Commissureless acts as a substrate adapter in a conserved Nedd4 E3 ubiquitin ligase pathway to promote axon growth across the midline

Abstract

In both vertebrates and invertebrates, commissural neurons prevent premature responsiveness to the midline repellant Slit by downregulating surface levels of its receptor Roundabout1 (Robo1). In Drosophila, Commissureless (Comm) plays a critical role in this process; however, there is conflicting data on the underlying molecular mechanism. Here, we demonstrate that the conserved PY motifs in the cytoplasmic domain of Comm are required allow the ubiquitination and lysosomal degradation of Robo1. Disruption of these motifs prevents Comm from localizing to Lamp1 positive late endosomes and to promote axon growth across the midline in vivo. In addition, we conclusively demonstrate a role for Nedd4 in midline crossing. Genetic analysis shows that nedd4 mutations result in midline crossing defects in the Drosophila embryonic nerve cord, which can be rescued by introduction of exogenous Nedd4. Biochemical evidence shows that Nedd4 incorporates into a three-member complex with Comm and Robo1 in a PY motif-dependent manner. Finally, we present genetic evidence that Nedd4 acts with Comm in the embryonic nerve cord to downregulate Robo1 levels. Taken together, these findings demonstrate that Comm promotes midline crossing in the nerve cord by facilitating Robo1 ubiquitination by Nedd4, ultimately leading to its degradation.

Keywords: D. melanogaster; Nedd4; Robo; Slit; axon guidance; commissureless; developmental biology; neuroscience; ubiquitin ligase.

Figure 1.. PY motifs are required for Comm to promote midline crossing.

(A) Comm variants used in this study including WT, and mutant forms in which the LPSY (1PY) or both the LPSY and PPCY (2PY) motifs are disrupted. (B–C) PY motifs are necessary for Comm to cause ectopic crossing in the ipsilateral FasII neuron population when driven pan-neuronally. (B) Micrographs of stage 16–17 nerve cords stained with HRP (pan-neuronal marker, magenta) and FasII (green) expressing no Comm (i), Comm WT (ii), Comm 1PY (iii), and Comm 2PY (iv) under the elav gal4 driver. (C) Quantification of ectopic FasII crossing in stage 16–17 embryos expressing wild type and PY-mutant Comm variants. Groups were compared using ANOVA with Tukey’s post-hoc test, and error bars represent mean with 95% confidence interval. (D–E) PY motifs are required for Comm to cell-autonomously promote ectopic midline crossing when driven in apterous neurons. (D) Micrographs of stage 16–17 nerve cords stained with HRP (magenta) and GFP (green), expressing GFP and no Comm (i), Comm WT (ii), Comm 1PY (iii), and Comm 2PY (iv) under the ap gal4 driver. (E) Quantification of ectopic apterous neuron crossing in stage 16–17 embryos expressing wild type and PY-mutant Comm variants. Groups were compared using ANOVA with Tukey’s post-hoc test, and error bars represent mean with 95% confidence interval. For all graphs, each data point represents one embryo. ** (p<0.01) **** (p<0.0001). Scale bars represent 20 μM.

Figure 1—figure supplement 1.. Driving high levels of Comm, including those lacking functional PY motifs, induces ectopic midline crossing in ipsilateral neurons.

(A–D) Nerve cords of embryos expressing no Comm (A), or 10 X UAS transgenes for Comm WT(B) Comm 1PY (C) and Comm 2PY (D). Embryos are stained with the pan-neuronal marker HRP (green) and an antibody against FasII (magenta), which labels a population of ipsilateral neurons. Scale bar represents 20 μM.

Figure 2.. PY motifs are required for Comm to downregulate Robo1 protein levels in vitro.

(A–C) Loss of PY motifs inhibits Robo1 degradation facilitated by Comm and increases Comm stability. (A) Western blot of S2R + cells transfected with HA-Robo1 and Comm-Myc WT, 1PY, or 2PY. (B) Fold change in Robo1 levels relative to the ‘Robo1 alone’ condition. Protein levels were calculated via densitometry and normalized to tubulin. (C) Fold change in Comm levels relative to the ‘WT Comm’ condition. Protein levels were calculated via densitometry and normalized to tubulin. For B and C, groups were compared using ANOVA with Tukey’s post hoc test. Error bars represent mean +/-SD. Each data point represents values calculated from a single experiment. * (p<0.05), ** (p<0.01), *** (p<0.001). (D–E) Comm requires PY motifs to facilitate Robo1 ubiquitination and degradation in the lysosome. (D) Immunoprecipitation of ubiquitinated Robo1 in control or chloroquine-treated S2R + cells transfected with FLAG-Ubiquitin and HA-Robo1 alone, or HA-Robo1 and WT, 1PY, or 2PY Comm-myc variants. Cells were incubated with Chloroquine for 2 hr. (E) Fold change of ubiquitinated Robo1 levels relative to the ‘Robo1 only’ condition. Ubiquitinated Robo1 levels were measured by quantifying immunoprecipitated FLAG blotting via densitometry and normalizing to IGG, lysate Robo1, and tubulin. For B and C, groups were compared using ANOVA with Tukey’s post hoc test. Error bars represent mean +/-SD. Each data point represents values calculated from a single experiment * (p<0.05), ** (p<0.01), **** (p<0.0001).

Figure 3.. PY motifs are necessary for Comm to downregulate Robo protein levels in vivo.

(A) Stage 15–16 embryos expressing no Comm or 5 X UAS Comm-myc variants under the pan-neuronal elav gal4 driver. Embryos are stained for endogenous Robo1 (i–iv), Myc (v-viii), and pan-neuronal marker HRP (ix-xii). Scale bar represents 20 μM. (B) Quantification of endogenous Robo1 levels in stage 15–16 embryos expressing Comm variants under elav gal4 driver. Robo1 levels were measured by creating a mask of the axonal scaffold, measuring 488 fluorescence within the scaffold, and dividing by scaffold area. (C) Quantification of Comm levels in stage 15–16 embryos expressing Comm variants under the elavGal4 driver. Comm levels were measured by creating a mask of the whole nerve cord, measuring Cy3 fluorescence within the nerve cord, and dividing by nerve cord area. For B and C, groups were compared using ANOVA with Tukey’s post-hoc test. ** p<0.01 **** p<0.0001. Error bars represent 95% confidence intervals around the mean. Each data point represents one embryo.

Figure 3—figure supplement 1.. Robo expression in 10XUAS Comm transgenic flies.

Stage 15–16 embryos expressing no Comm or 10 X UAS Comm-myc variants under the pan-neuronal elav gal4 driver. Embryos are stained for endogenous Robo1 (i–iv), Myc (v-viii), and pan-neuronal marker HRP (ix-xii). Scale bar represents 20 μM. (B) Quantification of endogenous Robo1 levels in stage 15–16 embryos expressing Comm variants under elav gal4 driver. Robo1 levels were measured by creating a mask of the axonal scaffold, measuring 488 fluorescence within the scaffold, and dividing by scaffold area. (C) Quantification of Comm levels in stage 15–16 embryos expressing Comm variants under the elavGal4 driver. Comm levels were measured by creating a mask of the whole nerve cord, measuring Cy3 fluorescence within the nerve cord, and dividing by nerve cord area. For B and C, groups were compared using ANOVA with Tukey’s post-hoc test. ** p<0.01 **** p<0.0001. Error bars represent 95% confidence intervals around the mean. Each data point represents one embryo.

Figure 4.. PY motifs are necessary for Comm to effectively traffic Robo1.

(A) COS-7 cells transfected with HA-Robo1 and Myc-Comm variants. Cells in which Comm and Robo1 are exclusively in puncta are outlined with dotted lines in merged images and labelled with arrowheads in images of individual color channels. Scale bar represents 15 μM. (B) Quantification of cells expressing Robo1 exclusively in intracellular puncta or in diffuse localization throughout the cell. N represents the number of individual cells counted, which were taken from seven images of cell fields chosen at random locations on the microscope slide. Distribution of exclusively punctate vs diffuse Robo1 localization in Comm-transfected conditions was compared to the Robo1 alone condition via Chi Square (** p<0.01). (C) Quantification of cells expressing Comm exclusively in intracellular puncta or in diffuse localization throughout the cell. N represents the number of individual cells counted, which were taken from seven images of cell fields chosen at random locations on the microscope slide. Distribution of exclusively punctate vs diffuse Comm localization in comm mutant conditions was compared to the WT Comm condition via Chi Square (** p<0.01). In both B and C, only cells expressing both Comm and Robo1 were counted. (D) Surface expression of Robo in nerve cords of stage 14–15 embryos. Embryos were live-dissected, left unpermeabilized, and stained with an antibody against endogenous Robo1 at 4 °C. After Robo1 staining, embryos were permeabilized and stained with HRP and Myc. Scale bar represents 20 μM.

Figure 5.. Comm requires intact PY motifs to localize appropriately in vivo.

(A) Micrographs of stage 16–17 embryos expressing CD8-GFP and WT or PY-mutant Comm-myc variants under the apgal4 driver. Embryos are stained with GFP to visualize apterous neurons, Myc (Comm), and HRP (pan-neuronal marker). Cell bodies are circled in panels showing Myc staining (iv-vi). In v, axonal Comm expression is indicated with arrowheads. Scale bar represents 20 μM. (B) Quantification of the ratio of axonal Comm to cell soma Comm in stage 17 embryos expressing WT and PY-mutant Comm variants in apterous neurons. Axonal and cell soma comm were measured within masks created of these two features, derived from images of apterous neurons in the GFP channel. Axon/cell soma ratios were compared via ANOVA with Tukey’s post-hoc test (**** p<0.0001). Error bars represent 95% confidence intervals around the mean. Each data point represents a single embryo. (C) Stage 15 embryos expressing Lamp1-GFP and WT or PY-mutant Comm myc variants under the apGal4 driver. Embryos are stained with GFP (Lamp1), Myc (Comm), and HRP (pan-neuronal marker). Cell bodies are outlined with white circles except those that are enlarged in inset images, which are outlined with yellow squares. Within inset images, Comm/Lamp1 co-positive puncta are indicated with yellow arrowheads and puncta containing Comm alone are indicated with white arrowheads. Scale bar represents 20 μM in the large image and 5 μM in the inset image. (D) Total colocalization of Comm and Lamp1. The proportion of areas of Comm expression that were also positive for Lamp1 was calculated via pearson’s coefficient within a mask created of the area of Comm expression using smoothened images from the Comm channel. (E) Colocalization of Comm and Lamp1 within cell bodies. The proportion of areas of Comm expression that were also positive for Lamp1 was calculated via Pearson’s coefficient within a mask created of cell bodies using smoothened images from the Lamp1 channel. For D and E, colocalization across conditions was compared via ANOVA with Tukey’s post hoc test (* p<0.05, ** p<0.01). Error bars represent 95% confidence intervals around the mean. Each data point represents one embryo.

Figure 5—figure supplement 1.. Robo colocalization with Comm is unaffected by mutations in Comm PY motifs.

(A) Stage 16–17 embryos expressing HA-Robo and WT or PY-mutant Comm myc variants under the apGal4 driver. Embryos are stained with HA (Robo), Myc (Comm), and HRP (pan-neuronal marker). Cell bodies are outlined with white circles except those that are enlarged in inset images, which are outlined with yellow squares. Within inset images, Robo/Comm co-positive puncta are indicated with yellow arrowheads and puncta containing Robo alone are indicated with white arrowheads. Axonal expression of Robo and Comm is indicated with hollow arrowheads. Scale bar represents 20 μM in the large image and 5 μM in the inset image. (B) Total colocalization of Robo and Comm. The proportion of areas of Robo expression that were also positive for Comm was calculated via Manders coefficient within a mask created of the area of Robo expression using smoothened images from the Robo channel, which contain both cell bodies and axons. (C) Colocalization of Comm and Lamp1 within cell bodies. The proportion of areas of Robo expression that were also positive for Comm was calculated via Manders coefficient within a mask created of cell bodies using smoothened images from the Robo channel, manually edited to remove axons. For B and C, colocalization across conditions was compared via ANOVA with Tukey’s post hoc test (* p<0.05, ** p<0.01). Error bars represent 95% confidence intervals around the mean. Each data point represents one embryo.

Figure 5—figure supplement 2.. Colocalization of Comm with the late endosomal marker Rab 7.

(A) Stage 17 embryos expressing Rab7-GFP and WT or PY-mutant Comm myc variants under the apGal4 driver. Embryos are stained with GFP (Rab7), Myc (Comm), and HRP (pan-neuronal marker). Cell bodies are outlined with white circles except those that are enlarged in inset images, which are outlined with yellow squares. Within inset images, Comm/Rab7 co-positive puncta are indicated with yellow arrowheads and puncta containing Comm alone are indicated with white arrowheads. Scale bar represents 20 μM. (B) Total colocalization of Comm and Rab7. Proportion of areas of Comm expression that were also positive for Rab7 was calculated via pearson’s coefficient within a mask created of the area of Comm expression using smoothened images from the Comm channel. (C) Colocalization of Comm and Rab7 within cell bodies. Proportion of areas of Comm expression that were also positive for Rab7 was calculated via pearson’s coefficient within a mask created of cell bodies using smoothened images from the Lamp1 channel. For B and C, colocalization across conditions was compared via ANOVA (* p<0.05, ** p<0.01). Error bars represent 95% confidence intervals around the mean. Each data point represents one embryo.

Figure 6.. Nedd4 is required for midline crossing in vivo.

(A–B) Loss of nedd4 enhances pan-neuronal crossing defects in a fra-/- background. (A) Stage 15–16 embryos with various combinations of wild-type and null fra and nedd4 alleles, stained with pan-neuronal markers BP102 (i–iv) and HRP (v). Commissures with crossing defects (either thin or completely missing) are indicated with yellow arrowheads. (B) Quantification of crossing defects in embryos within the genotypes indicated in A. Percentage of total crossing defects was compared across groups using ANOVA (** p<0.01, **** p<0.0001). N represents number of individual embryos. (C–D) Loss of nedd4 enhances EW crossing defects in a fra -/- background. (C) Stage 15–16 embryos with various combinations of wild-type and null fra and nedd4 alleles, expressing GFP under the egl gal4 driver. Embryos are stained with pan-neuronal marker HRP, and GFP to visualize eg neurons. Commissures with EW crossing defects are indicated with arrowheads. (D) Quantification of crossing defects in embryos within the genotypes indicated in (C). Percentage of crossing defects was compared across groups using ANOVA (** p<0.01, **** p<0.0001). Each data point represents an individual embryo and error bars represent 95% confidence intervals around the mean (E–F) Loss of nedd4 enhances EW crossing defects in the FraΔc background. (E) Embryos with nedd4 mutations and/or exogenous Nedd4 expression in the FraΔc background. (F) Quantification of EW crossing defects for the genotypes shown in (E). Percentage of crossing defects was compared across groups using ANOVA (** p<0.01, **** p<0.0001). Each data point represents an individual embryo and error bars represent95% confidence intervals around the mean. All scale bars represent 20 μM.

Figure 6—figure supplement 1.. Nedd4 and Smurf are expressed in the nerve cord at developmental stages when midline crossing occurs.

(A) Nerve cords of Stage 14–16 Nedd4 MiMIC embryos, in which endogenous Nedd4 protein is tagged with GFP. Nedd4 expression was visualized using an antibody against GFP and the axon scaffold was visualized with HRP. (B) In situ using an antisense probe against Smurf in Stage 14–16 embryos heterozygous (i) or homozygous (ii) for a chromosomal deletion covering the Smurf locus. Nerve cords are outlined with dotted lines. All scale bars represent 20 μM.

Figure 6—figure supplement 2.. Smurf does not appear to be required for midline crossing in vivo.

(A–B) Loss of smurf does not enhance pan-neuronal crossing defects in a fra-/- background. (A) Stage 15–16 embryos with various combinations of wild-type and null fra and smurf alleles, stained with pan-neuronal marker BP102 (green). Commissures with crossing defects (either thin or completely missing) are indicated with yellow arrowheads. (B) Quantification of crossing defects in embryos within the genotypes indicated in (A). Percentage of total crossing defects was compared across groups using ANOVA (** p<0.01, **** p<0.0001). N represents number of individual embryos. (C–D) Loss of smurf fails to enhance EW crossing defects in a fra -/- background. (C) Stage 15/16 embryos with various combinations of wt and null fra and nedd4 alleles, expressing GFP under the egl gal4 driver. Embryos are stained with pan-neuronal marker HRP, and GFP to visualize egl neurons. Commissures with EW crossing defects are indicated with arrowheads. (D) Quantification of crossing defects in embryos within the genotypes indicated in (C). Percentage of crossing defects was compared across groups using ANOVA (** p<0.01, **** p<0.0001). Each data point represents an individual embryo and error bars represent 95% confidence intervals around the mean. (E–F) Loss of smurf fails to enhance EW crossing defects in the FraΔc background. (E) Embryos with smurf mutations in FraΔc background. (F) Quantification of EW crossing defects for the genotypes shown in (E). Percentage of crossing defects was compared across groups using ANOVA (** p<0.01, **** p<0.0001). Each data point represents an individual embryo and error bars represent 95% confidence intervals around the mean. All scale bars represent 20 μM.

Figure 6—figure supplement 3.. Su(dx) does not appear to be required for midline crossing in vivo.

(A–B) Loss of Su(dx) fails to enhance pan-neuronal crossing defects in a fra -/- background. (A) Stage 15/16 embryos with various combinations of wild-type and null fra and Su(dx) alleles, stained with pan-neuronal marker BP102 (green). Commissures with crossing defects (either thin or completely missing) are indicated with yellow arrowheads. (B) Quantification of crossing defects in embryos within the genotypes indicated in A. Percentage of total crossing defects was compared across groups using ANOVA (** p<0.01, **** p<0.0001). N represents number of individual embryos. (C–D) Loss of Su(dx) fails to enhance EW crossing defects in a fra -/- background. C Stage 15/16 embryos with various combinations of wild-type and null fra and nedd4 alleles, expressing GFP under the egl gal4 driver. Embryos are stained with pan-neuronal marker HRP, and GFP to visualize egl neurons. Commissures with EW crossing defects are indicated with arrowheads. (D) Quantification of crossing defects in embryos within the genotypes indicated in (C). Percentage of crossing defects was compared across groups using ANOVA (** p<0.01, **** p<0.0001). Each data point represents an individual embryo and error bars represent 95% confidence intervals around the mean. (E–F) Loss of Su(dx) fails to enhance EW crossing defects in the FraΔc background. (E) Embryos with Su(dx) mutations in FraΔc background. (F) Quantification of EW crossing defects for the genotypes shown in E. Percentage of crossing defects was compared across groups using ANOVA (** p<0.01, **** p<0.0001). Each data point represents an individual embryo and error bars represent 95% confidence intervals around the mean. All scale bars represent 20 μM.

Figure 7.. Nedd4 interacts genetically and biochemically with slit/Robo1 pathway.

(A–C) Nedd4 is incorporated into a three-member complex with Robo1 and Comm. (A) Immunoprecipitation (i) and input (ii) of Robo1-GFP in S2R + cells co-transfected with Nedd4-HA and WT, 1PY, or 2PY Comm-myc variants. (B) Quantification of fold change in Nedd4 association with Robo1, relative to the Robo1 + Nedd4 condition, from the co-immunoprecipitation in (A). This was measured by quantifying immunoprecipitated Nedd4 via densitometry and normalizing to immunoprecipitated Robo1 and input Nedd4. Groups were compared using ANOVA with Tukey’s post-hoc test. (* p<0.05) Error bars represent mean +/-one standard deviation. (C) Schematic illustrating how Nedd4 relies on Comm PY motifs for recruitment into a ternary complex with Robo1 and Nedd4. (D–G) Exogenous Nedd4 enhances degradation of Robo1 in the presence of WT Comm in vivo. (D) Western blot of Lysates of 13-18hpf embryos expressing WT Comm-myc with or without Nedd4-HA under the elavGal4 driver. (E) Quantification of Robo1 (i) and Comm (ii) levels from the blot in (D). Protein levels were measured via densitometry and normalized to tubulin. Protein levels were compared across conditions using ANOVA with Tukey’s post-hoc test. (* p<0.05, *** p<0.001, **** p<0.0001) Error bars represent mean +/-one standard deviation. (F) Nerve cords of stage 15–16 embryos expressing WT Comm-myc with or without Nedd4-HA under the elavGal4 driver. Embryos are stained with the pan-neuronal marker HRP (magenta) and an antibody against endogenous Robo1 (green). Scale bar represents 20 μM. (G) Quantification of Robo1 levels from the conditions in (F). Robo1 levels were measured by creating a mask of the axonal scaffold, measuring 488 fluorescence within the scaffold, and dividing by scaffold area. Robo1 levels were compared across conditions using ANOVA with Tukey’s post-hoc test. (** p<0.01, **** p<0.0001) Error bars represent 95% confidence intervals around the mean.

Figure 7—figure supplement 1.. Robo, Comm, and Nedd4 can form a three-member complex in fly embryonic lysate.

Immunoprecipitation (A) and input (B) of lysates of 24hpf embryos expressing WT Comm-myc, with or without Nedd4 HA, under the pan-neural elav gal4 driver.

Figure 7—figure supplement 2.. Nedd4 enhances ectopic crossing phenotype induced by Comm overexpression.

(A) Nerve cords of stage 16–17 embryos expressing WT Comm-myc with or without Nedd4-HA, under the pan-neuronal elav gal4 driver. Embryos are stained with the pan-neuronal marker HRP. The width of a single body segment is marked with a white line. Scale bar represents 20 uM. Collapsed segments, described further in C, are indicated with yellow arrowheads. (B) Nerve cord width of stage 16–17 embryos expressing WT Comm-myc with or without Nedd4-HA, under the pan-neuronal elav gal4 driver. Nerve cord with is calculated by taking the average of the widths of the posterior eight body segments at their widest points the unit represented by a white line in (A). Nerve cord widths were binned into three different phenotypic classes and distribution of phenotypes between groups was compared using Fisher’s exact test with Freeman-Halton extension using raw counts per phenotypic class. (C–D) Nedd4 enhances collapse of nerve cord segments induced by Comm overexpression. (C) Diagram of Drosophila nerve cord axonal scaffolds from a WT embryo and one exhibiting some ectopic crossing. Nerve cord segments are indicated with brackets and negative space within the segments are indicated with asterisks. Nerve cord segments are considered collapsed when they have no negative space. (D) Percentage of collapsed segments in stage 16–17 embryos expressing WT Comm-myc with or without Nedd4-HA, under the pan-neuronal elav gal4 driver. Percentage of collapsed holes is calculated using the following formula: (number of segments lacking negative space)/ (total segments)*100. The mask of the axonal scaffold used for this analysis was created by taking micrographs of HRP-stained nerve cords, applying a smoothing filter, and generating a thresholded image from this smoothened microscope photo. Embryos were binned into the following phenotypic categories: no collapse (0% collapse), partial collapse (0<x < 100% collapse), and complete collapse (100% segments collapsed). Distribution of phenotypes between different genotypes was compared using Fisher’s exact test with Freeman-Halton extension, using the raw count of embryos within each phenotypic class. Differences were considered significant if p<0.05. **** p<0.0001, * p<0.05.

Figure 8.. A schematic illustrating our proposed model for Robo1 downregulation in pre-crossing commissural axons.

(A–B) Localization and repulsive activity of Robo1 in neurons when Comm is absent. (A) View of Robo1 trafficking down axons to the growth cone surface. In neurons not expressing Comm (such as ipsilateral or post-crossing neurons), Robo1 travels unimpeded to the growth cone surface. (B). At the growth cone surface, Robo1 induces a repulsive response to slit, preventing midline crossing. (C–D) Robo1 localization and repulsive activity in the presence of Comm. (C) Comm facilitates Robo1 ubiquitination by recruiting Nedd4 via its PY motifs. Robo1 ubiquitination drives the entire Robo1/Nedd4/Comm complex to the lysosome. (D) Due to lysosomal degradation, little Robo1 reaches the growth cone membrane. This lack of Robo1 renders axons insensitive to repulsion via slit, enabling them to cross the midline.