We focus on discovering mechanisms that regulate the development of tissues arising from the ectodermal region of vertebrate embryos. Our experimental model system is the frog Xenopus laevis. Promoter analysis, subtractive hybridization and microarray analysis, and in situ hybridization studies have identified genes encoding factors that both respond to secreted embryonic signals, such as bone morphogenetic proteins (BMPs), and mediate localized differentiation into ectodermal derivatives such as neural plate, neural crest, and epidermis. More specifically, the studies have identified two basic categories of regulatory factors: negatively acting Distal-less-like homeobox factors Dlx3, Dlx5, and Dlx6 and the related homeobox factor Msx1; and the positively acting transcription factor AP-2a. The Msx/Dlx genes respond differentially to a graded signal generated by antagonistic interactions between secreted BMPs and inducers originating in the Spemann Organizer region. This sets up different territories in the ectoderm in which different “codes” of Msx/Dlx gene expression appear to define specific tissue types. AP2a also is regulated by BMP signaling. In the epidermis, AP2a is necessary for expression of structural genes. A combination of BMPs and secreted factors in the Wnt class leads to a rise in the level of AP2a, which is essential in the induction of neural crest.

AP2a in Epidermal Development
Luo, Thomas, Sargent; in collaboration with Weeks
The group’s earlier work suggested that AP2a might be important in controlling the expression of certain epidermal genes in Xenopus. Two different types of experiment have now con-firmed and considerably extended our previous work. First, AP2a was artificially restored to ectoderm from embryos in which BMP signaling was blocked by over-expressing a BMP antago-nist (chordin). BMP signaling is required for epidermal development. This AP2 “rescue” experiment resulted in the reactivation of epidermal gene expression, which was inhibited by the BMP antagonist. We obtained the same result even when we inhibited endogenous protein synthesis with cycloheximide, indicating that AP2a is immediately upstream from epi-dermal genes and is a mediator of BMP signaling at the transcriptional level. A second experiment used antisense oligonucleotides designed to mediate the destruction of endogenous AP2a mRNA. The loss-of-function for AP2a resulted in suppression of all epidermal gene expression and activation of neural genes (as usually happens in Xenopus when epidermis is inhibited). Gastrulation movements were also disturbed in such embryos. Therefore, AP2a is required for epidermal cell differentiation, including both the expression of molecular phenotypic markers and the rearrangement of epidermal cells that accompanies gastrulation in the frog.

Dlx3 and AP2a in the Neural Crest
Luo, Sargent; in collaboration with Saint-Jeannet
It is generally accepted that some form of spatial gradient of BMP signaling provides an important cue for localized differentiation in early vertebrate embryos. However, little is known about how such spatial information is translated into region-specific gene expression. We have made some important progress in expanding our knowledge by showing that Msx1, Dlx3, and Dlx5 are specifically excluded from cement gland, neural crest, and neural plate, respectively, probably as the result of differential responsive-ness to graded BMP signaling within the ectoderm. When misexpressed by RNA injection, or in “animal cap” assays, these factors selectively inhibit the differentiation of cell types from which they are normally excluded. The inhibition of neural crest development by Dlx3 misexpression is consistent with our earlier observation that Dlx3 can act as an antagonist of Wnt/b-catenin signaling in frog embryo: Wnt signaling is required for neural crest induction. Thus, Dlx3 may function both to interpret intermediate BMP signal strength and to confine the response to Wnt signaling, spatially limiting the neural crest territory.

AP2a has been used as a marker for cranial neural crest for some time, and AP2a-null mice have major craniofacial abnormalities, suggesting a role for this factor in neural crest development, at some level. Using the same tools that revealed the need for AP2a function in the epidermis, we have shown a similarly essential role for AP2a in the initial specification of neural crest in Xenopus. We are currently testing the theory that positive control by AP2a and negative control by repressors including Dlx3 work together to specify this important tissue type.

Regulatory Networks
Luo, Lepperdinger, Khadka, Zhang, Sargent; in collaboration with Chandhoke, Grant, Christensen, Fryxell
In this project, we are using Dlx and AP2a genes as starting points to work out regulatory “wiring diagrams” for the induction and differentiation of epidermis and neural crest. The guiding hypothesis is that the control of these developmental programs is largely transcrip-tional in nature but depends on both direct interaction of regulatory proteins with target sequences in genomic DNA and protein-protein interactions with other control factors. We are thus taking two approaches to identifying genes that interact with Dlx3/5 and AP2a: target gene identification by microarray analysis and two-hybrid screens for interacting proteins.

In an initial study, we generated a subtracted library enriched in epidermis-specific genes. From this population, we performed 2,000 cDNA sequence reads, yielding about 1,000 unique expressed sequence tags (ESTs)s. Our collaborators at George Mason University amplified and spotted the ESTs. We then hybridized the arrays with probes derived from animal caps of embryos injected with BMP antagonists with the aim of identifying novel BMP-dependent genes that might be involved in regulating the formation of the epidermis during gastrulation. While the process led to the identification of several interesting ESTs and a number of previously characterized genes, it did not suggest any direct targets of AP2a or indicate that connecting the genes’ function with that of AP2a or other epidermal factors would be a straightforward proposition. A somewhat more focused microarray project is now under way. We are generating a subtracted and normalized cDNA library that should be enriched in early-expressed genes specific to the neural crest. Approximately 3,000 cDNA clones will be amplified and spotted onto glass microarrays. Probes will then be derived from animal cap ectoderm treated with various inducers and repressors of neural crest, including reagents specific for AP2a. At a minimum, we hope to identify novel neural crest marker genes that would augment the rather small collection currently available. Ideally, this approach will also identify candidate targets for the action of AP2a as a transcriptional activator in neural crest precursor cells. Subsequent gain and loss of function experiments can then be used to determine functional significance while transcriptional assays in embryos and animals caps can be employed to analyze the interactions between target DNA elements and AP2a protein.

To search for proteins that interact physically with either Dlx3/5 or AP2a, we are using a bacterial “two hybrid” system with both commercially available cDNA libraries and specialized libraries constructed in our laboratory. We hope that this prokaryotic system will circumvent the cell cycle inter-ference toxicity we previously encountered with Dlx3 in a yeast two-hybrid system. Any candidate interaction factors will be subjected to whole-mount in situ hybridization analysis and ultimately to functional studies using antisense or dominant negative overexpres-sion strategies.