Evolution of cis-regulatory modules for the head organizer gene goosecoid in chordates: comparisons between Branchiostoma and Xenopus

Abstract

Background: In cephalochordates (amphioxus), the notochord runs along the dorsal to the anterior tip of the body. In contrast, the vertebrate head is formed anterior to the notochord, as a result of head organizer formation in anterior mesoderm during early development. A key gene for the vertebrate head organizer, goosecoid (gsc), is broadly expressed in the dorsal mesoderm of amphioxus gastrula. Amphioxus gsc expression subsequently becomes restricted to the posterior notochord from the early neurula. This has prompted the hypothesis that a change in expression patterns of gsc led to development of the vertebrate head during chordate evolution. However, molecular mechanisms of head organizer evolution involving gsc have never been elucidated.

Results: To address this question, we compared cis-regulatory modules of vertebrate organizer genes between amphioxus, Branchiostoma japonicum, and frogs, Xenopus laevis and Xenopus tropicalis. Here we show conservation and diversification of gene regulatory mechanisms through cis-regulatory modules for gsc, lim1/lhx1, and chordin in Branchiostoma and Xenopus. Reporter analysis using Xenopus embryos demonstrates that activation of gsc by Nodal/FoxH1 signal through the 5' upstream region, that of lim1 by Nodal/FoxH1 signal through the first intron, and that of chordin by Lim1 through the second intron, are conserved between amphioxus and Xenopus. However, activation of gsc by Lim1 and Otx through the 5' upstream region in Xenopus are not conserved in amphioxus. Furthermore, the 5' region of amphioxus gsc recapitulated the amphioxus-like posterior mesoderm expression of the reporter gene in transgenic Xenopus embryos.

Conclusions: On the basis of this study, we propose a model, in which the gsc gene acquired the cis-regulatory module bound with Lim1 and Otx at its 5' upstream region to be activated persistently in anterior mesoderm, in the vertebrate lineage. Because Gsc globally represses trunk (notochord) genes in the vertebrate head organizer, this cooption of gsc in vertebrates appears to have resulted in inhibition of trunk genes and acquisition of the head organizer and its derivative prechordal plate.

Keywords: Cephalochordate; Chordin; Gene regulatory network; Genomics; Lim1; Nodal/FoxH1 signaling; Otx; Spemann-Mangold organizer; Vertebrate.

Fig. 1

Schematic representations of organizer gene expression patterns. a Expression patterns of lim1, otx2 (otx in amphioxus), goosecoid (gsc), and chordin (chrd) of Xenopus (top) and amphioxus (bottom) at the early gastrula stage (left) and the late neurula stage (right) are shown with colors as indicated. b–i Whole-mount in situ hybridization of gsc in B. japonicum embryos. In mid-gastrula (stage G5–6), Bj_gsc is expressed in the dorsal mesoderm (b–d). In late gastrula (stage G7–N0), Bj_gsc is still expressed throughout the dorsal mesoderm (e, f). In early neurula (stage N1), Bj_gsc expression is still strong in the posterior mesoderm but very weak in the anterior mesoderm (g, h). In mid-neurula (stage N2), Bj_gsc is expressed in the posterior mesoderm (arrowhead) and weakly in the dorsal endoderm (open arrowheads) (i). Embryos are shown in lateral view with dorsal to the top and anterior to the left (B, e, g, i), dorsal view with anterior to the left (c, f, h), or blastoporal view with dorsal to the top (d). *, blastopore

Fig. 2

CRM activities of the second intron of Xenopus chrd. a Sequence alignment of X. tropicalis and X. laevis chrd-D1 core regions (243 bp). Orange boxes, conserved Lim1 motifs; purple box, conserved FoxH1 motif (a major partner of Smad2/3 in Nodal signaling). Siamois may bind to the Lim1 site [50]. b–f Luciferase reporter assays. Xt_chrd-D1 (b, c), intron 2 sequences of Xl_chrd.L and .S (d), Xt_chrd-D1_104 bp (between light green brackets in a) (e), and Xt_chrd-D1_mt (all four conserved Lim1 motifs are mutated) were analyzed. Xt_chrd-D1 showed synergistic activation by Lim1, Ldb1, Ssbp3, and Otx2 (b). Strong activation through Xt_chrd-D1 was observed in Lim1/Ldb1/Ssbp3, Siamois and activin, but not in Wnt8 (c). Xl_chrd.L and .S intron2 showed conserved enhancer activity, which was activated by Lim1/Ldb1/Ssbp3 (d). No responsiveness of Xt_chrd –D1_104 bp to activin (e) suggests that Nodal signaling activates chrd-D1 through the conserved FoxH1 site. Reporter activation by Lim1/Ldb1/Ssbp3 was abolished by mutating four Lim1 motifs in Xt_chrd-D1 (f), indicating that Lim1 directly activates chrd through the intron2 enhancer. Bars represent mean ± s.e.m. *, P < 0.05; **, P < 0.01 (t-test, two tailed). Dosages of injected mRNAs are as follows: lim1, 100 pg/embryo; ldb1, 100 pg/embryo; ssbp3, 100 pg/embryo; otx2, 40 pg/embryo; simois, 100 pg/embryo; wnt8a, 25 pg/embryo; and activin A, 20 pg/embryo. g Transgenic reporter analysis of Xt_chrd-D1. Panels represent whole mount in situ hybridization of the reporter gene mVenus or endogenous chrd for transgenic embryos with dorso-ventral hemisections. Expression patterns were examined at the early (st. 10), middle (st. 11) and late (st. 12.5) gastrula stages as indicated. In total, 8 of 30 transgenic embryos showed reporter expression in the organizer. Embryos are shown with the animal pole at the top and dorsal to the right. Arrow heads, blastopore

Fig. 3

Epigenetic data from B.lanceolatum embryos and sequence comparisons between B.lanceolatum, B. floridae, B. belcheri, and B. japonicum. a ATAC-seq, H3K27ac ChIP-seq, and H3K4me3 ChIP-seq data from early gastrula (8 hpf) and early neurula (15hpf) in Bl_lim1 intron 1 are represented with Vista plot of Bl_lim1 intron 1 vs Bf_lim1 intron 1, Bb_lim1 intron1 and Bj_lim1 intron 1. Regions with 50–100% identity were shown and conserved non-coding sequences (CNSs) were colored in red. The number of Smad motifs and FoxH1 motifs in CNSs is shown as indicated by arrows. b Epigenetic data in Bl_chrd intron 2 are represented with Vista plot of Bl_chrd intron 2 vs Bf_chrd intron 2 Bb_chrd intron 2 and Bj_chrd intron 2. The number of Lim1 sites in CNSs is indicated by arrows. c Epigenetic data in Bl_chrd intron 2 are represented with Vista plot of the − 5 kb region of Bl_gsc vs those of Bf_gsc, Bb_gsc and Bj_gsc. The number of Lim1, bicoid, and FoxH1 sites in CNSs is shown as indicated by arrows. a–h, Regions used for reporter assays. Blue boxes indicate putative CRMs analyzed in reporter assays (Figs. 4 and 5). Additional file 1: Figure S1 shows sequence alignment of them (see Additional file 1)

Fig. 4

Reporter analyses of the lim1 intron 1 and chrd intron 2 in the Xenopus embryo. a, b Luciferase reporter assays of lim1 intron 1. Responsiveness of reporter constructs to Nodal signaling was tested with or without activin A mRNA (40 pg/embryo). A reporter construct mutated in two FoxH1 motifs (Fm), but not that in a Smad motif (Sm) exhibited no response to Nodal signaling (b), suggesting that Nodal signaling directly regulates Bj_lim1 through FoxH1 binding to the intron1 enhancer. c, d Luciferase reporter assays of Bj_chrd intron 2. Responsiveness of Bj_chrd intron 2 (region g) to Lim1 and Otx2 was tested with or without lim1, ldb1, ssbp3, and otx2 mRNAs (100, 100, 100 and 40 pg/embryo, respectively). Lim1/Ldb1/Ssbp3 significantly activated the reporter gene through Bj_chrd intron 2 (C), but the activation level was significantly reduced by mutating three Lim1 motifs (D). See Fig. 2f for details of the Lim1 motif mutation. Bars represent mean ± s.e.m. *, P < 0.05, **, P < 0.01; †, P < 0.1 (t-test, two tailed)

Fig. 5

Luciferase reporter analysis of the gsc 5′ region in Xenopus embryos. a Luciferase reporter assays of Bj_gsc 5′ regions for responsiveness to endogenous factors. Reporter constructs were injected into the animal pole (AP), ventral equatorial region (VER), or dorsal equatorial region (DER) at the four-cell stage to examine responsiveness of constructs to endogenous dorsal signals. Results were normalized with activity of embryos injected with reporter constructs into the animal pole. b Luciferase reporter assays of Xl_gsc-U1 and the Bj_gsc 5′ region for responsiveness to exogenous factors. Lim1/Ldb1/Ssbp3a strongly activated reporter gene expression through Xl_gsc-U1 but only slightly through the Bj_gsc 5′ region. While, Wnt and Nodal signaling synergistically activated reporter gene expression through the Bj_gsc 5′ region. c Luciferase reporter assays of Bj_gsc 5′ region with mutations of three FoxH1 motifs for responsiveness to Nodal signaling. The mutation construct greatly reduced responsiveness to activin, suggesting that Nodal/FoxH1 signaling directly regulates Bj_gsc through the 5′ region. See Fig. 4b for details of the FoxH1 motif mutation. Reporter constructs were injected into the animal pole with combinations of mRNAs with dosages as follows: lim1, 100 pg/embryo (Xl_gsc-U1) or 50 pg/embryo (Bj_gsc 5′ regions); ldb1, 100 pg/embryo (Xl_gsc-U1) or 50 pg/embryo (Bj_gsc 5′ regions); ssbp3, 100 pg/embryo (Xl_gsc-U1) or 50 pg/embryo (Bj_gsc 5′ regions); otx2, 50 pg/embryo; wnt8, 25 pg/embryo; and activin A, 40 pg/embryo. Bars represent mean ± s.e.m. *, P < 0.05; **, P < 0.01 (t-test, two tailed)

Fig. 6

Transgenic reporter analysis of the gsc 5′ region in Xenopus embryos. Transgenic reporter assays of the − 3 kb region of Xt_gsc and the − 4.5 kb region of Bj_gsc in Xenopus embryos. Panels represent whole mount in situ hybridization of the reporter gene mVenus (first and second rows) or the endogenous gsc gene (third row) in dorso-ventral hemisections. Expression patterns were examined at the early (st. 10), middle (st. 11) and late (st. 12.5) gastrula stages and late-neurula stage (st. 23) as indicated. In right panels of late gastrula embryos, hemisections were represented in the dorsal view with anterior to the top. Other panels of gastrula are shown with animal to the top and dorsal to the right. Neurula embryos are shown with dorsal to the top and anterior to the left. Neurula embryos cleared in BB/BA solution are shown in right panels. Arrowheads, blastopore; open arrowheads, mouth; and arrows, notochord. Additional file 1: Figure S2 shows results in more detail (see Additional file 1)

Fig. 7

Comparisons between amphioxus and Xenopus. a Comparisons of organizer formation and organizer gene regulatory networks (GRNs) between amphioxus (left) and Xenopus (right). GRNs in the head organizer of Xenopus are shown with a magenta circle. White boxes, CRMs of each gene; gray box, CRMs of amphioxus otx, which have not been identified yet; dotted lines, suggested regulation [26]. b Comparisons of expression domains of transcription factors (otx, lim1, gsc, and brachyury) in the dorsal endoderm and the dorsal mesoderm between amphioxus (a–c) and Xenopus (d–f). Colored bars represent expression domains of genes at the early gastrula stage (a,d), the late gastrula stage (b,e) and the late neurula stage (c,f) with anterior to the left, as indicated. The head organizer region and regulatory interactions between transcription factors are indicated in the late gastrula stage in Xenopus

Fig. 8

Evolutionary scenario of the vertebrate head organizer. Assuming the amphioxus-like chordate ancestor, the vertebrate ancestor should have adopted Wnt signaling for organizer formation and coopted gsc as a target of Lim1 and Otx2 to form the anteriorly enlarged brain by converting the anterior presumptive notochordal cells to the prechordal plate. Schematics of body plans are shown with anterior to the left and dorsal to the top. Orange, brain and neural tube; green, notochord; blue, prechordal plate