A relative shift in cloacal location repositions external genitalia in amniote evolution

Abstract

The move of vertebrates to a terrestrial lifestyle required major adaptations in their locomotory apparatus and reproductive organs. While the fin-to-limb transition has received considerable attention, little is known about the developmental and evolutionary origins of external genitalia. Similarities in gene expression have been interpreted as a potential evolutionary link between the limb and genitals; however, no underlying developmental mechanism has been identified. We re-examined this question using micro-computed tomography, lineage tracing in three amniote clades, and RNA-sequencing-based transcriptional profiling. Here we show that the developmental origin of external genitalia has shifted through evolution, and in some taxa limbs and genitals share a common primordium. In squamates, the genitalia develop directly from the budding hindlimbs, or the remnants thereof, whereas in mice the genital tubercle originates from the ventral and tail bud mesenchyme. The recruitment of different cell populations for genital outgrowth follows a change in the relative position of the cloaca, the genitalia organizing centre. Ectopic grafting of the cloaca demonstrates the conserved ability of different mesenchymal cells to respond to these genitalia-inducing signals. Our results support a limb-like developmental origin of external genitalia as the ancestral condition. Moreover, they suggest that a change in the relative position of the cloacal signalling centre during evolution has led to an altered developmental route for external genitalia in mammals, while preserving parts of the ancestral limb molecular circuitry owing to a common evolutionary origin.

Extended Data Figure 1

Two separable ventral cell populations give rise to the murine genital tubercle. a,b) Injection into the most distal ventral part of the embryo, the tailbud, marks cells posterior/ventral to the phallic part of the urethra (a, arrow; n=7), whereas injection closer to the allantois, into the infra-umbilical mesenchyme, labels cells anterior/dorsal to the phallic part of the urethra (b, arrow; n=4). Cells lining the peritoneal cavity are also being marked (arrowhead), due to accidental piercing of the coelom. Scale bars, 200 μm. gt: genital tubercle, ur: urethra.

Extended Data Figure 2

The squamate hemipenis mesenchyme initiates with limb-like cellular dynamics from the coelomic epithelium through a epithelial-to-mesenchymal transition (EMT) a) Injection of GFP-expressing lentiviruses into the coelom of chicken embryos at HH14 labels cells emerging from the epithelium that contribute to the hindlimb mesenchyme (arrowhead). b) In lizards, labeled cells leaving the coelomic epithelium contribute to the hemipenis mesenchyme (arrowheads). c) Dorsal view of the hindlimb region of an E10.0 mouse embryo. d) Transversal section of a limb bud, showing EMT of the coelomic epithelium (diffuse Laminin staining, empty arrowhead), as cells contribute to the limb-bud mesenchyme. e) Dorsal view of the budding hemipenis of a snake embryo, one day after egg deposition. f) Transversal section of the hemipenis region. The basement membrane of the coelomic epithelium is breaking down (empty arrowhead), while it is intact for both the nephric duct and the surface ectoderm (arrowheads). g-o) Expression of genitalia and limb genes during hemipenis initation. g-i) Tbx4 is expressed early and late during hemipenis initiation, in both the coelomic epithelium (arrowhead) and the hemipenis mesenchyme (arrow). j-l) Tbx5 is only expressed later, in the mesenchyme (arrow), but is absent from the coelomic epithelium (empty arrowheads). m-o) Limb marker gene Lhx9 (see also Fig. 4e) is absent from both epithelium (empty arrowhead) and mesenchyme (empty arrow), but can de detected in dI1 neurons (asterisk). All gene expression assessed in at least n=3. Scale bars, 50 μm. co: coelom; lb: limb; hp: hemipenis; nd: nephric duct; cl: cloaca.

Extended Data Figure 3

Heat maps of Pearson’s and Spearman’s rank correlation coefficients and cluster analysis of whole transcriptome data. a,b) Hierarchical clustering on pairwise correlation coefficients, for whole transcriptome data of anole (a) and mouse (b) samples. Numbers at nodes represent approximately unbiased p-values obtained by multiscale bootstrap resampling. Sample identifiers; a: anole; m: mouse; LB: limb; HP: hemipenis; GT: genital tubercle; e: early; l: late.

Extended Data Figure 4

Heat maps of Pearson’s and Spearman’s rank correlation coefficients and cluster analysis of transcription factor and signaling pathway data. a,b) Hierarchical clustering on pairwise correlation coefficients, of transcription factor and signaling pathway data of anole (a) and mouse (b) samples. Numbers at nodes represent approximately unbiased p-values obtained by multiscale bootstrap resampling. Sample identifiers; a: anole; m: mouse; LB: limb; HP: hemipenis; GT: genital tubercle; e: early; l: late.

Extended Data Figure 5

Heterotopic grafting of the cloacal signaling center leads to ectopic outgrowths and genitalia-like transcriptional changes. a-c) Schematics and close-up images of the cloacal grafting procedure. a) The cloaca of a stage HH17-19 GFP-transgenic chicken embryo (red rectangle) is transplanted into the proximal-ventral portion of the limb of a wild-type embryo. b,c) Only the ventral-most part of the cloaca, including the cloacal membrane, is dissected out (b, red box), and subsequently cleared of excess mesenchymal cells attached to the Shh-expressing endoderm (c, red outline). d-g) Grafting of beads soaked in Shh and Fgf can induce ectopic outgrowths on both limbs (e, n=6/48) and tail (g, n=3/31). h-k) Ectopic expression of genital markers Gata2 (h,i; arrowheads) and Runx1 (j,k; arrowhead) in limb buds, following cloaca-to-limb grafts. l-n) Ectopic expression of genital marker Gata2 (m; arrowheads) and Runx1 (n; arrowheads) in the tail region, following cloaca-to-tail grafts. All gene expression assessed in at least n=3. Scale bar, 200 μm. lb: limb; cl: cloaca; al: allantois.

Extended Data Figure 6

Pairwise differential expression analysis of limb and genitalia transcriptomes. a-d) Smear plot visualization of differential expression analyses of early anole (a), late anole (b), early mouse (c) and late mouse (d) limb versus genitalia transcriptomes. Genes used for the Venn diagram in Fig. 4d (|log2(fold-change)| > 1.5; p-value < 0.05) are highlighted in red, core 25 marker genes (see Fig. 4e and text) are highlighted and labeled in blue. e,f) Heat map of Z-score normalized expression values for all genes fulfilling Venn diagram criteria (n=2003), for anole (e) and mouse (f) data. Row-based hierarchical clustering was employed; core 25 marker genes are indicated on the right.

Extended Data Figure 7

Comparative marker gene expression analysis in mouse and squamate embryos. Genitalia markers Isl1 (a-d), Runx1 (e,f), Gata2 (g-j), Eya (k,l), Tbx5 (m-p) and Dkk2 (q,r). Gata2 only becomes visibly expressed at later stages of house snake hemipenis development (j, inset). Limb markers Lhx9 (s-v), Tbx18 (w,x) and Lmx1b (y-zi). All gene expression assessed in at least n=3. Scale bar, 200 μm. lb: limb; cl: cloaca; gt: genital tubercle; hp: hemipenis.

Extended Data Figure 8

Pairwise differential expression analysis of tailbud and genitalia transcriptomes. a,b) Smear plot visualization of differential expression analyses of early anole (a) and early mouse (b) tailbud versus genitalia transcriptomes. Genes used as input for the Venn diagram in (c) (|log2(fold-change)| > 1.5; p-value < 0.05) are highlighted in red, overlapping 257 marker genes are highlighted in blue. Top 25 genes in the two species, based on logCPM and logFC, are labeled. c) Venn diagram showing overlap of pairwise differential expression analysis results (log(fold change) > 1.5, p-value < 0.05) of tailbud versus genital tissues, for early budding stages in both anole and mouse.

Figure 1

A relative positional shift of limbs, genitalia and the cloaca in squamates. μCT-scans of mouse (a), anole (b), python (c) and house snake (d) lumbo-sacral regions (highlighted in sketch in white) at embryonic stages, illustrating the position of the developing external genitalia. (e-h) 3D-reconstructions of cloacal volumes. The cloaca is located at the same anterior-posterior position as the limb in squamates (f-h), however, is positioned more posteriorly in the mouse (e). (i-l) Transversal sections stained for β-catenin and Shh, indicating the conservation of a cloacal signaling center in all four species. Scale bars, 200 μm. gt: genital tubercle; hp: hemipenis; lb: limb; cl: cloaca.

Figure 2

Differential developmental origins of external genitalia in amniotes. (a-i) Transversal and sagittal views of GFP lentivirus injected embryos. Relative contribution of GFP-positive cells to respective organs is quantified on the right, normalized on tissue area. Error bars represent standard deviation in at least n=4 biological replicates. (a,b) Injection into the coelom at mouse E9.5 (n=48) labels the limb at E13.5, but excludes the genital tubercle (arrows). Only cells lining the peritoneal cavity are labeled (b, arrowhead), but none in the genital tubercle proper. c) Injection into the tailbud (n=101) labels cells in the genital tubercle. Accidental piercing of the coelom labels cells of the peritoneal cavity (arrowhead). (d,e) Coelom injection in HH14 chicken embryos (n=81) labels the limb and the genital tubercle at HH30. e) Sagittal and transversal close-up (inset) views. f) Sagittal and transversal close-up (inset) views of tailbud injected chick embryos (n=77), showing labeling in the genital tubercle. (g,h) Anole embryos injected into the coelom at stage 2-3 (n=94) show GFP labeling of both limb and hemipenis at stage 6-7. i) No hemipenis cells are labeled following tailbud injection (n=57), even though there are GFP-positive cells in the tail (inset, arrowhead). Scale bars, 200 μm, 50 μm (h,i). lb: limb; gt: genital tubercle; gf: genital fold; hp: hemipenis; tl: tail.

Figure 3

Molecular architecture of limbs and external genitalia in lizards and mice. a) Micro-dissected tissues for RNA-seq analysis, highlighted by color code and four-letter sample identifier. Early and late limb- and genitalia-buds were analyzed, for anole lizard and mouse embryos (n=2). b) Multidimensional scaling (MDS) analysis reveals greater overall transcriptome similarities in anole limb and genitalia datasets (triangles) than in their mouse counterparts (circles). (c,d) Hierarchical clustering of pairwise Pearson’s correlation coefficients, for whole transcriptome data of anole (c) and mouse (d) samples. Additional datasets are stage 2-3 anole and E9.5 mouse forelimb (turquoise) and tailbud (yellow). Numbers at nodes represent approximately unbiased p-values, obtained by multiscale bootstrap resampling. e) Principal component analysis (PCA). Species transcriptomes separate along principal component 1 (PC1), whereas organs are resolved along PC2. f) Absolute loading values for PC2, as shown in (e). g) GO term enrichment analysis using the top500 genes (red bar, f). Top hits include transcription factor- and signaling pathway-related terms. (h,i) Hierarchical clustering analysis of pairwise Pearson’s correlation coefficients, for transcription factors and signaling pathways data of anole (h) and mouse (i) samples. Sample identifiers; a: anole; m: mouse; LB: limb; HP: hemipenis; GT: genital tubercle; e: early; l: late.

Figure 4

The cloacal signaling center can recruit different mesenchymal cell populations for the outgrowth of external genitalia. a) Schematic of the hindlimb grafting procedure in chicken embryos. GFP-transgenic cloacae are transplanted into the proximal-ventral portion wild-type hindlimbs. b,c) Ectopic outgrowth (arrowheads) on limbs with cloacal grafts (n=30/118). d) Venn diagram of pairwise differential expression analysis results (log(fold change) > 1.5, p-value < 0.05) of limbs versus genital tissues, for early and late budding stages in anole and mouse. e) Heat map of Z-score normalized values of core 25 genes showing consistent differential expression between limbs and genitalia. Mouse row-based hierarchical clustering was re-used for anole samples. f-h) Analysis of limb- and genital-specific markers. Expression of limb markers Lhx9 (f) and Tbx18 (g) is down-regulated near the GFP-positive cloacal graft (empty arrowheads), while genital marker Isl1 (h) is expressed ectopically (arrowhead). i) Schematic of the tail bud grafting procedure in chicken embryos. GFP-transgenic cloacae are transplanted into ventral wild-type tailbuds. j) Ectopic outgrowth on tails with cloacal grafts (n=16/87). k) Genital marker Isl1 is up-regulated ectopically in the tail, close to the GFP-positive cloacal graft (arrowhead). All gene expression assessed in at least n=3. Scale bars, 200 μm.