Cables1 links Slit/Robo and Wnt/Frizzled signaling in commissural axon guidance

  1. Nikole R. Zuñiga
  2. Alexandre Dumoulin
  3. Giuseppe Vaccaro
  4. Esther T. Stoeckli

Reviewed by

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

Log in to save this article

Abstract

During neural circuit formation, axons navigate from one intermediate target to the next, until they reach their final target. At intermediate targets, axons switch from being attracted to being repelled by changing the guidance receptors on the growth cone surface. For smooth navigation of the intermediate target and the continuation of their journey, the switch in receptor expression has to be orchestrated in a precisely timed manner. As an alternative to changes in expression, receptor function could be regulated by phosphorylation of receptors or components of signaling pathways. We identified Cables1 as a linker between floor-plate exit of commissural axons, regulated by Slit/Robo signaling, and the rostral turn of post-crossing axons, regulated by Wnt/Frizzled signaling. Cables1 localizes β-catenin, phosphorylated at tyrosine 489 by Abelson kinase, to the distal axon, which in turn is necessary for the correct navigation of post-crossing commissural axons in the developing chicken spinal cord.

Article activity feed

  1. Version published to 10.1242/dev.201671
  2. Review Commons

    Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    Manuscript number: RC-2021-01111

    Corresponding author(s): Esther Stoeckli

    [Please use this template only if the submitted manuscript should be considered by the affiliate journal as a full revision in response to the points raised by the reviewers.

    If you wish to submit a preliminary revision with a revision plan, please use our "Revision Plan" template. It is important to use the appropriate template to clearly inform the editors of your intentions.]

    1. General Statements [optional]

    Dear editors at Review Commons

    Thanks for your patience. We have finally carried out a full revision of our originally submitted manuscript summarizing our findings on the role of Cables1 in axon guidance.

    In our study, we provide in vitro and in vivo evidence for a role of Cables1 as a linker between axon guidance signaling pathways. Commissural axons in the developing spinal cord leave their intermediate target, the floor plate, due to a switch from attraction to repulsion mediated by the specific trafficking of Robo1 receptors to the growth cone surface. The presence of Robo1 on growth cones after contact with the floor plate allows them to respond to Slit, the negative guidance cue associated with the floor plate. After leaving the floor plate on the contralateral side, growth cones respond to a Wnt gradient along the antero-posterior axis. The responsiveness to Wnt of post- but not pre-crossing axons is regulated by the trafficking of Fzd3 receptors to the growth cone membrane of post-crossing axons (Alther et al., 2016), but also by the specific phosphorylation of β-Catenin at tyrosine Y489 by Abl kinase. Cables1 mediates this phosphorylation by transferring Abl kinase from the C-terminus of Robo1 to β-Catenin (this study).

    The revised version of the manuscript contains additional experiments in vitro, in vivo and ex vivo combined with live imaging to further support our conclusion about the role of Cables1 as a linker between Robo/Slit and Wnt signaling.

    It took as longer than expected to carry out these new experiments, as Nikole Zuñiga, the first author of the paper, left the lab after her PhD defense to take up a job in industry. Unfortunately for the study, but fortunately for Giuseppe Vaccaro, he also got a job soon after taking over the project. Therefore, the revision was delayed again. We hope that the additional experiments will solve the issues that were raised by the reviewers. We thank them for their contributions and suggestions.

    Best regards

    Esther Stoeckli

    2. Point-by-point description of the revisions

    This section is mandatory. *Please insert a point-by-point reply describing the revisions that were already carried out and included in the transferred manuscript. *

    Point to point response to reviewers’ comments

    Reviewer #1 (Evidence, reproducibility and clarity (Required)): In this work by Zuñiga et al. the authors study the role of the adaptor protein Cables1 on the guidance of post-comissural spinal cord neurons. They hypothesize that commissural axons need Cables1 to leave the floor plate and turn to ascend to the brain. They propose that during this process, Cables1 acts as a linker of two key axon guidance pathways, Slit and Wnt. Cables1 would localize β-catenin phosphorylated at tyrosine 489 to the distal axon and this would be necessary for the correct turning and navigation of post-crossing commissural axons. Although the work may be potentially interesting, there are major issues that authors need to address in order to state their claims:

    -Fig. 2. To visualize the axonal phenotype after downregulation of Cables1 the authors use DiI labelling. This difficults the interpretation of the results as both electroporated and non- electroporated axons are labelled. Since the authors have a Math1::tdTomatoF reporter construct (as in Fig. 3), it would be desirable to use this construct Math1::tdTomatoF in combination with the dsCables1 plasmid to better visualize the phenotype. Alternatively and less preferred, GFP signal should be also shown in Fig.2B experiments.

    We respectfully disagree. Most likely, the reviewer thinks about a defined nerve that has a particular trajectory and then when labelled with a fluorescent marker, deviations from this pathway, or defasciculated growth can be easily visualized. However, in the spinal cord, the dI1 axons run ventrally more like a ‘curtain’. Therefore, the aberrant behavior of axons is difficult to see. We therefore, opted for the alternative suggestion and added the GFP images to visualize clearly that the axons labelled with DiI are from the injected area. We also would like to add that we are extremely careful in injecting DiI only to the dI1 population of commissural axons to avoid mixing populations with different trajectories. As the analysis is done by a person blind to the experimental condition, we are convinced that our way of analyzing the phenotype is valid. An approach that has been successfully used by many groups for decades now. Please also keep in mind that we are always comparing groups of embryos with each other. Furthermore, having axons traced by DiI which were not targeted by dsRNA electroporation would not increase but rather decrease the likelihood of aberrant behavior. Therefore, we are convinced that our method of quantification is valid.

    However, we have added new experiments using live-imaging which also demonstrate that many axons in embryos electroporated with dsCables1 fail to turn properly at the floor-plate exit site (see Movie 2). These experiments provide additional evidence for the validity of our results.

    -Fig. 2B and Supp.Fig.3. Comparable DiI labellings should be shown in the different conditions. The three examples shown in this panel despite different amount of DiI-labeled axons making it difficult to compare them.

    We have exchanged the image of the control-treated embryo in Figure 2 to have more comparable DiI injection sites. However, as we detail in our Material & Method section, the quantification was done in such a way that the number of axons does not matter. We rephrased this paragraph to make this point more clear (lines 630ff). Please also refer to the GFP-expressing control sample shown in Figure 6A.

    We counted a DiI injection site as showing floor-plate stalling when at least 50% of the fibers entering the floor plate failed to reach the exit site. Similarly, ‘No turn’ means that at least 50% of the axons at the exit site failed to turn rostrally. Because, these two phenotypes are not independent of each other (100% stalling prevents the analysis of the turning phenotype), we only did a statistical analysis for the DiI injection sites with correctly turning axons. We also would like to point out that we hardly had injection sites where it was difficult to decide whether the 50% threshold was reached or not.

    -Fig. 2D. An scheme depicting the different phenotypes: "normal", "FP stalling" and "no turn" would help to understand the results. They can use schemes similar to those shown in Fig. 2K Parra et al. 2010.

    We have added a scheme outlining the different phenotypes, as suggested to Figure 2A.

    -Fig. 3A. The open-book drawing is confusing. It seems that they are analyzing open-book preparations in this experiment when this is not the case.

    Now Figure 4: We have changed the schematic explaining our experimental design. We wanted to illustrate that we only took the dorsal-most part of the spinal cord, dissected from open-book preparations of the spinal cord, as explants to avoid the inclusion of other cell types.

    -Fig. 3B. Authors claim that Cables1 is not required in pre-crossing axons as dsCables electroporation does not affect axonal growth of DiI neurons taken at HH22. However, to be sure that Cables1 mRNA levels are downregulated in pre-crossing axons, relative levels of Cables1 mRNA and/or protein should be also determined at HH22 not only at HH25.

    We have clarified the quantification of downregulation efficiency. The qPCR data are taken from HH23, that is one day after electroporation. The Western blot data show differences in protein levels at HH25, that is 2 days after electroporation. In both cases, the downregulation efficiency is about 50%. This means that we got rid of all Cables1 mRNA, as we successfully transfected 50% of the cells in the targeted area (52.5% in n=4 embryos). The cell numbers were determined by counting the ratio of GFP-positive cells from transfected spinal cords in a single cell suspension.

    -Fig. 4. The incapacity of Slit to induce axonal retraction in dsCables1 neurons is used to conclude that Cables1 is required to respond to Slit. However, downregulation of Cables1 by itself is even more effective inhibiting axonal growth than Slit treatment. Upon this strong effect as a background, it is difficult to assay slit response. Authors should point this observation in the manuscript.

    We disagree. There is no significant difference between the neurite lengths between the control neurons in the presence of Slit and the neurons lacking Cables1 (dsCables1), p=023, or the neurons lacking Cables1 in the presence of Slit (dsCables1 and Slit), p>0.9999. As seen in the images and also from the measured neurite lengths, axons still show growth and further reduction would have been possible. We would also like to point out that the conclusion from this experiment is that Cables1 is required for the response of axons to either Slit or Wnt.

    To support our claims, we have added another experiment addressing the need for Cables1 for post-crossing axons’ responsiveness to Slit by downregulation of Robo receptors (Figure 10). These experiments confirmed that Slit/Robo signaling is required for the effect of Cables1 on post-crossing axons, in line with our final conclusion that Slit binding to Robo triggers internalization and Cables then transfers Abl from the C-term of Robo to β-Catenin. This results in phosphorylation of β-Catenin at tyrosine489 (β-Catenin pY489) and responsiveness to Wnt5a.

    -Fig. 5B. In this Figure they do not differentiate between FP stalling or no turn phenotypes. A quantification taking into account the different phenotypes as shown in Fig.2D should be included.

    Done, as suggested. This is Figure 6C in the revised manuscript.

    -Fig. 6D,E. As postulated in the manuscript and based on the Rhee, et al. paper, the β-catenin phosphorylation is triggered by Abl quinase upon Slit-Robo signaling. How the authors explain then that isolated cells with axons growing on a plate recapitulate specific distal phosphorilation of β-catenin at Y489 in the absence of Slit signaling? This experiment shows that postcrossing axons contain more phosphorylated β-catenin as an intrinsic characteristic rather than as a consecuence of contact with floor plate signals. Authors should try a similar experiment but exposing the neurons (or explants) to Slit. Also, why β-catenin phosphorylation was not measured at the growth cone?

    In Figure 6D and E (now Figure 7D,E), we compare pre- and post-crossing axons. Post-crossing axons do have ‘a memory’ of their contact with the floor plate, as this contact has changed the localization of Robo receptors to the surface (Philipp et al., 2012; Alther et al., 2016). Floor-plate contact also initiates differences in gene expression (e.g. Hhip expression in a Shh-and Glypican-dependent manner; Wilson and Stoeckli, 2013). The difference in Robo localization has also been described by others (Pignata et al., Cell Rep 29(2019)347).

    In fact, the distal localization of pY-489 β-Catenin is in perfect agreement with our results: The localization of Robo1 on the distal portion of the axon is in line with published data from our own lab but also from the Castellani and the Tessier-Lavigne lab. Our results suggest that Cables is recruited to Abl bound to the C-term of Robo. Cables transfers Abl then to β-Catenin which is phosphorylated by Abl. Thus pY-489 β-Catenin would be localized predominantly where Robo is localized, i.e. the distal axon. In support of these results, experiments added to the revised version of the manuscript indicate that the response to Slit is required for the increase in β-Catenin pY489 (Figure 10B).

    -Fig. 7. CAG::hrGFP electroporation is not specific for dl1 neurons. This experiment should be performed with Math1::tdTomatoF in order to analyze β-cat pY489 with or without dsCables1 specifically in dl1 neurons. Also, why GFP staining at the growth cones in Fig.7B is not visible in the axon?

    As indicated in our schematic drawing (Figure 7A) we only cultured explants from the dorsal-most part of open-book preparations of spinal cords, making sure that our cultures are not mixtures with more ventral populations of neurons. We opted for CAG::hrGFP because Math1 is a weak promoter and the expression of GFP was very difficult to see after dissociating cells and culturing them in vitro. We used a GFP version that is not farnesylated to avoid interference with axonal staining of pY-489 β-Catenin. Therefore, GFP is not visible in axons with the imaging conditions used.

    -Fig. 8. This experiment does not distinguish whether phosphorylated β-Cat is necessary for the correct navigation of post-crossing commissural axons (as it is claimed in the abstract) or it is also required for midline crossing. As it has been previously shown, correct navigation of post-crossing commisusal axons is a Wnt5 dependent process. As dsCables1 abrogates Wnt5a responsiveness (Fig.4B,C), does the phosphomimetic β-catenin Y489E construc rescue the Wnt5a response in dsCables1 electroporated neurons? Moreover, can the phosphomimetic β-catenin Y489E construc rescue the Slit response in dsCables1 electroporated neurons? Testing these effects on explants as in Fig. 4B,C but including phosphomimetic β-catenin, will help to understand to what extend phosphorylation of β-catenin is important for crossing, turning or both processes.

    Yes, the phosphomimetic Y489E version of β-Catenin reduces the percentage of DiI injections sites with aberrant axonal navigation to control levels (Figure 9 in the revised manuscript). In contrast, a mutant version of β-Catenin that cannot be phosphorylated, β-CateninY489F, cannot rescue the axon guidance phenotype seen in the absence of Cables1.

    -How do the authors envision the mechanism of Cables1/β-catenin mediated crossing and turning? A working model summarizing their hypothesis would help the reader to understand the results.

    **Minor points:** -Homogeneize the term "scale bars" or "bars" in the Figure Legends.

    done

    -Scale bar of insets in Fig.1C is missing.

    The scale bar is now added, we apologize for the mistake.

    -The antisense control for Cables probe should be shown at HH-22/24. Otherwise is not possible to distinguish whether they do not detect signal because is a negative control or because Cables1 is not expressed at HH25.

    We have added the image of an adjacent section hybridized with the sense probe for HH25, in addition to HH22 to clarify that Cables expression is higher during floorplate crossing, exiting and turning rostrally but then levels decrease when post-crossing axons have initiated their growth along the rostro-caudal axis.

    -Figure legend for Fig. 2D is missing

    corrected

    -Fig. 8B right panel is contaminated with growthing axons coming from the below DiI injection. Please replace the picture.

    We have changed the outline of this figure.

    -The quantification of the different phenotypes "FP stalling", "no turn" should be better explained in the Mat and Met section. The sentence " more than 50% of the axons...." is not clear. Was this measured by eye? Otherwise, please indicate the soIware used to measure.

    Yes, as mentioned above, it was hardly ever a close call. It is very easy for a person blind to the experimental condition to go through the DiI injection sites of an open-book preparation and to assess whether 50% or more of the axons that enter the floorplate reach the exit site, or not. Similarly, it is very easy to do the same for the turning behavior. We have changed the text describing this method of quantification to be more explicit (lines 630ff).

    -Provide the quantification of the WB in Supplementary Fig. 2B normalising to Gapdh.

    Added as Supplementary Figure 2C.

    Reviewer #1 (Significance (Required)): Previous results have demonstrated that Slit-induced modulation of adhesion is mediated by cables that links Robo-bound Abl kinase to N-cadherin-bound betacat (Rhee et al., 2007). Here the authors propose that a similar mechanism is operating in commissural neurons leave the midline after crossing and turn immediately after. The role of Cables in the process has not been previously addressed. Thus, after proper addressing of my main concerns, I consider this paper may advance in our knowlege of how growing axons navigate intermediate targets.

    We appreciate this positive evaluation of our study and hope that the additional experiments and more detailed explanations have helped clarify open questions of the reviewer.

    Reviewer #2 (Evidence, reproducibility and clarity (Required)): In this paper entitled "Cables1 links Slit/Robo and Wnt/Frizzled signaling in commissural axon guidance", authors aim to the find the mechanisms the coordinate the floor plate exit and the rostral turning of commissural axons. During development thousands of axons have to navigate long distances to reach their targets and build functional circuits. To facilitate their journey, their paths is divided into small portions by intermediated targets. The most studied intermediated target is the floorplate (FP) at the midline of the ventral spinal cord. Glia cells forming the FP express plethora of guidance cues. Commissural neurons, which have their cells bodies located in the dorsal part of the spinal cord, send their axons towards the FP. These axons are first attracted by the FP which facilitate their entry within the FP. However, they switch this attractive response into a repulsive one in order to exit the FP and turn rostrally to connect their brain targets. In order to ensure that this process will go smoothly, commissural axons have to adapt the composition of their receptors and the signaling pathways to switch from attractiveness to repulsion. So far, many processes have been involved such as the alternative splicing of receptors (Robo3; Chen et al Neuron 2008), protease regulation of receptor expression (Nawabi et al Genes & Dev 2010), trafficking of receptors, or their interaction profiles (Delloye et al Nat Neuro 2015). However, it is still not clear how 2 events (here exit from the FP and rostral turning) are linked. Authors propose an original mechanism that involved the adaptor protein Cables 1. This protein has been shown to link the Robo/Slit1 signaling to Cadherins. Cables regulates the repulsive response to Slit and adhesion by the phosphoryla4on of b-Catenin by the kinase Abelson (Rhee et al Nat Cell Biol 2007). The story developed here is very original and interesting: Cables would link the exit of FP (mediated by Robo/Slit signaling) and the rostral turning of the commissural axons (controlled by the Wnt/Fzd pathway. Below I'm proposing some experiments as many questions raised upon reading this beautiful work. The experiments are sound and could be reproducible. The statistic analysis looks fine.

    We thank the reviewer for this positive assessment of our study.

    I would suggest some experiments to strengthen the whole work: •Authors might want to consider to perform some biochemistry experiments to show that Cables is able to interact with Robo1 and Fzd3: are these proteins in the same molecular complex? They could do 2 experiments: one in vitro by transfecting a cell line (such as HEK293 or cos cells) with plasmids coding for Robo1, Cables and Fzd3 or at least Cables and Fzd3 (as for Robo1/Cables they could refers to Rhee et al 2007). Another one would be in vivo: extracting proteins from the pre-crossing stage, the FP and post crossing stage; immunoprecipitation of Cables1 and see whether Robo1 and/or Fzd are pull down with Cables 1.

    We decided not to do these experiments, as we felt that this would go beyond the current study. In fact, for our effects it is not necessary that Cables interacts physically with Robo or Fzd3. The important aspect is that Abl bound to Robo is transferred by Cables to β-Catenin. A direct interaction with Fzd3 is not necessary.

    From the pictures it seems that most of the axons are stalling in the FP when embryos are electroporated with dsCables1. It would be nice to show more examples of axons that are able to exit the FP but have turning problems. Given the data, as it is presented, it seems that Cables regulates more the FP exit (and therefore, as it was shown in Rhee et al, the responsiveness to Robo/Slit signaling).

    The major phenotype is ‘no turn’. However, as we describe in response to reviewer 1 and in the manuscript, the ‘floorplate stalling’ and the ‘no turn’ phenotypes are not independent of each other. At DiI injection sites, where almost all axons stall in the floorplate, the turning cannot be assessed. Thus, the ‘no turn’ phenotype tends to be underestimated in conditions where floorplate crossing is also affected, as is the case after silencing Cables1.

    In the same line, in Fig 4, Authors need to add a condition using dsCables and ds Fzd in order to see the effect of Cables on axon turning (response to Wnt). As it is this figure supports the role of Cables on FP exit but it's hard to make the link with commissural axon responsiveness to Wnt.

    We belief that experiment 4 clearly demonstrates the absence of the Wnt responsiveness, as axons fail to grow in response to Wnt when they extend from neurons transfected with dsCables1 (Figure 4C). Because dsCables1 alone already abolishes all responsiveness to Wnt, the removal of Fzd at the same time would not change anything.

    Authors aim to show that Cables is a linker between 2 events: maybe it should be nice to try to disconnect these events. One way would be (if technically possible) to modulated expression of Cables at different stages. What would happen if Cables was down regulated upon FP crossing? Would axons still be able to respond to Wnt? The question I'm wondering about is whether the responsiveness to Slit and Wnt is acquired at the same time or whether axons should become sensitive to Slit and this event will prime them to respond to Slit. In order to address the following experiment could be performed: explants from HH22-HH23 embryos, could be treated with medium containing Slit first and then Wnt or vice et versa and perform some collapse assay.

    Unfortunately, the experiment as proposed by the reviewer is not possible. The axons take on average 5.5 hours to cross the floorplate (entry – exit; Dumoulin et al., 2021). Most importantly, the protein that is already made before axons are at the exit site, could not be removed. Therefore, it is not possible to prevent the production of Cables only after axons have crossed the midline. As shown in Figure 1, Cables1 mRNA is present at HH22, that is when axons have reached and are about to enter the floorplate. We also do not belief that the in vitro experiment suggested by the reviewer would work. We would have to wash cell intensively to remove Slit added to the medium. This would interefere with their potential to grow in response to Wnt immediately after addition. However, we added experiments where we looked at the effect of Wnt after removal of Robo (Figure 10). These experiments demonstrate that responsiveness to Wnt can only be established when axons can respond to Slit, i.e. when Robo is activated.

    In Fig3 I was wondering whether post crossing axons were growing less because of the change in the regulation of adhesion: Rhee et al shows that Cables is able to modulate adhesion through N-cadherin. It would be interesting to perform immunostaining on these explant cultures to assess any change in adhesion molecules.

    We have not found any changes in the expression levels of Contactin-2 (Axonin-1), NrCAM, or most importantly β1-Integrin, as our cultures grow on laminin.

    It is not clear whether Robo1 and/or Fzd induces the phosphorylation of b-catenin: is the Robo1/Slit binding induce the phosphorylation of b-cat and this event will prime the axons to respond to Wnt/Fzd? Or Wnt/Fzd is also able to control b-cat phosphorylation?

    We have added an experiment, where we remove Robo1 from commissural neurons and compare pY489 β-Catenin levels (Figure 10). Furthermore, we demonstrate that in the absence of Robo1, Wnt has no stimulatory effect on axons (Figure 10C,D). These experiments supports our conclusion that Cables1 transfers Abl kinase from the C-terminal part of Robo to β-Catenin, which gets phosphorylated and thus is ready to act in the Wnt signaling pathway.

    The staining with the antibody needs to be detailed: as it is reported this antibody recognizes "a domain of Cables1 that is 90% identical to the corresponding region of Cables2": it seems that the Cables protein enrichment in the floor plate (around the central canal) is Cables 2 as its mRNA expression matches this profile of expression. The one expressed in the crossing axons might be Cables 1: one way to verify this, is to perform the staining on sections from embryos electroporated with dsCables 1. This is a very important control of the antibody to reinforce this point of the paper.

    We belief that the staining of the cells around the central canal could be due to endfeet of precursors spanning the neural tube from the apical to the basal side. All cells seem to express some Cables1 (Figure 1B,C). As we did not find any effect of Cables2 on commissural axon navigation and we do not use antibodies to functionally interfere with Cables1 function, we did not do this experiment, as the antibody is not able to distinguish the two proteins. Most likely, there is little, if any, Cables2 expressed in the spinal cord during this time window. We still did some functional analyses but found no effect on axon guidance (Supplementary Figure 3).

    In Figures 3-4: why not performing some co culture of spinal cord explants with COS or HEK 293 cells expressing Slit1 or Wnt? This experiment will provide a clear-cut response to see the role of Cables in axon guidance. As there it is, Fig3 shows a role of Cables in axon growth but not guidance.

    We respectfully disagree that in vitro experiment would help to show guidance versus growth. Guidance can only be shown in vivo. This is what we do. Our in vitro results are only included to address specific responsiveness of axons or expression changes in total β-Catenin or pY489 β-Catenin. But all our conclusions about the role of Cables in axon guidance are demonstrated in vivo. Experiments using co-cultures of axons with COS or HEK cells would be impossible to control for timing and amount of Slit or Wnt release.

    In Figure 6: my understanding of axon guidance is that every guidance decision happens at the level of the growth cone. However, it seems that in post crossing stage, there is a strong decrease of b-cat and phosphor- b cat within the growth cone compared to the precrossing stage. If beta cat is the effector of Cables to link Robo/Slit and Wnt/Fzd signaling I would expect it to be localized at the growth cone. I think authors should discuss this point. Regarding the normalization, it would be better to counterstaing the neurons with actin and use the measure of its fluorescence to normalize phopho-beta cat.

    There must be a misunderstanding. We do not demonstrate or claim that there is a decrease in β-Catenin or pY489 β-Catenin between pre- and post-crossing axons. We only demonstrate that the distribution of pY489 β-Catenin is clustered in distal post- but not pre-crossing axons. This change in localization of pY489 β-Catenin is supporting our model that Cables1 transfers Abl kinase to β-Catenin and phosphorylates it and prepare it for signaling in the Wnt pathway. And, as demonstrated pY489 β-Catenin and β-Catenin are in the growth cone. However, for quantification we concentrated on the axon, as the difference in growth cone morphology would have complicated the quantification.

    **Minor points:** •In figure 2: it seems that there are few axons labelled with DiI in the dsCables1 condition (Fig2B): it would be the choice of the picture or maybe the downregulation of Cables 1 interfere with the survival of dl1 neurons (even though in supp 1C it is shown that most of the populations are still there with no difference with the control side) or maybe some axons are delayed to reach to FP on time: the picture is focused on the FP: are there any axons still growing in the side of the open book preparation? Again, the picture that could be misleading.

    We have exchanged the images for alternatives with a better matched number of DiI-labelled axons. There is indeed no evidence for cell death, as axons are still there at normal numbers when we analyze open-book preparations a day later than usually. The difference in the number of axons labelled by DiI is only due to the variability in the amount of DiI injected per injection site.

    In Fig1 legends, maybe Authors wanted to write "At HH18 dl1 commissural neurons start to extend their axons in the ventral spinal cord"?

    No, what we mean is, as shown in Figure 1A, that axons emerge from the cell body at this time. They reach the ventral spinal cord by HH21 and the floor plate by HH22.

    I would also remove the yellow shadow on the Fig1A: it could be misleading as at first glance the reader might wonder whether there are 2 populations of dl1 neurons.

    We have done as suggested to make the image clearer.

    Reviewer #2 (Significance (Required)): It is still not clear how axons cross the midline. So far, many processes have been involved such as the alternative splicing of receptors (Robo3; Chen et al Neuron 2008), protease regulation of receptor expression (Nawabi et al Genes & Dev 2010), trafficking of receptors, or their interaction profiles (Delloye et al Nat Neuro 2015). However, it is still not clear how 2 events (here exit from the FP and rostral turning) are linked. Authors propose an original mechanism that involved the adaptor protein Cables 1. This protein has been shown to link the Robo/Slit1 signaling to Cadherins. Cables regulates the repulsive response to Slit and adhesion by the phosphorylation of b-Catenin by the kinase Abelson (Rhee et al Nat Cell Biol 2007). The audience that will be interested in this work is the neurodevelopment filed, axon regeneration field and overall people interested in neuronal circuit formation and function. My field of expertise is molecular and cellular neuroscience applied to axon guidance (crossing the FP) in mice models, axon regeneration and circuit formation.

    We are happy to learn about the positive assessment of our work by a specialist.

    Reviewer #3 (Evidence, reproducibility and clarity (Required)): In their manuscript, Zuniga and Stoeckli characterize the role of Cables in commissural axon guidance in the developing chick spinal cord. Based on a combination of in vitro outgrowth assays and in vivo dye-tracing experiments, the authors propose that Cables participates in both normal repulsive responses to Slit and attractive responses to Wnt5. Using combinations of low-does knock down of cables/robo and or B-catenin, the author suggest an in vivo link between these pathways. Using IF with phospho-specific antibodies to B-catenin, the authors suggest that there is elevated P-Bcatenin in the post-crossing segments of distal axons. While potentially interesting, the present study falls short of adequately supporting the major claims. In addition, there are several instances where experiments lack appropriate controls.

    **Specific Comments** The conclusions reached by the authors are over-stated given the experiments performed. For example, the authors describe 'silencing' cables throughout the paper; however, the knock down that they achieve is approximately 50%. Indeed, it is quite surprising that such strong effects on growth/guidance can be achieved with a two-fold depletion of the gene product. Nevertheless, the rescue experiments provide nice evidence that dsRNA for Cables is causing a phenotype. This partial knockdown precludes strong conclusions, like for Figure 3, where they state that 'Cables is not required for pre-crossing.' The language needs to be tempered.

    We rephrased the paragraph where we describe the effect of Cables 1 and the efficiency of downregulation to stress that the parameters that we use for electroporation result in around 50% of the cells successfully transfected (lines 154 – 162, and legend of Supplementary Figure 2). Therefore, to find mRNA levels and protein levels reduced to about half indicates that our method is extremely efficient and removes the targeted mRNA and the protein almost completely. We need to point out here that we always analyze the temporal expression pattern to electroporated embryos before the protein of interest has accumulated, as in ovo RNAi obviously does not remove protein but only prevents translation and therefore the synthesis of new protein. As proteins can be extremely stable compared to the time line of embryonic development, we inject and electroporate dsRNA before we find expression of mRNA.

    Figure 4: the authors use bath application of Slit and Wnt to test effects of cables on Slit and Wnt responses. The observed effect sizes are very small and a single assay of this type does not allow such strong conclusions like 'loss of Cables prevents responsiveness.' Again, it is difficult to imagine that 50% reduction would completely prevent responses, raising questions about the suitability of this assay for measuring responsiveness- perhaps growth cone collapse would give more convincing results.

    As mentioned above, we are almost completely eliminating the targeted protein in the transfected neurons. For the explants, we only looked at the neurons expressing td-Tomato driven by the Math1 promoter. Thus, these neurons were transfected. Obviously, we cannot be sure that 100% of our cells took up the plasmid and the dsRNA, but the chances are very high that this is the case based on the ration between plasmid and dsRNA.

    Figure 5: The authors should more clearly document the effects they are seeing in these manipulations. As written, all we know is that there are 'significant effects on axon guidance.' What are these effects? Do they see the predicted differences between robo/cables and Bcatenin/cables phenotypes? e.g re-crossing defects in the case of robo and anterior turning defects in the case of B-catenin?

    We have added the analysis of the detailed axon guidance problems seen in the absence of Robo1, Cables1, βCatenin, or combinations, now Figure 6C. Indeed, we find that the phenotype ‘no turn’ is more prevalent in the condition with loss of both Cables and βCatenin. However, as mentioned above in response to a question raised by Reviewer 2, the two phenotypes are not independent of each other. Stalling in the floor plate of the majority of axons prevents the analysis of the turning phenotype. That is why we only use the ‘normal’ DiI injection sites for the statistical analysis.

    Also related to Figure 5: The authors do not validate the dsRNA knockdown of either Robo or B catenin. It is unclear what the interpretation or expectation of the triple knock down condition is.

    We have used the same ESTs to produce dsRNA derived from Robo and βCatenin in our previous publications (Alther et al., Development 143(2016)994; Avilés and Stoeckli Dev Neurobiol 76(2016)190). Therefore, we only repeated the functional experiments to verify reproducibility of the effect but we did not quantify the efficiency of downregulation in detail again.

    Figure 6: For this reviewer images showing enhanced P-Catenin in post-crossing distal axons is not convincing. The differences are not obvious by eye and the quantification suggests an ~30% increase. In contrast a nearly 4-fold increase is reported in Figure 7 for this same measurement. This raises concerns about the reproducibility of this 'phenotype.'

    Staining intensities are subject to batch-to-batch variability. Therefore, the experiments shown in Figure 7 (Figure 6 in the original manuscript) cannot be directly compared to the levels in Figure 8 (previously Figure 7). However, within the experiments, we carefully normalized data. We do not make any claims about absolute staining intensities.

    Also related to Figure 6: No validation of antibody specificity is provided or described.

    Again, please keep in mind that we do not make any claims about absolute values. All are results are based on stainings with the same antibody and comparison between different areas of the same axons. Therefore, the specificity of the antibody is important but not a fundamental aspect of our results.

    Figure 8: As for figure 5, phenotypic documentation is incomplete. In addition, no controls are shown to assure that the different mutant forms of B-catenin are comparably expressed, nor is there an unmutated wild-type control. The authors state that expression of these constructs alone has no effect on normal guidance; however, the supplemental data 6B would seem to indicate that both forms lead to increases abnormal phenotypes.

    There is an increase in the number of injection sites with aberrant axon guidance, however, this was not significant. We cannot exclude the possibility that premature expression, or overexpression of βCatenin pY489E or βCatenin pY489F does interfere with the endogenous βCatenin pY489. We still decided to keep these experiments in the revised version of the manuscript as they support our conclusion that Cables1 is required for axonal responsiveness to Slit and Wnts, and that this effect is mediated by phosphorylation of βCatenin at Y489. We are aware that this experiment in isolation is not sufficient.

    Reviewer #3 (Significance (Required)): The work builds on in vitro observa4ons from Rhee, 2007 about links between Robo signaling and Cables func4on. If adequately demonstrated, integra4on and coordina4on of Robo and Wnt axon guidance pathways is quite significant.

    We thank the reviewer for this positive assessment.

  3. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #3

    Evidence, reproducibility and clarity

    In their manuscript, Zuniga and Stoeckli characterize the role of Cables in commissural axon guidance in the developing chick spinal cord. Based on a combination of in vitro outgrowth assays and in vivo dye-tracing experiments, the authors propose that Cables participates in both normal repulsive responses to Slit and attractive responses to Wnt5. Using combinations of low-does knock down of cables/robo and or B-catenin, the author suggest an in vivo link between these pathways. Using IF with phospho-specific antibodies to B-catenin, the authors suggest that there is elevated P-Bcatenin in the post-crossing segments of distal axons. While potentially interesting, the present study falls short of adequately supporting the major claims. In addition, there are several instances where experiments lack appropriate controls.

    Specific Comments

    The conclusions reached by the authors are over-stated given the experiments performed. For example, the authors describe 'silencing' cables throughout the paper; however, the knock down that they achieve is approximately 50%. Indeed, it is quite surprising that such strong effects on growth/guidance can be achieved with a two-fold depletion of the gene product. Nevertheless, the rescue experiments provide nice evidence that dsRNA for Cables is causing a phenotype.

    This partial knockdown precludes strong conclusions, like for Figure 3, where they state that 'Cables is not required for pre-crossing.' The language needs to be tempered.

    Figure 4: the authors use bath application of Slit and Wnt to test effects of cables on Slit and Wnt responses. The observed effect sizes are very small and a single assay of this type does not allow such strong conclusions like 'loss of Cables prevents responsiveness.' Again, it is difficult to imagine that 50% reduction would completely prevent responses, raising questions about the suitability of this assay for measuring responsiveness- perhaps growth cone collapse would give more convincing results.

    Figure 5: The authors should more clearly document the effects they are seeing in these manipulations. As written, all we know is that there are 'significant effects on axon guidance.' What are these effects? Do they see the predicted differences between robo/cables and Bcatenin/cables phenotypes? e.g re-crossing defects in the case of robo and anterior turning defects in the case of B-catenin?

    Also related to Figure 5:

    The authors do not validate the dsRNA knockdown of either Robo or B catenin. It is unclear what the interpretation or expectation of the triple knock down condition is.

    Figure 6: For this reviewer images showing enhanced P-Catenin in post-crossing distal axons is not convincing. The differences are not obvious by eye and the quantification suggests an ~30% increase. In contrast a nearly 4-fold increase is reported in Figure 7 for this same measurement. This raises concerns about the reproducibility of this 'phenotype.' Also related to Figure 6:

    No validation of antibody specificity is provided or described.

    Figure 8: As for figure 5, phenotypic documentation is incomplete. In addition, no controls are shown to assure that the different mutant forms of B-catenin are comparably expressed, nor is there an unmutated wild-type control. The authors state that expression of these constructs alone has no effect on normal guidance; however, the supplemental data 6B would seem to indicate that both forms lead to increases abnormal phenotypes.

    Significance

    The work builds on in vitro observations from Rhee, 2007 about links between Robo signaling and Cables function. If adequately demonstrated, integration and coordination of Robo and Wnt axon guidance pathways is quite significant.

  4. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    In this paper entitled "Cables1 links Slit/Robo and Wnt/Frizzled signaling in commissural axon guidance", authors aim to the find the mechanisms the coordinate the floor plate exit and the rostral turning of commissural axons. During development thousands of axons have to navigate long distances to reach their targets and build functional circuits. To facilitate their journey, their paths is divided into small portions by intermediated targets. The most studied intermediated target is the floorplate (FP) at the midline of the ventral spinal cord. Glia cells forming the FP express plethora of guidance cues. Commissural neurons, which have their cells bodies located in the dorsal part of the spinal cord, send their axons towards the FP. These axons are first attracted by the FP which facilitate their entry within the FP. However, they switch this attractive response into a repulsive one in order to exit the FP and turn rostrally to connect their brain targets.

    In order to ensure that this process will go smoothly, commissural axons have to adapt the composition of their receptors and the signaling pathways to switch from attractiveness to repulsion. So far, many processes have been involved such as the alternative splicing of receptors (Robo3; Chen et al Neuron 2008), protease regulation of receptor expression (Nawabi et al Genes & Dev 2010), trafficking of receptors, or their interaction profiles (Delloye et al Nat Neuro 2015). However, it is still not clear how 2 events (here exit from the FP and rostral turning) are linked.

    Authors propose an original mechanism that involved the adaptor protein Cables 1. This protein has been shown to link the Robo/Slit1 signaling to Cadherins. Cables regulates the repulsive response to Slit and adhesion by the phosphorylation of b-Catenin by the kinase Abelson (Rhee et al Nat Cell Biol 2007). The story developed here is very original and interesting: Cables would link the exit of FP (mediated by Robo/Slit signaling) and the rostral turning of the commissural axons (controlled by the Wnt/Fzd pathway. Below I'm proposing some experiments as many questions raised upon reading this beautiful work. The experiments are sound and could be reproducible. The statistic analysis looks fine.

    I would suggest some experiments to strengthen the whole work:

    •Authors might want to consider to perform some biochemistry experiments to show that Cables is able to interact with Robo1 and Fzd3: are these proteins in the same molecular complex? They could do 2 experiments: one in vitro by transfecting a cell line (such as HEK293 or cos cells) with plasmids coding for Robo1, Cables and Fzd3 or at least Cables and Fzd3 (as for Robo1/Cables they could refers to Rhee et al 2007). Another one would be in vivo: extracting proteins from the pre-crossing stage, the FP and post crossing stage; immunoprecipitation of Cables1 and see whether Robo1 and/or Fzd are pull down with Cables 1.

    •From the pictures it seems that most of the axons are stalling in the FP when embryos are electroporated with dsCables1. It would be nice to show more examples of axons that are able to exit the FP but have turning problems. Given the data, as it is presented, it seems that Cables regulates more the FP exit (and therefore, as it was shown in Rhee et al, the responsiveness to Robo/Slit signaling). In the same line, in Fig 4, Authors need to add a condition using dsCables and ds Fzd in order to see the effect of Cables on axon turning (response to Wnt). As it is this figure supports the role of Cables on FP exit but it's hard to make the link with commissural axon responsiveness to Wnt.

    •Authors aim to show that Cables is a linker between 2 events: maybe it should be nice to try to disconnect these events. One way would be (if technically possible) to modulated expression of Cables at different stages. What would happen if Cables was down regulated upon FP crossing? Would axons still be able to respond to Wnt? The question I'm wondering about is whether the responsiveness to Slit and Wnt is acquired at the same time or whether axons should become sensitive to Slit and this event will prime them to respond to Slit. In order to address the following experiment could be performed: explants from HH22-HH23 embryos, could be treated with medium containing Slit first and then Wnt or vice et versa and perform some collapse assay.

    •In Fig3 I was wondering whether post crossing axons were growing less because of the change in the regulation of adhesion: Rhee et al shows that Cables is able to modulate adhesion through N-cadherin. It would be interesting to perform immunostaining on these explant cultures to assess any change in adhesion molecules.

    •It is not clear whether Robo1 and/or Fzd induces the phosphorylation of b-catenin: is the Robo1/Slit binding induce the phosphorylation of b-cat and this event will prime the axons to respond to Wnt/Fzd? Or Wnt/Fzd is also able to control b-cat phosphorylation?

    •The staining with the antibody needs to be detailed: as it is reported this antibody recognizes "a domain of Cables1 that is 90% identical to the corresponding region of Cables2": it seems that the Cables protein enrichment in the floor plate (around the central canal) is Cables 2 as its mRNA expression matches this profile of expression. The one expressed in the crossing axons might be Cables 1: one way to verify this, is to perform the staining on sections from embryos electroporated with dsCables 1. This is a very important control of the antibody to reinforce this point of the paper.

    •In Figures 3-4: why not performing some co culture of spinal cord explants with COS or HEK 293 cells expressing Slit1 or Wnt? This experiment will provide a clear-cut response to see the role of Cables in axon guidance. As there it is, Fig3 shows a role of Cables in axon growth but not guidance.

    •In Figure 6: my understanding of axon guidance is that every guidance decision happens at the level of the growth cone. However, it seems that in post crossing stage, there is a strong decrease of b-cat and phosphor- b cat within the growth cone compared to the precrossing stage. If beta cat is the effector of Cables to link Robo/Slit and Wnt/Fzd signaling I would expect it to be localized at the growth cone. I think authors should discuss this point. Regarding the normalization, it would be better to counterstaing the neurons with actin and use the measure of its fluorescence to normalize phopho-beta cat.

    Minor points:

    •In figure 2: it seems that there are few axons labelled with DiI in the dsCables1 condition (Fig2B): it would be the choice of the picture or maybe the downregulation of Cables 1 interfere with the survival of dl1 neurons (even though in supp 1C it is shown that most of the populations are still there with no difference with the control side) or maybe some axons are delayed to reach to FP on time: the picture is focused on the FP: are there any axons still growing in the side of the open book preparation? Again, the picture that could be misleading.

    •In Fig1 legends, maybe Authors wanted to write "At HH18 dl1 commissural neurons start to extend their axons in the ventral spinal cord"?

    •I would also remove the yellow shadow on the Fig1A: it could be misleading as at first glance the reader might wonder whether there are 2 populations of dl1 neurons.

    Significance

    It is still not clear how axons cross the midline. So far, many processes have been involved such as the alternative splicing of receptors (Robo3; Chen et al Neuron 2008), protease regulation of receptor expression (Nawabi et al Genes & Dev 2010), trafficking of receptors, or their interaction profiles (Delloye et al Nat Neuro 2015). However, it is still not clear how 2 events (here exit from the FP and rostral turning) are linked.

    Authors propose an original mechanism that involved the adaptor protein Cables 1. This protein has been shown to link the Robo/Slit1 signaling to Cadherins. Cables regulates the repulsive response to Slit and adhesion by the phosphorylation of b-Catenin by the kinase Abelson (Rhee et al Nat Cell Biol 2007). The audience that will be interested in this work is the neurodevelopment filed, axon regeneration field and overall people interested in neuronal circuit formation and function.

    My field of expertise is molecular and cellular neuroscience applied to axon guidance (crossing the FP) in mice models, axon regeneration and circuit formation.

  5. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    In this work by Zuñiga et al. the authors study the role of the adaptor protein Cables1 on the guidance of post-comissural spinal cord neurons. They hypothetize that commissural axons need Cables1 to leave the floor plate and turn to ascend to the brain. They propose that during this process, Cables1 acts as a linker of two key axon guidance pathways, Slit and Wnt. Cables1 would localize β-catenin phosphorylated at tyrosine 489 to the distal axon and this would be necessary for the correct turning and navigation of post-crossing commissural axons. Although the work may be potentially interesting, there are major issues that authors need to address in order to state their claims:

    -Fig. 2. To visualize the axonal phenotype after downregulation of Cables1 the authors use DiI labelling. This difficults the interpretation of the results as both electroporated and non-electroporated axons are labelled. Since the authors have a Math1::tdTomatoF reporter construct (as in Fig. 3), it would be desirable to use this construct Math1::tdTomatoF in combination with the dsCables1 plasmid to better visualize the phenotype. Alternatively and less preferred, GFP signal should be also shown in Fig.2B experiments.

    -Fig. 2B and Supp.Fig.3. Comparable DiI labellings should be shown in the different conditions. The three examples shown in this panel despite different amount of DiI-labeled axons making it difficult to compare them.

    -Fig. 2D. An scheme depicting the different phenotypes: "normal", "FP stalling" and "no turn" would help to understand the results. They can use schemes similar to those shown in Fig. 2K Parra et al. 2010.

    -Fig. 3A. The open-book drawing is confusing. It seems that they are analyzing open-book preparations in this experiment when this is not the case.

    -Fig. 3B. Authors claim that Cables1 is not required in pre-crossing axons as dsCables electroporation does not affect axonal growth of DiI neurons taken at HH22. However, to be sure that Cables1 mRNA levels are downregulated in pre-crossing axons, relative levels of Cables1 mRNA and/or protein should be also determined at HH22 not only at HH25.

    -Fig. 4. The incapacity of Slit to induce axonal retraction in dsCables1 neurons is used to conclude that Cables1 is required to respond to Slit. However, downregulation of Cables1 by itself is even more effective inhibiting axonal growth than Slit treatment. Upon this strong effect as a background, it is difficult to assay slit response. Authors should point this observation in the manuscript.

    -Fig. 5B. In this Figure they do not differentiate between FP stalling or no turn phenotypes. A quantification taking into account the different phenotypes as shown in Fig.2D should be included.

    -Fig. 6D,E. As postulated in the manuscript and based on the Rhee, et al. paper, the β-catenin phosphorylation is triggered by Abl quinase upon Slit-Robo signaling. How the authors explain then that isolated cells with axons growing on a plate recapitulate specific distal phosphorilation of β-catenin at Y489 in the absence of Slit signaling? This experiment shows that postcrossing axons contain more phosphorylated β-catenin as an intrinsic characteristic rather than as a consecuence of contact with floor plate signals. Authors should try a similar experiment but exposing the neurons (or explants) to Slit. Also, why β-catenin phosphorylation was not measured at the growth cone?

    -Fig. 7. CAG::hrGFP electroporation is not specific for dl1 neurons. This experiment should be performed with Math1::tdTomatoF in order to analyze β-cat pY489 with or without dsCables1 specifically in dl1 neurons. Also, why GFP staining at the growth cones in Fig.7B is not visible in the axon?

    -Fig. 8. This experiment does not distinguish whether phosphorylated β-Cat is necessary for the correct navigation of post-crossing commissural axons (as it is claimed in the abstract) or it is also required for midline crossing. As it has been previously shown, correct navigation of post-crossing commisusal axons is a Wnt5 dependent process. As dsCables1 abrogates Wnt5a responsiveness (Fig. 4B,C), does the phosphomimetic β-catenin Y489E construc rescue the Wnt5a response in dsCables1 electroporated neurons? Moreover, can the phosphomimetic β-catenin Y489E construc rescue the Slit response in dsCables1 electroporated neurons? Testing these effects on explants as in Fig. 4B,C but including phosphomimetic β-catenin, will help to understand to what extend phosphorylation of β-catenin is important for crossing, turning or both processes.

    -How do the authors envision the mechanism of Cables1/β-catenin mediated crossing and turning? A working model summarizing their hypothesis would help the reader to understand the results.

    Minor points:

    -Homogeneize the term "scale bars" or "bars" in the Figure Legends.

    -Scale bar of insets in Fig.1C is missing.

    -The antisense control for Cables probe should be shown at HH-22/24. Otherwise is not possible to distinguish whether they do not detect signal because is a negative control or because Cables1 is not expressed at HH25.

    -Figure legend for Fig. 2D is missing

    -Fig. 8B right panel is contaminated with growthing axons coming from the below DiI injection. Please replace the picture.

    -The quantification of the different phenotypes "FP stalling", "no turn" should be better explained in the Mat and Met section. The sentence " more than 50% of the axons...." is not clear. Was this measured by eye? Otherwise, please indicate the software used to measure.

    -Provide the quantification of the WB in Supplementary Fig. 2B normalising to Gapdh.

    Significance

    Previous results have demonstrated that Slit-induced modulation of adhesion is mediated by cables that links Robo-bound Abl kinase to N-cadherin-bound betacat (Rhee et al., 2007). Here the authors propose that a similar mechanism is operating in commissural neurons leave the midline after crossing and turn immediately after. The role of Cables in the process has not been previously addressed. Thus, after proper addressing of my main concerns, I consider this paper may advance in our knowleged of how growing axons navigate intermediate targets.

  6. Version published to 10.1101/2021.10.19.464945 on bioRxiv