Coupled local translation and degradation regulate growth cone collapse

Local translation mediates axonal responses to Semaphorin3A (Sema3A) and other guidance cues. However, only a subset of the axonal proteome is locally synthesized, while most proteins are trafficked from the soma. The reason why only specific proteins are locally synthesized is unknown. Here we show that local protein synthesis and degradation are linked events in growth cones. We find that growth cones exhibit high levels of ubiquitination and that local signaling pathways trigger the ubiquitination and degradation of RhoA, a mediator of Sema3A-induced growth cone collapse. Inhibition of RhoA degradation is sufficient to remove the protein-synthesis requirement for Sema3A-induced growth cone collapse. In addition to RhoA, we find that locally translated proteins are the main targets of the ubiquitin-proteasome system in growth cones. Thus, local protein degradation is a major feature of growth cones and creates a requirement for local translation to replenish proteins needed to maintain growth cone responses.

a montage of 3 separate images of the cell body compartment, microfluidics channels, and axon compartment. Scale bar represents 100 m.
(b) Additional controls for Figure 1a. Representative phase images of a growth cone after two sequential treatments with vehicle (top row) and of a growth cone that was first treated with vehicle and then with Sema3A (bottom row).
(c) Sema3A-induced growth cone collapse is protein synthesis dependent in microfluidic chambers. Pre-treatment of DRG axons with the protein synthesis inhibitor anisomycin (Aniso, 40 M) for 30 min before Sema3A treatment inhibits growth cone collapse *p<0.05, Student's ttest, n≥30 per condition from at least three independent experiments.
(d) The second Sema3A-induced growth cone collapse requires RhoA activity. Treatment of axons with the RhoA inhibitor Y-27632 (10 M) completely abolishes the second Sema3Ainduced growth cone collapse. **p<0.01, ***p<0.001, one-way ANOVA, n≥31 per condition from at least three independent experiments.
(e) Morphological recovery is not affected by treatment with proteasomal inhibitors or low NGF. Growth cones were treated with Sema3A for 1 h and then allowed to morphologically recover in medium containing the indicated treatments. Treatment with proteasomal inhibitors for 1 h did not significantly prevent the morphological recovery of growth cones. One-way ANOVA, n≥33 per condition from at least three independent experiments.
(g) MG132 pretreatment leads to a protein synthesis independent growth cone collapse in hippocampal cells. Anisomycin treatment prevents Sema3A-induced growth cone collapse, but not when axons are pre-treated with MG132 for 1.5 hr. *p<0.05, **p<0.01, n≥37 per condition from at least three independent experiments.

Supplementary Fig.2. NGF induces ubiquitin-dependent proteasomal degradation of growth cone proteins via Smurf1
(a) NGF increases K48-linked ubiquitin conjugates in growth cones. A 10-min NGF application induces rapid accumulation of K48-linked ubiquitin protein conjugates (K48-Ub, green). Cells were treated with MG132 (5 M) to prevent proteasomal degradation of the ubiquitin conjugates. Phalloidin, red. Scale bar represents 5 m. **p<0.01, two-tailed Student's t-test, n50 growth cones per condition from at least three independent experiments.
(b) Quantification of results in (a). Shown is the normalized K48-Ub or K63-Ub signal in growth cones. The K48-Ub and K63-Ub signals were normalized to tau-1. **p<0.01, two-tailed Student's t-test, n50 growth cones per condition from at least three independent experiments.
(c) Validation of the specificity of the anti-Smurf1 antibody used in immunofluorescence experiments. Rat DRG neurons were transduced with a lentiviral construct expressing GFP and a control (shLacZ) or a Smurf1-targeting shRNA (Smurf1 sh1) and grown for 6 DIV. Scale bar represents 20 m.
(d,e) Quantification of the changes in Smurf1 localization within growth cones in response to NGF. NGF treatment increases the levels of Smurf1 in growth cones (d), with a greater number of Smurf1 puncta along filopodia (light gray bar) and at filopodial tips (terminal Smurf1, dark gray bar) (e). Smurf1 intensity was normalized to GAP43. **p<0.01, two-tailed Student's t-test, n48 growth cones per condition from at least three independent experiments.
(f) Smurf1 is frequently found at sites of membrane protrusion. DRG neurons transduced with a lentiviral construct encoding Cherry-Smurf1 were imaged every 400 msec after application of 50 ng per ml NGF. The images are frames from a representative experiment, taken at the times indicated. In the phase-contrast images, growth cones are outlined for easier identification. A filopodia (black arrow) forms upon arrival of Cherry-Smurf1 (white arrow) to the membrane. The last image shows an overlay between the phase-contrast image after the time course and the fluorescence image from the previous time point to localize the Smurf1 puncta within the growth cone. Scale bar represents 5 m, inset scale bar represents 2 m. See Supplementary Movie 1.   Supplementary Fig.3. Smurf1 is required for NGF signaling (a) Schematic of a tripartite microfluidic chamber (top). Montage of images from DIV5 DRG neurons transduced with a lentiviral construct expressing GFP and then immunostained with an anti-GFP antibody (bottom). A growth cone can be seen in the inset. Scale bar represents 100 m, inset scale bar represents 20 m.
(b) Western blot validation of the shRNAs used to knock down Smurf1. Neurons were transduced upon plating and grown for 6 days. Endogenous Smurf1 levels were evaluated by Western blotting. Two distinct shRNAs targeting Smuf1 reduced the levels of Smurf1 expression by at least 75% compared to non-targeting control shRNA (shLacZ).
(c) Quantification of the results in (b) from three independent experiments. ***p<0.001, one-way ANOVA with Turkey's multiple comparison test.
(d) Schematic diagram of electroporation of spinal cord explants. Open book preparations of E10.5 mouse spinal cords were electroporated with plasmids expressing EGFP to visualize axons. After 2 DIV, explants were fixed and analyzed for EGFP fluorescence. We divided the distance between cell bodies and the midline into three equal parts (Zones 1-3) and counted the number of axon terminals in each of the three areas.
(e) Smurf1 and RhoA degradation are required for proper axon growth in vivo. Spinal cord explants were co-electroporated withplasmids expressing the catalytically inactive mutant Smurf1 C699A or the non-degradable mutant RhoA K6/7/51Rtogether with EGFP. The three areas where axons can extend are indicated at the top of the figure by red lines (Zones 1-3). Axons expressing Smurf1 C699A or RhoA K6/7/51R display reduced growth. Axon terminals are labeled with white arrows. Scale bar represents 50 m.
(f) Quantification of results in (e). The number of axons with terminals in Zones 1-3 is indicated as a percentage of the total number of electroporated axons. While the majority of axons reach the midline in the control electroporations (Zone 3, ~56%), only a small fraction of the axons expressing Smurf1 C699A or RhoA K6/7/51R (~16% and ~12%, respectively) reach Zone 3. p<0.001, chi-square test, n=14 distinct spinal cord explants per condition.

Supplementary Fig.4. Smurf1 is required to maintain protein synthesis-dependent growth cone collapse
(a) Knockdown of Smurf1 in DRG neurons increases basal collapse. Basal collapse of DRG neurons infected upon plating with lentiviruses expressing a nontargeting shRNA (shLacZ) or two distinct shRNAs targeting Smurf1 was measured. We found that knockdown of Smurf1 leads to a higher basal collapse. This may reflect increased levels of RhoA that accumulates in neurons kept in culture for 5-6 days to allow efficient knockdown. **p<0.01, one-way ANOVA, n≥35 per condition from at least three independent experiments.
(b) Overexpression of Smurf1 or Smurf1 dominant negative does not affect basal collapse. Basal collapse of neurons infected upon plating with lentiviral constructs expressing GST, wild type Smurf1, or the catalytically inactive mutant Smurf1 C699A. We found that overexpression of these proteins does not significantly affect the number of growth cones that exhibit a collapsed morphology under basal culturing conditions ("basal collapse"). One-way ANOVA, n≥24 per condition from at least three independent experiments.

Supplementary Fig.5. NGF degrades RhoA via Smurf1
(a) RhoA is ubiquitinated in neurons. Neurons were treated with DMSO or MG132 for 12 h and the appearance of ubiquitinated RhoA was detected by Western blotting for ubiquitin in RhoA immunoprecipitates. MG132 treatment allows for accumulation of RhoA ubiquitinated conjugates. It should be noted that ubiquitin is limiting in cells, so proteasome inhibition does not cause all proteins to become ubiquitinated. Proteasome inhibition prevents ubiquitinated proteins from being degraded, sequestering ubiquitin in proteins 3 . The size of the ubiquitinated products can be highly variable [3][4][5][6] . Anti-ubiquitin antibodies preferentially label poly-ubiquitin chains. Variability in the types of ubiquitin conjugates that are seen can come from the amount of time that the cells are treated with proteasome inhibitors. Additionally, both the degree of ubiquitination, chain length, and chain topology is different in different cell types. IP, immunoprecipitation; WB, Western blotting.
(b) Endogenous RhoA is polyubiquitinated by Smurf1. Histidine-tagged ubiquitin (His 6 -Ub) was expressed in HEK293 cells and purified using a nickel-nitriloacetic (Ni-NTA) column. Inhibition of the proteasome with LLnL (50 µM) allowed accumulation of ubiquitinated RhoA. Western blotting of the immunoprecipitate for endogenous RhoA shows a signal consistent with polyubiquitination when Smurf1 is expressed (right lane). AP, affinity purification; WB, Western blotting.
(c) Full Western blots for Fig. 4d (d) Smurf1 is required for NGF-dependent ubiquitination of RhoA. Anti-ubiquitin Western blotting of the immunoprecipitate for endogenous RhoA from PC12 cells transfected with Smurf1 (+) or control (-) siRNA. Cells were treated with 100 ng per ml NGF for 20 min after overnight incubation in serum-free media. Western blotting of the immunoprecipitate for endogenous RhoA shows that NGF treatment leads to an increase in RhoA ubiquitination only when Smurf1 is present, but not in Smurf1-knockdown cells.
(e) Smurf1 knockdown increases the levels of RhoA in growth cones. Immunostaining was used to quantify endogenous RhoA levels in growth cones of DRG neurons expressing a Smurf1 shRNA. Knockdown of Smurf1 results in a 20% increase in axonal RhoA levels. For quantification purposes, RhoA immunofluorescence was normalized to GAP43. ***p<0.001, Student's t-test, n≥33 per condition from at least three independent experiments.