Mesoderm formation in Eleutherodactylus coqui: body patterning in a frog with a large egg.

The direct developing frog, Eleutherodactylus coqui, develops from a large egg (diameter 3.5 mm). To investigate the effect of egg size on germ-layer formation, we studied mesoderm formation in E. coqui and compared it to that of Xenopus laevis (diameter 1.3 mm). First, we identified the position of prospective mesoderm in the 16-cell E. coqui embryo by cell-lineage tracing. Although the animal blastomeres are small, they form most of the blastocoel roof and make extensive contributions to some mesodermal tissues. Second, we performed recombinant analysis with X. laevis animal caps to define the distribution of mesoderm-inducing activity. Mesoderm-inducing activity in E. coqui was restricted around the marginal zone with strong activity in the superficial cells. Neither the vegetal pole nor the blastocoel floor had activity, although these same regions from X. laevis induced mesoderm. Third, we cloned Ecbra, a homologue of Xbra, an early mesoderm marker in X. laevis. Ecbra was expressed in the marginal ring close to the surface, similar to X. laevis, but E. coqui had weaker expression on the dorsal side. Our results suggest that mesoderm formation is shifted more animally and superficially in E. coqui compared to X. laevis.

RNA anchoring in the vegetal cortex of the Xenopus oocyte.

The body plan of the embryo is established by a polarized source of developmental information in the oocyte. The Xenopus laevis oocyte creates polarity by anchoring mRNAs in the vegetal cortex, including Vg1 and Xwnt-11, which might function in body plan specification, and Xcat-2, which might function in germ cell development. To identify components of the RNA anchoring mechanism, we used the manually isolated vegetal cortex (IVC) to assay loss or change in spatial arrangement of mRNAs caused by disruption of cortical elements. The role of cytoskeleton in mRNA anchoring was tested by treating oocytes with inhibitors that selectively disrupted actin microfilaments and cytokeratin filaments. Treatment of oocytes with cytochalasin B caused clumping of Vg1 and Xwnt-11 as revealed by in situ hybridization of the IVC, but did not cause their release, as confirmed by RT-PCR analysis. These mRNA clumps did not match the distribution of actin microfilament clumps, but were distributed similarly to the remnant cytokeratin filaments. Treatment of oocytes with monoclonal anti-cytokeratin antibody C11 released these mRNAs from the cortex. C11 altered the texture of the cytokeratin network, but did not affect the actin meshwork. These results show that Vg1 and Xwnt-11 are retained by a cytokeratin filament-dependent mechanism, and that organization of the cytokeratin network depend on an intact actin meshwork. Colcemid did not disrupt Vg1 and Xwnt-11 retention in the IVC, so anchoring of these mRNAs are independent of microtubules. Membrane disruption in the IVC by Triton X-100 decreased Vg1 and Xwnt-11. Loss of these mRNAs was due mainly to ribonuclease activity released from membrane components. However, when ribonuclease activity was suppressed under cold temperature, a higher amount of Vg1 and Xwnt-11 was recovered in the supernatant. This result suggested that a fraction of these mRNAs required membranes to be retained in the cortex. By contrast, Xcat-2 mRNA was neither released nor degraded following treatments with cytochalasin B, C11, colcemid and Triton X-100 under cold temperature, so no cortical element could be implicated in its anchoring.

Frogs without polliwogs: evolution of anuran direct development.

Direct development is the assumption of the adult morphology without progression through an intervening, morphologically distinct, free-living larval phase. We discuss the ecological factors contributing to the evolution of this derived life-history strategy in frogs, and the developmental modifications that facilitate such an unusual mode of embryogenesis. Studies on the Puerto Rican tree frog, Eleutherodactylus coqui, have identified several such modifications, including developmental adaptations for dealing with increased egg size, and loss of tadpole structures. Surprisingly, this direct developer still undergoes a thyroid hormone-dependent metamorphosis, which occurs before hatching. We suggest how the ancestral biphasic developmental pattern may have been rearranged during the evolution of direct development.

Thyroid hormone-dependent metamorphosis in a direct developing frog.

The direct developing anuran, Eleutherodactylus coqui, lacks a tadpole, hatching as a tiny frog. We investigated the role of the metamorphic trigger, thyroid hormone (TH), in this unusual ontogeny. Expression patterns of the thyroid hormone receptors, TRalpha and TRbeta, were similar to those of indirect developers. TRbeta mRNA levels increased dramatically around the time of thyroid maturation, when remodeling events reminiscent of metamorphosis occur. Treatment with the goitrogen methimazole inhibited this remodeling, which was reinitiated on cotreatment with TH. Despite their radically altered ontogeny, direct developers still undergo a TH-dependent metamorphosis, which occurs before hatching. We propose a new model for the evolution of anuran direct development.

Opercular development and ontogenetic re-organization in a direct-developing frog.

The direct-developing frog, Eleutherodactylus coqui, has eliminated the tadpole stage from its ontogeny, and lacks many larval characters. We demonstrate that the dermal folds of E. coqui are homologous with the opercular folds of metamorphosing frogs. In both E. coqui and its metamorphic counterparts the opercular folds grow over the developing forelimb before perforating to free the entrapped limb. Opercular perforation in E. coqui occurs even in the absence of the forelimb but shows no signs of thyroid hormone dependence. The condensation of E. coqui development appears due to the excision of the extended larval period of developmental stasis. Analysis of opercular development, when viewed in conjunction with other developmental characters, suggests the ontogenetic period in the ancestral Eleutherodactylus life-history from which the tadpole was likely eliminated.

Ectopic expression of Xenopus noggin RNA induces complete secondary body axes in embryos of the direct developing frog Eleutherodactylus coqui.

We tested the effects of noggin RNA from Xenopus laevis on axis induction in embryos of a direct developing frog, Eleutherodactylus coqui. We microinjected noggin RNA into one blastomere of 4-cell embryos at the site close to the animal pole, and found that overexpression of noggin RNA is not only sufficient to induce additional axes but also induces heads with eyes. We also injected noggin RNA into 8-cell or 16-cell embryos in various sites, including the marginal zone, above the marginal zone, and the vegetal pole, and found the formation of a complete secondary axis in all three types of injection. These effects of X. laevis noggin RNA on the E. coqui embryo are remarkably different from those found in the X. laevis embryo itself. It has been shown previously that overexpression of noggin RNA on the ventral side of the normal X. laevis embryo induces only a partial axis, with no head structures. We show here that the failure of noggin induction of a complete axis when overexpressed on the ventral side of the X. laevis embryos is not due to an insufficient amount of RNA injected. Also, the failure is unlikely due to inhibition from the primary axis since noggin RNA can induce duplicated head structures on opposite sides of UV-irradiated X. laevis embryos. There appear to be fundamental differences in the responses of E. coqui and X. laevis embryos to exogenous noggin RNA. We propose that these differences stem from an alteration in cytoplasmic arrangements that occurred during evolution of this large egg.

Novel structural elements identified during tail resorption in Xenopus laevis metamorphosis: lessons from tailed frogs.

When the tail of the Xenopus laevis tadpole resorbs at the end of metamorphosis, various cell types, including muscle, fibroblasts, skin, and spinal cord, are lost at about the same time. However, feeding frogs with tails can be produced by inhibiting thyroid hormone production at the climax of metamorphosis with the goitrogen methimazole. These tails lose their fast muscle preferentially, showing that the different cell types of the tail have different fates and confirming that more than one cell death program is involved in tail resorption. Both normal and methimazole tails contain "cords," novel structures that consist of two dorsal and two ventral parallel rows of slow muscle bundles joined by collagen fibers that run the length of the tail. The cords persist until the very end of tail resorption, being the last structure to dissolve. When thyroid hormone induces expression of proteolytic enzymes in the notochord sheath, the notochord, a structural rod that runs the length of the tail, begins to buckle, demonstrating that the tail is under tension. When sections of the tail that contain cords are surgically separated from the notochord, they contract in vitro, suggesting that the cords contribute to the tension that augments tail resorption.

Relationship of vegetal cortical dorsal factors in the Xenopus egg with the Wnt/beta-catenin signaling pathway.

In Xenopus, the dorsal factor in the vegetal cortical cytoplasm (VCC) of the egg is responsible for axis formation of the embryo. Previous studies have shown that VCC dorsal factor has properties similar to activators of the Wnt/beta-catenin-signaling pathway. In this study, we examined the relationship of the VCC dorsal factor with components of the pathway. First, we tested whether beta-catenin protein, which is known to be localized on the dorsal side of early embryos, accounts for the dorsal axis activity of VCC. Reduction of beta-catenin mRNA and protein in oocytes did not diminish the activity of VCC to induce a secondary axis in recipient embryos. The amount of beta-catenin protein was not enriched in VCC compared to animal cortical cytoplasm, which has no dorsal axis activity. These results indicate that beta-catenin is unlikely to be the VCC dorsal axis factor. Secondly, we examined the effects of four Wnt-pathway-interfering constructs (dominant-negative Xdsh, XGSK3, Axin, and dominant-negative XTcf3) on the ability of VCC to induce expression of the early Wnt target genes, Siamois and Xnr3. The activity of VCC was inhibited by Axin and dominant negative XTcf3 but not by dominant negative Xdsh or XGSK3. We also showed that VCC decreased neither the amount nor the activity of exogenous XGSK3, suggesting that the VCC dorsal factor does not act by affecting XGSK3 directly. Finally, we tested six Wnt-pathway activating constructs (Xwnt8, Xdsh, dominant negative XGSK3, dominant negative Axin, XAPC and beta-catenin) for their responses to the four Wnt-pathway-interfering constructs. We found that only XAPC exhibited the same responses as VCC; it was inhibited by Axin and dominant negative XTcf3 but not by dominant negative Xdsh or XGSK3. Although the connection between XAPC and the VCC dorsal factor is not yet clear, the fact that APC binds Axin suggests that the VCC dorsal factor could act on Axin rather than XGSK3.

Evolutionary alteration in anterior patterning: otx2 expression in the direct developing frog Eleutherodactylus coqui.

The homeobox gene otx2 is a key regulator for specifying the rostral part of the vertebrate head. In Xenopus, otx2 directly controls the differentiation of the cement gland, the anterior-most organ formed in the tadpole. Since embryos of a direct developing frog, Eleutherodactylus coqui, lack a cement gland, we are interested in whether altered expression of the otx2 gene is involved in this evolutionary change. We have cloned the E. coqui homologue of otx2, Ecotx2. The homeodomain of the Ecotx2 protein is identical to the mouse and zebrafish Otx2 proteins and differs by a single amino acid from the Xenopus Otx2 protein. Study of the spatiotemporal expression pattern shows that Ecotx2 RNA is progressively restricted to the anterior region of the embryo during gastrulation and becomes further restricted to the future forebrain and midbrain during neural development. In Xenopus, in addition to the conserved expression in the anterior neuroectoderm, the expression in ectoderm expands more anteriorly to the cement gland primordium. This anterior expansion of otx2 expression is not found in E. coqui, correlating with the loss of a cement gland. When misexpressed in Xenopus laevis ectoderm, Ecotx2 can activate expression of the cement-gland-specific genes XCG and XAG1, indicating that the function of activating the pathway of cement gland formation is retained by the Ecotx2 protein. Our results indicate that there are modifications in the pathway of cement gland formation, upstream of otx2 expression, in the development of E. coqui.

Secondary coverage of the yolk by the body wall in the direct developing frog, Eleutherodactylus coqui: an unusual process for amphibian embryos.

Eleutherodactylus coqui develops directly from a large 3.5-mm egg to a froglet, without an intervening tadpole stage. We have examined the development of the body wall, a structure whose behavior has been altered in this derived development. In an event that is unusual for amphibian embryos, the yolk mass is secondarily surrounded by the body wall, which originates near the embryo's trunk. The epidermis of the body wall is marked by melanophores, and the rectus abdominis, which will form the ventral musculature, is near its leading edge. As the body wall expands, the epidermis, melanophores, and rectus abdominis all move from the dorsal side to close over the yolk at the ventral midline. The original ectoderm over the yolk undergoes apoptosis, as it is replaced by body wall epidermis. Intact muscles are not required for ventral closure of the body wall, despite their normal presence near the advancing edge. Comparative examination of embryos of Xenopus laevis and Rana pipiens suggests that ventral closure does not occur in species with tadpoles. The expansion of dorsal tissues over the yolk, as illustrated by E. coqui, may have been important in the origin of amniote embryos.

beta-TrCP is a negative regulator of Wnt/beta-catenin signaling pathway and dorsal axis formation in Xenopus embryos.

The Wnt/beta-catenin signaling pathway is responsible for the establishment of dorsoventral axis of Xenopus embryos. The recent finding of the F-box/WD40-repeat protein slimb in Drosophila, whose loss-of-function mutation causes ectopic activation of wingless signaling (Jiang, J., Struhl, G., 1998. Nature 391, 493-496), led us to examine the role of its vertebrate homolog betaTrCp in the Wnt/beta-catenin signaling and dorsal axis formation in Xenopus embryos. Co-injection of betaTrCp mRNA diminished Xwnt8 mRNA-induced axis formation and expression of Siamois and Xnr3, suggesting that betaTrCP is a negative regulator of the Wnt/beta-catenin signaling pathway. An mRNA for a betaTrCp mutant construct (DeltaF), which lacked the F-box domain, induced an ectopic axis and expression of Siamois and Xnr3. Because this activity of DeltaF was suppressed by co-injection of DeltaF TrCP mRNA, DeltaF likely acts in a dominant negative fashion. The activity of DeltaF was diminished by C-cadherin, glycogen synthase kinase 3 and Axin, but not by a dominant negative dishevelled. These results suggest that betaTrCp can act as a negative regulator of dorsal axis formation in Xenopus embryos.

Limb development and evolution: a frog embryo with no apical ectodermal ridge (AER).

The treefrog Eleutherodactylus coqui is a direct developer--it has no tadpole stage. The limb buds develop earlier than in metamorphosing species (indirect developers, such as Xenopus laevis). Previous molecular studies suggest that at least some mechanisms of limb development in E. coqui are similar to those of other vertebrates and we wished to see how limb morphogenesis in this species compares with that in other vertebrates. We found that the hind limb buds are larger and more advanced than the forelimbs at all stages examined, thus differing from the typical amniote pattern. The limb buds were also small compared to those in the chick. Scanning and transmission electron microscopy showed that although the apical ectoderm is thickened, there was no apical ectodermal ridge (AER). In addition, the limb buds lacked the dorsoventral flattening seen in many amniotes. These findings could suggest a mechanical function for the AER in maintaining dorsoventral flattening, although not all data are consistent with this view. Removal of distal ectoderm from E. coqui hindlimb buds does not stop outgrowth, although it does produce anterior defects in the skeletal pattern. The defects are less severe when the excisions are performed earlier. These results contrast with the chick, in which AER excision leads to loss of distal structures. We suggest that an AER was present in the common ancestor of anurans and amniotes and has been lost in at least some direct developers including E. coqui.

The legacy of lithium effects on development.

The simple element lithium causes specific effect on early embryonic development. In amphibians, treatment during early cleavage enhances anterior and dorsal development at the expense of posterior and ventral development. Conversely, treatment at late cleavage reduces head development. While an effect of lithium on the phosphoinositide cycle was suspected, the enhancement of dorsoanterior development is now thought to result primarily from stimulation of the Wnt/beta-catenin pathway. The multiple uses of this pathway in development may account for lithium's various specific effects.

Dorsal determinants in the Xenopus egg are firmly associated with the vegetal cortex and behave like activators of the Wnt pathway.

The Xenopus egg contains maternal dorsal determinants that are specifically located at the vegetal cortex. To study physical and functional properties of the dorsal determinants, we took advantage of the animal-vegetal reversed embryo. The animal-vegetal reversed embryo is produced by inversion of the fertilized egg, which results in formation of ectoderm and endoderm from the unpigmented and the pigmented halves, respectively [Neff et al. (1983). Dev. Biol. 97, 103-112; Black and Gerhart (1985). Dev. Biol. 108, 310-324]. We demonstrated by cytoplasmic transplantation that the dorsal activity was specifically localized to the unpigmented cortical cytoplasm of the inverted egg, which is segregated into the future ectodermal lineage. This result suggests that the dorsal determinants are associated with the unpigmented cortex and are not dislodged by the inversion. In addition, we found that two vegetally localized transcripts, Xcat2 and Vg1 mRNAs, were present in the reversed animal pole of the inverted egg, suggesting their association with the unpigmented cortex. In order to compare the dorsal determinant activity with known dorsalizing molecules, we examined the expression pattern of Xnr3 and Siamois in the reversed embryo because these two genes are activated by the Wnt-pathway activators (Xwnt-8, beta-catenin, etc.) but not by other dorsalizing molecules (noggin, BVg1, etc.). Animal cap of the reversed embryo, which received the unpigmented cortex of the egg, expressed Xnr3 and Siamois. However, Mix.1, a marker expressed in endoderm and mesoderm in the normal embryo in response to mesodermal inducers, was not detected in the animal cap of the reversed embryo. In addition, we found that beta-catenin protein accumulated in nuclei of unpigmented animal pole cells of the reversed embryo. These results suggest that the maternal dorsal determinants behave more similarly to the Wnt-pathway activators than noggin or BVg1.

Animal and vegetal pole cells of early Xenopus embryos respond differently to maternal dorsal determinants: implications for the patterning of the organiser.

The maternal dorsal determinants required for the specification of the dorsal territories of Xenopus early gastrulae are located at the vegetal pole of unfertilised eggs and are moved towards the prospective dorsal region of the fertilised egg during cortical rotation. While the molecular identity of the determinants is unknown, there are dorsal factors in the vegetal cortical cytoplasm (VCC). Here, we show that the VCC factors, when injected into animal cells activate the zygotic genes Siamois and Xnr3, suggesting that they act along the Wnt/beta-catenin pathway. In addition, Siamois and Xnr3 are activated at the vegetal pole of UV-irradiated embryos, indicating that these two genes are targets of the VCC factors in all embryonic cells. However, the consequences of their activation in cells that occupy different positions along the animal-vegetal axis differ. Dorsal vegetal cells of normal embryos or VCC-treated injected animal cells are able to dorsalise ventral mesoderm in conjugate experiments but UV-treated vegetal caps do not have this property. This difference is unlikely to reflect different levels of activation of FGF or activin-like signal transduction pathways but may reflect the activation of different targets of Siamois. Chordin, a marker of the head and axial mesoderm, is activated by the VCC/Siamois pathway in animal cells but not in vegetal cells whereas cerberus, a marker of the anterior mesendoderm which lacks dorsalising activity, can only be activated by the VCC/Siamois pathway in vegetal cells. We propose that the regionalisation of the organiser during gastrulation proceeds from the differential interpretation along the animal-vegetal axis of the activation of the VCC/beta-catenin/Siamois pathway.

Patterns of distal-less gene expression and inductive interactions in the head of the direct developing frog Eleutherodactylus coqui.

The direct developing frog Eleutherodactylus coqui exhibits radical changes in its embryogenesis. A frog-like head forms directly with no appearance of a cement gland or several jaw cartilages characteristic of tadpoles, and limbs appear early in development. The numerous differences in the embryogenesis of E. coqui provide an opportunity to examine developmental causes for the evolutionary shift from biphasic to direct development. We have cloned DNA fragments corresponding to four E. coqui genes related to the Drosophila distal-less gene Dll. While the expression patterns of the distal-less genes are generally conserved, there are some spatiotemporal differences when embryos of E. coqui are compared to those of Xenopus laevis. The changes in gene expression are correlated with the embryonic changes in head structures including craniofacial cartilages and in particular, the cement gland. We have then examined inductive interactions involved in cement gland formation by interspecific transplants and recombinants. E. coqui embryos can generate signaling that culminates in cement gland formation, but E. coqui ectoderm appears to be incapable of a cement gland response. These results show here that inductive interactions in the anterior region of the E. coqui embryo have been modified during the evolution of direct development, and that changes in the competence of the E. coqui ectoderm may be responsible for the loss of certain tadpole-specific structures, such as cement gland.

Developmental regulation of the urea-cycle enzyme arginase in the direct developing frog Eleutherodactylus coqui.

Direct developing organisms obviate the larval intermediary from their ontogeny, hatching as miniature adults. To investigate this phenomenon, we have examined the developmental expression of arginase in the direct developing frog Eleutherodactylus coqui. An enzyme in the ornithine-urea cycle, the activation of liver arginase is necessary for the switch from ammonotelism to ureotelism which occurs when many frogs metamorphose and assume a terrestrial existence. Arginase enzyme activity is detectable at low levels in late prehatching stages of E. coqui, and increases at hatching, at which point the protein becomes detectable on Western blots. The activity increases gradually during posthatching development, reaching maximal levels at approximately the same time as yolk resorption is completed. Thyroid hormone is responsible for upregulating arginase activity during metamorphosis in Rana, but the role of thyroid hormone in direct developing frogs is unknown. A high dose (250 nM) of the thyroid hormone analogue 3,3'5-triiodo-L-thyronine (T3) caused precocious induction of arginase protein and activity, showing that even in a direct developing frog, some level of responsiveness to the metamorphic trigger, thyroid hormone, has been retained.

Identification of new localized RNAs in the Xenopus oocyte by differential display PCR.

We have identified localized transcripts in full-grown Xenopus oocytes by differential display PCR. One clone, An4a, has two transcripts, which localize to the animal half of the stage VI oocyte. The transcripts are expressed throughout early development, with embryonic expression primarily in anterior neural tissues. An4a has a high degree of sequence identity to a human cDNA clone of unknown function. Another clone, the previously identified beta-transducin repeat containing protein (beta-TrCP), has three transcripts with a unique pattern of localization, one localized to the animal half and two localized primarily to the vegetal cortex. This cDNA has previously been shown to rescue a yeast cell division cycle mutant, raising the possibility that the different Xenopus transcripts are involved in animal and vegetal cell cycles. Embryonic expression is primarily in the cement gland. These new localized transcripts contribute to the general observation that the vegetal cortex, but not the animal cortex, is a specific site for RNA localization.