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

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.


Fig. S1. In the developing spinal cord, Cables2 is expressed at very low levels, if at all
qRT-PCR analysis showed high expression levels of Cables1 isoformX1 in comparison to Cables1 isoformX2 and isoformX3, as well as Cables2 at stages HH22 and HH25.All mRNA levels were normalized to Cables1 isoformX1 at HH22 (A).If at all, Cables2 mRNA was found at very low levels throughout the developing neural tube during the time window, when dI1 commissural axons cross the floor plate and turn rostral along the contralateral floor-plate border (compare sections hybridized with anti-sense probe to the section hybridized with the sense probe (last panel labeled S)).In contrast to what we observed for Cables1 mRNA (Figure 1), Cables2 mRNA was not upregulated in dI1 commissural neurons during the time when their axons cross the midline (B).Scale bar: 50 µm.Effective downregulation of Cables mRNA was demonstrated with qRT-PCR for isoform X1 (A) with 4 pools of embryos sacrificed at HH23 (A).*p=0.03,paired t-test.Western blotting of proteins isolated from HH25 spinal cords confirmed effective downregulation of Cables1 in three independent pools of embryos electroporated with dsCables1 at HH17/18 (B).Levels were reduced by 50-60%.With the parameters used to silence Cables1, we successfully electroporated on average 53% (range 45-60%) of the cells in the targeted area of the neural tube.Therefore, the observed reduction in Cables1 protein indicate a more or less complete removal of Cables1 from the electroporated cells.

Development • Supplementary information
To exclude a slower growth rate as the reason for the failure of axons to turn into the longitudinal axis in the absence of Cables1, we analyzed dI1 axonal navigation at the floor plate in open-book preparations of spinal cords dissected from embryos at HH29/30, that is 1.5 days older than those shown in Figure 2. The fact that axons still stalled at the floor-plate exit site and failed to turn into the longitudinal axis confirmed that axon guidance defects could not be explained by a slower growth rate, but had to be due to a failure to respond to guidance cues provided by the floor plate (A).(B) When compared to control-treated embryos, aberrant axon guidance was found at 68.4±17.9% of the DiI injection sites in embryos electroporated with dsCables1 (n=56 injection sites in N=5 embryos), compared to control-injected embryos (GFP plasmid only), where aberrant axonal trajectories were seen only at 25.4±7.7% of the injection sites (* p=0.0338; n=74 injection sites in 7 control-injected embryos; unpaired t-test).In contrast to GFP-expressing control embryos, embryos electroporated with dsCables showed an increase in the number of DiI injection sites with axons either stalling in the floor plate, or not turning into the longitudinal axis at the floor-plate exit site (C).Scale bar: 50 µm.Immunostaining of HH25 spinal cord sections with antibodies against Lhx2 (marker for dI1 neurons), Pax2 (interneurons), Islet-1 (motoneurons), and Hnf3β (floor-plate cells) did not reveal any differences in neuronal differentiation between control-treated, GFP-expressing controls, and experimental embryos electroporated with dsRNA derived from Cables1 at HH18 (D).Scale Bar: 100 µm.Using the regular concentrations of dsRNA derived from Robo1, β-Catenin (300 ng/µl), or Cables1 (500 ng/µl) to silence target genes effectively interfered with commissural axon guidance, as demonstrated earlier (Philipp et al., 2012;Alther et al., 2016;Avilés and Stoeckli, 2016).Knockdown of Robo1 was done by injection and electroporation of dsRobo1 at HH17-18, that is at a time that is not resulting in a maximal effect, as most Robo1 protein is made before that stage and stored in vesicles before being inserted into the growth cone surface by specific trafficking (Alther et al., 2016).However, we wanted to use the same protocol for all groups.Axonal trajectories were aberrant at 51.1±7.8% of the DiI sites in dsRobo1-treated embryos (n=143, N=14), at 55.5±7.0% of the DiI injection sites after silencing Cables1 (n=123 injection sites in N=11 embryos), and at 82.3±4.9% of the injection sites after silencing β-Catenin (n=91, N=8).Embryos injected with the plasmid encoding GFP were used as controls (same group as the one shown in Figure 6).Pathfinding was affected only at 26.2±4.8% of the injection sites in control-treated embryos (n=125 injection sites in N=12 embryos).*p=0.0207,**p=0.0094,****p<0.0001ANOVA with Tukey's multiple-comparisons test.(B) Looking at the individual phenotypes indicated that axons were both stalling more often and failed to turn at the exit site in experimental compared to control embryos.To verify specificity of our approach and staining of either β-Catenin or β-Catenin pY489 we used a construct expressing a short-hairpin directed against β-Catenin followed by an IRES and bluefluorescent protein (EBFP).After electroporation at E3, embryos were sacrificed two days later, at HH26, and dI1 neurons were isolated and kept in vitro for 2 days before fixation and staining.Panel A demonstrates that dI1 neurons (green, identified by their expression of Robo3) expressed β-Catenin (magenta; arrowhead).However, those cells that were efficiently targeted by electroporation (indicated by expression of EBFP; arrow) did not exhibit any β-Catenin staining.Similarly, using the monoclonal antibody specific for β-Catenin pY489 (magenta) revealed staining in Robo3-positive dI1 neurons (green) only, if they were not efficiently targeted by the plasmid (arrowhead; compare magenta staining in panel B with blue staining).Those that were efficiently targeted (arrows) did not show any β-Catenin pY489 staining.Bar: 50 µm.In cultured post-crossing commissural axons, β-Catenin pY489 was found at higher levels in the distal axon (A; see also Figures 7 and 8).Contactin2 was used to stain axons and growth cones of dI1 neurons (B).β-Catenin pY489 is localized in the central part of the growth cone (A'), but not found in filopodia  Hoechst staining shows nuclei.

Fig. S2 .
Fig. S2.Downregulation of Cables1 by in ovo RNAi effectively reduces levels of Cables1 mRNA and protein.

Fig. S4 .
Fig. S4.The axon guidance defects observed after silencing Cables1 are not explained by a reduction in neurite growth speed or due to defects in patterning.

Fig. S6 .
Fig. S6.A phosphorylated form of β-Catenin, β-Catenin pY489, accumulates in distal post-crossing axons.Transverse sections of spinal cords from HH26 embryos (A, B, D ,E) or dissected neurons (C, F) were stained for total β-Catenin (A, B, C) or β-Catenin pY489 (D, E, F).With an antibody recognizing all forms of β-Catenin, pre-and post-crossing axons were stained (A, B).In contrast, an antibody specific for β-Catenin that is phosphorylated at Y489 revealed higher levels of -β-Catenin pY489 on post-crossing axons (arrowhead), very low levels or no β-Catenin pY489 was found on pre-crossing axons (arrows) (D, E).Commissural axons and axons from dorsal root ganglia (DRG) neurons are visualized with an anti-Contactin 2 (Axonin1) antibody (green, A, D).Staining of cultures of dissociated neurons dissected from embryos sacrificed at HH26 demonstrates the accumulation of β-Catenin pY489 in the distal axon (F), whereas levels of total β-Catenin are more homogenous along the axon (C).Robo3, a marker for dI1 commissural axons is distributed equally along the axon (C, F).Scale bar: 50 µm in A,B,D,E; 20 µm in C,F.

Fig
Fig. S8.β-Catenin pY489 localizes predominantly to the distal axon, the transition zone to the growth cone and in the central part of the growth cone.

Movie 1 .
mRuby3-mCables1 localizes to dI1 growth cones at the floor-plate exit site.Mouse Cables1 (mCables1) was fused to mRuby3 (shown in magenta) and electroporated in ovo into commissural neurons together with a Math1::EGFP-F plasmid to specifically label dI1 neurons (shown in green).This video shows a dI1 growth cone exiting the floor plate and turning rostrally (white arrowhead) in an intact cultured spinal cord using live imaging (see Figue 3A).In the lower right panel a heat map of pseudo-colored mRuby3-mCables together with the edge of the growth cone traced as a white line are shown.A clear signal of mRuby3-mCables could be seen in this growth cone during the exit of the floor plate as well as the rostral turn (white arrowhead).One stack was taken every 10 min for 90 min.Hi, high; Lo, low.Rostral is up.Development: doi:10.1242/dev.201671:Supplementary information Development • Supplementary information Movie 2. Twenty-four-hour time-lapse recording of dI1 axons crossing the floor plate in cultured intact spinal cords.dI1 neurons were visualized by in ovo electroporation of the Math1::tdTomato-F plasmid (shown in magenta).In the control condition, dI1 axons turned rostrally in a well-organized manner after exiting the floor plate.However, knockdown of Cables1 induced aberrant phenotypes at the floor-plate exit site with axons turning caudally (yellow arrow) or having problems to extend rostrally (yellow arrowhead).These aberrant phenotypes clearly induced the formation of a disorganized ventral axon bundle as shown by all commissural axons expressing EGFP-F (green) compared to the tightly organized control funiculus (white arrows).R, rostral; C, caudal.Development: doi:10.1242/dev.201671:Supplementary information Development • Supplementary information Development: doi:10.1242/dev.201671:Supplementary information