The integration mechanism of the four positional cues—anterior, posterior, dorsal, and ventral—in axolotl limb regeneration

Reviewer #1 (Public review):

Summary:

The manuscript by Yamamoto et al. presents a model by which the four main axes of the limb are required for limb regeneration to occur in the axolotl. A longstanding question in regeneration biology is how existing positional information is used to regenerate the correct missing elements. The limb provides an accessible experimental system by which to study the involvement of the anteroposterior, dorsoventral, and proximodistal axes in the regenerating limb. Extensive experimentation has been performed in this area using grafting experiments. Yamamoto et al. use the accessory limb model and some molecular tools to address this question. There are some interesting observations in the study. In particular, one strength is the potent induction of accessory limbs in the dorsal axis with BMP2+Fgf2+Fgf8, which is very interesting.

Strengths:

The manuscript presents some novel phenotypes generated in axolotl limbs due to Wnt signaling. This is generally the first example in which Wnt signaling has provided a gain-of-function in the axolotl limb model. They also present a potent way of inducing limb patterning in the dorsal axis by the addition of just beads loaded with Bmp2+Fgf8+Fgf2.

Weaknesses:

Although interesting, the study makes bold claims about determining the molecular basis of DV positional cues, but the experimental evidence is not definitive and does not take into account the previous work on DV patterning in the amniote limb. Also, testing the hypothesis on blastemas after limb amputation would be needed to support the strong claims in the study. There are several examples of very strong claims, but the evidence lacks support for these claims.

Reviewer #2 (Public review):

Summary:

This study explores how signals from all sides of a developing limb, front/back and top/bottom, work together to guide the regrowth of a fully patterned limb in axolotls, a type of salamander known for its impressive ability to regenerate limbs. Using a model called the Accessory Limb Model (ALM), the researchers created early limb regenerates (called blastemas) with cells from different sides of the limb. They discovered that successful limb regrowth only happens when the blastema contains cells from both the top (dorsal) and bottom (ventral) of the limb. They also found that a key gene involved in front/back limb patterning, called Shh (Sonic hedgehog), is only turned on when cells from both the dorsal and ventral sides come into contact. The study identified two important molecules, Wnt10B and FGF2, that help activate Shh when dorsal and ventral cells interact. Finally, the authors propose a new model that explains how cells from all four sides of a limb, dorsal, ventral, anterior (front), and posterior (back), contribute at both the cellular and molecular level to rebuilding a properly structured limb during regeneration.

Strengths:

The techniques used in this study, like delicate surgeries, tissue grafting, and implanting tiny beads soaked with growth factors, are extremely difficult, and only a few research groups in the world can do them successfully. These methods are essential for answering important questions about how animals like axolotls regenerate limbs with the correct structure and orientation. To understand how cells from different sides of the limb communicate during regeneration, the researchers used a technique called in situ hybridization, which lets them see where specific genes are active in the developing limb. They clearly showed that the gene Shh, which helps pattern the front and back of the limb, only turns on when cells from both the top (dorsal) and bottom (ventral) sides are present and interacting. The team also took a broad, unbiased approach to figure out which signaling molecules are unique to dorsal and ventral limb cells. They tested these molecules individually and discovered which could substitute for actual dorsal and ventral cells, providing the same necessary signals for proper limb development. Overall, this study makes a major contribution to our understanding of how complex signals guide limb regeneration, showing how different regions of the limb work together at both the cellular and molecular levels to rebuild a fully patterned structure.

Weaknesses:

Because the expressional analyses are performed on thin sections of regenerating tissue, they provide only a limited view of the gene expression patterns in their experiments, opening the possibility that they could be missing some expression in other regions of the blastema. Additionally, the quantification method of the expressional phenotypes in most of the experiments does not appear to be based on a rigorous methodology. Therefore, performing alternate expressional analysis, using RNA-seq or qRT-PCR (for example) on the entire blastema would help validate that the authors are not missing something.

Overall, the number of replicates per sample group is quite low (sometimes as low as 3), which is especially risky with challenging techniques like the ones the authors employ. The authors don't appear to have performed a power analysis to calculate the number of animals used in each experiment that are sufficient to identify possible statistical differences between groups. Increasing the sample sizes would substantially increase the rigor of their experiments.

Likewise, the authors' use of an AI-generated algorithm to quantify symmetry on the dorsal/ventral axis, and this approach doesn't appear to account for possible biases due to tissue sectioning angles. They also appear to arbitrarily pick locations in each sample group to compare symmetry measurements. There are other methods, which include using specific muscle groups and nerve bundles as dorsal/ventral landmarks, that would more clearly show differences in symmetry.

Reviewer #3 (Public review):

Summary:

After salamander limb amputation, the cross-section of the stump has two major axes: anterior-posterior and dorsal-ventral. Cells from all axial positions (anterior, posterior, dorsal, ventral) are necessary for regeneration, yet the molecular basis for this requirement has remained unknown. To address this gap, Yamamoto et al. took advantage of the ALM assay, in which defined positional identities can be combined on demand and their effects assessed through the outgrowth of an ectopic limb. They propose a compelling model in which dorsal and ventral cells communicate by secreting Wnt10b and Fgf2 ligands, respectively, with this interaction inducing Shh expression in posterior cells. Shh was previously shown to induce limb outgrowth in collaboration with anterior Fgf8 (PMID: 27120163). Thus, this study completes a concept in which four secreted signals from four axial positions interact for limb patterning. Notably, this work firmly places dorsal-ventral interactions upstream of anterior-posterior, which is striking for a field that has been focussed on anterior-posterior communication. The ligands identified (Wnt10b, Fgf2) are different from those implicated in dorsal-ventral patterning in the non-regenerative mouse and chick models. The results in the context of ALM/ectopic limb engineering are impressive, but the authors do not extend their experiments to assay 'normal' regeneration after amputation.

Strengths:

(1) The ALM and use of GFP grafts for lineage tracing (Figures 1-3) take full advantage of the salamander model's unique ability to outgrow patterned limbs under defined conditions. As far as I am aware, the ALM has not been combined with precise grafts that assay 2 axial positions at once, as performed in Figure 3. The number of ALMs performed in this study deserves special mention, considering the challenging surgery involved.

(2) The authors identify that posterior Shh is not expressed unless both dorsal and ventral cells are present. This echoes previous work in mouse limb development models (AER/ectoderm-mesoderm interaction), but this link between axes was not known in salamanders. The authors elegantly reconstitute dorsal-ventral communication by grafting, finding that this is sufficient to trigger Shh expression (Figure 3 - although see also the Weaknesses section.)

(3) Impressively, the authors discovered two molecules sufficient to substitute dorsal or ventral cells through electroporation into dorsal- or ventral-depleted ALMs (Figure 5). These molecules did not change the positional identity of target cells. The same group previously identified the ventral factor (Fgf2) to be a nerve-derived factor essential for regeneration. In Figure 6, the authors demonstrate that nerve-derived factors, including Fgf2, are alone sufficient to grow out ectopic limbs from a dorsal wound. Limb induction with a 3-factor cocktail without supplementing with other cells is conceptually important for regenerative engineering.

(4) The writing style and presentation of results are very clear.

Weaknesses:

(1) The expression data are the weakest part of this study.

• Despite being a central message, I found the Shh in situs unconvincing (e.g. Figure 2I, 3C, 5C), especially without sense probe controls. An additional assay would be essential to make the Shh data convincing - perhaps like in Figure 5D (qPCR?), RNA-sequencing, or a downstream target gene.

• It is not clear what the n numbers mean for the in situ data (slides analysed / number of biological samples / other?). This is crucial to understanding the reliability of the results.

• The authors do not assay where and when Wnt10b and Fgf2 are expressed beyond the bulk RNA-sequencing (which presumably contains both epidermis and mesenchyme cells). This is a shame, as understanding which cell types express these molecules, and when, would be important for understanding the mechanism.

(2) It is important to consider that the ALM is not 'regeneration', even if the authors have previously argued that ALM bumps and regenerating blastemas are equivalent (PMID: 17959163). The start- and end- points of ALM are different from regeneration, even though there are undoubtedly common principles involved. Thus, I find the word 'regeneration' in the title and last sentence of the abstract unsubstantiated unless evidence is provided that the same mechanisms (Wnt10b/Fgf2/Shh) function during normal limb regeneration.

(3) Drawing the exact boundaries of the Ant/Pos/Dor/Ven BL and grafts in the cartoon in Figure 1 (with respect to anatomical landmarks) would help to better understand the experiments in Figures 3 and 4.

(4) I find the 'positional cue' and 'positional value' terminology confusing, despite the authors' efforts. It is not clear if they refer to cell autonomous or secreted signals, and, as the authors mention, the definitions partially overlap. Lmx1b is defined as a positional value, even though it is necessary and sufficient for dorsal identity (so, isn't it positional information?). Much simpler would be to describe Wnt10b and Fgf2 as what they are: dorsally or ventrally expressed signals that substitute for dorsal or ventral tissue without inducing changes in positional information.

Overall appraisal:

This is a logical and well-executed study that creatively uses the axolotl model to advance an important framework for understanding limb patterning. The reliability of the Shh expression data is a weak point in this otherwise impressive study. The relevance of the mechanisms to normal limb regeneration is not substantiated.

Author response:

We sincerely thank the editor and all three reviewers for their constructive comments. We deeply appreciate the reviewers’ efforts in highlighting both the strengths and the weaknesses of our study. To enhance the quality and clarity of our work, we plan to address the concerns raised in the public reviews through the following actions:

(1) Improving the tone and language of the manuscript

We will revise the manuscript thoroughly, incorporating additional explanations and clarifications where necessary, and improving the tone and language to enhance readability and precision. Especially, we will pay careful attention on the terms “positional information,” “positional value,” and “positional cue,” and we plan to explain them in a historical context.

(2) Extending analysis to regular blastemas

To validate the applicability of our proposed model beyond the accessory limb model (ALM), we will examine the gene expression patterns of key signaling molecules in regular blastemas generated by limb amputation. This will allow us to test whether the mechanisms we describe are also active during normal limb regeneration.

(3) Increasing sample sizes in critical experiments

In order to ensure reproducibility and statistical reliability, we will increase the number of biological replicates in key experiments within the limitations regulated by our animal ethics approval. Additionally, we will collect data that clearly defines the dorsal/ventral axis within the structures, as far as possible. We will also revise the manuscript to pay closer attention to the anterior/posterior/dorsal/ventral axis in the existing data, ensuring that it is clearly described.

(4) Adding quantitative gene expression data

To support and reinforce our in situ hybridization results, we will include additional quantitative gene expression analyses (e.g., qRT-PCR), thereby strengthening the conclusions drawn from our expression data.

We are grateful for the reviewers’ insights and are confident that these revisions will significantly strengthen our manuscript.