ALM experiments at the four orientations

(A) Schematic image of anatomy at the stylopod level of axolotl limb. HE staining (bright field) and acetylated alpha tubulin, visualized by immunofluorescence (green, dark field), are shown in the right panels. Black and white arrows, respectively, indicate major blood vessels and nerves. H: humerus, AHL: Anconaeus humeralis lateralis, HAB: Humeroantebrachialis, CBL: Coracobrachialis longus. (B‒I) Blastemas induced at the anterior, posterior, dorsal, or ventral region by skin wounding plus nerve deviation without (B‒E) or with (F‒I) skin grafting from the opposite side of the limb. (F‒I) Limb patterning was observed (n=8/9 for F, 4/7 for G, 7/14 for H, and 7/11 for I, see Table 1 for more detail). Images were captured at 10 and 60 dps. Scale bar = 3 mm.

The induction rate of bump/limb formation in ALM

Gene expression patterns of the ALM-induced blastemas

Sections of anteriorly (A‒D), posteriorly (F‒I), dorsally (K‒N), or ventrally (P‒S) induced blastemas at 10 dps. Acetylated alpha tubulin (A‒P) was visualized by immunofluorescence. Expression of Lmx1b (B‒Q), Fgf8 (C‒R), and Shh (D‒S) in the regions indicated by white boxes in (A‒P) was visualized by in situ hybridization. Images of the entire blastema are provided in Fig. S1. Fgf8 expression was observed in AntBL (C), DorBL (M), and VentBL (R) (n=4/5, 4/6, and 4/6, respectively. Shh expression was observed in PostBL (I) (n=5/5). In each case, these expression patterns of Fgf8 and Shh were focal and only a few cells expressed Fgf8 or Shh. Black and white arrowheads indicate the signals of Fgf8 and Shh expression, respectively. The dotted line indicates the epithelial-mesenchyme border. (E‒T) Schematic images of gene expression patterns. (U, V) Quantitative analysis of Fgf8 and Shh expression in ALM blastemas (n=5 for all groups). n.s.; no significant difference, *p<0.05, **p<0.005 (two-tailed Welch’s t-test). Scale bar in (A) = 700 μm.

Co-existence of dorsal and ventral cells induces Shh expression

(A) Experimental scheme. Posterior half of dorsal (PDgfp) or ventral (PVgfp) GFP-expressing skin was grafted on VentBL (VentBL+PDgfp; C, D / VentBL+PVgfp; G, H), or DorBL (DorBL+PDgfp; E, F / DorBL+PVgfp; I, J) region. (B) Induced blastemas at 10 dps; images of bright and dark fields are merged. (C‒J) Dark and bright fields of the same sections of induced blastemas at 10 dps. Red boxes in (C‒I) indicate the corresponding regions of lower images. Expression of Shh and Lmx1b were visualized by in situ hybridization. GFP signals were visualized by immunofluorescence. Arrowheads indicate the cells expressing Shh. Shh expression was observed in VentBL+PDgfp (C) and DorBL+PVgfp (I) (n=4/6 and 3/6, respectively), but not in DorBL+PDgfp (E) and VentBL+PVgfp (G) (n=6/6 and 7/7, respectively). The dotted line indicates the epithelial-mesenchyme border. For all samples, we collected serial sections spanning the entire blastema. For blastemas in which Shh expression was observed, we present representative sections showing the signal. For blastemas without detectable Shh expression, we present a section from the central region that contains GFP-positive cells. Scale bar = 3 mm (B) and 700 μm (C).

Limb formation without the co-existence of dorsal and ventral cells by Shh overexpresion

(A) Experimental scheme. Shh-p2A-GFP- or GFP-containing pCS2 vector was electroporated (EP) into AntBL, DorBL, and VentBL. (B) Induced blastemas at 14 dps; images of bright and dark fields are merged. (C‒G) Phenotypes at 90 dps. Limb patterning was observed in (E‒G) (n=10/19, 6/18, and 8/14, respectively). (H‒K) Histological analysis of intact and induced limbs. Standard Masson’s trichrome staining was performed on the transverse sections. The dotted boxes indicate the regions shown in (L). (L) Upper panels: analyzed regions for calculating symmetry scores. Lower panels: images after pixel classification by machine learning. (M‒R) Symmetry scores of each class. Scores obtained from the same limb are plotted at the same x-coordinates. n.s.; no significant difference, *p<0.05, **p<0.005 (two-tailed Welch’s t-test). Scale bar = 2 mm (B), 4 mm (C‒G), and 1 mm (H‒K).

Identification of candidate molecules of the dorsal- and ventral-mediated signals

RNA-Seq was performed on DorBL and VentBL at 10 dps. (A) MA plot of the result. (B) List of DEGs annotated as “intercellular signaling molecules.” (C) Bright and dark fields of sections of DorBL and VentBL at 10 dps with candidate genes introduced. Expression of Shh and Lmx1b was visualized by in situ hybridization, and GFP signals were visualized by immunofluorescence. Arrowheads indicate the cells expressing Shh. The white boxes indicate the regions of the lower panels. Images of dark and bright fields of Shh are obtained from the same section. For samples with detectable Shh expression, the window was placed in the region where the signal was observed, and for conditions without detectable Shh expression, the window was positioned in a comparable region containing GFP-positive cells. Shh expression was observed in Wnt10b-electroporated VentBL (n=4/5) and Fgf2-electroporated DorBL (n=5/7), but not in Wnt10b-electroporated DorBL (n=6/6), Fgf2-electroporated VentBL (n=5/5), Fgf7-electroporated DorBL (n=5/5), Tgfb2-electroporated DorBL (n=5/5), or in Wnt4-electroporated VentBL (n=6/6). (D, E) Quantitative analysis of Shh expression in Fgf2- or GFP-electroporated DorBL and Wnt10b- or GFP-electroporated VentBL (n=7 for both). In each case, Fgf2 or Wnt10b was electroporated into the DorBL or VentBL induced in the left limb, and GFP was electroporated into the contralateral right limb of the same animal. (F, G) Quantitative analysis of Axin2 and Lef1 expression in Wnt10b- or GFP-electroporated VentBL (n=7 for both). (H, I) The limbs formed from VentBL with Wnt10b and DorBL with Fgf2 at 90 dps. Histological analysis and pixel classification was performed in the same way as in Fig. 4. The dotted boxes indicate the regions shown in the right panels. (J) Symmetry scores of each class. Scores obtained from the same limb are plotted at the same x-coordinates. The plots of “int” are the same plots as in Fig. 4. n.s.; no significant difference, *p<0.05, **p<0.005 (two-tailed paired t-test for D‒ G, and two-tailed Welch’s t-test for J). Scale bar = 700 μm (C), 4 mm (H, I, upper panels), 1 mm (H, I, lower panels).

Limb formation at the dorsal region by BMP2+FGF2+FGF8 supplementation

(A, B) 14 and 60 dps phenotypes of BMP2+FGF2+FGF8 supplementation by bead grafting at the dorsal (A) or ventral (B) region. Neither nerve deviation nor skin grafting was performed. Limb patterning was observed in the dorsa group (A) (n=12/20) but not in the ventral group (B) (n = 0/17). (C‒H) Expression patterns of Fgf8, Shh and Lmx1b of induced blastemas at 10 dps. Gene expression was visualized by in situ hybridization. The dotted line indicates the external shape of the blastema. Fgf8 expression was detected in both dorsal and ventral groups (n=8/8 for C and 5/5 for F), whereas Shh expression was detected only in the dorsal group (D) (n=8/8) and not in the ventral group (G) (n= 0/5). (I‒J) Phenotype obtained by BMP2+FGF2+FGF8 supplementation to two dorsal regions of an identical limb. (K, L) Phenotype obtained by BMP2+FGF2+FGF8 supplementation to five dorsal regions of an identical limb (n=16/35). Scale bar = 3 mm (B), 700 μm (H), 2 cm (J), 1 cm (K, L).

The dorsoventral-mediated Shh induction mechanism

Schematic images of a normal blastema (A), AntBL (B), PostBL (C), DorBL (D), and VentBL (E). Green, red, blue, and yellow boxes within the blastema represent cells derived from anterior, posterior, dorsal, and ventral regions, respectively. Colored arrows indicate the presence of the corresponding signals mediated by these cells. In this model, limb patterning requires both FGF8 and SHH, and Shh expression in posteriorly derived cells is induced by the co-existence of WNT10B and FGF2.

Gene expression patterns across the entire ALM-induced blastemas

Sections of ALM blastemas induced at the anterior (A‒C), posterior (D‒F), dorsal (G‒I), or ventral (J‒L) region at 10 dps. Lmx1b (A, D, G, J), Fgf8 (B, E, H, K), and Shh (C, F, I, L) expression was detected by in situ hybridization. These panels show the full blastema regions corresponding to the higher-magnification views in Fig. 2. Scale bar = 700 μm (A).

Histological analysis of intact limbs and limbs induced by Shh electroporation

(A‒D) Transverse sections taken at multiple levels along the proximodistal axis from an intact limb and from the limbs shown in Fig. 4. Sections from limbs induced from DorBL and VentBL exhibit dorsoventrally symmetric internal structures, whereas intact limbs and limbs induced from AntBL do not. Scale bar = 1 mm.

Limb patterning from BIO-treated VentBL

(A, C) VentBL treated with 1μM BIO (C) or DMSO (A) at 90 dps. (E) Quantitative analysis of Shh expression in VentBL treated with 1μM BIO and with DMSO. RNA was prepared from 4 identical VentBL for both. Technological replicates are plotted in the same color. (C‒D) Histological analysis and pixel classification was performed in the same way as in Fig. 4. The dotted boxes indicate the regions shown in (D). (E) Symmetry scores of each class. Scores obtained from the same limb are plotted at the same x-coordinates. The plots of “int” are the same plots as in Fig. 4. n.s.; no significant difference, *p<0.05, **p<0.005. Scale bar = 4 mm (B), 1 cm (C).

Histological analysis of limbs induced by Fgf2 or Wnt10b electroporation

(A, B) Transverse sections taken at multiple levels along the proximodistal axis from the limbs shown in Fig. 5. The induced limbs exhibit dorsoventrally symmetric internal structures. Scale bar = 1 mm.

Gene expression patterns during normal limb regeneration

(A) Sections of amputation-induced blastemas at the early bud (EB), middle bud (MB), and late bud (LB) stages. Lmx1b, Wnt10b, and Fgf2 expression was visualized by in situ hybridization. (B) Quantitative analysis of Lmx1b, Wnt10b, and Fgf2 expression in MB-stage blastemas during normal regeneration (n=13). Gene expression was quantified by RT-qPCR on manually microdissected dorsal and ventral halves of each blastema. *p<0.05, **p<0.005 (two-tailed paired t-test). (C) Quantitative analysis of Wnt10b, Fgf2, and Shh expression across stages (intact, EB, MB, LB, and early digit [ED]; n = 5 for all stages). Wnt10b and Fgf2 expression peaked at the MB stage, whereas Shh expression peaked later, at the LB stage. One-way ANOVA detected a stage-dependent difference (p < 0.05 for Wnt10b and Fgf2; p < 0.001 for Shh), followed by Dunnett’s test. *p<0.05, **p<0.005.

Gene expression patterns in a normal blastema assessed by reanalysis of axolotl single-cell RNA-seq data

Reanalysis of single-cell RNA-seq data from a middle bud (MB) stage blastema (Li et al., 2021). (A‒G) FeaturePlot visualizations of Prrx1 (mesenchyme marker), Krt17 (epithelium marker), Lmx1b, Wnt10b, Fgf2, Fgf8, and Shh. Lmx1b, Fgf2, Fgf8, and Shh expression are detected in the mesenchymal cluster, whereas Wnt10b expression is detected in both mesenchymal and epithelial clusters. (H‒M) “Co-expression” maps generated by classifying cells as expressing gene A only, gene B only, both genes, or neither gene, and overlaying these classes on the UMAP using DimPlot.