An autoinhibitory clamp of actin assembly constrains and directs synaptic endocytosis

Abstract

Synaptic membrane-remodeling events such as endocytosis require force-generating actin assembly. The endocytic machinery that regulates these actin and membrane dynamics localizes at high concentrations to large areas of the presynaptic membrane, but actin assembly and productive endocytosis are far more restricted in space and time. Here we describe a mechanism whereby autoinhibition clamps the presynaptic endocytic machinery to limit actin assembly to discrete functional events. We found that collective interactions between the Drosophila endocytic proteins Nwk/FCHSD2, Dap160/intersectin, and WASp relieve Nwk autoinhibition and promote robust membrane-coupled actin assembly in vitro. Using automated particle tracking to quantify synaptic actin dynamics in vivo, we discovered that Nwk-Dap160 interactions constrain spurious assembly of WASp-dependent actin structures. These interactions also promote synaptic endocytosis, suggesting that autoinhibition both clamps and primes the synaptic endocytic machinery, thereby constraining actin assembly to drive productive membrane remodeling in response to physiological cues.

Keywords: D. melanogaster; actin; cell biology; endocytosis; neuroscience; synapse.

Figure 1.. Synaptic actin patches are dynamic WASp-dependent structures.

(A) Representative maximum intensity projections (MaxIPs) of single spinning disc confocal microscopy time points, showing C155-Gal4-driven actin probes GFP::actin, GMA, and Lifeact::Ruby. (B–D) Automatic detection and analysis of movies acquired at 0.25 Hz of F-actin patch intensity amplitude (B), frequency (C), and duration distribution (D) show similar dynamics for different reporters. (E, F) Single plane Airyscan image of a live muscle 6/7 neuromuscular junction (NMJ) expressing Lifeact::Ruby (magenta) and Arp3::GFP (green). Actin patches colocalize extensively with Arp3::GFP. (F) Quantification of colocalization by Pearson’s coefficient. Arp3 colocalizes with Lifeact significantly more than BRP::GFP, a similarly punctate and membrane-associated negative control. Graph shows mean ± sem; n represents NMJs. (G–I) Patch assembly requires the Arp2/3 activator WASp. GMA patch dynamics in control and WASp mutant animals imaged at 0.25 Hz. (G) MaxIPs of single spinning disc confocal microscopy time points, showing pan-neuronally expressed GMA localization in control and wsp¹ mutant muscle 6/7 NMJs. (H) Quantification of patch frequency. Graph shows mean ± sem; n represents NMJs. (I) Quantification of patch-duration distribution. Bins are 20 s; X-axis values represent bin centers. n represents patches. Scale bars in (A) and (G) are 5 µm, and scale bar in (E) is 2.5 µm. Associated with Figure 1—figure supplement 1, Figure 1—figure supplement 2, and Video 1.

Figure 1—figure supplement 1.. Additional characterization of actin patches.

(A) Images from Video 1 highlighting patch structures and dynamics. All patch and track regions of interest are color coded identically in all panels. Left - last frame of the GMA video from Video 1 showing all patch tracks (colored lines) and current patch detections (colored circles). Center - sum intensity projection of the video with locations of representative patches circled. Right - representative kymographs and intensity traces of patches. Kymograph represents entire timelapse from time 0 (top) to 6 min (bottom), with patch dynamics highlighted by color. As discussed in the text, not all patches are tracked due to overlapping signals or missing frames. (B) Quantification of patch frequency in neuromuscular junctions (NMJs) expressing GMA and imaged at 1 Hz. Data are from the same experiment as controls in Figure 6A–C. (C) Representative maximum intensity projections (MaxIPs) of single spinning disc confocal microscopy time points of neuronally expressed GMA in control and WASp RNAi (expressed in neurons via C155-GAL4) NMJs. (D, E) Quantitative analysis of patch frequency and patch-duration distribution (imaged at 0.25 Hz) shows that presynaptic WASp RNAi recapitulates the decreased patch frequency observed at WASp mutant NMJs. Histogram bins are 20 s; X-axis values represent bin centers. (F) Validation of wsp RNAi knockdown by quantification of presynaptic myc::WASp accumulation. Scale bars in (A) and (C) are 5 µm. Graphs show mean ± sem; n represents NMJs. Associated with Figure 1, Video 1.

Figure 1—figure supplement 2.. Actin dynamics analysis is robust to tracking parameters.

(A) Quantification of patch frequency in the control genotype of Figure 1G–I across a range of patch detection and tracking parameters. Link distance and patch detection threshold have a small effect on patch frequency. Lower graph highlights the effect of linking distance vs detection threshold specifically at 0.3 quality filter (as used in all analyses). (B) Quantification of the percent difference in patch frequency between WASp mutant neuromuscular junctions (NMJs) and controls, from the same experiment shown in Figure 1G–I. Across the majority of the parameter space, WASp mutants show a decrease in patch frequency, as reported. Results diverge only at extreme tracking and detection parameters. Lower graph highlights the effect of linking distance vs detection threshold at 0.3 quality filter. Associated with Figure 1.

Figure 2.. Periactive zone proteins accumulate broadly across the NMJ.

(A) The periactive zone (PAZ) proteins Nwk (magenta) and Dap160 (green) accumulate in a micron-scale mesh adjacent to active zones (AZ) (Bruchpilot, blue). Image shows maximum intensity projection (MaxIP) of a structured illumination microscopy (SIM) Z-stack. (B) Surface projection (top) and medial optical section (bottom) SIM images of live-imaged endogenous Nwk::GFP showing abundant and specific membrane recruitment, similar to fixed imaging. (C–F) PAZ proteins partially colocalize with actin patches. Optical slices of SIM micrographs showing F-actin (labeled with GMA) localization with presynaptically expressed WASp::Myc and Nwk (C) or Dap160 (E). (D, F) Quantification of colocalization between GMA and WASp::Myc, and Nwk (D) or Dap160 (F). (D, F) Quantification (Pearson correlation coefficient R) of colocalization between the indicated pairs of proteins. Graphs show mean ± sem; n represents neuromuscular junctions (NMJs).

Figure 3.. Distinct SH3-SH3 and SH3-BAR domain interactions drive Dap160-Nwk association in vitro and at synapses.

(A) Model for autoinhibition of Nwk membrane binding and WASp activation. Neither membrane-bound nor membrane-free Nwk efficiently activates WASp-mediated actin polymerization, due to persistent SH3b-mediated autoinhibitory interactions, suggesting that an SH3b domain ligand is required for activation. (B) Dap160^SH3CD exhibits electrostatic and hydrophobic interactions with the Nwk F-BAR and SH3 domains, respectively. Glutathione-S-transferase (GST) fusion proteins were immobilized on glutathione agarose and incubated with the indicated purified proteins. Pellets and supernatants were fractionated by sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS-PAGE), Coomassie stained, and quantified by densitometry. Graphs show the average ± sem of three independent reactions. [Nwk^F-BAR] = 1.5 μM, [Nwk] = 0.8 μM, [GST-Dap160^SH3CD] = 1.6 μM, [GST-Dap160^SH3C/D] = 1.2 μM. (C) Summary of Dap160^SH3CD-Nwk^SH3ab interactions. Gray and black arrows indicate electrostatic and hydrophobic interactions, respectively. (D, E) Maximum intensity projection (MaxIP) spinning disc confocal (D) or single Z-plane structured illumination microscopy (SIM) micrographs (E) of muscle 4 neuromuscular junctions (NMJs) expressing C155-GAL4-driven UAS-Dap160 rescue transgene variants in a dap160 null background (dap160^Δ1/Df). Loss of the Dap160^SH3CD domains (Dap160^ΔSH3CD), but not the SH3D domain alone (Dap160^ΔSH3D), decreases the abundance of Nwk (D, right) and Dap160-Nwk colocalization (E, right) at synapses. Contrast-matched panels in (E) are displayed with the same brightness/contrast. Adjacent panels are contrast-adjusted per image to facilitate comparison of Nwk-Dap160 colocalization. Graphs show mean ± sem; n represents NMJs. Scale bars in (D) and (E) are 5 μm and 2.5 μm, respectively. Associated with Figure 3—figure supplements 1–2.

Figure 3—figure supplement 1.. Domain specific interactions between Dap160 and Nwk.

(A, B) Glutathione-S-transferase (GST) fusion proteins were immobilized on glutathione agarose and incubated with the indicated purified proteins. Pellets and supernatants were fractionated by sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS-PAGE), immunoblotted (A) or Coomassie stained (B), and quantified by densitometry. (A) Nwk^SH3b interacts directly with Dap160^SH3C, Dap160^SH3CD, or Dap160^SH3ABCD, but not with Dap160^SH3D alone. [Nwk^SH3b] = 7 μM. (B) The Nwk F-BAR domain interacts directly with Dap160^SH3, Dap160^SH3D, and Dap160^SH3CD. [Nwk^F-BAR] = 1.5 μM, [GST-Dap160^SH3CD] = 8.5 μM, [GST-Dap160^SH3C/D] = 11 μM. (C) Representative Coomassie-stained gels for Figure 3B. Associated with Figure 3.

Figure 3—figure supplement 2.. Validation of Dap160 transgene rescue and knockdown experiments.

(A) Schematic of Dap160 rescue transgenes used in in vivo and Nwk fragments used in in vitro experiments. (B, C) Quantification of Dap160 variant transgene expression (mCherry signal) at neuromuscular junctions (NMJs) from datasets in Figure 3D and Figure 3E, respectively. All transgenes were expressed in neurons by C155-GAL4 in a dap160 null background (dap160^Δ1/Df). (D) Maximum intensity projections (MaxIPs) (left) and quantification (right) of spinning disc confocal micrographs of control and dap160 RNAi-expressing NMJs showing knockdown of Dap160 protein levels, using the same conditions as in Figure 3D. Associated with Figure 3.

Figure 4.. Nwk, Dap160, and PI(4,5)P₂ potentiate WASp-mediated actin assembly at membranes.

(A, B) Pyrene-actin assembly assay (2.5 µM actin, 5% pyrene-labeled). Curves are representative single experiments demonstrating actin assembly kinetics; graphs represent rates calculated from the linear range of assembly from at least two independent experiments. (A) The combination (red trace) of Nwk and Dap160^SH3CD enhances WASp-Arp2/3-mediated actin assembly. Either alone (magenta and blue traces) has no effect on WASp activity. [Nwk] = 500 nM, [Dap160] = 2 µM, [WASp] = 50 nM, [Arp2/3] = 50 nM. (B) PI(4,5)P₂ enhances Nwk-Dap160 activation of WASp-mediated actin assembly. Nwk alone or in combination with 10% PI(4,5)P₂ liposomes fails to activate WASp, while the addition of Dap160^SH3CD and PI(4,5)P₂ synergistically enhances WASp-mediated actin assembly. [Nwk] = 100 nM, [Dap160] = 500 nM, [WASp] = 50 nM, [Arp2/3] = 50 nM. (C) Single slices from spinning disc confocal micrographs of water-droplet actin assembly assay: SNAP-labeled Nwk constructs (red) and Oregon Green actin (green) were mixed with the indicated proteins in an aqueous solution and emulsified in 97.5% 1,2-diphytanoyl-sn-glycero-3-phosphocholine (DPHPC), 2.5% PI(4,5)P₂ in decane. Both deregulated Nwk^ΔSH3b and Nwk + Dap160^SH3CD promote F-actin assembly in droplets. However, while Nwk - Dap160^SH3CD derived F-actin associates with the lipid interface, de-regulated Nwk^ΔSH3b promotes actin assembly from asters that do not associate with membranes. [Nwk^1-xxx] = 500 nM, [Dap160] = 2 µM, [WASp] = 50 nM, [Arp2/3] = 50 nM. Graph indicates percentage of droplets with observable actin filament assembly. Scale bar in (C) is 10 µm.

Figure 5.. Dap160^SH3CD and WASp promote Nwk membrane association.

(A–C) Liposome cosedimentation assays between the indicated purified proteins and liposomes composed of [mol% = DOPC/DOPE/DOPS/PI(4,5)P₂ = 80-x/15/5/x], with x representing PI(4,5)P₂ concentration as noted. Quantification from Coomassie-stained gels represents the mean fraction of total protein that cosedimented with the liposome pellet ± sem. (A) 1:3 Nwk:Dap160^SH3CD saturates enhancement of Nwk membrane association at 5% PI(4,5)P₂, but not to the level of the isolated Nwk F-BAR alone (Nwk^F-BAR, black bar). [Nwk^F-BAR] = 3 µM, [Nwk] = 1.125 µM, [Dap160^SH3CD] = 1.7–6.8 µM. (B) Dap160^SH3CD (but not Dap160^SH3C) enhances Nwk association with membranes at a range of PI(4,5)P₂ concentrations. Maximum binding (comparable to Nwk^F-BAR) occurs only at 5–10% PI(4,5)P₂ concentrations. [Nwk^1-xxx] = 2 µM, [Dap160] = 6 µM. (C) Nwk, WASp, and Dap160^SH3CD mutually enhance membrane recruitment. Addition of Dap160^SH3CD and WASp additively enhances Nwk membrane association, while Dap160^SH3CD and WASp show maximum recruitment to 10% PI(4,5)P₂ liposomes in the presence of both other proteins. [Nwk] = 1 µM, [WASp] = 1 µM, [Dap160^SH3CD] = 3 µM. (D) Giant unilamellar vesicle (GUV) decoration assay, with 10% PI(4,5)P₂ GUVs labeled with <1% TopFluor-PE. The addition of WASp to Nwk (red) and Dap160^SH3CD (blue) enhances the recruitment of Dap160^SH3CD to the membrane (green, note diffuse blue signal in (-) WASp condition). [Nwk] = 250 nM, [WASp] = 250 nM, [Dap160^SH3CD] = 1 µM. Scale bar is 10 μm. (E, F) Fluorescence recovery after photobleaching (FRAP) assay of endogenously labeled Nwk in control and C155-GAL4/UAS-Dicer-driven Dap160^RNAineuromuscular junctions (NMJs). Images show individual medial optical sections of Airyscan confocal images at the indicated time point. Control Nwk signal shows membrane association (see strong peripheral signal) and slower recovery kinetics, while loss of Dap160 eliminates the strong peripheral accumulation of Nwk::GFP and increases the recovery kinetics of Nwk::GFP in the bleached region (dashed circles). Graph shows mean ± sem; n represents NMJs. Scale bar is 5 μm. Associated with Figure 5—figure supplement 1.

Figure 5—figure supplement 1.. Membrane binding and bending by Dap160-WASp-Nwk.

(A) Cosedimentation assay between the indicated purified proteins and liposomes composed of [mol% = DOPC/DOPE/DOPS/PI(4,5)P₂ = 75/15/5/5]. Dap160^SH3CD is unable to enhance membrane binding of Nwk^ΔSH3b (which lacks the autoinhibitory and Dap160-binding SH3b domain) or WASp. Conversely, Nwk^ΔSH3b is unable to promote membrane recruitment of Dap160^SH3CD. Quantification from Coomassie-stained gels represents the mean fraction of total protein that cosedimented with the liposome pellet ± sem. [Nwk^ΔSH3b] = 1 µM, [WASp] = 1 µM, [Dap160] = 3 µM. Graph shows mean ± sem. (B) Representative Coomassie-stained gel from (A). (C) Representative Coomassie-stained gel from Figure 5C. (D) Confocal timelapse of deformation of GUVs (10% PI(4,5)P₂, labeled with <1% TopFluor-PE) decorated with the indicated proteins [Nwk^1-xxx] = 250 nM, [WASp] = 250 nM, [Dap160] = 1.25 µM. The Nwk-Dap160-WASp complex retains the same membrane-remodeling activity as Nwk alone. Associated with Figure 5.

Figure 6.. Loss of the Dap160-Nwk interaction disrupts actin patch dynamics at synapses in vivo.

(A, D) Maximum intensity projections (MaxIPs) of live spinning disc confocal micrographs of presynaptically expressed GMA in muscle 6/7 neuromuscular junctions (NMJs) of the indicated genotypes, imaged at 1 Hz. Graphs quantify patch frequency (B, E) and distribution of patch durations (C, F). Loss of nwk (A–C) or of the Nwk-interacting Dap160^SH3CD domain (D–F) increases the frequency of actin patch assembly. In both cases, there is no change in the distribution of patch durations (C, F). Scale bars in (A, D) are 5 μm. Associated with Figure 6—figure supplements 1–2, Video 2.

Figure 6—figure supplement 1.. Analysis of actin dynamics is robust to tracking parameters at 1 Hz imaging.

(A) Quantification of patch frequency in the control genotype of Figure 6A–C across a range of patch detection and tracking parameters. Consistent with the 0.25 Hz imaging (Figure 1—figure supplement 2), link distance and patch detection threshold have a small effect on patch frequency. Lower graph highlights the effect of linking distance vs detection threshold specifically at the 0.3 quality filter (as used in all analyses). (B) Quantification of the percent difference in patch frequency between nwk mutant neuromuscular junctions (NMJs) and controls, from the same experiment shown in Figure 6A–C. Across the majority of the parameter space, nwk mutants show an increase in patch frequency, as reported. Lower graph highlights the effect of linking distance vs detection threshold at the 0.3 quality filter. Associated with Figure 6.

Figure 6—figure supplement 2.. Validation of actin particle analysis in nwk mutants.

(A) Coefficient of variation (CoV) projections of timelapse videos of GMA dynamics in control and nwk mutant neuromuscular junctions (NMJs) (analysis of the same dataset as shown in Figure 6A–C; different representative images shown).(B) Average temporal CoV values across NMJs. nwk mutants exhibit greater temporal variation in GMA signal than controls. (C) Quantification of the fraction of NMJ area covered by ‘high’ CoV pixels (identified by automated thresholding using either Li (Li and Tam, 1998) or Moments (Tsai, 1995) algorithm). nwk mutants show no difference in the fraction of high CoV pixels, suggesting that actin dynamics are confined to a restricted region of the synapse and vary more over time rather than space. (D) Example of results from temporal CoV analysis of synthetic data, in which the frequency and lifetime were systematically varied. (Left) Representative images of synthetic data created using a custom FIJI script. (Right) Quantification of CoV over time. Shaded green region indicates values that most closely match in vivo measurements drawn from Patchtracker particle-based analysis. In this regime, our model indicates that the nwk mutant effect size in panel (B) corresponds to a 43% increase in patch frequency, slightly higher than the particle-based measurement of 28%. Associated with Figure 6.

Figure 7.. Actin patches and the Nwk-Dap160 interaction are associated with synaptic endocytosis.

(A, B) Sum intensity projection (A) and representative kymographs (B) of spinning disc confocal timelapse of presynaptically expressed Lifeact::Ruby and clc::GFP. (A) Sum projection of 41 frames (82 s) highlights overlapping intensities of clc and Lifeact (circles and arrowheads). Circles indicate locations of kymographs in panel (B). (B) Kymographs of clc and Lifeact signals. Kymographs span the full duration of the movie from left (0 s) to right (82 s). Intensity profiles were normalized per channel from the minimum to the maximum value of each profile. (C) Quantification of colocalization between Lifeact::Ruby and Clc::GFP. Presynaptically expressed Lifeact::Ruby was co-expressed with either presynaptically expressed Clc::GFP or a BRP::GFP knockin and imaged in 3D stacks by Airyscan. Bruchpilot (BRP) control is the same dataset as in Figure 1F (all data were acquired contemporaneously). (D, E) Normal patch assembly requires dynamin activity. (D) Maximum intensity projections (MaxIPs) of single spinning disc confocal microscopy time points, showing pan-neuronally expressed GFP::actin in control and shi^TS1 mutant muscle 6/7 neuromuscular junctions (NMJs), imaged at 1 Hz, at the restrictive temperature of 31^oC under stimulating conditions to drive the terminal shi^TS1 phenotype (45 mM KCl, 2 mM CaCl₂). Graph shows mean ± sem. n represents NMJs. (E) Quantification of patch frequency. (F–I) FM dye uptake assays at muscle 6/7 NMJs following 5 min of 90 mM potassium stimulation at 36°C. (F, H) MaxIPs of spinning disc confocal micrographs of FM dye uptake assays. Dotted lines highlight synapse outline in shi^TS1 NMJs. (F, G) nwk mutants exhibit partially defective FM4-64fx dye uptake relative to shi^TS1 mutants. (H, I) Loss of Dap160-Nwk interactions in a Dap160^ΔSH3CD truncation (but not Dap160^ΔSH3D) exhibits partially defective FM1-43 dye uptake relative to shi^TS1, similar to nwk mutants. Graphs show mean ± sem; n represents NMJs. Scale bars are 2.5 µm (A, B) or 5 µm (D, F, H). Associated with Figure 7—figure supplements 1–3, Video 3.

Figure 7—figure supplement 1.. Colocalization of actin with AP2α and Clathrin light chain.

(A) Clc::GFP and AP2α::GFP are enriched at the synaptic plasma membrane. Images show single plane Z-slices of the mid-bouton region (effectively a cross-section through the bouton) from live Airyscan imaging of presynaptically expressed Clc::GFP (green) and endogenously tagged AP2α::GFP (magenta). Right: radial intensity profile from the outside edge (left) to the center (right) of boutons similar to those shown (values are mean ± Std Dev across n = 6 (Clc::GFP) or 7 (AP2 α::GFP) boutons). (B, C) Spinning disc confocal imaging maximum intensity projections (MaxIPs) of Airyscan micrographs of endogenously tagged AP2α::GFP (green) and Lifeact::Ruby (magenta) in live (B) or fixed (C) preparations. (C) Actin and AP2α partially colocalize; note a significant fraction of endogenously labeled AP2α::GFP is postsynaptic. Graph shows quantification of colocalization by Pearson’s coefficient. Scale bars in (A) and (B) are 2.5 μm. Associated with Figure 7.

Figure 7—figure supplement 2.. All Dap160 transgenes rescue dap160 satellite bouton phenotype.

(A) Maximum intensity projections (MaxIPs) of epifluorescence micrographs of muscle 4 neuromuscular junctions (NMJs) stained for anti-HRP (magenta) and anti-Dlg (green). Dap160 RNAi-expressing NMJs exhibit many satellite boutons, which are restored to normal levels by presynaptic expression of Dap160^FL, Dap160^ΔSH3D, or Dap160^ΔSH3CD (in dap160^Δ1/Df larvae). (B) Quantification of satellite boutons. Graph shows mean ± sem; n represents NMJs. Scale bar in (A) is 10 µm. Associated with Figure 7.

Figure 7—figure supplement 3.. nwk and dap160 domain mutants do not disrupt FM dye unloading.

Loading and unloading of FM dyes in nwk and dap160 domain mutants by stimulation in 90 mM KCl + 2 mM CaCl₂. (A, D) Maximum intensity projections (MaxIPs) of spinning disc confocal stacks acquired after loading (left) and unloading (right) of FM dyes. (B, C) Quantification of FM4-64 loading (B) and unloading (C) in nwk mutants. (E, F) Quantification of FM1-43 loading (E) and unloading (F) in dap160 domain mutants.