Slit and Receptor Tyrosine Phosphatase 69D Confer Spatial Specificity to Axon Branching via Dscam1

Abstract

Axonal branching contributes substantially to neuronal circuit complexity. Studies in Drosophila have shown that loss of Dscam1 receptor diversity can fully block axon branching in mechanosensory neurons. Here we report that cell-autonomous loss of the receptor tyrosine phosphatase 69D (RPTP69D) and loss of midline-localized Slit inhibit formation of specific axon collaterals through modulation of Dscam1 activity. Genetic and biochemical data support a model in which direct binding of Slit to Dscam1 enhances the interaction of Dscam1 with RPTP69D, stimulating Dscam1 dephosphorylation. Single-growth-cone imaging reveals that Slit/RPTP69D are not required for general branch initiation but instead promote the extension of specific axon collaterals. Hence, although regulation of intrinsic Dscam1-Dscam1 isoform interactions is essential for formation of all mechanosensory-axon branches, the local ligand-induced alterations of Dscam1 phosphorylation in distinct growth-cone compartments enable the spatial specificity of axon collateral formation.

Figure 1. RPTP69D and Dscam1 play opposite roles in midline-directed collateral formation

(A) Axonal branching of ms-neurons leads to a increase in network complexity by allowing a neuron to connect to different postsynaptic targets. (B) Stereotyped branching pattern of Drosophila ms-neurons. (C-I) Representative pSC neurons dye-filled with carbocyanine dyes. (C) WT pSC ms-neurons displaying the stereotyped branching pattern. (D) Dscam1 LOF phenotypes pSC neurons using MARCM lead to the formation of a “clump” of entangled processes. (E) Dscam1 GOF phenotypes resulting from a loss of isoform diversity. (F, G) RPTP69D LOF partially phenocopies Dscam GOF null clones (G) and RPTP69D strong hypomorph combinations (F). For allele designations and additional quantifications, see Figures S1K, T. (H, H’) RPTP69D¹/RPTP69D¹⁰ and RPTP69D¹/RPTP69D²⁰ mutants display a characteristic absence of the anterior midline-crossing collateral. (I, I’, I”) Heterozygosity of Dscam1 in RPTP69D mutants suppresses RPTP69D branching defects. Dscam1²¹/+; RPTP69D¹/RPTP69D¹⁰ (I), Dscam1²¹/+; RPTP69D¹/RPTP69D²⁰ (I’) and Dscam1³⁹/+; RPTP69D¹/RPTP69D²⁰ (I”). (J) Quantification of of ms-neurons defects (Chi-square test, **** P ≤ 0.0001). (K) Dscam1 heterozygosity in RPTP69D hypomorphic background also improves adult survival. For all figures numbers between parentheses represent the numbers of neurons scored. Normal targets area of anterior midline-crossing collaterals are indicated by blue circles. Dotted yellow line indicates the midline. Scale bars are 50µm unless specified otherwise. See also Figures S1 and S2.

Figure 2. Dscam1 is an in vitro substrate for RPTP69D

(A) Design of the chimeric Met-Dscam1 receptor. (B, C) HGF-mediated activation of Met-Dscam1 leads to an increase of its Y-phosphorylation in S2 or BG3C2 cells. (B) Met-Dscam1 expressed in S2 cells was immunoprecipitated and phosphorylation assessed by semiquantitative Western blot (WB) analysis. All measurements normalized to baseline phosphorylation (Summary of 6 experiments) (C) HGF mediated activation of Met-Dscam1 is stronger and lasts longer in BG3C2 cells than in S2 cells (n=6 for 10 min, n=4 for 20, 60, 90 and 120 min, n=5 for 30 min, n=3 for 40 min). (D) Met-Dscam1 IP from BG3C2 cells before and 30 minutes after HGF addition in the presence or absence RPTP69D shows that RPTP69D KD leads to increased baseline phosphorylation of Met-Dscam1. Top: Representative WB of one experiment. Bottom: Quantification of multiple experiments. (n=6; paired t-test: * P ≤ 0.05.) (E) RNAi tests show that RPTP69D is the only neuronal RPTP affecting Met-Dscam1 baseline phosphorylation (see also Figure S1). Top: Summary of multiple experiments for each dsRNA (n=7 for RPTP69D and Lar, n=2 for RPTP 4E, 10D and 52F). ANOVA/Dunnett: *** P ≤ 0.001. Bottom: Knockdown efficiency assessed by quantitative RT-PCR (n=6 for PTP69D, n=3 for Lar, RPTP10D, 4E and 52F. ANOVA/Dunnett: ** P ≤ 0.01; *** P ≤ 0.001). (F) Endogenous Dscam1 binds to substrate trapping mutants in BG3C2 cells. Top: Mutations introduced in the WPD loop of domain 1 (DA1), or 2 DA2 or both (DA12). Bottom: Co-IP of Dscam1 from S2 cells transiently expressing HA-tagged RPTP69D. Only DA1 and DA2 mutants are able to co-IP Dscam1 (G) Mutational analysis of RPTP69D phosphatase domains. Dscam1 IPs from S2 cells expressing V5-tagged RPTP69D WT and Y>F mutant proteins. Y-phosphorylation status evaluated by semiquantitative WB analysis. Top: WB of one representative experiment. Bottom: Summary of multiple experiments (n=6 for WT, n=4 for DA1 and DA2, n=2 for DA12; ANOVA/Dunnett: ** P ≤ 0.01; *** P ≤ 0.001). Error bars: SEM. See also Figure S3.

Figure 3. Expression of a RPTP69D substrate-trap mutant phenocopies Dscam1 LOF

(A-F’) Representative dye-fills of pSC neurons (B) Defects in Dscam³⁹/Dscam⁴⁷ reveal “clump” formation (arrowheads) and longitudinal axon growth defects (asterisks). (C-C’) RNAi knock-down of Dscam1 results in “clumps” (arrowheads) and longitudinal axon growth defects (asterisks), higher magnification in (C’). (D-F’) Expression of RPTP69DA12 mutant in ms-neurons phenocopies Dscam1 LOF defects. Expression of UAS-RPTP69D-WT (D) and UAS-RPTP69D-DA12 (E-F’) in ms-neurons using pnr-Gal4. Higher magnification (F’).

Figure 4. RPTP69D targets several tyrosine residues of Dscam1 CT

(A) Position of 15 conserved tyrosines in Dscam1 IC (in invertebrates). (B-C) Five Y>F mutations lead to significant decrease of baseline Y-phosphorylation. (B) IP of WT and Met-Dscam1 Y>F mutants before and after HGF addition. Semiquantitative WB analysis (C) Summary of multiple experiments. Mean values compared to the baseline phosphorylation of WT Met-Dscam1. Baseline phosphorylation is reduced in all five Y>F mutants, but significant reduction in response to HGF is only observed for mutants in SH2-binding sites (Y1857 and Y1890). (n=5 for YF1707 and YF1911; n=4 for YF1857; n=3 for YF1890 and YF1981; ANOVA/Dunnett to compare the phosphorylation state of Met-Dscam1 in presence and absence of HGF: * or ° P ≤ 0.05; ** P ≤ 0.01; *** P ≤ 0.001). (D) Sequence alignment of the 5 candidate RPTP69D and known PTP tyrosine substrate residues (Selner et al. 2014). Substrate tyrosines (magenta). Acidic or large hydrophobic residues (green). (E, F) RPTP69D interacts with at least two tyrosines of Dscam1 IC. (E) An overexpression-based phosphatase assay to examine RPTP69D-mediated dephosphorylation of Dscam1. Top: Representative WB of the IP of WT Dscam1-HA from S2 cells in the presence and absence of WT RPTP69D-V5. Bottom: Quantification of phosphorylation of mutant Dscam1-HA proteins. Co-expression of RPTP69D significantly reduces the Y-phosphorylation of WT Dscam1-HA as well as Y1707, Y1857 and Y1911 Dscam1-HA mutants. Y-phosphorylation is abolished in YF1890 and YF1981 mutants. Bars represent the mean of several experiments normalized to the baseline phosphorylation of a given construct (in absence of RPTP69D). (n=3 for YF1707; n=11 for YF1857; n=7 for YF1890 and YF1981; n=4 for YF1911; n=6 for WT; paired t-tests: * P ≤ 0.05; ** P ≤ 0.01; *** P ≤ 0.001) (F) Representative WBs showing single experiments of the overexpression based phosphatase assay described in (E) Each lane represents data from one gel with lanes from other YF mutations omitted. Error bars: SEM. See also Figure S4.

Figure 5. The RPTP69D-target sites are important for Dscam1 function in vivo

(A-E) Representative confocal images of pDC dye-fills (A, A’) WT pDC neurons have a stereotypical branching pattern similar to pSC neurons. (B-E) Over-expression of WT or mutated Dscam1 isoforms in single pDC neurons (see Figure S1A) results in branching defects. While expression of a WT Dscam1 leads to truncation or shortening of the anterior midline-crossing collateral (B,B’), expression of the Y>F1857 Dscam1 mutant does not (C) or only weakly (C’). Expression of the Y>F1890 Dscam1 mutant (D, D’) results in weak defects(D’). Expression of the Y>F1981 Dscam1 mutant leads to dominant defects that affects all axon collaterals (E,E’). (F) Schematics of the different Dscam1 mutants. (G) Quantification of the phenotypic defects. Statistics: Chi-Square, *** p<0.001.

Figure 6. Slit is required for midline collateral formation and enhances RPTP69D-Dscam complex formation by binding to Dscam1

(A-E) Representative confocal images of pSC dye-fills (B, C) Reduction of Slit expression in Sli^dui/Sli^dui (B) or Sli²/Sli^dui (C) mutants results in defects similar to those observed in RPTP69D mutants. Removal of one copy or RPTP69D enhances the penetrance of midline collateral defects in Sli²/Sli^dui flies (D). Removal of one copy of Dscam1 reduces the defects seen in Sli^dui/Sli^dui flies (E). (F) Quantifications of the genetic interactions.(Statistics: Chi-Square test, * P ≤ 0.05; ** P ≤ 0.01, *** P < 0.001). (G) Slit is expressed at the VNC midline during pupal development. Expression is strongly reduced in Sli^dui/Sli^dui flies. (H, I) MARCM clones for Robo1 and Robo2 in ms-neurons show that neither Robo1 nor Robo2 LOF affects branching of ms-neurons (for Robo3, see Figure S7F). Dscam1 binding was detected by WB analysis. (J, K) Schematic representation of the protein expression constructs (J). Top: Alkaline phosphatase (AP) was fused to the Slit-N fragment consisting of the N-terminal 4 LRR-repeats (boxes) and the 5 EGF-like domains (circles) and expressed by baculovirus mediated infection of High Five cells (see Experimental Procedures). The optimal multiplicity of infection (MOI, pfu/cell) was found to be 10. Bottom: A Dscam1 EC4 construct consisting of the N-terminal 4 Dscam1-IG domains fused to Fc tag. (K) Representative binding curve for AP-Slit-N and EC4. Five independent experiments with an average dissociation constant (K_d) for Slit-N/Dscam1-EC4 binding of 22.2 ± 2.85 nM. (L) Slit promotes the formation of a Dscam1-RPTP69D complex. Co-IP of Dscam1 and RPTP69D from a stable cell line shows small amounts of Dscam1-RPTP69D complex, which is significantly increased when cells were incubated with Slit-conditioned medium (Experimental Procedures). Complexes were assessed by semiquantitative WB analysis. Top: Representative WB of one experiment. Bottom: Quantification of multiple independent experiments (n=3 experiments; paired t-test: * P ≤ 0.05). (M) Slit incubation leads to Dscam1 dephosphorylation in S2 cells. Endogenous Dscam1 was immunoprecipitated from S2 cells incubated with or without Slit. Phosphorylation levels were evaluated by semiquantitative WB analysis. Top: representative WB from one experiment. Bottom: Quantification of multiple independent experiments (n=5 experiments; paired t-test: * P ≤ 0.05). (N) Slit and the extracellular domain of Dscam1 can physically interact in vitro. IP of Slit and Dscam1 extracellular domain (EC10) from conditioned S2 cell media (Experimental Procedures). Error bars: SEM. See also Figures S5 and S6.

Figure 7. RPTP69D and Slit are required for specific axon branch consolidation and extension but not branch initiation

(A-L) Single Growth cone analysis. Representative confocal images of single pSC axons expressing GFP under a pSC-specific Gal4 driver, at early (first column), intermediate (second and third columns) and later stages (last column). All samples were collected between 23h and 30h after puparium formation (APF). Midline-position indicated by yellow squares. Stochastic ectopic GFP expression in unrelated neurons/axons occurs occasionally and is indicated by asterisks. WT pSC axons (A) as well as RPTP69D (E) and Slit (I) mutant growth cones are highly complex, displaying a great number of filopodia and micropodia processes (circles) during early stages when the growth cones start to expand and sprout. (B, C). In intermediate stages, a separation into a posterior and anterior axon branch compartment is recognizable (B, C, F, G, J, K). Numerous micropodia can be observed, some of which show small globular tips (satellite growth cones) and secondary short filopodia (arrowheads in B,C,F,K). A subset of micropodia are midline-directed (arrows, B, C, D) and form stable axon branches (white arrow C, D). In WT, sometimes two anterior midline-directed collateral branches are present (yellow and white arrows in C,D), but only one is stabilized subsequently. The axons of RPTP69D (E-H) and Slit (I-L) mutants are still able to form many processes early (E, I), but at intermediate stages, midline-directed projections of micropodia with satellite growth cones are shorter and stunted (yellow arrows in G, K). An impairment of the ability to extend towards the midline persists into later stages (H, L, open arrowheads). In most cases mis-directed (H) or stunted (L) micropodia fail to mature into an anterior midline-directed collateral projection. (M) Quantification of the total number of processes (filopodia and micropodia) at different developmental stages in control, RPTP69D and Slit mutants. No significant differences in total process numbers between control and RPTP69D or Slit mutant axons (M). (N, O) Quantifications of midline-directed micropodia with satellite growth cones (intermediate and late stages). Reduced number of midline-directed collateral growth cones in RPTP69D and Slit mutant axons. Maximum length of the main shaft was used assign samples to three stage groups: early: <30µm, intermediate: 30 and 60 µm, and late: >90 µm. (Statistics: One-Way ANOVA followed by Dunnett multiple comparisons, ** P ≤ 0.005; *** P ≤ 0.001 (M, N) Fisher’s exact test, * P ≤ 0.05 (O)). (P) Schematic of the defects observed in RPTP69D and Slit mutant growth cones: while processes are still formed, their midline-directed extension and consolidation is impaired. Scale bars represent 25µm. Error bars: SEM. See also Figure S7.