Project 1

Establishing an infrastructure for genetic research on X. tropicalis.

Project 2

Generating stable transgenic X. tropicalis lines to streamline current embryological procedures and create novel experimental paradigms.

Project 3

Using transgenic reporter assays to identify, dissect, and compare cis-acting genomic regulatory regions in sets of genes with overlapping temporal and spatial regulation (promoter bashing).

Project 4

Identifying genes involved in axial extension, morphogenetic movements, axial patterning, and eye development via

Summary 

Establishing an infrastructure for genetic research on X. tropicalis, by:

For X. tropicalis to succeed as a genetic model system, certain basic tools must be prepared, not the least of which is an optimal set of protocols for conveniently and inexpensively maintaining healthy colonies, producing large numbers of embryos on demand, and minimizing generation time.

Additional useful tools minimize the number of generations required to perform an experiment. Development of healthy, genetically identical (isogenic) strains would provide a consistent background upon which to examine complex processes or identify mutants. Crossing different strains can locate genes with respect to a physical map of DNA polymorphisms, which will be neccessary in order to categorize genes and alleles with similar phenotypes. A small set of scorable mutants in X. tropicalis will facilitate fine-tuning mutagenesis strategies.

Practical considerations, rather than unique attributes, often make or break study systems. While we have established X. tropicalis in culture, genetic approaches will require very large-scale manipulations of eggs and embryos. Among the parameters we have identified for further optimization are 1) minimzing generation time; 2) in vitro fertilization; 3) production of completely homozygous gynogenetic diploids by suppression of first cleavage, and 4) optimization of the transgenic procedure for the smaller X. tropicalis egg.

a. live food diet: Supplementing pelleted frog food with live, moving prey items seems to greatly increase feeding enthusiasm. We are investigating several live foods, including bloodworms, X. laevis tadpoles, and brine shrimp.

b. hormome treatment: For some purposes, such as multi-generation experimental designs, hormonal treatments which further accelerate sexual maturity may be useful. Ovary development may be hastened by injections of low doses of follicle-stimulating hormone-related hormones (such as PMSG) or luteinizing hormone-related hormones (such as HCG), which accelerate vitellogenin production and uptake^82,83. Estradiol treatment also may accelerate sexual maturity of female X. laevis (D. Reinschmidt, pers. comm), and can be used to manipulate sex of tadpoles. Corticotropin-releasing hormone-like peptides have been shown to accelerate metamorphosis in other anurans (Scaphiopus and Rana)⁸⁴; however, it is not known whether the smaller froglets produced would reach sexual maturity more quickly than untreated controls. PMSG, HCG, and estradiol will be administered at intervals in a range of doses to immature frogs and compared with untreated controls.

Highly-inbred, or 'isogenic' strains reduce the variability or 'noise' complicating studies of many biological processes. A diverse set of different (polymorphic) isogenic strains is also useful for creating genetic and physical maps and locating mutations. Such strains can be generated by inbreeding sibling frogs for successive generations (we are currently inbreeding F4 sib groups) or by artificially manipulating ploidy to produce completely homozygous embryos (gynogenetic diploids).

Our breeding population of X. tropicalis may not be sufficiently genetically diverse to provide a useful set of genetically polymorphic strains for Project 1.C). We have obtained animals from geographically disjunct populations in from Pacific Biological Supply, which we believe will provide sufficient genetic variation.

Eggs from these animals and from our current colony will be collected and "fertilized" with UV-irradiated X. tropicalis sperm to make gynogenetic haploid embryos. Approximately 40 minutes after fertilization (five minutes before appearance of the first cleavage furrow), embryos will be subjected to 6000 pounds per square inch for three to six minutes in a pressure bomb apparatus.

Successful suppression of first cleavage and generation of homozygous diploid embryos is morphologically visible both by delay in the formation of the first cleavage furrow, and by rescue of the haploid phenotype (ventral edema, kinked anteroposterior axis), and will be confirmed by karyotype analysis. Creating completely isogenic populations requires clonal expansion of the progeny of individual homozygous diploid animals. This has been accomplished either by formation of gynogenetic diploids at each successive generation (using the easier "early pressure" suppression of polar body formation), or by treating the tadpoles from primary gynogenesis with estrogenic steroids, which reliably accomplishes full sex reversal in genetically-male ZZ animals77. These animals can then be cloned by gynogenesis; half the resulting progeny will then be treated with estrogens to develop as females, and half allowed to develop as males. The resulting clonal population can then reproduce sexually as long as some ZZ animals are sex-reversed.

Recessive alleles present in the maternal genome will be uncovered by production of gynogenetic diploids. Unless X. tropicalis populations have significantly different mutation loads than X. laevis, we expect to uncover naturally-occuring recessive mutations in the process of creating isogenic strains: gynogenesis experiments using eight wild-caught female X. laevis resulted in the identification of 12 developmental mutations74(see preliminary results).

Scorable mutations, regardless of their phenotype, will be very useful in subsequent genetic analysis, e.g. titrating reagents for mutagenesis and marking linkage groups. Since our current colony may not be genetically diverse, we will screen additional genomes by obtaining additional wild-caught animals from disjunct populations in West Africa in in order to identify mutants in gynogenetic progeny. Mutations that give rise to axial defects or eye phenotypes will be subjected to extensive further characterization (see Specific Aim #4.A for details).

Genetics denotes a way to order the genome, classically by determining recombination frequencies among visible phenotypic markers. Such a genetic map permits categorization of alleles with distinct phenotypes and different mutations with similar phenotypes, as well as facilitating molecular characterization of mutants by positional cloning or candidate gene approaches. In the first phase of this project, only a crude linkage map is required in order to sort mutants, with a resolution on the order of 10cM. If interesting mutants are isolated, we hope to employ alternative cloning strategies (see specific aim #4) and minimize our dependence on positional cloning methods involving fine-structure genetic mapping and chromosomal walks.

A decade ago, proposing to create a genetic map for a vertebrate with no characterized visible phenotypic markers would have been grounds for psychiatric evaluation. However, a polymerase chain reaction (PCR) application has been developed which simultaneously detects large numbers of random nucleotide sequence polymorphisms usable as genetic markers. This technique produces a map with abundant, well-distributed, highly polymorphic, readily-detectable markers56, 57, and requires relatively little effort: the bulk of the map of the zebrafish genome was prepared by a single undergraduate in a year56(Postlethwait, pers. comm.).

In this procedure, called Random Amplified Polymorphic DNA (RAPD) mapping, PCR is performed with genomic DNA templates and single ten-nucleotide-long primers of arbitrary sequence, which amplify six to twelve DNA fragments in similarly-sized genomes; distinct sets of DNA fragments are amplified by different arbitrary tenmers. If two strains of a species are sufficiently polymorphic, this will be reflected in the differential amplification of some random DNA fragments (due to failure of the primers to find identical binding sites or to insertions or deletions between sites); presence or absence of a given fragment is scored as a strain-specific genetic marker.

After collecting a set of several hundred such strain-specific markers, the RAPDs of the haploid progeny of a hybrid (linkage map cross) between two strains can be subjected to linkage analysis. Haploids are extremely useful in this procedure, since recessive markers (i.e. absence of a particular PCR fragment) are obscured in heterozygotes. Using this technique, a single undergraduate used the progeny of a single zebrafish to organize a genetic map of 401 RAPD markers into 29 linkage groups (zebrafish have a haploid complement of 25 chromosomes) with only four scorable loci remaining unlinked56(Postlethwait pers. comm).

Mutations with visible phenotypes can be placed on a RAPD map by 'bulked segregant analysis'. An animal carrying the mutation is crossed to a different strain, and haploid offspring are separated by phenotype (which should present a 50:50 ratio). DNA is purified from individuals, pooled, and computer-assisted RAPD analysis is performed as above. Genetic markers unlinked to the mutation will be present in both pools; individual linked markers will segregate with either wild-type or mutant phenotypes. By then repeating the assay using primers that produce linked markers and DNA from individual haploid embryos, it is possible to more precisely locate the mutation on the RAPD map.

Identity of a cloned gene with a given mapped mutation can be evaluated by mapping the candidate gene, using bulked segregant analysis with specific primers56. To use bulked segregant analysis to map candidate genes, specific primer sets are prepared to amplify fragments containing introns or 3' untranslated regions of cloned genes, which often contain strain-specific polymorphisms. Identified polymorphisms can then be mapped with respect to RAPD markers using previously prepared DNA from a linkage map cross. Alternatively, using another feature that is unique to amphibians, candidate genes may be mapped by fluorescence in situ hybridization (FISH) on lampbrush chromosomes (J. Gall, pers. comm.). By also performing FISH with a small set of RAPD marker probes, candidate gene positions could be correlated with the genetic map.

If we are successful in isolating a number of mutations with interesting phenotypes but alternative cloning strategies (see specific aim #4) fail, positional cloning may become unavoidable. Although not a goal during the course of the proposed work, in this case it would be useful to increase the resolution of the map to less than 0.5 cM by identifying more oligonucleotide primers which produce additional polymorphisms. RAPD fragments that are linked to mutations can be readily cloned and used as probes to initiate chromosomal walks.

 Back to top

To generate stable transgenic X. tropicalis lines with the goal of streamlining current embryological procedures or creating novel experimental paradigms, specifically

A. to facilitate preparation of genetic mosaics and supplant laborious gene expression assays with rapid in vivo reporter detection, by establishing transgenic reporter lines tissue-specifically expressing Green Fluorescent Protein (GFP) driven by characterized X. laevis regulatory elements (brachyury, cardiac actin, N-tubulin, and goosecoid promoters) with useful domains of expression;

B. to simplify tissue labeling for transplantation, by establishing transgenic lines expressing GFP ubiquitously using the cytoskeletal actin and cytomegalovirus promoters

C. to exert temporal control expression over transgenic gene function, by assaying inducible systems

D. to combine transgenesis with transplantation techniques in order to implement genetic mosaic assays

 Back to top

Stable transgenic lines expressing GFP, a reporter marker which can be assayed in vivo, have the potential to transform embryological research, where much of the labor is currently invested in assaying gene expression. Conventional detection of specific gene expression in whole mount assays (e.g. in situ hybridization, immunocytochemistry) takes several days, considerable labor, and expensive reagents. GFP can be detected instantly in living embryos or explants using fluorescence microscopy.

Transgenic reporter constructs in X. laevis have been shown to faithfully reproduce the endogenous pattern of expression of several promoters^{1, 2} (see Preliminary Results). Using the short-generation X. tropicalis instead of X. laevis facilitates breeding transgenic founders to generate populations of animals with identical reporter integrations, which should give more consistent reporter patterns for experimental analysis.

A panel of readily-assayable stable transgenic lines will also be invaluable as substrates for chemical mutagenesis (see specific aim #4). Identifying mutations in particular processes, such as eye development and axial patterning, will be much more feasible if workers can visually screen embryos for abnormalities in easily recognizable patterns of relevant gene regulation. The promoters that we propose to use have all been characterized either by plasmid injection or transgenic assays in X. laevis, and are likely to give appropriate and useful non-mosaic tissue-specific expression in X. tropicalis 1, 89. Goosecoid (gsc) is expressed in Spemann's organizer, notochord, and head mesoderm; N-tubulin is a marker for differentiating neurons; the brachyury (Xbra) elements which have been characterized reproduce the ventral mesoderm component of expression; cardiac actin is expressed in developing heart and somites.

Finally, analysis of targeted mouse mutants is often complicated by early embryonic lethality (precluding study of later phenotypes), or by pleiotropy (obscuring primary roles for gene products in particular processes). In other cases, dominance or haploinsufficiency makes propagation of mutants difficult. Even when it is possible to carry mutants as heterozygotes, only 1/4 of embryos from a cross of heterozgyotes will be homozygous; it is frequently neccessary to begin complex manipulations before homozygotes can be phenotypically identified, leading to much wasted effort. Simultaneous analysis of multiple recessive lethal alleles is even more impractical for similar reasons: only one out of sixteen embryos from a cross of two double heterozygotes will be a double homozygote. Such analysis is possible in Drosophila because of the extensive genetic tools available, such as visible markers and balancers. In some cases, the embryological tools available in amphibians may facilitate similar genetic analyses. A set of manipulations, described in section D below, are feasible with the X. tropicalis system should allow us to circumvent these problems in many cases, and underline the potential usefulness of an amphibian genetic model system.

X. tropicalis sperm nuclei will be incubated with DNA constructs fusing previously characterized functional gsc, xbra, cardiac actin, and N-tubulin regulatory regions with a highly-fluorescent mutant of GFP90 followed by an SV40 polyadenylation sequence (SV40pA), and injected into unfertilized eggs according to the transgenic procedure (see Preliminary Data 1). Embryos will be assayed by fluorescence microscopy following the onset of expression of the cognate gene, and embryos displaying tissue-specific reporter patterns comparable to in situ hybridization signals will be cultured to maturity. Populations of marked animals will be generated either by using transgenic females to make gynogenetic diploids, creating a strain bearing identical integrations, or conventionally inbred to fix reporter loci.

Preliminary evidence indicates that occasional transgenic embryos display unexpected tissue-specific expression patterns, possibly through an "enhancer trap" mechanism in which transgenes integrate near cis-acting regulatory regions powerful enough to supersede regulation by transgene promoter sequences (see Preliminary Results, Plate Fig. 6E-G). If stable, these rare events may also provide useful tissue-specific reporter lines.

The cytomegalovirus promoter gives high-level expression in plasmid injection and transient transgenic assays, but it is not known whether it will undergo methylation repression in the germ line as do many other viral promoters. Vector construction and transgenesis will be as above. Lines of X. tropicalis expressing GFP under the control of these promoters would be useful in several ways: 1) to unambiguously identify specific tissues for transplantation early in development; 2) to identify changes in cell fate as a result of tissue rearrangements or local expression of genes from injected RNAs; and 3) as easily observed markers to identify mutants of leading edge mesodermal cell patterning or migration (goosecoid), axial and paraxial defects in patterning, convergent extension, and differentiation (brachyury, cardiac actin), and neural patterning and morphogenesis (N-tubulin).

Reliable inducible expression systems circumvent lethality and pleiotropy of early ubiquitous expression of some gene products. These are common problems in mRNA injection assays, but are also likely to complicate transgenic assays for gene function, at least until a suitable 'toolbox' of tissue-specific promoters is assembled. Inducible systems would be particularly useful when combined with transplantation assays: inducibility permits temporal regulation of gene expression; transplantation (as well as the use of tissue-specific promoters) permits spatial control (see section D below).

One strategy for creating inducible expression, which has been successfully employed in a variety of other organisms, involves establishing two strains of transgenic animals. In this "binary inducible" design, one strain expresses a sequence-specific transcriptional activator transgene, which has been modified by fusion to some kind of regulatory protein domain which restricts activity until an inducing agent is provided. One such example would be the yeast Gal4 transcriptional activator regulated by fusion to the ligand-binding domain of the glucocorticoid receptor, which sequesters the protein in the cytoplasm until the cognate ligand is added. This construct might be expressed under the control of a tissue-specific or ubiquitously-expressed promoter.

The second strain is the "target", in which the gene whose expression is being experimentally manipulated is placed under the control of yeast sequences which bind Gal4 and allow transcriptional activation. In order to actually conduct the experiment, the transactivator strain is crossed to the target strain. Embryos are allowed to develop until experimental expression of the target gene is desired, and then the inducing ligand is introduced, binding to the steroid receptor moiety of the transactivator, allowing Gal4 to enter the nucleus and activate transcription.

Transgenesis is a recent development, and relatively few tools, such as tissue-specific or inducible promoters, are currently available. In this section, we will propose experiments that combine well-characterized embryological manipulations, such as transplantation, with multigeneration transgenic analysis using the limited available reagents, in an attempt to demonstrate the practical utility of the system. Development of additional tools for controlling gene expression will make a much wider range of experiments practical.

The tolerance of the amphibian embryo for transplantation makes it possible to create genetic mosaics by replacing a piece of tissue from an embryo of one genotype with that of another. For instance, the cells that will give rise to the germline have been fatemapped to vegetal endoderm, and it is feasible to transplant prospective germline tissue at early stages into a wildtype embryo from one carrying a mutation that will cause a lethal somatic defect. Similar transplantations have been used in the axolotl to rescue the eyeless mutation⁹⁸. For many loci, wildtype gene function will not be required in the germline, and a mosaic animal, carrying a lethal allele in its germline, will be viable. Gynogenesis can then be used to ensure that all of its progeny express the mutant phenotype. Those mutant embryos can then be assayed directly, or by transplantation of specific tissues into wildtype hosts to circumvent obscuring pleiotropy or lethality.

We propose to test the utility of this approach using available reagents. Understanding the role of transcription factors like Pax-6 in eye determination has been one of our major interests. This gene has been argued to be a master regulator of eye formation⁹⁹. Homozygous mutant Pax-6 mouse embryos do not develop differentiated eye tissue, but interpretation of the phenotype is complicated by its pleiotropic effect on both retina and lens ectoderm¹⁰⁰: failure in lens formation could be directly due to the absence of Pax-6 in the lens itself, or indirectly due to effects on the retina, which is thought to produce lens-inducing signals.

To properly test this hypothesis, transcriptional activation by Pax-6 must be blocked specifically in lens ectoderm, but not retina. Activity of transcriptional activators such as Pax-6 can be blocked by creating a fusion of the DNA-binding domain of the gene being assayed (Pax-6) to the repressor domain of the Engrailed gene, generating a site-specific repressor¹⁰¹. Delivery of this dominant inhibitory construct to lens ectoderm could be accomplished in X. tropicalis by expressing the chimaeric Pax-6-Engrailed gene from a constitutive promoter in donor embryos, then transplanting prospective lens ectoderm to the site of lens formation in wildtype host embryos (see Preliminary Results). Using this transplantation strategy to create genetic mosaics, targeted inactivation of Pax-6 gene function in the prospective lens area should allow us to test whether the inhibition of lens formation is due to an autonomous effect of Pax-6 or not, avoiding the complicating potential retinal pleiotropy.

However, this useful experiment may suffer from another common transgenic complication: the dominant inhibitory form of Pax-6 described here is likely to be lethal in tadpoles, since homozygous Pax-6 mutant mouse pups die shortly after birth¹⁰². Lethality might be avoided by conditional expression of dominant inhibitory gene function (see section C above), but at present this technology is not perfected.

A solution to this problem involves making genetic mosaics in a way that is not possible in other vertebrates like mouse or fish. The Pax-6 gene is expressed only in ectodermal tissues⁷². It should be possible to rescue transgenic embryos bearing dominant inhibitory Pax-6 constructs by exchanging the 'animal cap' ectoderm of the mutant transgenic for wildtype ectoderm, leaving the mutant transgene viably carried in the endodermal germline cells. This animal cap exchange procedure results in normal development, though situs inversus does occur in a fraction of the tadpoles that develop (R.M. Grainger, unpublished;¹⁰³). Such chimeric animals should develop normally, but since the mutation is present in the germ line, all of their progeny may be used as donors in the lens transplantation experiments described above to assay the tissue autonomy of Pax-6. Note that this general strategy of germline mosaicism by transplantation may simplify analysis of many lethal mutations with neural or mesodermal defects but no germline requirements; the potential to carry homozygous lethal genotypes will also simplify analysis of double mutants.

We will pursue this experimental design using stable lines of transgenic X. tropicalis and not primary transgenic X. laevis for several reasons. First, stable lines give more consistent levels of gene expression, and may be selected for optimal qualities in a given assay. Second, it may be difficult to identify the primary transgenic embryos without assaying for specific gene activity in the transplanted tissue.

We are currently evaluating the hypothesis that Pax-6 activates the early stage of the lens determination process. This proposal is also testable by transplantation experiments assaying the lens-forming ability of Pax-6-expressing ectoderm heterotopically transplanted to wildtype embryos. Lines of X. tropicalis expressing the normal Pax-6 gene in an inducible construct that is available now (see Specific Aim 2.C) could be used in a similar manner to examine the effect of controlled mis-expression of this gene.

In addition to transplantation, transgenesis with tissue-specific promoters may be used to experimentally manipulate spatial and temporal aspects of gene function. The menu of functionally characterized promoters in frogs is currently expanding (see section A and C above); one experiment using available reagents is described briefly. Experiments with dominant negative forms of the extracellular matrix protein fibrillin have suggested an important role in the morphogenetic movements associated with somite and notochord formation, including the boundary between these two tissues (Skoglund et al., 1997 in preparation). To ask whether fibrillin in somites helps form the boundary with notochord, the xbra promoter (see Preliminary Results) will be used to drive somite expression of dominant negative fibrillin constructs in transgenic embryos. Results from RNA injections (ibid) suggest that this will block convergent extension of the somitic mesoderm and perhaps that of the notochordal mesoderm as well, since the notochordal behavior seems dependent on the somitic tissue¹⁰⁴. As in the previous example, it is likely to be helpful to use transplantation to produce viable genetic mosaic transgenics carrying the lethal dominant negative fibrillin allele in the germline.

A. genes expressed early (Six-3) and late (g-crystallin) in eye development

B. genes expressed in the Spemann organizer (xNR3, noggin, and goosecoid)

 Back to top

Transgenic assays for cis-acting regulatory sequences have been used to identify enhancers in vivo¹⁰⁵, and clearly present the most reliable way of developing new reagents for driving tissue-specific expression in transgenics. Using mice for this kind of analysis is costly and time-consuming. Transgenic frogs can be prepared quickly, inexpensively, and in large numbers (see preliminary results), and reporter constructs reflect cognate patterns of gene expression. This system is ideally suited to analysis of gene regulation.

We will initially study two different groups of genes with overlapping temporal and spatial patterns of expression, one set which is expressed in an early inducing center, the Spemann organizer, and one set which is expressed in a developmental sequence in a later-developing tissue, the eye. We may be able to identify cis-acting elements common to each set of genes (e.g. "organizer" or "eye" enhancers) as well as elements which regulate more subtle differences in spatial (e.g. organizer subdivision^{106, 107})or temporal (six-3 early, g-crystallin late in the eye) expression pattern.

Identification of important regulatory elements is useful in two ways: first, regulatory elements which function reliably in transgenic reporter assays will be valuable reagents for driving tissue-specific expression in other transgenic gene function paradigms. Second, these sequences can be used to characterize the trans-acting factors with which they interact, and hence further our understanding of regulation of embryonic inductions, signal transduction, and differentiation pathways. Several of the genes being analyzed are themselves trans-acting factors and candidates for interacting with regulatory elements in their own or other promoters being analyzed.

While these nuclear injections and transient reporter assays could be performed in X. laevis, it is very useful to harvest cis-acting regulatory sequences from S. tropicalis genomic DNA, since many of the duplicate genes in the pseudotetraploid X. laevis may no longer be functional. Such pseudogenes can be difficult to distinguish from functional alleles on the basis of their coding regions, but may have crippled regulatory regions, complicating reporter analysis. Conducting the transgenic reporter assays in the short-generation S. tropicalis also leaves open the possibility of culturing successful tissue-specific reporter integrations to establish stable transgenic reporter lines for use in subsequent gene expression assays (see Specific Aim 2.A)

We have isolated several genomic clones encoding Xenopus g-crystallins, the major embryonic lens proteins, and sequenced a number of putative promoters⁷³. After cloning these genes from an X. tropicalis genomic library, comparison of promoter sequences between active genes and pseudogenes from the two related species may help identify key conserved regulatory elements. Transgenic analysis to functionally map regulatory sequences may reveal binding sites for genes like the transcription factors Pax-6 and Sox-2 thought to be important in regulating lens development¹⁰⁸. The fact that mouse g-crystallin promoters drive high lens-enriched expression of reporter genes in frogs¹⁰⁹ leads us to believe that these experiments are highly feasible. In addition, stable crystallin promoter-GFP S. tropicalis lines will be extremely useful reagents with which to focus a mutagenesis screen as proposed in Specific Aim 4.A. Second, transgenic reporter lines will be useful in studies of induction where time-consuming assays hamper progress (see Specific Aim 2).

Another gene that is likely to be important in early eye determination is the Six-3 gene, a vertebrate relative of the Drosophila sine oculis gene that is required for eye formation (Oliver et al., 1996). A Xenopus Six-3 cDNA clone has been isolated (Hirsch and Grainger, unpublished); expression of this gene is restricted to the eye rudiment at developmental stages known to be important in eye determination (see Preliminary Results). Functional dissection of the S. tropicalis Six-3 promoter may identify elements common to later promoters, like g-crystallin, and provide probes to characterize trans-acting factors which regulate Six-3 and other steps early in the eye genetic hierarchy.

Understanding the embryonic inductions orchestrated by the Spemann organizer is of special interest to us, and recent advances have clarified some of the molecular mechanisms involved^{110 111 9, 112-114}. This important axial inducing center is itself induced at blastula stages by the adjacent dorsal endoderm, the Nieuwkoop center, and by the onset of gastrulation, gene expression within the organizer shows considerable subdivision, perhaps indicating functional regionalization into separate head and trunk/tail inducing centers^107,115.

The mechanisms which induce and subdivide the organizer may be studied using transgenic reporter constructs to dissect the regulation of organizer-specific genes. By isolating DNA sequences which confer organizer-specific expression and using those sequences to describe transcription factors which are involved in the regulation of a set of organizer genes, it may be possible to start climbing up the signal transduction pathway from the Spemann organizer towards the signaling region that induces it, the Nieuwkoop center. We have selected three organizer genes to characterize in transgenic reporter assays. One, xNR3, is expressed only at gastrulation in the dorsal epithelial layer of the organizer¹⁰⁶; the others, noggin and goosecoid, are expressed in the deep layers of the organizer, but also have later domains of expression in notochord and other tissues whose regulation will be useful to dissect. Genomic X. laevis xNR3 and noggin clones have already been obtained and subjected to preliminary analysis, and the X. laevis goosecoid promoter has already been characterized in plasmid injection assays⁸⁹.

B. insertional mutagenesis; and

 Rationale: There are several reasons to believe that the labor required for limited screens is outweighed by the likelihood of obtaining useful research tools and insights.

First, mutagenisis by transgenesis (insertional mutagenesis) needs to be quantitated if only to optimize the transgenic procedure for generating stable lines, and offers a direct path to molecular characterization of mutant loci.

Second, combining chemical mutagenesis with transgenesis- especially tissue-specific reporter lines- offers ways for us to focus our screens on the pathways that interest us, and may also make discerning subtly aberrant phenotypes easier.

Third, the riskiest proposals, such as testing gene targeting in frogs, have very large potential payoffs relative to the investment of labor, since preliminary assays of feasibility can be conducted with available reagents (e.g. nuclear transfer from extant X. laevis euploid cell lines).

Fourth, some of the same techniques that made early zebrafish screens feasible, including the ready production of gynogenetic haploids and diploids, likewise make it possible for us avoid carrying large numbers of animals for the full duration of the screen. Should the need arise, we have the facilities to maintain up to 20,000 S. tropicalis in a new facility recently constructed at the University of Virginia.

Rationale: Conventional chemical mutagenesis is likely to be more efficient and easier to titrate than insertional mutagenesis, with better prospects for generating a varied set of single-gene mutants for analysis. Molecular characterization of chemically-induced point mutants remains a stumbling block. However, our screens include design elements to help focus on particular developmental processes; we are not planning to evaluate the very large number of lines required for a saturation mutagenesis. Instead, we hope to use streamlined assays for genes expressed in recognizable patterns indicative of normal

early development (see ‘focused screens’ below) to help isolate mutants in relevant pathways. Gene expression assays may also create a more readily-recognizable, patterned "search image" to help workers spot subtle aberrations in development, much as the regular patterns of larval denticle belts facilitated the identification of axis formation mutants in Drosophila. Methods

One- and two-generation Screens Following determination of the specific locus mutation rate at a range of ENU doses, an optimal dose for delivering one mutagenic ‘hit’ causing a visible embryonic phenotype per haploid sperm genome will be delivered to five males (F0) by immersion, and following a sperm maturation period, the males will allowed to mate with five hormonally-induced females.

From the eggs of five females crossed to mutagenized F0 males, we plan to raise 1000 F1 frogs to maturity, of which half are expected to be female. The 500 F1 "candidate heterozygous" females will be individually marked with a coded brand, and then simultaneously outbred and assayed by gynogenesis at a rate of 25 females per day. ~300 eggs per female will be fertilized with UV-irradiated sperm to make F2 haploids, and with unirradiated sperm to generate F2 outbred progeny. Half of the F2 haploids will be subjected to late pressure treatment to inhibit first cleavage and restore diploidy (F2 gynogenetic diploids). Haploids and homozygous diploids will be scored for mutant phenotypes at gastrula, neurula, and tailbud stages. Assuming normal survival rates at early stages, this will require sorting and visual screening of ~10,000 embryos per day for a team of five or six people, and perhaps two months to score the progeny of 500 F1 females.

Formally, half of the haploid and gynogenetic diploid progeny of parents carrying a single scorable recessive mutation should be phenotypically mutant. We are proposing to score both treatments. At later tailbud stages haploid syndrome abnormalities may obscure some phenotypes, and pressure treatment rarely rescues all of the embryos, so it is difficult to unambiguously identify mendelian ratios of subtle phenotypes. A haploid screen allows us to discard lines that are phenotypically wildtype or carry "uninteresting" phenotypes. Lines that carry "interesting" tissue-specific phenotypes will be maintained and characterized further by conventional inbreeding. We estimate that we will discard >90% of the lines at this point, and will be keeping <50 F2 families per F0 mutagenized male. 20 to 25 F2 outbred siblings of putative mutant haploid and gynogenetic diploid F2 embryos will be cultured to maturity, and inbred in natural matings. At one mutagenic "hit" per sperm, on average every other F2 sibling natural mating will result in 25% mutant phenotype, and will be free of the haploid syndrome background. The presence of additional mutations will result in higher and more variable phenotypic ratios, and will require additional segregation in order to characterize the individual mutations.

Focused screens: How can we hope to characterize genes in pathways in which we are interested without the enormous investment involved in screening and maintaining large numbers of random mutations? We propose two strategies to focus our screens and make it easier to spot the rare mutation that affects a specific process in a background of wildtypes, less-relevant mutants, and epigenetic variations:

1. By mutagenizing and screening isogenic lines bearing tissue-specific GFP reporter activity. For example, lines homozygous for neural tubulin-GFP, which is expressed in the neural plate in a striped pattern during neural tube closure should provide a distinct search image on which to discern mutations affecting morphogenesis of the neural tube.

2. By screening haploid and gynogenetic diploid F2 embryos by in situ hybridization with various probe cocktails which reflect patterned gene expression in the wild-type embryo. Embryos screened in this fashion obviously cannot be rescreened later in development, and the labor involved is considerably greater than unaided visual scoring, but the available spectrum of molecular probes is much larger. Using mixed probes facilitates screening for phenotypes in multiple pathways simultaneously.

Target phenotypes:

1) mutations affecting movements of axial convergent extension, which define and elongate the anterior-posterior axis of the vertebrate body; screened on a neural tubulin-GFP background to enhance detection of axial defects: a. blastopores that stall in closure at the midgastrula stage, when convergent extension normally begins; b. short axes and widened notochordal and somitic tissues, indicating decreased rate or amount of convergent extension^, c.arched backs, indicating retarded mesodermal convergent extension; d. blastopores that re-open, indicating locally fractured circumblastoporal arrays of converging cells, forming the classical ring embryo phenotype^,.

2) mutations affecting eye formation: a. small or missing lenses, assayed by diminished or missing GFP expression in crystallin promoter-GFP-bearing transgenic lines of X. tropicalis. b. altered, reduced or missing retinal tissue, assayed by monitoring distribution of GFP expression in Six-3 promoter-GFP bearing transgenic lines of S. tropicalis.

We feel that it is important to be fairly stringent in restricting the phenotypes that we choose for extensive characterization. Our screen is designed so that visible morphological defects can be readily correlated with aberrations in relevant gene expression. As other developmental phenotypes which do not meet these criteria emerge in screens, they will be briefly evaluated and distributed to other Xenopus laboratories with cognate research interests.

Cloning mutant loci: The conventional method to isolate chemically-induced point mutations is positional cloning: isolating closely linked markers and mapping by recombination followed by chromosome walking. Even with a fine-structure genetic map, this is laborious. We will test the following alternatives:

1. candidate gene approaches: Many genes with activity in pathways of interest have been identified. Assays used to evaluate candidates for a given mutation will include rescue of the mutant phenotype by injection of candidate gene mRNA and transgenic rescue of mutants with candidate wild-type gene driven by a suitable characterized promoter, and correlation of RAPD map position of the mutation and the candidate gene (see Specific Aim 1.D).

2. expression cloning approaches: Specific biological activities have been identified in X. laevis mRNA pool injection and sib-selection assays by their ability to rescue the ‘phenotype’ of eggs exposed to ultraviolet light . It may be possible to similarly rescue certain zygotic mutant phenotypes, by injecting pools of synthetic mRNA to deliver the missing gene product. Pools that rescue can then be split and re-assayed until the activity is cloned. X. laevis cDNA libraries in expression vectors designed for this purpose have already been constructed. Correlation of RAPD map position of mutations and sib-selected genes will then be used to confirm identity (see Specific Aim 1.D).

Specific aim #4.B: Insertional Mutagenesis The remaining projects are feasibility studies, the results of which will determine subsequent investment of effort. While all of these projects have challenging goals, each is built around a preliminary experiment which does not require significant preparation, and which should reliably predict a project’s long-term prospects.

The difficulty of cloning chemically-induced mutants is a drawback to most of the recently completed zebrafish screens. A small-scale zebrafish screen has avoided this difficulty by using a retroviral insertional mutagenesis approach; virtually all of the mutants they have characterized have been rapidly cloned. In frogs, the ease with which large numbers of embryos transgenic at multiple loci can be generated suggests that insertional mutagenesis could be feasible.

We propose to assess the feasibility of gene traps as insertional mutagens. Gene traps are splice acceptor-containing constructs which, if inserted into an exon or intron, can simultaneously inactivate a locus and provide locus-specific reporter expression. This should enable us to screen F0 heterozygotes (the immediate products of a transgenic procedure using a gene trap vector) for candidate insertions in interesting pathways, greatly reducing the number of lines that we will carry to homozygosity for phenotypic analysis.

Large-scale gene-trap mutagenesis based on mouse ES cells have been performed. One major effort involved screening ~40,000 integration-bearing ES colonies, approximately 1% of which expressed the gene-trap reporter and were assayed for expression pattern as chimeras, yielding 36 lines giving tissue-specific expression patterns at the developmental time point assayed . The frequency of transgenic frog embryos expressing gene trap reporters may be somewhat higher than 1%, since not all genes are expressed in ES cells, and since the frog transgenic procedure typically gives integrations at multiple loci. Balanced against this is the evidence that gene trap insertions in ES cells seem to favor actively transcribed regions, and it is not known how transgenic insertions are biased in sperm nuclei. Not all of the mutagenic insertions will give detectable reporter expression, however; integrations into exons, into genes with weak promoters, or in the wrong orientation or reading frame all could inactivate alleles without giving a reporter signal.

We believe that the frog system is amenable to this insertional mutagenesis strategy, but at greatly reduced expense relative to mice. We can generate multiple gene trap insertions in hundreds of transgenic animals per day for prompt reporter screening, discarding non-expressing tadpoles within a week. In addition, all stages of development in the frog are accessible and screening for marker gene expression can be done in living embryos at any stage.

Gene trap design: All gene trap vectors contain a splice acceptor (SA) sequence followed by a marker gene. Since the marker gene lacks a promoter, it can only be transcribed and translated if it integrates properly into an exon or intron of an endogenous gene. The marker gene we have chosen is a very bright version of green fluorescent protein (GFP), which, unlike other reporter genes, is visible in living embryos. This allows us to detect expression at any stage of development without having to fix and stain the embryos; only embryos that express GFP in an interesting pattern need be cultured to maturity. In addition, since the trapped (and prospectively mutated) gene is always marked by GFP, one can easily distinguish between embryos carrying the mutation from ones that do not.

Preliminary experiments: Since improving transgenesis is a major technical focus of the X. tropicalis genetics project, preliminary experiments to optimize efficient, multi-site integration in viable embryos are already underway. As with transgenesis using other vectors, it will be important to establish a balance between maximizing integrations and maintaining optimal long-term transgenic viability, since treatments that encourage incorporation of foreign DNA may also cause chromosome damage. Southern analysis to determine the absolute number of integration sites produced by each set of conditions provides an indirect assay for the prospective mutagenicity of gene traps under various conditions as an alternative to calculating the specific locus mutation rate.

While maxima have yet to be determined, published data show 4-8 integrations per transgenic frog. Extrapolating from the mouse ES gene trap screens, in which 1% of integrations led to expression of the reporter, we might expect ~4-8% of transgenic embryos to display locus-specific reporter expression. Thus a reasonable preliminary study might involve generation and fluorescence assay of ~1000-2000 transgenic embryos, or roughly one week of nuclear injections. The rate of production of embryos expressing GFP in specific tissues would determine whether injections would be scaled up.

Gene trap screen: If preliminary experiments result in >0.5% embryos expressing GFP in a regulated fashion, ~10,000 transgenic embryos carrying the gene trap vector will be generated over the course of several months and screened for GFP fluorescence at stage 11, 17, 24, and 44. F0 transgenic embryos not showing detectable expression will be discarded; those displaying reporter expression, especially temporally- or spatially-regulated, will be categorized based on their reporter expression and sorted for culture to sexual maturity. Extrapolating a rough expected frequency of reporter expression from the mouse ES cell screens and transgenic integration rates, we might anticipate detecting 50-500 expressing lines. In order to determine whether an expressed gene trap has disrupted an essential gene, embryos will be cultured to sexual maturity and subjected to gynogenesis or androgenesis in order to score homozygous phenotypes, and outbred to carry the gene trap allele.

maternal effect genes: Expressed insertions that do not display a recessive zygotic phenotype may have maternal effects. Viable homozygous F1 gynogenetic diploid females will be raised to sexual maturity, and haploid, gynogenetic diploid, and outbred progeny will be assayed for mutant phenotypes.

Cloning trapped genes: Identification of loci with insertions is straightforward. cDNA is prepared from tissues expressing the reporter and Rapid Amplification of cDNA Ends (RACE ), using GFP sequence for the specific primer, is used to generate PCR fragments of GFP fusion transcripts. While there is a low probability of transcribed multiple insertions persisting after segregation, it should be possible to tell which are expressing the reporter by comparing GFP expression with whole mount in situ hybridizations using RACE probes. Multiple expressed traps would likely be reflected in pleiotropic and variable F2 haploid and gynogenetic diploid phenotypes; outcrossing for another generation and re-examing homozygous progeny should permit examination of single-gene phenotypes.

Project 4.C Gene Targeting

Rationale:

A large number of candidate genes for participation in developmental processes have already been identified in Xenopus and other systems. Gene function in many of these processes could be effectively characterized in advanced frog embryological assays if specific mutants were available. While candidate gene mutations may be characterized through the insertional and chemical mutagenesis screens described in specific aims #4A and B, the efficiency with which a mutant for any given candidate is identified is likely to be very low. Currently, mouse embryonic stem cell (ES cell) technology offers the only system for targeted mutagenesis in higher eukaryotes.

4.C.i. Nuclear transfer from cultured cells:

ES cell equivalents are unlikely to be reproduced in amphibian systems soon. However, late-stage embryos and in some cases fertile adult frogs can be produced by nuclear transfer from a surprisingly wide variety of donor cells, including endoderm, melanophores, keratinized skin cells, neural cells, erythrocytes, and intestinal epithelial cells^127-129. If a cultured cell system could likewise be developed as a source of nuclei, then identifying and expanding cells that carry specific mutations and using them to make transgenic mutants may become feasible. What are the requirements for developing homologous recombination using nuclear transfer?

1. A cell line whose nuclei are competent for transfer into enucleated eggs and direction of subsequent development;

2. an efficient transfection system; and

3. a method for identifying and cloning cells which have correctly integrated targeting constructs. Selection regimes simplify this last task, but are not formally required.

Of these requirements, the first is likely to be the most stringent. Several cell lines have been assayed in serial nuclear transfer protocols without producing feeding-stage tadpoles^{130, 128,131}. The high frequency of aneuploidy in long-term cultured cells¹³⁰ has been a significant stumbling block; in only one of these cases, using a haploid cell line, were the cells identifiably euploid¹³¹. Developmental potential of ES cells likewise becomes restricted with multiple passages²⁸; by focusing our manipulations on low-passage cultures, karyotypic abnormalities may be minimized.

Encouragingly, we have recently generated and characterized two clonal X. laevis cell lines, one with fibroblast-like morphology, one with epithelial morphology, from dorsal explants of neurulae. Both of these lines are karyotyically euploid with respect to chromosome number (60% of mitotic nuclei displaying thirty-six chromosomes) (see Preliminary Results). Even when euploid somatic nuclei are transplanted, visible chromosome aberrations appear in a large fraction of the resulting embryos¹²⁷, perhaps as a result of the rapid post-fertilization cell cycle, and are thought to be a principal impediment to normal development.

Strategies that may help protect nuclei from these stresses include serial transplantation¹²⁸, incubation in meiotic oocytes or activated egg extracts^1, and induction of quiescence by serum starvation¹²⁹. The latter was implicated in the recent success in cloning by nuclear transfer from a sheep cultured cell line, in which it was hypothesized that the chromatin of quiescent nuclei was more readily modified by the oocyte cytoplasm.

We propose to test the prospects of gene targeting in X. tropicalis by generating and karyotyping cell lines and assaying their developmental potential in a nuclear transfer procedure adapted from the transgenic technique. If successful, we will elaborate an effective electroporation procedure, and test selection protocols and methods to identify homologous recombinants in vitro and in transgenic frogs.

Two clonal cell lines have already been established from X. laevis neurulae and shown to have a euploid complement of chromosomes (see Preliminary Results). A modified transgenic protocol will be used to assay nuclei from these lines for their ability to direct development of eggs that have been UV-irradiated to inactivate the maternal genome¹³⁴. Methods for nuclear transfer from cell lines have been streamlined¹³⁰; variations such as serum starvation and mitotic extract incubation may be evaluated with relatively little effort. Identification of a developmentally-competent source of X. laevis cultured cell nuclei would encourage us to establish comparable euploid X. tropicalis lines, and to develop transfection, targeting strategy, and selection protocols with which to isolate clones bearing targeted alleles. Two positive selective regimes, neomycin resistance and hygromycin resistance, have already been worked out in Xenopus cell lines^{126; 130}. Additional cell lines may be established from isogenic X. tropicalis neurulae using established methods^{79, 135, 136} as described in Preliminary Results. Further experiments with cell lines will be predicated on the identification of euploid cell lines with developmentally-competent nuclei.

In an exploratory venture like the one proposed here, not all of the projects will succeed. Even if more challenging technologies described here (e.g. targeted gene inactivation) turn out not to be feasible in the near future, we expect that the tools that we develop, such as a molecular genetic map of X. tropicalis, a panel of tissue-specific regulatory elements driving GFP expression, and a set of mutants, will be of significant utility to the molecular embryology research community.

Given the vertebrate genome's propensity for redundant function, the next phase of unraveling biological pathways is likely to depend on studies of multigene phenotypes, double and triple mutants, which are very impractical to study in mice and fish. As unlikely as it may seem at present, amphibian classical embryology- for example, transplantation techniques to implement genetic mosaics- may be a tool which advances vertebrate molecular genetics well into the next century.