Experimental analysis of the control of expression of the homeobox-gene Msx-1 in the developing limb and face

Genes of the Antennapedia-type homeobox series of the Hox A-D (Scott, 1992) complexes appear to encode axial position within the vertebrate embryo (reviewed by Duboule, 1991; Hunt and Krumlauf, 1992). In contrast, two more divergent homeobox-containing genes, Msx-1 and Msx-2, are expressed in many different regions of the embryo and particularly at sites where epithelial-mesenchymal interactions are occurring (reviewed by Davidson and Hill, 1991). In the limb the expression patterns of Msx-1 and Msx-2 suggest a role in interactions that mediate bud outgrowth, and grafting experiments show that mesenchymal expression of the genes is position-dependent (Davidson et al., 1991). The pattern of expression of the genes in the facial primordia shows a striking parallel with expression in the limb bud. This raises the possibility that these homeobox genes play a similar role in both limb and facial development and that the same mechanisms control expression of the genes in both regions of the embryo.

Mouse Msx-1 and Msx-2, previously known as Hox-7 and Hox-8, share homology with the Drosophila Msh homeobox gene (Hill et al., 1989; Robert et al., 1989; Monaghan et al., 1991; Mackenzie et al., 1991, 1992). More recently, homologous genes in birds have been isolated (Takahashi and Le Douarin, 1990; Coelho et al., 1991a; Yokouchi et al., 1991; Suzuki et al., 1991; Robert et al., 1991). The patterns of expression of these two genes during embryonic development are complex, but a striking feature of their expression is the overlapping pattern of transcripts in the tissues of the developing face (Mackenzie et al., 1991, 1992; Takahashi and Le Douarin, 1990) and limbs (Suzuki et al., 1991; Yokouchi et al., 1991; Robert et al., 1991). In the early buds of both limbs and facial primordia (frontonasal mass, maxillary, mandibular and lateral nasal processes) Msx-1 and Msx-2 are expressed (Hill et al., 1989; Mackenzie et al., 1991) and, as the buds elongate, mesenchymal expression becomes restricted to the tips of the limb buds and facial primordia.

In the limb, two lines of evidence show that mesenchy-mal expression of the Msx-1 and Msx-2 genes at the tip is controlled by signals emanating from the apical ectodermal ridge, the thickened rim of epithelium that mediates bud outgrowth. Firstly, in the chick mutant, limbless, which lacks an apical ridge, mesenchymal expression of Msx-1 and Msx-2 is reduced but can be restored to normal levels by grafting an apical ridge from a normal embryo (Robert et al., 1991; Coelho et al., 1991b). Secondly, in a different approach, heterospecific grafts of limb bud mesenchyme were carried out between mouse and chick embryos (Davidson et al., 1991). When grafts of mouse limb cells are placed beneath the apical ridge in chick wing buds, the grafted cells express both genes irrespective of whether they were taken from distal (expressing), or proximal (non-expressing) regions; cells placed at a distance from the apical ridge do not express the genes.

Outgrowth of the major facial primordia also involves epithelial-mesenchymal interactions. When the ectoderm is removed from facial mesenchyme and the mesenchyme is then grafted to the dorsal surface of limb buds to continue development, no outgrowth occurs (Wedden, 1987; Richman and Tickle, 1989). This is similar to the effects of removing the apical ectodermal ridge from limb buds, which gives limb truncations. However, in the face, unlike the limb bud, there is no readily identifiable apical ridge rimming the facial buds (Wedden et al., 1988).

Here we use mouse/chicken heterospecific grafts to investigate further the control of Msx-1 in the mesenchyme of both the developing limb and face. Using probes that recognize species-specific transcripts, we are able to compare the level of Msx-1 expression in the grafted mouse cells and the surrounding host chick cells. This approach provides a way of testing whether expression of the gene in mesenchyme cells is regulated by local conditions in different regions of the embryo. The results show that, in the face, as in the limb, expression of Msx-1 appears to be position-dependent. However, although facial mesenchyme can respond to signals in the limb bud, Msx-1 is not activated when limb bud cells are placed in the face.

10- to 10.5-day p.c. C57 Black and Tan mouse embryos were used as donor embryos for the tissue grafting experiments. The embryos were removed from the decidua, transferred to a tissue culture medium (Minimum Essential Medium + 4 mM glutamine + 10% foetal calf serum + 200 units ml^-1 penicillin, 200 μg ml^-1 streptomycin and 0.5 μg ml^-1 fungizone (GIBCO Biocult)) and the extraembryonic membranes removed. The fore-limb buds, the maxillary, mandibular and medial nasal processes were dissected from the embryos, treated with 2% trypsin on ice for 1 hour and then the ectoderm removed. Pieces of tissue (200×200 μm) were cut from distal and proximal mesenchyme of the buds and facial processes using sharpened tungsten needles. The pieces of mes-enchyme were then grafted either into wing buds of stage 20 or 22 chick embryos (Hamburger and Hamilton stages), or into the maxillary primordium of stage 24 chick embryos (Fig. 1).

Grafts to the wing were made in two positions. Tissue was placed distally by positioning the graft under the apical ectodermal ridge, which had been eased away from the distal mesenchyme to make a loop. Tissue was placed proximally by either inserting the graft under a flap of ectoderm or into a hole cut in the dorsal surface of stage 22 wing buds. In the face, grafts were placed in the right maxillary primordium since this is the most readily accessible of the facial processes. A hole was cut into the dorsal surface of the primordium of stage 24 chick embryos (Fig. 1) and the graft of mouse mesenchyme inserted.

After grafting, the chick embryos were incubated at 38°C for either 5 hours or 24-26 hours. They were then removed from the egg, washed in PBS and fixed overnight in 4% paraformaldehyde at 4°C. The tissue was then either processed and sectioned as described by Davidson et al. (1988), or prepared for whole-mount in situ hybridization (Wilkinson, 1992). The sections containing the grafted mouse tissue were identified by staining every eighth section with Biebrich Scarlet. The remaining sections were processed for in situ hybridization as described by Davidson et al. (1988) using ³⁵S-labelled riboprobes for the 3’ end and untranslated regions of the Msx-1 gene (Hill et al., 1989) using species-specific chick and mouse probes on alternating, adjacent, sections. Sections were exposed for autoradiography at 4°C for 3-5 weeks prior to developing and then stained with Malachite Green for dark- and bright-field photography. Whole-mount in situ hybridization was as described by Wilkinson (1992), using digoxigenin-labelled riboprobes for the mouse and chick Msx-1 genes.

In fore-limb buds of stage 19 chick and of 10.5 day mouse embryos, Msx-1 is expressed in a distally resticted fashion, with the highest levels of transcripts in the distal mes-enchyme immediately under the apical ectodermal ridge and lower levels more proximally (Fig. 2A,B). As the bud elongates, the distal mesenchymal domain is maintained (Fig. 2C,D). In the faces of stage 24 chick and 10.5 day mouse embryos, Msx-1 transcripts are found in distal mes-enchyme at the tips of the facial primordia and are present in a gradient, with the highest levels of expression distally (Fig. 3A,B).

In order to compare the domains of Msx-1 expression in graft and host, mouse mesenchyme was placed in chick wing buds and gene activity assayed by in situ hybridization using species-specific probes on adjacent sections (Table 1). In tissue fixed 24 hours after grafting, the pattern of Msx-1 expression in the grafts matches that in the adjacent host cells irrespective of the original position from which the graft cells were taken. In distally placed grafts from both distal and proximal regions of mouse limb buds, Msx-1 transcripts were expressed in a graded fashion with the highest level of transcripts in the mesenchyme immediately under the apical ridge. In some cases, the graft straddled the proximal border of the host domain and showed clearly the correspondence between expression in the graft and expression in host cells (Fig. 4A-D). The region of the graft within the host domain expressed Msx-1, but no transcripts could be detected in cells in the region of the graft that lies outside the host domain (Fig. 5A-D). In proximally placed grafts from both distal and proximal regions of mouse limb buds, which ended up completely surrounded by non-expressing cells, there was no detectable expression of Msx-1.

The rapidity of the response to a change in the position of cells in the limb was investigated in grafts after 5 hours. In 2/2 cases when proximal mouse cells were placed distally in the chick wing, Msx-1 expression was observed only in those cells closest to the apical ridge (Table 1) although the entire graft lay within the host domain of expression (data not shown). In a graft of distal cells placed proximally, Msx-1 transcripts were still present although the surrounding chick mesenchyme cells did not detectably express the gene.

To investigate whether the expression of Msx-1 in facial mesenchyme is also determined by position, mesenchyme from either proximal or distal regions of three mouse facial primordia was grafted to the chick maxillary primordium. After 24 hours, 8/15 grafts (4 taken from the maxillary process, 3 from the medial nasal process and one from the mandibular process) had ended up in a region of the primordium where the host cells were expressing Msx-1 (Table 1). In all 8 cases, irrespective of the origin of the cells, mouse transcripts could be detected within the graft (Fig. 5E-H). In the remaining 7 cases, the grafts were found in regions of low host Msx-1 expression and none of these expressed Msx-1 after 24 hours. All of these 7 grafts (4 from the maxillary primordium, 2 from the medial nasal process and 1 from the mandibular primordium) had been taken from a region in the mouse embryo that would normally show strong expression. This clearly demonstrates that expression of Msx-1 is strictly position-dependent within the primordium. One additional graft taken from the distal tip of the medial nasal process ended up protruding from the side of the primordium and there was no detectable expression of Msx-1 (data not shown).

To study whether Msx-1 expression in any given tissue can respond to signals present in different regions of the embryo, mouse facial cells were grafted to chick wing buds and mouse limb bud cells were grafted to chick maxillary process. Expression of Msx-1 in facial cells, when placed in the limb bud, is position-dependent. Just as in limb bud grafts, distally placed grafts of facial mesenchmye express Msx-1 whereas proximally placed grafts do not. Some grafts lay in a region of the host limb where the level of transcripts was graded and there was corresponding graded expression of Msx-1 in the graft (Fig. 6A-D). The similarity of the response of mouse limb and facial cells also extends to the rapidity of the response. After 5 hours, there was detectable expression in 3/4 grafts of facial mesenchyme placed beneath the apical ridge.

In contrast to the responsiveness of mouse facial cells in the limb bud, mouse limb bud cells do not appear to respond readily to local conditions in the facial primordia. 24 hours after grafting, 3/5 grafts of proximal limb cells ended up in regions of the maxillary primordium where Msx-1 expression is high but in none of these grafts was there any detectable expression of Msx-1 (Table 1; Fig. 6E-H). 24 hours after grafting distal limb bud mesenchyme into the maxillary primordium one graft (1/3) continued to express Msx-1 at low levels and was surrounded by chick cells that expressed Msx-1. The other two grafts were not in a region of high host Msx-1 expression and did not detectably express Msx-1 transcripts (Table 1). In a further series of experiments distal mouse limb tissue was grafted into the maxillary primordium and, after 24 hours, Msx-1 was assayed in the grafted cells by whole-mount in situ hybridization. In all cases (8/8), no Msx-1 transcripts were seen in the graft (Table 1). These results constrast with an experiment done at the same time and assayed in the same manner, in which distal limb bud tissue was grafted to the tip of the chick limb bud. Here, Msx-1 expression was clearly seen in the grafted tissue at 24 hours.

Mouse/chicken heterospecific grafts allow Msx-1 expression in graft and host cells to be monitored separately and compared. Mouse limb bud cells express Msx-1 in a position-dependent fashion in the chick wing bud, such that Msx-1 is only expressed when the tissue lies beneath the apical ectodermal ridge and is within the host expression domain. The same technique reveals position-dependent expression of Msx-1 in mouse facial mesenchyme grafted to the chick maxillary primordium. Transcripts of the gene are only detected when the graft comes to lie within the domain of Msx-1 expression of the host primordium. Grafts from all three major facial primordia of mouse embryos show the same behaviour when placed in the maxillary primordium. When mouse facial tissue is placed in the chick limb bud, transcripts are found only in those grafts that are beneath the apical ridge and lying alongside host cells expressing the gene. In contrast, Msx-1 expression is not activated in limb cells placed in the face.

The pattern of Msx-1 expression in the grafted mouse tissue is consistent with it being regulated by signals from the apical ectodermal ridge (see also Davidson et al., 1991; Robert et al., 1991; Coelho et al., 1991b). After 24 hours, a graded pattern of transcripts has been established in the graft that matches almost exactly that present in the host chick mesenchyme, suggesting that the response of mouse and chick cells to local signals is very similar. Msx-1 expression appears to be switched off more slowly in proximal grafts. When mouse cells from a region of high expression are grafted to an area of low expression, transcripts are still present after 5 hours but have disappeared by 24 hours. This delay may reflect transcript stability rather than a slower cellular response.

The behaviour of heterospecific grafts of mouse facial cells placed into the chick maxillary primordium suggests that Msx-1 expression is also position-dependent in the face. The correspondence between levels of Msx-1 transcripts in both graft and host tissue suggests that not only are signals qualitatively conserved but that the whole signalling and response system shows remarkable quantitative conservation between chick and mouse.

The signals that lead to the pattern of Msx-1 expression at the tips of the facial primordia may emanate from the overlying ectoderm. Since mesenchyme from three different primordia show the same behaviour when grafted to the chick maxillary process this suggests that the signals are either identical or follow the same pathways in the different primordia. In this context, it is interesting that epithelial-mesenchymal recombinations of tissues from different facial primordia undergo normal development and morphogenesis (Richman and Tickle, 1989).

Signals from the limb apical ectodermal ridge can regulate Msx-1 expression in facial mesenchyme cells. In all het-erospecific grafts made to the limb bud, Msx-1 transcripts are found in cells that lie close to the apical ridge and the graded distribution of transcripts matches the pattern of transcripts in the neighbouring limb bud cells. Msx-1 is switched on as rapidly in face mesenchyme grafted to the tip of limb buds as the limb-to-limb grafts, suggesting that the face tissue can respond efficiently to local signals in the limb. However, suprisingly Msx-1 expression is not activated in proximal limb mesenchyme cells grafted to the face. In addition, when Msx-1-expressing tissue is grafted into the maxillary primordium, Msx-1 transcripts are down-regulated by 24 hours. This suggests that signals in the face are not able to support high level Msx-1 expression in the limb bud cells. Recombination experiments in which tissues are exchanged between limb bud and facial processes also suggest that epithelial-mesenchymal interactions are not identical in the face and limb (Richman and Tickle, 1992). Facial mesenchyme from each of the three major primordia can maintain the limb apical ectodermal ridge (Richman and Tickle, 1992; see also Cairns, 1965) and recombinations between the frontonasal mass and the limb tissues produced substantial outgrowths. However, recombinations between tissue from the two different sites do not undergo as extensive morphogenesis and patterning as when either limb tissues or facial tissues are recombined.

One possibility is that there are either quantitative or qualitative differences in the extracellular signals that control Msx-1 expression in the limb and face mesenchyme. For example, if limb bud cells require a higher level of signal to activate Msx-1 than facial cells, this would explain the difference in behaviour of the face and limb grafts. A second possibility is that the signalling system is the same in both parts of the embryo but the response of the cells involves other factors. For example, although BMP and Wnt genes, in addition to Msx-1 and Msx-2, are expressed in both face and limb, genes of the HOX complexes are expressed in the limb but not the face. Activation of Msx-1 genes in the face cells grafted to the limb could be dependent on changes in expression of other genes within the grafted cells. In this context, it is interesting that when anterior limb bud cells are placed posteriorly in the limb bud 5’ Hox D genes are activated. It is possible that this behaviour has a parallel with the activation of Msx-1 in face to limb grafts where tissue is moved in an anterior to posterior direction in the embryo (Izpisúa-Belmonte et al., 1992).

Research of J. M. B., L. G. R. and C. T. is supported by grants from the Medical Research Council; G. H. M. was in receipt of a Wellcome vacation studentship. We thank Anne Crawley for help with histology and photography.