Developmental origins of a novel gut morphology in frogs

Abstract

Phenotypic variation is a prerequisite for evolution by natural selection, yet the processes that give rise to the novel morphologies upon which selection acts are poorly understood. We employed a chemical genetic screen to identify developmental changes capable of generating ecologically relevant morphological variation as observed among extant species. Specifically, we assayed for exogenously applied small molecules capable of transforming the ancestral larval foregut of the herbivorous Xenopus laevis to resemble the derived larval foregut of the carnivorous Lepidobatrachus laevis. Appropriately, the small molecules that demonstrate this capacity modulate conserved morphogenetic pathways involved in gut development, including downregulation of retinoic acid (RA) signaling. Identical manipulation of RA signaling in a species that is more closely related to Lepidobatrachus, Ceratophrys cranwelli, yielded even more similar transformations, corroborating the relevance of RA signaling variation in interspecific morphological change. Finally, we were able to recover the ancestral gut phenotype in Lepidobatrachus by performing a reverse chemical manipulation to upregulate RA signaling, providing strong evidence that modifications to this specific pathway promoted the emergence of a lineage-specific phenotypic novelty. Interestingly, our screen also revealed pathways that have not yet been implicated in early gut morphogenesis, such as thyroid hormone signaling. In general, the chemical genetic screen may be a valuable tool for identifying developmental mechanisms that underlie ecologically and evolutionarily relevant phenotypic variation.

Fig. 1

Gut development in omnivorous and carnivorous anuran larvae. Ventral views of the developing gut of an omnivorous tadpole (Xenopus laevis) at Nieuwkoop and Faber (NF) stages 41 (A), 43 (D) and 46 (G) are compared to the developing guts of carnivorous Ceratophrys cranwellii and Lepidobatrachus laevis tadpoles at comparable Gosner stages (GS) 21 (B and C), 23 (E and F) and 25 (H and I). In Xenopus (A) the GD loop (arrow) is located in a proximal position along the length of the gut tube, the foregut (FG) is small relative to the midgut (MG), and the pancreas is located within the GD concavity. The GD loop is similarly positioned in Ceratophrys (B), although the pancreas is not visible early in development. In Lepidobatrachus the GD loop forms more distally, which leaves the portion of the gut tube proximal to the GD loop of more equal proportion to the prospective midgut (C). The relative positions of the developing stomach (s), liver (L) and pancreas (p) are indicated, where visible. (The pancreas remains dorsal in Lepidobatrachus and is not visible in these ventral views.) Dashed lines in G, H, and I indicate the approximate position of the embryonic midline and the left and right sides of each embryo. Images are not to scale. The cladogram (J) illustrates the relationships among Xenopus and three ceratophryine genera, including Ceratophrys and Lepidobatrachus.

Fig. 2

Treatment of anuran embryos with a retinoic acid synthesis inhibitor results in the formation of a more derived/carnivore-like GD loop morphology. Xenopus laevis embryos were subjected to an acute chemical treatment with solvent control (ethanol, EtOH; A) or an RA synthesis inhibitor, DEAB (0.4 mM; B). The Xenopus GD loop (arrowhead; NF46) shifts posteriorly upon exposure to DEAB (B) and the final foregut anatomy appears similar to the normal morphology of Lepidobatrachus laevis (C; G 23; Although Xenopus NF46 is most equivalent to GS25, the relative anatomical topology of the foregut organs is already established by GS23 and is more easily visualized at this stage, i.e., before stomach expansion.). D: Effects on gut morphogenesis after treatment with small molecule inhibitors of retinoic acid synthesis (DEAB) or signaling (Ro-41-5253) are concentration dependent. The percentage of embryos with the derived/carnivore-like GD loop and organ placement (NF46) is indicated for different concentrations of each molecule. Embryos that exhibit severely disrupted development (e.g., massive edema, tail curvature) or abnormal, uninterpretable phenotypes not resembling either species are classified as “teratogenized.” Results are pooled from 5 different experiments. DMSO was used as the solvent control for Ro-41-5253. Ceratophrys cranwellii embryos were subjected to an acute chemical treatment with solvent control (EtOH; E) or an RA synthesis inhibitor, DEAB (0.5 mM; F). As observed in Xenopus, the Ceratophrys GD loop (arrowhead; GS25) shifts posteriorly upon exposure to DEAB (F) and the intestine (*) does not elongate, phenotypes remarkably similar to the morphological features found in Lepidobatrachus (C; GS23). The relative positions of the developing stomach (s), liver (L) and pancreas (p) are indicated, where visible.

Fig. 3

Treatment of Lepidobatrachus laevis with ectopic RA results in a more ancestral/omnivore-like GD loop morphology. Lepidobatrachus embryos were exposed to solvent control (ethanol, EtOH; A) or RA (B). RA-treated Lepidobatrachus embryos (B; GS23) exhibit a lack of midgut elongation, a known teratogenic effect of RA exposure in vertebrates. In this context, the GD loop (arrowhead) shifts anteriorly upon exposure to ectopic RA (B) and foregut morphology appears remarkably similar to that of Xenopus laevis at a comparable degree of midgut elongation (C; NF41, reproduced from Fig. 1A). D: Effects on gut morphogenesis after treatment with ectopic RA at successively later developmental stages are both concentration- and stage-dependent. Results are pooled from two different breedings. “Teratogenized” classification is as in Fig. 2. The relative positions of the developing stomach (s), liver (L) and pancreas (p) are indicated, where visible.

Fig. 4

Chemically modulating RA signaling in anuran embryos shifts the location of the GD loop along the anterioposterior axis of the gut tube. The stomach-duodenal boundary in each species is indicated by expression of the homeobox transcription factor Nkx2.5. The GD loop (red arrowhead) is located adjacent to the Nkx2.5 expression domain (black arrowhead) in Xenopus controls (A; NF42), but it is shifted posteriorly following DEAB treatment (C; NF45). Conversely, the GD loop is located posterior to Nkx2.5 expression in Lepidobatrachus controls (B; GS23), but it is shifted anteriorly following RA treatment (D, GS23).

Fig. 5

Chemically modulating RA synthesis and signaling shifts the expression of Pitx2, a left-side determinant of asymmetric gut looping. Compared to the domain of Pitx2 expression (purple) revealed by in situ hybridization in Xenopus embryos (A), the Pitx2 domain is shifted posteriorly and ventrally in the Lepidobatrachus embryo (B). C–D: Pitx2 expression is shifted posteriorly and ventrally in Xenopus embryos exposed to DEAB (C), and anteriorly and dorsally in Lepidobatrachus embryos exposed to RA (D). Arrows in A and C indicate the posterior limit of Pitx2 expression; those in B and D indicate the anterior limit.

Fig. 6

An evolutionary hypothesis regarding the sequence in which increased thyroid hormone (TH) or decreased retinoic acid (RA) signaling arose in ceratophryine lineages. The domain of Pitx2 expression (depicted with dark shading on the embryo diagrams) is anteriorly restricted in Xenopus and Ceratophrys, but begins and extends more posteriorly in Lepidobatrachus (as highlighted by the horizontal bars over each diagram). These domains are correlated with the position of the gastroduodenal loop, which is positioned more dorsally in Lepidobatrachus. Lepidobatrachus possesses the most extreme larval carnivore morphology, with an enlarged and transversely oriented stomach and a severely reduced, dorsally positioned pancreas. Lepidobatrachus and Ceratophrys share several thyroid hormone dependent traits, such as rapid development and precocious pepsinogen production (see Discussion; data not shown), which suggests that increased TH signaling occurred in the ancestors of all ceratophryine lineages. In contrast, only Lepidobatrachus larvae possess a posteriorally shifted Pitx2 domain (a pattern that can be reproduced in Xenopus using an RA synthesis inhibitor), and Ceratophrys respond to inhibited RA synthesis by developing the more derived, carnivore phenotype, which suggests that decreased RA signaling occurred only in Lepidobatrachus. The domain of Pitx2 expression is unknown for Chacophrys.