During early vertebrate development coordinated movements of groups of cells lead to the formation of the three germ layers, the ectoderm, the mesoderm and the endoderm. This process is called gastrulation, and by the end of it the main regions of the embryo - head, trunk and tail - become determined. Our work is aimed at identifying the genetic mechanisms that direct the formation of the body plan in frog or mouse embryos. A foundation for understanding gastrulation was provided by an experiment carried out by Spemann and Mangold 70 years ago involving grafting of the dorsal lip. The dorsal blastopore lip of the amphibian embryo is the region where gastrulation begins. When they transplanted the dorsal lip into the ventral (opposite) side of a host embryo a secondary body axis, or siamese twin, was obtained. The grafted tissue was able to recruit, or organize, cells of the host into the twinned axis. This "organizer" experiment has been very influential, having led to our current view of development as a series of cell-cell interactions that determine cell differentiation.
The amphibian embryo affords many advantages to the molecular embryologist. The embryo is large, can be easily dissected to make region-specific cDNA libraries, and can be experimentally manipulated by microinjection and transplantation at early stages before the body plan is determined. However, definite proof of the function of a candidate gene requires lines of transgenic animals and targeted gene disruptions that can at present only be achieved in the mouse. In addition, in mammals gastrulation is the least understood period of development, even though most spontaneous conceptus losses occur at this crucial stage (14-20 days in humans). Studying the mammalian gastrula is rather difficult due to the small size and uterine implantation of the embryo. Our strategy is to isolate genes of interest from embryos of the frog Xenopus and then to study the function of their mammalian homologues. It is hoped that this shuttling between frog and mouse will help better understand the elements of gastrulation that are common to all vertebrates.
Homeobox-containing genes in the organizer
Homeobox genes are one of the families of genes involved in the control of development. They were so named because the sequence of their DNA-binding domain is conserved among homeotic genes that determine the identity of body regions in Drosophila. The isolation of homeobox-containing genes from vertebrates - initially those of the Hox family in Xenopus - using probes derived from fruit fly DNA had already provided a way for isolating developmental control genes from organisms lacking the sophisticated genetics available in Drosophila. More recent results indicate that certain vertebrate homeobox genes appear to be also important in the control of gastrulation.
To isolate genes involved in the organizer phenomenon a cDNA library was prepared from manually dissected Xenopus dorsal lips. An organizer-specific homeobox gene, called goosecoid (gsc), was identified. The gsc protein binds to DNA with a specificity similar to bicoid, which is the gene responsible for anterior development in the fruit fly. The expression of gsc closely matches the location of Spemann's head organizer tissue in development (first in the dorsal lip and then in the prechordal plate or head mesoderm). When gsc synthetic mRNA is injected into the ventral part of an early Xenopus embryo, a twinned axis results. The injected cell is able to recruit neighboring non-injected cells into the secondary axis. Because gsc is a DNA-binding protein these inductive effects should be mediated by target genes downstream of gsc, some of which should encode secreted signalling factors. The gsc product also has potent effects on the injected cell itself, triggering cell movements in the anterior direction and dorsalizing the differentiation of ventral mesoderm. Thus, one can mimic some of the properties of Spemann's organizer by transplanting the goosecoid gene product rather than a group of cells. However, gsc does not by itself convert ectodermal cells into dorsal mesoderm, and thus the cooperation of other factors is required.
In addition to gsc, several other genes encoding DNA-binding proteins have been found to be expressed specifically in the organizer region, namely Xlim-1, XFH-1 and Xnot, which have been identified by the laboratories of I. Dawid, M. Jamrich and D. Kimmelman, respectively. All these genes are involved in organizer function, and understanding their precise roles and how they cooperate with each other will be the focus of much research in coming years. It is already clear, however, that inductive signals mediated by growth factors such activin and Wnts activate the transcription of homeobox genes in the organizer region. These DNA-binding proteins in turn should activate secreted target genes that mediate cell-cell interactions. Identifying these target genes is of great interest, for the organizer is the source of the signals that pattern the vertebrate body plan.
Chordin, a downstream target of organizer-specific homeobox genes
Using differential screening of cDNA libraries and functional assays in Xenopus embryos, we are currently searching for genes downstream of gsc. One interesting candidate has been isolated so far. This gene, chordin, is expressed at the right time and place to regulate cell-cell interactions in the organizing centers of head, trunk and tail development. Its expression can be activated by microinjection of gsc or Xnot mRNA (Xlim-1 and XFH-1 have not been yet tested). Microinjection of chordin mRNA induces twinned axes, recruiting neighboring cells as in Spemann's experiment, and can completely rescue dorsal axis development in organizer-deficient embryos. Since there are multiple organizer-specific transcription factors, it will be of interest to determine whether they converge on the regulation of a common dorsalizing factor or whether they regulate parallel, redundant pathways. Studies on the chordin promoter and its regulatory factors may provide answers to this question.
Chordin encodes a new secreted protein of 941 amino acids. It has four cysteine-rich domains found in extracellular proteins such as thrombospondin, von Willebrand factor and procollagen. The conservation is limited to the spacing of the cysteine residues, but these repeats could have a role in modulating signalling by growth factors; in the case of thrombospondin its cys-rich domain has been shown to possess anti-angiogenic activity. chordin has structural homology with a gene that was isolated concurrently in Drosophila, called short-gastrulation (sog). sog is expressed ventrally in the fly, where it functions as an antagonist of a dorsally-produced TGF-ß factor called dpp. By microinjecting sog mRNA into Xenopus embryos we have been able to show that sog has the same dorsalizing effects as chordin. This raises the possibility, to be explored further, that chordin may function by inhibiting signalling by BMP-4, which is a growth factor related to dpp expressed ventrally in the Xenopus embryo. The results also suggest that regions of the embryo that are dorsal in vertebrates correspond to regions that are ventral in arthropods. Presumably the molecular mechanisms that determine dorso-ventral polarity appeared very early in metazoan evolution. It is hoped that the study of organizer-specific genes will shed light not only on how the cellular migrations of gastrulation pattern the vertebrate body plan, but also how other body plans arose in the course of evolution.
A grant from the NIH provided support for the projects described above