Cytoneme delivery of Sonic Hedgehog from ligand-producing cells requires Myosin 10 and a Dispatched-BOC/CDON co-receptor complex

Essential revisions:

While the reviewers find the data to be overall convincing and supporting the author's conclusions, they have raised some points that the authors should address in a revision.

1) The authors show that SHH has a dramatic stimulatory effect on cytoneme occurrence. Is this effect due to SHH signaling through the canonical Patched-Smoothened pathway? This could be tested using a synthetic Smo antagonist such as vismodegib (not cyclopamine). Along the same lines, does the Smoothened activator, SAG1, cause a similar increase in cytoneme number?

We quantified cytoneme occurrence rates in GFP (control) or SHH-expressing NIH3T3 cells following treatment with SAG and Vismodegib. These results are presented in Figure 1—figure supplement 1G. SAG treatment did not significantly increase cytoneme occurrence rates in GFP (white bars) or SHH (gray bars) expressing cells. We tested whether SMO loss would impact cytoneme occurrence by generating SMO-null NIH3T3 cells using CRISPR/Cas9 technology. Smo^-/- NIH3T3 cells showed increased cytoneme occurrence in response to SHH expression, albeit to a slightly lesser extent than what was observed in wild type NIH3T3 cells. Thus, SMO signaling is not required for cytoneme induction, but may enhance cytoneme occurrence rates under SHH-stimulated conditions. As requested, we also examined the effect of Vismodegib, and found that cytoneme dynamics were altered in both the absence and presence of SHH. Vismodegib-treated cells showed elevated baseline occurrence rates and blunted SHH-stimulated cytoneme occurrence, resulting in a modest ligand-induced occurrence rate change in drug treated cells. Nevertheless, given the ability of SHH to increase cytoneme occurrence in a statistically significant manner in Smo^-/- cells, we are confident that canonical PTCH-SMO signaling is not required for the SHH effects.

2) The neural tube phenotype of MYO10^-/- embryos is very subtle (if it exists at all). A good comparison is the phenotype of Disp^-/- embryos (see Figure 8 in PMID: 12421714), where the SHH ligand fails to exit its site of synthesis in the notochord. In Disp^-/- embryos the floor plate (marked by FoxA2) fails to be specified and so fails to secrete SHH. In contrast, in MYO10^-/- embryos, floor plate specification seems fine, showing normal SHH transport. Also the embryo sections seem to be from slightly different stages, making it difficult to determine if the slight change in the nkx.2 and olig2 domains is due to a decrease in SHH responsive or just due to differences in timing. For example, embryos in B appear to be for an earlier stage (narrower tube) compared to embryos in A. It is critical in these embryos to show (1) antibody stain for SHH to show that its spread is compromised and (2) in situ hybridization for the direct target gene Gli1 to show reduction in Hh target gene induction. Unlike the progenitor domains, these are the most direct readouts of SHH transport in the tube and the downstream effect on Shh signal reception.

We agree with the reviewers that shifts in progenitor domains in the E10.5 neural tubes shown in the initial submission were difficult to appreciate. To address this concern, we performed several additional crosses to examine stage-matched embryos at E9.5, rather than E10.5. All sections shown were taken from equivalent cardiac level regions of the embryos.

As suggested, we examined the SHH activity domain surrounding the floor plate and performed in situ hybridization to track Gli1 expression. We also include new neural tube sections showing shifted Olig2 and Pax6-marked progenitor domains. As discussed in the text, the severity of the Myo10^m1J/m1J phenotype is variable, likely due to compensation by other factors. So, we examined only Myo10^m1J/m1J embryos showing overt exencephaly to control against phenotypic variability.

We attempted to stain for SHH using available antibodies, but could not obtain quality results due to pronounced signal to noise issues. So, to address the question about SHH spread in the neural tube, we obtained the SHH::GFP allele originally published by the McMahon lab (B6.129X1 (Cg)-Shh^tm6Amc/J), which tags endogenously-expressed bioactive SHH with GFP. We crossed the SHH::GFP allele into the Myo10^m1J/+ background, and analyzed the SHH-GFP signal in E9.5 Myo10^m1J/+ and Myo10^m1J/m1J embryos. Analysis of SHH^GFP/+; Myo10 ^m1J/m1J embryos with exencephaly revealed the domain of SHH-GFP signal to be compressed as compared to neural tubes of Shh^GFP/+; Myo ^m1J/+ controls (Figure 5A-C and quantified in D using 3-5 cardiac region sections from 7 individual animals). In addition, Myo10 ^m1J/m1J embryos consistently showed defective initiation of notochord regression, indicative of reduced GLI activity (Park et al., 2000). We speculate near-normal floor plate induction in Myo10 ^m1J/m1J embryos may result from sustained close proximity of the non-regressed notochord to the neural tube, which may allow for SHH to be directly transferred/delivered through cytoneme-independent mechanisms. We briefly discuss this possibility in the Discussion.

As requested, we performed in situ hybridization experiments for the direct SHH target Gli1 in wild type, Myo10 ^m1J/+ and Myo10 ^m1J/m1J E9.5 embryos. Myo10 ^m1J/m1J animals showed clear reduction in Gli1 expression. E9.5 Myo10 ^m1J/m1J embryos also showed compressed Olig2 progenitor domains, consistent with reduced SHH target gene induction.

We think the inclusion of these additional in vivo experiments strengthens our conclusion that Myo10 contributes to SHH deployment in vivo, and thank the reviewers for these helpful suggestions.