the forebrain

ELSEVIER

Mechanisms of Development 52 (1995) 5144

Expression patterns of the paired-related homeobox genes MHoxlPrxl and WPrx2 suggest roles in development of the heart and the forebrain Bertil Leussinka, Antje Brouwera, Mohamed El Khattabia, Rob E. Poelmannb, Adriana C. Gittenberger-de Grootb, Frits Meijlir&,* aHubrecht Laboratory, Netherlands Institute for Developmental Biology, Uppsalalaan 8, 3584 CT Vtrecht, The Netherlands bLeiden UniversiQ, Department of Anatomy and Embryology, PO Box 9602,230O RC Leiden, The Netherlands

Received 8 March 1995; revision received 4 April 1995; accepted 4 April 1995

Abstract Prxl and Pm2 (previously called MHox and S8, respectively) are the members of a small subfamily of vertebrate homeobox genes expressed during embryogenesis from gastrulation onwards. We directly compared the expression domains of the Prx genes in detail in mouse and in addition some aspects of these patterns in chicken. In addition to the superficially similar expression patterns of Prxl and Prx2 in cranial

mesenchyme, limb buds, axial mesoderm, and branchial arches and their derivatives, we detect major differences at many sites particularly in heart and brain. Our analysis indicated in several cases a correlation with regions developing into connective tissues. From at least day 8.5, Prx-1 expression was observed in the heart, initially in the endocardial cushions and later in the developing semilunar and atrioventricular valves. Prx2 develops early on a diffuse myocardial expression pattern and is later higher expressed in the ventricular septum and in particular in the ductus arteriosus. Pm2 is never expressed in the brain, whereas Prxl is expressed, from at least day 9.5 onwards, in a unique distinct domain in the ventral part of the hypothalamus, as well as in a broader region of the telencephalon. Keywords:

Homeobox genes; Limb development; Heart valves; Connective tissue; Diencephalon; Telencephalon

1. Introduction Homeobox genes play a part in many different aspects of development (see for instance De Robertis, 1994). They can be classified, on the basis of sequence comparisons, in many different families (Kappen et al., 1993; Biirglin, 1994). The homeodomain proteins encoded are thought to be transcriptional regulators that are able to specifically interact with DNA. In vertebrates, the 3% member Hox family found in mouse and human constitutes the most frequently studied and best understood set of homeobox genes. These Hox genes are the homologues of the clustered homeotic genes (HOM-C) of Drosophila melanogaster, and they have convincingly been shown to function as transmitters of anteroposterior positional information (Krumlauf, 1994).

* Corresponding author, Tel.: +31 30 510211; Fax: +31 30 516464; E-mail: [email protected].

The function of many vertebrate homeobox genes belonging to the other classes has often remained rather elusive. The related MHox and S8 genes are among these poorly understood genes. We propose here the names Prxl and Prx2, respectively (see Section 4) for these genes. S8fPrx2 was discovered in a screen for pairedrelated homeobox genes (Kongsuwan, 1988). Prxl has previously been termed MHox (Cserjesi et al., 1992), K2 (Kern et al., 1992) and Pmx (Kern et al., 1994) for the mouse gene, Phox-1 (Grueneberg et al., 1992) for the human homologue, and Prx-I (Nohno et al., 1993) and gMHox (Kuratani et al., 1994) for the chicken homologue. It was independently isolated on the basis of diverse cloning strategies: (1) as a gene encoding a transcription factor potentially interacting with the muscle creatine kinase promoter (Cserjesi et al., 1992, 1994); (2) on the basis of sequence similarity with the segmentation gene paired (Kern et al., 1992; Nohno et al., 1993) or S8 (now Prx2) gene (F.M. and A.B., unpublished data); and (3) as a gene encoding a factor capable of intensifying the

09254773/95/$09.50 0 1995 Elsevier Science Ireland Ltd. All rights reserved 0925-4773(95)00389-I

SSDI

52

B. Leussink et al. I Mechanisms of Development 52 (1995) 51-64

interaction of MADS-type transcription factors like MCMl and Serum Response Factor with their targets (Grueneberg et al., 1992). The Prx genes encode proteins with homeodomains that are about 67% identical to those of the segmentation genes paired and gooseberry. Genes convincingly homologous to Pm have not yet been found in Drosophila. In contrast with the Drosophila paired and gooseberry genes and the vertebrate Pax genes, the Prx genes do not contain a paired box, and at amino acid 50 of the homeodomain contain a glutamine rather than the paired-specific serine. This amino acid position has been reported to be critical for the specificity of DNA binding of the paired protein (Treisman et al., 1989). More recently it was shown to be crucial, in the isolated pairedtype homeodomain, principally for the mode of protein dimerization which influences DNA-binding specificity (see Wilson et al., 1993). Previously we showed that Prx2 is expressed in a complex pattern during mouse embryogenesis (Opstelten et al., 1991; De Jong and Meijlink, 1993) in a mostly mesenchyme-specific pattern. Although it has been reported that Prxl has an expression pattern distinct from, but generally similar to that of Prx2 (Cserjesi et al., 1992; Kern et al., 1992) this has not been investigated in detail. We therefore decided to compare these patterns directly, in most cases in serial sections. Such comparative expression studies have previously supplied suggestive evidence as to how the Hox genes and to some extent the Msx genes, share a task by their coordinated expression in overlapping (Hox) or partly overlapping and partly complementary (Msx) domains (see Davidson et al., 1991). We found that the expression domains of Prxl and Prx2 differ considerably in many details, in spite of a major overlap in loose mesenchyme of cranium, branchial arches, body wall and limbs. The expression pattern of Prxl fails to confirm a role in muscle-specific gene expression (except locally in smooth muscle) but rather points to a correlation with differentiation into a subset of connective tissues, including valves of the heart, and skeletal tendons. Unexpectedly, the Prxl expression pattern includes a rather broad part of the developing telencephalon, and a sharply delineated expression domain in the ventral part of the diencephalon.

quence of Prxl and a part of the Prx2 sequence. In the course of this work, the chicken Prxl sequence was twice published by other groups (Nohno et al., 1993; Kuratani et al., 1994). The Prx2 sequence data determined by us was completely in accordance with that of Kuratani et al. (1994) and differed from that of Nohno et al. (1993) only in that we found an additional alanine triplet three residues downstream from the homeobox (not shown). The nucleotide sequence and putative reading frame of chicken Prx2 as far as available now, is shown in Fig. IA. Fig. 1B shows a comparison between chick and mouse

2. Results 2.1. Structure of the Prx genes; cloning of chicken Prx homologues To investigate the conservation of the structure and

(A) Nucleotide sequence of chicken Prx2 with the translation indicated above. The homeobox is underlined. (B) Comparison of

expression patterns of the Prx genes between diverse vertebrate species, we screened two chicken embryo cDNA libraries with mouse Prx2 probes. This led to the isolation of incomplete cDNAs specific for Prxl and Prx2, which were subsequently used in further screenings of the same libraries. From the set of overlapping clones thus obtained, we could determine the entire coding se-

available inferred protein sequences of chicken and mouse Prx2. The homeodomain is in bold. (C) Comparison of homeodomain sequences of the two Prx genes with five related homeodomains. The number in parentheses indicates the number of identical amino acids, compared to the Prx2 homeodomain. References: Phox2, ValarchC et al., 1993; Smox-3, Webster and Mansour, 1992; Aristaless (abbreviated Arist), Schneitz et al., 1993; paired, Frigerio, et al. 1986; Pax-3, Goulding et al.. 1991.

B. Leussink et (11.I Mechanisms

Prx2 proteins, demonstrating identity of the homeodomains and very high similarity in the amino acid stretch towards the carboxy terminus (71 identical and six similar amino acids out of 90). Fig. 1C compares the Prxl and Prx2 homeodomains with the homeodomains that are most similar to them. The homeodomains encoded by the Phox2 gene, the Schistomania Smox3 gene, as well as those in the Drosophila paired gene and its Drosophila and vertebrate homologues have 40-45 identical residues among the 60 amino acids of the homeodomain. Clearly, Prxl and Prx2 belong to a unique common sub-class. The striking similarity between Smox-3 and Phox-2 (resulting in S/60 identities in the homeodomain) suggests that these genes are true homologues. We do not feel that classification of Phox-2 in the same subfamily as Prxl alias Phoxl - is appropriate as suggested by its name. In total RNA isolated from whole chick embryos of 3 and 6 days of hatching and dissected embryos of 9 days, we detected, by RNA blot analysis, a Prxl mRNA of 2.5 kb and a Prx2 mRNA of about 4 kb. By contrast, in the mouse Prxl messengers of 3.6 and 4.0 kb (Cserjesi et

@Development

52 (1995) 51-64

53

al., 1992) and a major Prx2 mRNA of 1.4-1.6 kb (Opstelten et al., 1991) are detected (not shown). This indicates major changes in the untranslated regions of these genes. 2.2. Expression surrounding

in primitive-streak

stage embryos and

decidua

We sectioned whole decidua from mice at 6.58.5 days of pregnancy. Previously we described in detail Prx2 expression detected in such sections in mesoderm posterior from and lateral to the primitive streak (De Jong and Meijlink, 1993) and in a characteristic pattern in the (maternal) deciduum. In addition to confirming these data we now show that Prxl is expressed in a different pattern in mesenchymal cells directly underlying the remnants of the epithelium of the uterus and in addition in the periphery of the deciduum in a subset of the Prx2 expressing cells. This is shown in Fig. 2, along with the Prx2 pattern for comparison, for a deciduum with a tangentially cut head-fold-stage embryo. Expression at similar stages in decidual cells adjacent to the uterine lumen has been re-

Fig. 2. Prx expression in presomite embryos and decidua. Due to the small size of these embryos in this case no serial sections were made. The probe used is indicated in the upper right comer of the dark field panels. (A) Parasagittal section through decidua containing 7.5day embryo. The signal near the uterine lumen is located in the mesenchymal cells underneath it. (B) Parasagittal section through decidua with different but similar embryo as in (A). (CD) Higher magnification of Prxl expression patterns in 75day embryos showing expression in mesoderm, ectoderm and in a part of the allantois. all, allantois; d. deciduum; ec, ectoderm; lu, remnant of uterine lumen; m, mesoderm; ps, primitive streak, Bar 200pm.

54

B. Leussink et al. I Mechanisms of Development 52 (1995) 5164

ported for several genes including for instance Activin PA (Manova et al., 1992; Albano et al., 1994), tenascin (Julian et al., 1994), and the apoptosis-related genes transglutaminase, Bcl-2, and c-myc (see Piacentini and Autuori, 1994) but usually these patterns occurred in association with expression in the decidual and uterineepithelial cells surrounding the embryo, which we did not observe in the case of Prxl. Longer exposures demonstrated that Prxl is also expressed in the outer layers of the decidua at these stages (not shown). Expression of Prx2 in the embryo proper at this stage was lower than that of Prxl, and longer exposures were required to ob-

tain clear signals, unlike the situation at later stages. A major difference between the two Prx expression patterns observed at early headfold stages was that in addition to widespread expression in the embryonic and allantoic mesoderm, similarly to the Prx2 pattern, Prxl is also expressed in the ectoderm. Expression in ectoderm as well as in mesoderm is absent from the primitive streak and its surroundings, giving a region-specific (as opposed to tissue-specific) nature to this early expression pattern. Due to the limited resolution of these radioactive hybridizations, we cannot exclude expression in the one-layered endoderm at this early stage. In 8.5-day embryos, highest expression of Prxl is seen in the extraembryonic (yolk sac) mesoderm and in the mandibular arch that begins to form at this stage. Prx2 expression is also high in the mandibular arch. but lower in the extraembryonic mesoderm (not shown). 2.3. Expression at 9.0-10.5 days of gestation Fig. 3 shows a comparison of expression of both Prx genes at 9.5 and 10.5 days of gestation. These results can be considered semi-quantitative, since similar probes and exposure times were used. Characteristic parallels are the expression in the branchial arches and somatopleure. (Limbs are not visible in these near-midsagittal sections.) Typical differences include the Prxl expression in the dermis and the dermomyotomes (see Cserjesi et al., 1992; Kern et al. 1992; Kuratani et al. 1994). In addition Prxl is expressed in a broader part of the cranial mesenchyme. Prxl but not Prx2 is expressed in mesenchyme surrounding the mesonephric duct (Fig. 3B,C). Fig. 4 shows the results of whole-mount hybridizations of embryos of 9.5 and 10.5 days, and of a chicken embryo of stage 15-16. While these pictures demonstrate the overall pattern of the Prx genes, a number of the details are described below based on radioactive experiments on sections. In the 9-day embryo high expression of both genes in the first visceral arches, but not in its epithelium, was clearly seen, as well as expression in the somites and the somatopleure. Expression was visible in the remaining part of the base of the removed allantois.

Fig. 3. Overview of expression at 9.5-10.5 days of expression. The probes used are indicated in the upper right corner of the dark field panels, (A) 9.5.day embryo; nearby sections were hybridized with either probe. Note high expression in first arches and (lower) Prxl expression in telencephalic vesicle. (B) 10.5day embryo; nearby sections were hybridized with either probe. Both genes are expressed in the visceral arches including the branchial pouches. Note differences in cranial mesenchyme and somites. al, first branchial arch; dm, dermomyotomes; mn, mesonephtic duct; op, olfactory pit; sm. somatopleure; u, ureter. Bar 500pm.

2.4. Expression in the developing heart We followed the expression of the Prx genes in the developing heart from the head-fold stage until 17.5 days of development. The broad expression domains at 7.5 days (Fig. 2) include the area from which the heart will develop (see also De Jong and Meijlink, 1993). Expression of both Prxl and Prx2 at days 8.5 and 9.5 is abundant in the developing visceral arches and the splanchnic mesoderm surrounding the gut. Fig. 5A-D shows Prxl expression in two transverse sections of an 8.5-day embryo. Hybridization was seen at the venous pole of the heart in the mesenchyme of the dorsal mesocardium and in the inner curvature. The dorsal mesocardium is known to be continuous with the develop-

B. Leussink et (11.I Mechanisms

of Development 52 (1995) 5144

5.5

Fig. 4. Whole-mount analysis of embryos. Whole mount hybridization was performed with 9.0, 9.5 and 10.5day mouse embryos and a stage 15-16 chicken embryo, as indicated in the lower left comer of the pictures. The 9.0-day embryo was staged as such on the basis of its morphology, but was actually isolated at 8.5 days. In the upper comers the probe used is indicated (GPrxl = chicken &cl). The absence of Prx2 signal in the heart and Prxl in the somites of the 9.5-day embryos is probably caused by the lower sensitivity of detection in these experiments. alla2, first/second branchial arch; all, allantois; fl, forelimb; h, heart; hl, hind limb; md/mx, mandibular/maxillary component of first arch; o, otocyst; lm, lateral plate mesoderm; s, somites.

ing atrioventricular cushions at later stages via the inner curvature of the heart tube. At 8.5 and 9.5 days of gestation, the myocardium of the developing atria and ventricles were negative, as was the endocardium lining of the heart. At day 12.5 (Fig. 5E,F) we detected strong hybridization with the Prxl probe of the atrioventricular and outflow tract cushions including their overlying endocardium. In the outflow tract the wall of the developing great arteries was weakly positive. The marked expression domain of the outflow tract cushions, however, was seen to be continuous with that in the condensed mesenchyme of the aorto-pulmonary septum (see Bartelings and Gittenberger-de Groot, 1988). This structure is known to have

contributions of neural crest cells (Kirby and Waldo, 1990). Recent investigations with retroviral markers show these cells also to extend into the outflow tract cushions, at least in birds (Noden et al., 1995). Prxl was relatively weakly expressed in the great vessels outside the heart, viz. the ascending aorta and pulmonary trunk, whereas the dorsal aorta and the cardinal veins had a marked expression in the smooth-muscle cells of the walls of the tunica media as well as in the lining endothelium. Pm2 was more diffusely present from days 8.5-12.5 in the myocardium of the heart (Opstelten et al., 1991, and our unpublished data). At day 14.5 (Fig. 6) comparison of Prxl (Fig. 6D-F) and Prx2 (Fig. 6G-I) showed conspicu-

56

B. Leussink et ul. I Mechanisms qf Development 52 (1995) 51-64

Fig. 5. Prxl expression in developing heart in embryos of 8.5-12.5 days of gestation. (A-D) Bright (A,C) and dark (B,D) field pictures of two sections of the same 8.5.day embryo hybridized with the Prxl probe. In the heart there is expression at the inner curvature in the atrioventricular region. This tissue is continuous with the dorsal mesocardium (not shown). Expression in the splanchnic mesoderm is indicated with arrows in (C) and (D). (E,F) At 12.5 days, the expression is very marked and restricted present in the atrio-ventricular and outflow tract cushions. Note high expression in the outflow tract ridge (Ott), but absence of signal in the aortic wall (ao; indicated with triangle in (F)). avc, atrio-ventricular cushions; bc, bulbus cordis; ca, common atrium; cm, cephalic mesenchyme; cv, common ventricle; da, dorsal aorta; ec. endocardial cushion; fg, foregut pocket; fl, fore limb; is, interventricular septum; Iv, left ventricle; nc, notochord; nt, neural tube; oe. oesophagus; otr, outflow tract ridge; rv, right ventricle; s, somite; sm. splanchnic mesenchyme; so, somatopleure; sp, splanchnopleure; sv. sinus venosus; tr, trachea; tw, thoracic bodywall. Bars 200pm.

ous differences. Prxl expression was now almost exclusively restricted to the developing semilunar and atrioventricular valves (Fig. 6D-F). Outside the heart, the signal in the arterial pole was weak. The ductus arteriosus, the dorsal aorta and the cardinal veins showed a more clear signal (Fig. 6D-F). At the venous pole of the heart the sinus venosus area, including the pulmonary veins and the connection of the dorsal mesocardium through the spina vestibuli (the lower rim of the atria1 septum) with the atrioventricular valves was still seen to be positive as well. Prx2 expression at day 14.5 (Fig. 6G-I) was, in

addition to its presence in the valves, more extensive in the ventricular septum. Prx2 is very unusual in its massive expression in the ductus arteriosus (Fig. 6G). Fig. 7 shows dark field pictures of sections through a day-17.5 heart. Here, Prxl staining was even more restricted to the developing valves and fibrous skeleton of the heart. Prx2 also shows this pattern with some staining of myocardium which may be the developing chordae of the atrioventricular valves. Otherwise these patterns were generally in agreement with the earlier expression patterns.

B. Lmssink et al. I Mechunisms of Development 52 (1995) 51-64

57

Fig. 6. Comparison of Prxl and Prx2 expression in the heart at 14.5 days. (A-C) Bright field pictures of transverse sections hybridized to a Prxl probe. (D-F) Dark field exposures of panels A-C. (H-J) Hybridizations with Prx2 probe of sections close to A-C, respectively. Prxl is markedly expressed in the developing atrioventricular and semilunar valves, while lower expression levels are seen in the ascending aorta and pulmonary trunk. The ductus arteriosus-dorsal aorta continuation shows a higher expression (D). (F) Conspicuous signal in the splanchnic mesoderm at the level of the venous pole of the heart. The Prx2 probe also hybridizes to the semilunar and atrioventricular valves, but additionally to the myocardium of the ventricular septum. (G) The massive Prx2 expression in the ductus arteriosus. (I) The posterior wall of the atrium, which is of sinus venosus origin, shows high PrxZ expression. ao, aorta; av, aotic semilunar valve; br, bronchi; ec, endocardial cushions; is, interventricular septum; lu, lung; mv, mitral valve; pt, pulmonary trunk; pv, pulmonary valve; r, rib; m/la, right/ left atrium; sp, septum primum; ss, septum secundum; tv, tricuspid valve; vv, venous valve. Bar 400Pm.

2.5. Expression

in developing

limbs

In early limb buds a number of transient structures have been identified such as the zone of polarizing activity and the apical ectodermal ridge, which have a crucial role in pattern formation (see for review Tickle and Eichele, 1994). Recently a number of genes have been identified including Hex and other homeobox genes, whose function could be linked to the production and interpretation of signals produced by these structures (see for instance Maden, 1994). Since both Prxl and Prx2 are highly expressed early on in the developing limb buds, we decided to compare these patterns in more detail. The high expression in undifferentiated mesenchymal cells is followed by down-regulation as cells begin to differentiate into cartilage cells (Opstelten et al., 1991; Nohno et al., 1993; Kuratani et al., 1994). As a first approach to sorting out the possible role of the Prx genes in limb development we looked at the earliest time points of expression in the limb field. The early expression is best

illustrated by the result of a whole-mount hybridization with a chick Prxl probe, that showed expression in the flank of a stage-15 chick embryo at a site where the limb is expected to form, but prior to discernible bud formation (Fig. 4, ‘GPrxl’). Fig. 4 also shows whole-mount in situ hybridization of 9.5 and 10.5 day embryos, with conspicuous expression in the developing limbs. A striking difference between Prxl expression in fore- and hind limbs was seen, expression in hind limbs lacking in the distal zone. We then studied Prx expression in limbless chicken mutants. Homozygous mutant limbless embryos form a limb bud which fails to elongate, apparently as a consequence of a defect in the apical ectodermal ridge (Fallon et al., 1983). Fig. 8 shows expression in transverse sections of stage-21 wild type and limbless chicken embryos. The expression patterns in limbs, lateral plate mesoderm and dermomyotome typical for Prxl and Prx2 were clearly seen and no differences specific for limbless were

B. Leussink et ul. I Mechanisms @‘Development 52 (1995) 51-64

Fig. 7. Comparison of Prxl and Prx2 expression in frontal sections of hearts dissected from 17.5-day embryos, Only dark field exposures are shown. Both Prxl and Prx2 expression is mainly concentrated in the semilunar and atrioventricular valve areas. Analysis of the sections shows that valves as well as the future fibrous heart skeleton including the fibrous trigone are stained, This correlates with the finding of signal in the area at the base of the atria1 septum that connects the fibrous trigone to the dorsal mesocardium. It is remarkable that Prx-I expression in this area disappears (A) whereas Prx-2 expression is sustained (D) during development. av, aortic semilunar valve; is, intraventricular septum; la, left atrium; mv, mitral valve; pv, pulmonary

valve; rv, right ventricle;

tv, tricuspid

valve. Bar 400~m.

Fig. 8. Prx expression in wild type and limbless chicken limb buds. Transversal sections of stage-21 chicken embryos, WTl, wild type chick embryo hybridized with Prxl. WT2. wild type embryo hybridized with Prx2. 111,2, limbless embryo hybridized with Prxl and Prx2. respectively.

observed, indicating independence of Prx expression from the AER. The possibility of a role of Prx-I in muscle development (Cserjesi et al., 1992) was recently nuanced (Cserjesi et al., 1994; Kuratani et al., 1994). We hybridized Prx probes to sections of relatively late stages, in which muscle can be readily distinguished from connective tissues. Fig. 9 shows examples of sections through a forelimb of a 17.5-day embryo hybridized with either probe. Expression was absent from differentiated cartilage and bone and present in a thin layer of one or two cells around the chondrogenic centers, in or near the perichondrium. In addition, especially Prxl was expressed in the dermis, and both genes were highly expressed in the footpads and in the tendons. Occasionally we observed a thin layer of expression skirting the muscle. which may correspond to expression in the presumptive fascia. We conclude here that expression of both Pm

B. Leussinket al. I MechanismsqfDevelopment52 (1995) 5144

59

Fig. 9. Prx expression in mouse limbs at 17.5 days. (A) Prxl hybridization of longitudinal section through forelimb of 17.5-day embryo, (B), Prxl expression of section transversally through the distal part of a forelimb at the level of the metacarpal bones. (CD) Nearby sections hybridized with Prx2 probe. Note hybridization in tendons, footpads, and near perichondrium. m3,4, third and fourth metacarpal bone; t, tendon; h, humerus; p, footpad; u, ulna. Bar 400pm.

genes tends to be associated with developing connective tissue. Interestingly, Kuratani et al. (1994) provided evidence that the Prxl expression originally described as is linked to the dermatome dermomyotome-specific, rather than to the myotome. 2.6. Prxl expression

in the developing

central nervous

system

In accord with our previously published data, Prx2 was not expressed, at any stage, in neural tissues. Unexpectedly, we found that Prxl is expressed in a specific pattern in the central nervous system from at least 9.5 days after conception. The potential of Prxl to be expressed in brain seems to be foreshadowed already at mid to late streak stages, when we saw expression in most of the ectoderm, including the precursors of the brain (Fig. 2). In contrast Prx2 was not expressed in ectoderm of I.5day embryos, and never in brain. Fig. 10 shows Prxl expression in the brain in sections of a number of stages between 9.5 days of gestation and newborn. Two areas showed expression, the telencephalon in a rather broad domain at 9.5 days evolving in a more restricted pattern at later stages. Fig. IOA-G shows expression in the ventral forebrain at 9.5 and 10.5 days of gestation and in an initially small and distinct region, located in the mammillary area. The level of expression was higher than in the adjacent mesenthyme, but substantially lower than in regions like the branchial arches and developing limbs, where Prxl expression is particularly high. Only a narrow and ventral region expressed Prxl, as is evident from the lack of signal in lateral sections and from transverse sections as shown in Fig. lOD,E for a 10.5-day embryo. A growing number of genes including many homeobox genes have recently been reported to be expressed in apparently segment-related domains in the forebrain, and this has contributed to novel hypotheses concerning the

development and organization of the forebrain, and in particular its postulated segmentation (Bulfone et al., 1993; Figdor and Stern, 1993; Puelles and Rubenstein, 1993; Rubenstein et al., 1994). Bulfone et al. (1993) propose on the basis of gene expression studies and other data that diencephalon and telencephalon are organized in six forebrain neuromeres, the prosomeres, which are numbered Pl-P6, P6 being the anteriormost segment. In parallel, Figdor and Stern (1993) have proposed an organization of the diencephalon in four diencephalon segments (Dl-D4) and a telencephalic segment, based primarily on observations in experiments related to cell lineage, and combined with other data. The most direct way to correlate the Prxl expression domain with these hypotheses, is to compare it in serial sections with some of the expression patterns that have contributed to the models. Therefore, to look for conformation of our impression that Prxl expression is in the ventral zone of the P4 prosomer (Bulfone et al., 1993), we hybridized adjacent sections of 10.5-day and 12.5-day mouse embryos with Prxl, O&-l, and Dlx-I probes. Fig. lOF-G shows the comparison of Prxl with Otx-I in sagittal sections of a 10.5-day embryo. The homeobox gene Otx-I (Simeone et al., 1993) is expressed in broad regions of the mid- and forebrain and in limited regions of the hindbrain. The sharp rostra1 limit of its expression domain in the diencephalon defines the P4lP5 prosomer boundary in the model of Bulfone et al. (1993). Prxl clearly shares with Otx-1 the ventral part of its expression boundary near the midline in 10.5-day embryos (Fig. lOF-H). Since Prxl has a somewhat butterfly-like expression domain with expression extending laterally more rostrally, however, Prxl expression crosses the P4/PS boundary adjacent to the midline meaning that it only partly respects this prosomerboundary. In the model of Figdor and Stern (1993) this expression boundary is not considered to be a segment

60

B. Leussink et al. I Mechanisms

border: it is localized within the singular large telencephalic segment Tl. Transverse sections (shown for 12.5day embryos, Fig. 101-K) demonstrated the overlap between Prxl and Otx-I in the mammillary area where expression of both genes was restricted to the ependymal

qf Development 52 (I 995) 5164

layer. Otx-I expression extended, in the ependymal layer, to the dorsal part of the diencephalon. Comparison with Dlx-I in a transverse section at the level of prosomer 4 (Fig. IOL-N) showed that the expression domains of these genes is separated by a gap

B. L.eussink et al. I Mechanisms of Development 52 (1995) 5164

which expresses

neither gene. The ventral border of the domain is the sulcus limitans (Bulfone et al., 1993) and runs parallel with the dorsal border of the Prxl domain. More anteriorly, in prosomer 5, Dlx-I is at day 12.5 also expressed in the dorsal half of the basal plate, whereas Prxi was at this level not detectable in the midline area and is present more laterally. From comparison of the expression patterns in transverse serial sections we learned that the posterior border of the Prxl domain was a few sections anterior from that of the Dlxl domain (not shown). This indicates that the posterior limit of Prxl expression is between the postulated P4 prosomer and the P2/P3 boundary (the zona limitans intrathalamica, which is the Dlx-I border). It remains, however, uncertain whether it coincides with the P3/P4 border. The broad expression with diffuse borders that was visible very antero-dorsally in the prosencephalon at day 9.5 (Fig. 10B) and in the telencephalic vesicles at day 10.5 (Fig. lOD-G) evolves in a more complex pattern at later stages. Generally this expression was limited to the mitotically active cells in the ependymal zone which surrounds the lumen; see Fig. 100,P. From day 11.5 on, expression is detected in an area surrounding the entrance point of the optic nerve in the diencephalon. This is shown for a 12.5-day embryo in Fig. 100,P and at higher magnification in Fig. lOQ,R. Expression continues after birth; in brains isolated from a newborn mouse we observed a highly specific area of expression antero-ventrally from the third ventricle (see Fig. lOS,T) in an area extending from the infundibular recess to the preoptic recess. High expression was seen in the ventral ventricular zone of the third ventricle and in the ventromedial hypothalamic nucleus. Presumably these structures originate from the region in the developing thalamus (the mammillary area) where Prxi is expressed at mid-gestation. Expression was also seen just beneath the mammillary bodies which were themselves negative. Expression was also prominent in the meninges, a site where both Prx genes are expressed throughout development. These are the oldest brains that we analyzed by in situ analysis; in brains of adult mice we detected Prxl RNA by blot analysis (not shown).

61

3. Discussion

Dlx-Z expression

Sequence comparison of Prx genes from chick and mouse indicates that Prxl and Prx2 have arisen from a gene duplication well before the diversion of aves and mammals. They may therefore share similar functions in all vertebrates. The highly specific expression patterns presented in this paper indicate that functions of the Prx genes go beyond a general cellular function, but their complexity and heterogeneity opens the possibility that they have been recruited for diverse biological roles. At the molecular level, Prxl may, apart from functioning as a transcription factor by itself, have an additional role linked to its potential to influence the interaction between serum response factor (and perhaps other MADS transcription factors) and its targets. This implies that its function may depend on co-localization with other factors. Evidence for homo- and hetero-dimerization between paired-like proteins including non-Pax proteins with homeodomains containing a glutamine at position 50 was supplied by Wilson et al. (1993). Our own in vitro studies on the Prx2 protein (De Jong et al., 1993) have not indicated the potential of this homodimerization, also making heterodimerization less probable. In spite of the intricacy of the patterns, temporal expression of particularly Prxl in the heart and brain constitute a coherent line of association with the morphogenesis of the associated structures. From the earliest stages at which the atrioventricular cushions can be distinguished, Prxl is expressed in the progenitors of the valves in which they remain active during valve formation. We do not know whether expression persists in adult heart valves, but Prxl mRNA is known to be present in RNA preparations from total adult heart (Cserjesi et al., 1992). Similarly we observed expression in a specific part of the ventral forebrain region throughout development culminating in a specific expression pattern in brains of newborn mice. Our data do not support a role of Prxl in striated muscle development as proposed by Cserjesi et al. (1992) who isolated Prxl (MHox) by virtue of its binding to a regulatory element in the muscle creatine kinase promoter

Fig. 10. Prxf expression in the developing brain. Numbers in upper right corner indicate the age of the embryo and gene names the probe used; no indication of probe means that Prxf was used; nb indicates new-born mouse. (A-C) Expression in a 9.5day embryo, as shown in sagittal sections. The sharply delineated expression domain in the diencephalon is clearly seen. (C) Higher magnification of same section shown in (A) and (B). (D,E) Transverse section through brain of 10.5day embryo. Note lack of signal in an area corresponding to a few cells in the midline of the diencephalic expression domain. (F-K) Comparison of Prxl and Ofxl expression boundaries. (F-H) Sagittal sections of 10.5day embryo, The white arrows indicate the common ‘P3/P4’ border. It should be noted that in adjacent sections (not shown), the boundaries do not co-locate. (I-K) adjacent transversal sections of 12.5-day embryo. (J-N) comparison of DZxl and Prxl expression in transversal sections of a 12.5-day mouse embryo. The Dlxl expression boundary is used as a marker of the sulcus limitans. Clearly, Prxl expression does not extend to this border. (O-R) Prxl hybridization with transversal section through 12.5.day embryo. (0,P) Low power magnification showing signal in ependymal layer of telencephahc vesicle, and expression in diencephalon near entrance point of optical nerve, which is more clearly seen in the high-power magnification shown in (Q) and (R). Note the much higher level of expression in cranial mesenchyme and sclera of the eye. (S,T) Example of expression in brains of newborn embryo, This is a relatively long exposure, explaining why the expression in the meninges, which was observed at all stages for both genes, seems higher. 3v/4v, third and fourth ventricle; cb, cerebellum; di, diencephalon; ellez; ependymal layer/zone; ir, infundibular recess; Iv, left ventricle; mf, mesencephalic flexure; oc, optic chiasma; OS, optic stalk; po, preoptic area; rh, rhombencephalon; t, telencephalon; tv, telencephalic vesicle; vz, ventricular zone. Bars 200pm.

62

B. L.eussink et al. I Mechanisms ofDevelopment 52 (1995) 51-64

and detected Prxl in preparations of skeletal, heart and smooth muscles. We did observe, however, expression in the smooth muscle tissue of certain heart vessels. We see in general that Prxl expression is frequently associated with prospective connective tissues, for example the future meninges, the ‘skeleton of the eye’ in the eye lids, the cartilage of the trachea, the pads and at least some tendons of the feet and possibly the fascia of limb muscles. In the developing heart too, Prxl expression appears to correlate with the non-muscular parts, in particular all valves. Pm1 and Pm2 are markedly expressed in the mesenchyme that is present at the arterial and venous poles of the developing heart. In view of the observation that the Prx genes are expressed in a number of areas known to be derived from the neural crest (in particular the visceral arches and many of its derivatives) an obvious question would be whether neural crest-contributed mesenchyme in the heart expresses Prx genes. Recently the venous pole of the heart, which was not known to have a neural crest cell contribution, has been redescribed for pulmonary vein development. In early development in the avian embryo this area is known to stain with the HNKl monoclonal antibody and in the mouse it is Leu-7 positive (Luider et al., 1993; Gittenberger-de Groot et al., 1995). As HNKl has been described to stain migrating neural crest cells it has been postulated that this staining is indicative for neural crest. Chicken-quail chimera studies have, however, shown that neural crest cells do not arrive in this area at the time indicated. In the outflow tract we see that the condensed mesenchyme of the aortopulmonary septum and the prongs that extend into the endocardial outflow tract ridges express particularly Prxl. From studies with chimeras (Kirby and Waldo, 1990) and retroviral marking (Noden et al., 1995) we know that neural crest cells are present in this mesenchyme. On the other hand, not only neural crest cells, but also splanchnic mesenchyme contributes to these structures. The very high expression of Prx2 in the ductus arteriosus is intriguing. The morphogenetics of this muscular artery, bound to close spontaneously after birth, remains elusive (Gittenberger-de Groot et al., 1995). Strikingly, the adjoining elastic arteries, which are also rich in smooth muscle, do not express Prx2. It is noteworthy that disruption of Hoxa3 (Chisaka and Capecchi, 1991) has been reported to lead to abnormalities in structures expressing Prxl and Prx2 (e.g. ductus arteriosus, semilunar valves and great thoracic veins). It could be informative to study Prx expression in Hoxa3-Imice. The expression pattern of Prxl in the heart resembles that of Msx-I (Chan-Thomas et al., 1993) in that these genes are both expressed in endocardial cushion tissue and subsequently in developing valves. In contrast, Msx-2 is not expressed in the epicardial covering of the heart and is expressed in the conduction system of the developing heart, in which we did not detect transcripts of either Prx gene.

In the developing limb, Prx expression appears initially to be linked to an undifferentiated state of the mesenchyme. Upon differentiation into cartilage cells expression is down-regulated except in certain connective tissues (see also Opstelten et al., 1991; Kuratani et al., 1994). Our observation that Prxl expression precedes limb bud formation opens the possibility that it also functions at some level in the cascade between the anteroposterior signals that determine the site of the limb development and the initiation of limb bud formation. Unlike Msx-I (Coelho et al., 1991; Robert et al., 1991) Prx genes are unlikely to be regulated by signals the apical ectoderma1 ridge since first, expression starts prior to detectable ridge formation, and second, expression in limbless mutant material was normal. In view of the general tendency of both Prx genes to be expressed in mesenchyme, Prxl expression in the brain may reflect recruitment of the gene for a completely different biological role for which the biochemical properties of this transcription factor were suitable. The significance of the brain pattern is that it appears to predict early on a segmentation process which becomes morphologically manifest only later, rather than being merely a consequence of a differentiation process. Identification of signals responsible for the early Prxl pattern in the developing hypothalamus may reveal factors that have a function in the early patterning of the brain. The analysis of gene expression patterns in the brain has strongly modified the way of thinking about the organization of the brain. The co-location of rhombomeric boundaries with Hox expression patterns has demonstrated that morphological boundaries and expression boundaries of important regulatory genes may coincide. These and other observations have resulted in a speculative hypothesis of Bulfone et al. (1993) and Puelles and Rubenstein (1993) who propose a model for the organization of the brain, on the basis of the published expression patterns of a series of regulatory genes integrated with morphological evidence and findings from lineage experiments. The segment boundaries of the six ‘prosomeres’ (forebrain neuromeres) described in this model are only in part respected by the Prxl expression domain. It is unclear which role in the patterning of the brain, if any, is played by the ventral midline area anterior from the floorplate where Prx-1 is expressed. The floorplate is traditionally considered to end around the midbrainlforebrain barrier since it is known to be induced by the notochord. The Sonic Hedgehog (Shh) gene (Echelard et al., 1994) has recently been shown to be functionally linked to patterning of the floor plate, and is expressed accordingly in notochord and floor plate. Interestingly, it is also expressed in the forebrain, however dorso-laterally from the midline region, in the sulcus limitans (see Rubenstein et al., 1994). The unique segment-like expression domain of Prxl in the forebrain appears to be a segment of a re-

B. Leussink et (11.I Mechanisms of’ Development 52 (1995) 5144

gion more ventral than the basal plate which directly overlays the prechordal plate. It is unclear whether there is a role of this ventral midline structure in the forebrain in the organization of the brain. It would be interesting to investigate whether this aspect of Prxl expression depends on signals from the prechordal plate. In conclusion, the Prx genes have unusually complex expression patterns, making them candidates to have roles in diverse aspects of development. Obviously, to answer the question what these biological roles are, mutational analysis of these genes in vivo will be indispensable, and is currently being undertaken. The expression studies described here should be helpful in the analysis and understanding of the phenotypes that will be observed. 4. Experimental

procedures

4.1. Embryos Mouse embryos were F2 from crosses between C57BY6 and CBA mice. Gestation was assumed to have begun at midnight before the plug was observed. For embryos younger than 8 days, whole decidua were prepared and sectioned. Fertilized White Leghorn chicken eggs were purchased from Drost, Loosdrecht, The Netherlands. Staging of chicken embryos was based on Hamburger and Hamilton (1992). 4.2. In situ hybridization Preparation and sectioning of embryos and subsequent treatment and hybridization of sections with radioactive probes was essentially as described by Wilkinson et al. (1987) using modifications described previously (Vogels et al., 1990; Opstelten et al., 1991). Whole-mount, nonradioactive hybridizations were essentially as described by Wilkinson (1992). 4.3. Probes To generate radioactively labeled probes we used the following plasmids: for Prxl, a 1.8-kb EcoRI cDNA fragment cloned in pBluescript, which was obtained from P. Cserjesi and E.N. Olson, Houston, TX (Cserjesi et al., 1992); this plasmid was cut with Bgll and transcribed with T3 polymerase resulting in a probe not containing the homeobox region; Prx2, a 576-bp cDNA fragment cloned in pBluescript (see Opstelten et al., 1991); Dlx-I, a 164-bp PstI-Kpnl genomic fragment cloned in pGEM-3 obtained from D. Duboule (see Price et al., 1991); Utx-Z, a 400-bp cDNA fragment obtained from E. Boncinelli (see Simeone et al., 1993). 4.4. NucLeic acid procedures Two chicken cDNA libraries, both from lo-day embryos, were screened. One was a gift of G. AB (Groningen) and originated from B. Venstrijm (Stockholm). The RNA used for it was from 10 day-old decapitated chicken embryos with most of the intestine removed. The second

63

library, a gift of M.-J. Vaessen (Rotterdam), was originally from Clontech. Sequencing was performed using an ABI 373 automatic sequencer. Other nucleic acid procedures were according to standard procedures (Sambrook et al., 1989). The accession number of the reported chicken Prx2 cDNA fragment is X79695. 4.5. Nomenclature We propose and use in this paper a modified nomenclature for S8/MHox which is in better conformity with the official rules and guidelines and expresses the strong structural relationship between these genes (Committee on Standardized Genetic Nomenclature for Mice, 1994). The multitude of names (MHox, Phox, K2 and Pmx) that have been used for Prxl are either misleading or lack system, the general consent being now that the abbreviation ‘Hox’ be used for the clustered homeotic-like Hox[ad] genes (Scott, 1993). Furthermore, the name S8 has been previously used for a ribosomal protein. Nohno et al. (1993) have used the name Prx-I (for paired related homeobox) for the chicken homologue of MHox. Although the S8 gene was described prior to its family member, we chose to conform to the name used by Nohno et al. in their published paper. The name Pmx for paired related mesoderm homeobox recently coined by Kern et al. (1994) for Prxl seems unfortunate since we show in this paper that it is expressed in the brain. Acknowledgements We thank P. Cserjesi and E.N. Olson for the mouse MHox cDNA, and G. AB and M.-J. Vaessen for chicken cDNA libraries. We are also grateful to E. Boncinelli and D. Duboule for Dlx and Otx probes, respectively, and to several other investigators for probes used for experiments not presented in this paper. We thank B. Robert and J. Fallon for limbless embryo material. We thank O.H.J. Destree and K.A. Lawson for discussions and interest, and to J. Deschamps for advice concerning wholemount in situ hybridization and for reading the manuscript. We finally would like to acknowledge J. Vervaart for sequence analysis, and F. Vervoordeldonk and J. Heinen for photography. References Albano, R.M., Arkell, R., Beddington, R.S.P. and Smith, J.C. (1994) Development 120,803-8 13. Bartelings, M.M. and Gittenberger-de Groot, A.C. (1988) Anat. Embtyol. 117,537-542. Bulfone, A., Puelles, L., Porteus, M.H., Frohman, M.A., Martin, G.R. and Rubenstein, J.L.R. (1993) J. Neurosci. 13,3155-3172. Btirglin, T.R. (1994) In Duboule, D. (eds.), A Comprehensive Classification of Homeobox Genes, Oxford University Press, Oxford. pp. 27-7 1. Ghan-Thomas, P.S., Thompson, R.P., Robert, B., Yacoub, M.H. and Barton, P.J.R. (1993) Dev. Dynam. 197.203-216. Chisaka, 0. and Capecchi, M.R. (1991) Nature 350, 473-479.

64

B. Lassink et ai. I Mechanisms

Coelho. C.N.D., Krabbenhoff, K.M., Upholt, W.B., Fallon, J. and Kosher, R.A. (1991) Development 113, 1487-1493. Committee on Standardized Genetic Nomenclature for Mice (Davisson, M.T., chairperson) (1994) Mouse Genome 92(2), vi-xxiii. Cserjesi, P., Lilly, B., Bryson, L., Wang, Y., Sassoon, D.A. and Olson, E.N. (1992) Development 115, 1087-l 101. Cserjesi, P., Lilly, B., Hinkley, C., Perry, M. and Olson, E.N. (1994) J. Biol. Chem. 269, 1674&16745. Davidson, D.R., Crawley, A.. Hill, R.E. and Tickle, C. (1991) Nature 352,429-431. De Jong, R. and Meijlink, F. (1993) Dev. Biol. 1.57,133-146. De Jong, R., Van der Heijden, J. and Meijlink, F. (1993) Nucleic Acids Res. 20,471 l-4720. De Robertis, E.M. (1994) In Duboule, D. (eds.), The Homeobox in Cell Differentiation and Evolution, Oxford University Press, Oxford. pp. 1 l-23. Echelard, Y., Epstein, D.J., St Jacques, B., Shen, L., Mohler, J., McMahon, J.A. and McMahon, A.P. (1993) Cell 75, 1417-1430. Fallon. J.F., Frederick. J.M., Carrington, J.L., Lanser, M.E. and Simandl. B.K. (1983) In Fallon, J.F. and Caplan, AI. (eds.), Studies on a Limbless Mutant in the Chick Embryo, Alan R. Liss, New York. pp. 3343. Figdor, M.C. and Stem, C.D. (1993) Nature 363,630634. Frigerio, G.. Burri. M., Bopp, D., Baumgartner, S. and Noll, M. (1986) Cell 47,735-746. Gittenberger-de Groot, A.C., Bartelings, M.M. and Poelmann, R.E. (1995) In Clark, E.B., Markwald, R.R. and Takas, A. (eds.), Developmental Mechanisms of Heart Disease, Future Publishing, Armonk, NY. In press. Goulding, M.D., Calepakis, G., Deutsch, U., Erselius, J.R. and Gruss. P. (1991) EMBO J. 10, 1135-1147. Grueneberg, D.A., Natesan, S., Alexandre, C. and Gilman, M.Z. (1992) Science 257, 1089- 109.5. Hamburger, V. and Hamilton, H.L. (1992) Dev. Dynam. 195, 231-272. Julian, J., Chiquet-Ehrismann, R., Erickson, H.P. and Carson, D.D. (1994) Development 120,661-671. Kappen, C., Schughart, K. and Ruddle, F.H. (1993) Genomics 18, 5470. Kern, M.J., Witte, D P., Valerius, M.T., Aronow, B.J. and Potter, S.S. (1992) Nucleic Acids Res. 20, 5189-5195. Kern, M.J., Argao. E.A , Birkenmeier, E.H., Rowe, L.B. and Potter. S.S. (1994) Genomics 19, 334-340. Kirby, M.L. and Waldo, K.L. (1990) Circulation 82,332-340. Kongsuwan, K., Webb. E., Housiaux, P. and Adams, J.M. (1988) EMBO J.7.2131-2138. Krumlauf, R. (1994) Cell 78, 191-201. Kuratani, S., Martin, J.F., Wawersik, S., Lilly, B., Eichele, G. and Olson, E.N. (I 994) Dev. Biol. 16 1, 357-369.

of Development 52 (1995) 5144 Luider, T.M., Bravenboer, N., Meijers, C., Van der Kamp, A.W.M., Tibboel, D and Poelmann, R.E. (1993) Anat. Embryol. 188, 307316. Maden, M. (1994) Nature 371, 560-561. Manova, K., Paynton, B.V. and Bachvarova, R.F. (1992) Mech. Dev. 36.141-152. Noden, D.M., Poelmann, R.E. and Gittenberger-de Groot, A.C. (1995) Trends Cardiovasc. Med., in press. Nohno, T., Koyama, E., Myokai, F., Taniguchi, S., Ohuchi, H.. Saito, T. and Noji, S. (1993) Dev. Biol. 158254-264. Opstelten, D.E., Vogels, R., Robert, B., Kalkboven, E., Zwartkruis, F., de Laaf, L., Destree, O.H., Deschamps, J., Lawson, K.A. and Meijlink, F. (1991) Mech. Dev. 34,29-42. Piacentini, M. and Autuori, F. (1994) Differentiation 57, 5161. Price, M., Lemaistre, M., Pischetola, M., Di Lauro, R. and Duboule, D. (1991) Nature 351,748-750. Puelles, L. and Rubenstein, J.L.R. (1993) Trends Neurosci. 16, 472479. Robert. B., Lyons, G., Simandl, B.K., Kuroiwa, A. and Buckingham, M. (1991) Genes Dev. 5,2363-2374. Rubenstein, J.L.R.. Martinez, S., Shimamura, K. and Puelles, L. (1994) Science 266,578-580. Sambrook, J., Fritsch, E.F. and Mamatis, T. (1989) Molecular Cloning. A Laboratory Manual, 2nd edn., Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. Schneitz, K., Spielmann, P. and Noll, M. (1993) Genes Dev. 7, 114129. Scott, M.P. (1993) Nucleic Acids Res. 21, 1687-1688. Simeone, A., Acampora, D., Mallamaci, A., Stomaiulo, A., D’Apice, M-R., Nigro, V. and Boncinelli, E. (1993) EMBO J. 12, 2735-2747. Tickle. C. and Eichele, G. (1994) Annu. Rev. Cell Biol. 10, 121-152. Treisman, J., Gonczy, P., Vashishtha, M., Harris, E. and Desplan, C. (1989) Cell 59,553-562. Valarche, I., Tissier-Seta, J-.P., Hirsch, M.R., Martinez, S.. Goridis, C. and Brunet, J.F. (1993) Development 119, 881-896. Vogels, R., de Graaff, W. and Deschamps, J. (1990) Development 110, 1159-1168. Webster, P.J. and Mansour, T.E. (1992) Mech. Dev. 38. 25-32. Wilkinson, D.G. (1992) In Wilkinson. D.G. (eds.), Whole Mount In Situ Hybridisation of Vertebrate Embryos. IRL Press, Oxford. pp. 75-83. Wilkinson, D.G., Bailes, J.A., Champion, J.E. and McMahon, A.P. (1987) Development, 93,493-500. Wilson, D., Sheng, G., Lecuit, T., Dostatni, N. and Desplan, C. (1993) Genes Dev., 7,2 120-Z 134.