Secondary Axis Inducing Nodal and BMP Double Clones Produce a Localized Region of pSmad2 Activity Overlapping with a Broad Domain of pSmad5 Activity

(A) Nodal and BMP form orthogonal overlapping gradients in zebrafish embryos. Transplanting ectopic sources of Nodal and BMP induces the formation of a secondary axis, which contains both anterior and posterior structures such as the hindbrain, otic vesicles, notochord, and tail.

(B) Double clones of Bmp2b/7-sfGFP and Squint-mVenus imaged 30 min and 180 min post-transplantation. The first row depicts confocal microscopy images of Bmp2b/7-sfGFP (red) and Squint-mVenus (green). The second row shows light-sheet microscopy images of embryos immunostained with anti-pSmad2 (green) or anti-pSmad5 (red) antibodies as well as a cross-reactive anti-GFP antibody to detect Bmp2b/7-sfGFP and Squint-mVenus (blue). The third row shows comparable wild-type embryos. Nodal clones are traced in cyan and BMP clones are traced in white. Scale bar, 150 μm.

(C) Higher magnification of images shown in (B) with separate fluorescent channels. Scale bar, 150 μm.

(D) Images showing Nodal/BMP double clones with different spacings of transplanted cells taken immediately after transplantation. Scale bar, 150 μm.

(E) Nodal/BMP double clones were transplanted with different spacings into blastula-stage zebrafish embryos: narrow (~0 μm between clones, n = 60), moderate (40–50 μm between clones, n = 44), wide (120–150 μm between clones, n = 29), and very wide (>170 μm between clones, n = 20). Narrow to wide spacings support the formation of secondary axes, whereas secondary axis formation fails with extremely wide spacing between Nodal and BMP clones. Quantification was performed at 24 h post-transplantation.

Nodal and BMP Form Similar Protein Gradients but Have Different Signaling Ranges during Secondary Axis Formation

(A) Bmp2b/7-sfGFP as well as Squint-mVenus and Bmp2b/7-sfGFP double clones in wild-type or maternal-zygotic swirl mutant (MZswr) embryos at 1 day post-transplantation, with untransplanted embryos for comparison. The arrowheads point to ectopic secondary axes. Scale bar, 150 μm.

(B) Bmp2b/7-sfGFP clones compared to uninjected mock clones 30 min and 180 min post-transplantation in wild-type or MZswr embryos. Embryos were immunostained with anti-pSmad5 (red) and anti-GFP (blue) antibodies. Mock sources were labeled with cascade blue-dextran (blue). Scale bar, 150 μm.

(C) pSmad5 distributions in embryos with single Bmp2b/7-sfGFP clones in MZswr embryos at 30 min (n = 9), 60 min (n = 8), 120 min (n = 10), and 180 min (n = 9) post-transplantation. Shaded regions indicate 95% confidence intervals around the mean (lines). Scale bar, 150 μm.

(D) Squint-mVenus as well as Squint-mVenus and Bmp2b/7-sfGFP double clones in wild-type or maternal-zygotic squint and cyclops double mutant (MZsqt;cyc) embryos 1 day post-transplantation, with untransplanted embryos for comparison. The arrowheads point to ectopic structures or secondary axes. Scale bar, 150 μm.

(E) Squint-mVenus clones compared to uninjected mock clones 30 min and 180 min post-transplantation in wild-type or MZsqt;cyc embryos. Embryos were immunostained with anti-pSmad2 (green) and anti-GFP (blue) antibodies. Mock sources were labeled with cascade blue-dextran (blue). Scale bar, 150 μm.

(F) pSmad2 distributions in embryos with single Squint-mVenus clones in wild-type embryos at 30, 60, 120, and 180 min post-transplantation (n = 11 each). Shaded regions indicate 95% confidence intervals around the mean (lines).

(G) BMP protein gradients in wild-type embryos with single Bmp2b/7-sfGFP clones at 30, 60, 120, and 180 min post-transplantation. The same embryos were imaged throughout the time course (n = 14). Fluorescence intensity was converted to concentration based on a calibration curve using recombinant sfGFP imaged with the same microscope settings. Shaded regions indicate 95% confidence intervals around the mean (lines).

(H) Nodal protein gradients in wild-type embryos with single Squint-mVenus clones at 30, 60, 120, and 180 min post-transplantation. The same embryos were imaged throughout the time course (n = 12). Fluorescence intensity was converted to concentration based on a calibration curve using recombinant mVenus imaged with the same microscope settings. Shaded regions indicate 95% confidence intervals around the mean (lines).

See also Figures S2–S4.

Different BMP and Nodal Signaling Ranges Arise from Differential Signaling Activation Kinetics

(A) Single clones expressing mouse BMP4 (mBMP4) induce the formation of a secondary axis in zebrafish embryos (arrowhead). Scale bar, 150 μm.

(B) Mathematical modeling shows that a difference in signaling activation kinetics could explain how a single gradient of mBMP4 induces pSmad5 and pSmad2 at different ranges.

(C) Wild-type zebrafish embryos with clones expressing mBMP4 30 min and 180 min post-transplantation immunostained with anti-pSmad5 (red) and anti-pSmad2 (green) antibodies. The clones were labeled with cascade blue-dextran (blue). Scale bars, 150 μm.

(D) Higher magnification of images shown in (C) with separated fluorescent channels. Scale bar, 150 μm.

(E) MZsqt;cyc embryos with clones expressing mBMP4 30 min and 180 min post-transplantation immunostained with anti-pSmad5 (red) and anti-pSmad2 (green) antibodies. The clones were labeled with cascade blue-dextran (blue). Scale bars, 150 μm.

(F) Higher magnification of images shown in (E) with separated fluorescent channels. Scale bar, 150 μm.

(G) Zebrafish Bmp2b/7-sfGFP and Squint-mVenus clones in morphotrap-expressing wild-type embryos 30 min post-transplantation. Scale bar, 150 μm.

(H) Double clones with fluorescently tagged or untagged zebrafish Nodal and BMP and with narrow or wide spacing were generated in morphotrap-expressing embryos. The frequency of the different structures induced by the clones was assessed 24 h post-transplantation.

Ectopic Expression of Different Amounts of smad2-CA and smad5-CA mRNA Generates Distinct Embryonic Structures

(A) Ectopic structures were generated by injecting three adjacent blastomeres in 64- to 128-cell-stage embryos.

(B) The floorplate marker shha is expressed throughout the axis (gray arrowhead) of wild-type embryos 24 h post-fertilization (hpf). Injection of smad2-CA mRNA into animal pole blastomeres results in the formation of an ectopic axial structure (black arrowhead) that expresses shha, similar to the results with a squint-mVenus-expressing clone.

(C) krox20 is expressed as a pair in rhombomeres 3 and 5 in the hindbrain (gray arrowhead) of wild-type embryos at 24 hpf. Injection of smad2-CA and smad5-CA mRNA in animal pole blastomeres results in the formation of anterior trunk structures with paired krox20 expression (black arrowhead), similar to the outcome with a squint-mVenus and bmp2b/7-sfGFP-expressing double clone.

(D) hoxc13b is expressed in the tail tip (gray arrowhead) in wild-type embryos at 24 hpf. Injection of smad2-CA and four times more smad5-CA mRNA into animal pole blastomeres results in the formation of a tail structure expressing hoxc13b (black arrowhead), similar to the outcome with a double clone expressing low squint-mVenus and high bmp2b/7-sfGFP (Figure S5).

(E) shha is not expressed in embryos exposed to the Nodal receptor inhibitor SB-505124. Injection of smad2-CA mRNA into animal pole blastomeres results in ectopic shha-positive axial structures despite Nodal receptor inhibition.

(F) krox20 remains expressed (gray arrowheads) in embryos exposed to SB-505124. Injection of smad2-CA and smad5-CA mRNA into animal pole blastomeres stage results in the formation of anterior trunk structures with paired krox20 expression (black arrowheads). Nodal and BMP receptor inhibition by combined exposure to SB-505124 and Dorsomorphin generates embryos with reduced tails compared to the treatment with SB-505124 alone, but krox20 expression persists (gray arrowhead). Injection of smad2-CA and smad5-CA mRNA into animal pole blastomeres results in the formation of anterior trunk structures with paired krox20 expression (black arrowhead) despite Nodal and BMP receptor inhibition. Scale bar in all images, 150 μm.

Mutual Antagonism of Smad2 and Smad5 for Specific Cell Fates

(A) gsc is expressed at the dorsal margin (Stachel et al., 1993), while foxi1 is expressed on the ventral side but excluded from the margin (Dal-Pra et al., 2006), and eve1 is expressed at the ventral margin (Joly et al., 1993) where Nodal and BMP signaling overlap (Figure 1A).

(B) Average gsc fluorescence in situ hybridization (FISH) intensity in 6-hpf embryos that were injected with the indicated smad2-CA and smad5-CA mRNA amounts at the one-cell stage (n = 10, 7, and 7).

(C) Average foxi1 FISH intensity in 6-hpf embryos that were injected with the indicated smad2-CA and smad5-CA mRNA amounts at the one-cell stage (n = 7, 6, and 7).

(D) Average eve1 FISH intensity of 6-hpf embryos that were injected with the indicated smad2-CA and smad5-CA mRNA amounts at the one-cell stage (n = 5, 6, 7, and 8).

(E) Embryos with Nodal and BMP double clones subjected to FISH with gsc (left, blue), foxi1 (middle, blue), or eve1 (right, blue) probes followed by pSmad2 (green) and pSmad5 (red) immunostaining. Blue dotted lines trace Nodal clones, and white dotted lines trace BMP clones. Scale bar, 150 μm. Error bars indicate 95% confidence intervals around the mean (horizontal lines) in (B)–(D).

Selective Mutual Antagonism of pSmad2 and pSmad5 Allows Cells to Respond to Different Ratios of Nodal and BMP Signaling

Schematic of a parsimonious model explaining the present findings. The antagonism of pSmad2 and pSmad5 to foxi1 and gsc induction, respectively, allows cells with both high pSmad2 and pSmad5 to express eve1 without expressing foxi1 or gsc. The activation and inhibition arrows are an abstraction, and the underlying mechanisms may be direct or indirect.