Volume 93, Issue 4, Pages 854-866.e4 (February 2017) Synaptic Vesicle Endocytosis Occurs on Multiple Timescales and Is Mediated by Formin-Dependent Actin Assembly  Tolga Soykan, Natalie Kaempf, Takeshi Sakaba, Dennis Vollweiter, Felix Goerdeler, Dmytro Puchkov, Natalia L. Kononenko, Volker Haucke  Neuron  Volume 93, Issue 4, Pages 854-866.e4 (February 2017) DOI: 10.1016/j.neuron.2017.02.011 Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 1 SV Proteins Are Endocytosed on Multiple Timescales Ranging from Less Than a Second to Several Seconds at Physiological Temperature (A) Scheme for pHluorin experiment coupled to external acid pulse. SynaptopHluorin-expressing cultured neurons are subjected to a brief phase of external acidic buffer 15 s before (Q0) and t number of seconds after (Q1) a train of stimulation, in the presence of folimycin. The relative amounts of synaptopHluorin released (the steep rise in fluorescence upon stimulation) and retrieved within the t seconds after stimulation (already internalized proteins not affected by surface quenching, shaded in yellow) is derived from the normalized fluorescence values as indicated. Representative SVs and presynaptic plasma membrane are depicted for each step of the protocol. Blue shade represents acidic pH, black and green outlines represent quenched and unquenched fluorescence of pHluorin, respectively. (B and C) The acid quench protocol reveals the fraction of SV proteins retrieved immediately after stimulation. The average traces for synaptotagmin 1-pHluorin-expressing neurons exposed to external acidic buffer 15 s before and 1 s after a 40 Hz 5 s stimulation at 37°C (B) and 25°C (C). The fractions corresponding to fast retrieval of pHluorin are shaded in yellow. Traces represent the mean ± SEM of n = 15 (25°C) and 13 (37°C) neurons measured in N = 3 independent experiments. (D) SV protein endocytosis is temperature dependent. Decreased levels of synaptotagmin 1-pHluorin retrieval in 1 s after 40 Hz 5 s stimulation at 25°C (11.0% ± 1.6%), as compared to 37°C (19.2% ± 0.1%, p < 0.01). Data shown represent the mean ± SEM of N = 3 independent experiments, n = 13 and 15 neurons in total for 37°C and 25°C. (E) Percentage of synaptotagmin1-pHluorin retrieved in 1 s after 10 Hz 5 s (23.7% ± 1.6%), 40 Hz 1 s (14.2% ± 0.7%), and 40 Hz 5 s (22.5% ± 0.9%) and synaptophysin-pHluorin retrieved in 1 s after 40 Hz 5 s (28.1% ± 2.0%) stimulation. Data shown represent the mean ± SEM of N = 3 or 4 independent experiments, n = 7, 10, 10, and 12 neurons measured in total. Significance evaluated using one-way ANOVA. (F and G) The average traces for synaptophysin-pHluorin-expressing neurons exposed to external acidic buffer 15 s before and 1 s after a 10 Hz 1 s (F) and 4 Hz 0.5 s (G) stimulation in the presence of folimycin and 4 mM extracellular calcium. The fractions corresponding fast retrieval of synaptophysin-pHluorin are shaded in yellow. Traces represent the mean ± SEM of n = 15 (2 AP) and 19 (10 AP) neurons measured in N = 4 independent experiments. (H) Percentage of synaptophysin-pHluorin retrieved in 1 s after 10 Hz 1 s (18.1% ± 1.9%) and 4 Hz 0.5 s (45.7% ± 2.8%, p < 0.001). Data shown represent the mean ± SEM of N = 4 experiments, n = 15 and 19 neurons measured for 2 AP and 10 AP, respectively. (I) The time course of synaptophysin-pHluorin retrieval after 10 Hz 1 s and 4 Hz 0.5 s stimulation measured by applying an acid pulse at 1, 3, 5, and 10 s after the end of the stimulation. The data points were fit with biexponential decay curves, yielding two time constants τfast = 0.76 s (19%) and τslow = 25.9 s for 10 AP and τfast = 0.72 s (54%) and τslow = 12.0 s for 2 AP stimulation. Please note that the determination of τfast is associated with some imprecision as the assay is “blind” to endocytosis that occurs during the stimulation. Statistically significant estimates of data shown were obtained from N independent experiments. See also Figure S1. Neuron 2017 93, 854-866.e4DOI: (10.1016/j.neuron.2017.02.011) Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 2 Endocytosis of SV Proteins Is Largely Independent of Clathrin and AP-2 and, Thus, of CME in Hippocampal Neurons (A) Primary hippocampal neurons from WT mice transfected with ubiquitin-promoter controlled synaptotagmin 1-pHluorin (green) and scrambled or CHC shRNA (red) using calcium-phosphate or lipofectamine were immunostained with CHC antibodies (magenta). Scale bar, 20 μm. (B) Levels of clathrin (CHC) quantified from (A). Transfection with CHC shRNA leads to depletion of CHC down to 26.6% ± 4.5% (p < 0.01) in calcium-phosphate (CaP) and 43.5% ± 3.6% (p < 0.05) in lipofectamine-transfected (LF 2000) neurons. Data shown represent the mean ± SEM of N = 5 with n = 135 and n = 118 neurons in total for lipofectamine-transfected scrambled shRNA and CHC shRNA and N = 5 with n = 84 and n = 86 neurons in total for calcium-transfected scrambled shRNA and CHC shRNA. (C–E) Depletion of clathrin does not affect endocytic retrieval during mild stimulation. Average normalized traces for calcium-phosphate transfected neurons coexpressing synaptophysin-pHluorin and scrambled shRNA or clathrin (CHC) shRNA in response to 10 APs with 10 Hz, 1 s stimulation in 4 mM extracellular calcium (C) and 50 APs (10 Hz, 5 s, (D). (E) Endocytic decay constant of calcium-phosphate-transfected neurons coexpressing synaptophysin-pHluorin and scrambled shRNA (20.2 ± 5.0 s) or clathrin (CHC) shRNA (28.6 ± 7.4 s) at 10 APs and neurons coexpressing synaptophysin-pHluorin and scrambled shRNA (17.3 ± 1.2 s) or clathrin (CHC) shRNA (20.2 ± 2.9 s) at 50 APs. Data shown represent the mean ± SEM of N = 6 independent experiments with n = 26 and n = 17 neurons in total for scrambled shRNA and CHC shRNA for 10 APs stimulation and N = 6 independent experiments with n = 40 and n = 26 neurons in total for scrambled shRNA and CHC shRNA for 50 APs. (F–H) Depletion of clathrin affects the time course of endocytosis in lipofectamine-transfected neurons. Average normalized traces for calcium-phosphate-transfected (F) and lipofectamine-transfected (G) neurons coexpressing synaptotagmin 1-pHluorin and scrambled shRNA or clathrin (CHC) shRNA in response to strong stimulation with 200 APs (40 Hz, 5 s). (H) Endocytic decay constant of calcium-phosphate-transfected neurons coexpressing synaptotagmin 1-pHluorin and scrambled shRNA (36.6 ± 5.0 s) or clathrin (CHC) shRNA (36.6 ± 9.1 s) and lipofectamine-transfected neurons coexpressing synaptotagmin 1-pHluorin and scrambled shRNA (45.9 ± 6.7 s) or clathrin (CHC) shRNA (87.6 ± 9.2 s, p < 0.01) in response to 200 APs (40 Hz, 5 s). (I) Depletion of clathrin leads to reduced peak amplitude of synaptotagmin 1-pHluorin, independent of the method of transfection. Relative peak amplitude of synaptotagmin 1-pHluorin in calcium phosphate-transfected neurons expressing scrambled shRNA (100% ± 22.5%) or clathrin (CHC) shRNA (57.1% ± 9.1%, p < 0.05) and lipofectamine-transfected neurons expressing scrambled shRNA (100% ± 11.2%) or CHC shRNA (61.7% ± 7.1%, p < 0.05) in response to 200 APs (40 Hz, 5 s). (F–I) Data shown represent the mean ± SEM of N = 6 with n = 49 and n = 30 neurons in total for lipofectamine-transfected scrambled shRNA and CHC shRNA and N = 6 with n = 38 and n = 37 neurons in total for calcium-transfected scrambled shRNA and CHC shRNA. (J) Average normalized traces for lipofectamine-transfected neurons coexpressing synaptotagmin 1-pHluorin and scrambled shRNA or clathrin (CHC) shRNA in response to strong stimulation with 400 APs (40 Hz, 10 s) as in Nicholson-Fish et al. (2015). (K) Endocytic decay constant of lipofectamine-transfected neurons coexpressing synaptotagmin 1-pHluorin and scrambled shRNA (67.8 ± 4.6 s) or CHC shRNA (143.9 ± 25.3 s, p < 0.05) in response to 40 Hz 10 s stimulation. Data shown represent the mean ± SEM of N = 4 independent experiments with n = 33 and n = 14 neurons in total for scrambled shRNA and CHC shRNA. Statistically significant estimates of data shown were obtained from N independent experiments. See also Figure S2. Neuron 2017 93, 854-866.e4DOI: (10.1016/j.neuron.2017.02.011) Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 3 SV Endocytosis Is Driven by Formin-Mediated Actin Assembly and Myosin II Activity in Hippocampal Neurons (A) Schematic illustrating the mode of action of the actin modulating drugs used in the study. Latrunculin A binds to actin monomers and prevents their incorporation into existing actin filaments. SMIFH2 inhibits formins and blocks linear actin polymerization. Wiskostatin inhibits N-WASP and blocks branched actin nucleation. Blebbistatin inhibits the ATPase activity of Myosin II. Jasplakinolide stabilizes actin filaments by preventing polymer disassembly. (B) Averaged normalized traces for neurons expressing synaptophysin-pHluorin in response to 10 Hz 5 s stimulation, treated with 0.1% DMSO or 20 μM para-nitroblebbistatin. (C) τDMSO = 27.0 ± 4.2 s, τpara-nitro-Blebbistatin = 57.5 ± 6.5 s, p < 0.05. Data shown represent the mean ± SEM of N = 5 independent experiments, n = 61 and 49 neurons in total for DMSO and para-nitroblebbistatin. (D) Averaged normalized traces for neurons expressing synaptophysin-pHluorin in response to 10 Hz 5 s stimulation, treated with 0.1% DMSO or 10 μM Wiskostatin. (E) τDMSO = 21.9 ± 2.8 s, τWiskostatin = 22.8 ± 4.7 s, p = 0.71. Data shown represent the mean ± SEM of N = 3 independent experiments, n = 28 and 27 neurons in total for DMSO and Wiskostatin. (F) Averaged normalized traces for neurons expressing synaptophysin-pHluorin in response to 10 Hz 5 s stimulation, treated with 0.5% DMSO or 5 μM Jasplakinolide. (G) τDMSO = 22.3 ± 2.1 s, τJasplakinolide = 15.5 ± 2.6 s, p < 0.05. Data shown represent the mean ± SEM of N = 3 independent experiments, n = 30 and 26 neurons in total for DMSO and Jasplakinolide. (H) Averaged normalized traces for neurons expressing synaptophysin-pHluorin in response to 10 Hz 5 s stimulation, treated with 0.1% DMSO or 30 μM SMIFH2. (I) τDMSO = 19.2 ± 3.0 s, τSMIFH2 = 65.5 ± 2.8 s, p < 0.001. Data shown represent the mean ± SEM of N = 3 independent experiments, n = 22 and 18 neurons in total for DMSO and SMIFH2. (J) Averaged normalized traces for neurons expressing synaptotagmin 1-pHluorin in response to 40 Hz 5 s stimulation, treated with 0.1% DMSO or 30 μM SMIFH2. (K) τDMSO = 32.3 ± 1.4 s, τSMIFH2 = 58.3 ± 9.6 s, p < 0.05. Data shown represent the mean ± SEM of N = 4 independent experiments, n = 34 and 26 neurons in total for DMSO and SMIFH2. (L) Averaged normalized traces for neurons coexpressing synaptophysin-pHluorin and shRNA plasmid targeting mDia1 (shmDia1) or a non-target shRNA (shControl) in response to 10 Hz 5 s stimulation. (M) τshControl = 20.1 ± 3.1 s, τshmDia1 = 34.0 ± 3.9 s, p < 0.05. Data shown represent the mean ± SEM of N = 5 independent experiments, n = 32 and 45 neurons in total for shControl and shmDia1. Statistically significant estimates of data shown were obtained from N independent experiments. See also Figure S3. Neuron 2017 93, 854-866.e4DOI: (10.1016/j.neuron.2017.02.011) Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 4 Formin and Myosin II Are Required for SV Endocytosis at the Calyx of Held (A–D) The calyx of Held terminals (P8–P12) were patch clamped and depolarized from −80 mV to 0 mV for 50 ms to deplete the RRP of synaptic vesicles. Application of blebbistatin and SMIFH2 leads to slower capacitance decays (absolute in A and C and normalized in B and D). Traces represent the mean ± SEM of n = 9, 8, and 7 calyces measured for control, blebbistatin, and SMIFH2, respectively. See also Figure S4. Neuron 2017 93, 854-866.e4DOI: (10.1016/j.neuron.2017.02.011) Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 5 Formins Are Required for Formation of Endosome-like Vacuoles (A) Electron micrographs of endocytic intermediates at hippocampal synapses treated with DMSO or SMIFH2 and non-stimulated or stimulated with 200 APs (40 Hz 5 s) and chemically fixed 20 s afterward. Endosome-like vacuoles (ELVs) and non-coated invaginations are marked with ∗ and arrowheads, respectively. Postsynaptic compartment and the postsynaptic density are highlighted in yellow and orange, respectively. Scale bar, 500 nm. (B) Example images of clathrin-coated vesicles (CCV), clathrin-coated pits (CCP), non-coated invaginations, and ELVs. Scale bar, 250 nm. (C–F) Formin inhibition leads to reduced ELV formation and the accumulation of non-coated presynaptic plasma membrane invaginations in stimulated boutons. (C) Average number of CCPs per μm2 in non-stimulated (DMSO: 0.02 ± 0.01, SMIFH2: 0.03 ± 0.01) and stimulated (DMSO: 0.20 ± 0.05, SMIFH2: 0.28 ± 0.06) boutons. (D) Average number of CCVs per μm2 in non-stimulated (DMSO: 0.07 ± 0.03, SMIFH2: 0.08 ± 0.03) and stimulated (DMSO: 0.24 ± 0.05, SMIFH2: 0.18 ± 0.04) boutons. (E) Average number of non-coated invaginations per μm2 in non-stimulated (DMSO: 0.22 ± 0.06, SMIFH2: 0.19 ± 0.05) and stimulated (DMSO: 0.17 ± 0.04, SMIFH2: 0.49 ± 0.09, p < 0.05) boutons. (F) Average number of ELVs per μm2 in non-stimulated (DMSO: 3.46 ± 0.23, SMIFH2: 3.84 ± 0.28) and stimulated (DMSO: 6.81 ± 0.37, SMIFH2: 4.39 ± 0.30, p < 0.05) boutons. Data shown represent the mean ± SEM of N = 4 independent experiments, n > 200 boutons analyzed per condition. Neuron 2017 93, 854-866.e4DOI: (10.1016/j.neuron.2017.02.011) Copyright © 2017 Elsevier Inc. Terms and Conditions

Figure 6 A Model for the Mechanism of SV Endocytosis A schematic summarizing clathrin-independent endocytosis (CIE) of synaptic vesicles (SVs). Exocytic fusion of SVs is followed by actomyosin-driven membrane invagination, regulated by membrane-associated formins such as mDia1, independent of clathrin and adaptors. The invaginated membrane undergoes fission by the combined forces of dynamin and myosin II motors to form an endosome-like vacuole (ELV). This process starts rapidly (i.e., less than 1 s) after a brief stimulus, but depending on the number of fused SVs may take up to several seconds to complete. Concurrently, adaptor proteins and clathrin accumulate on invaginating membranes and ELVs to cluster SV proteins and generate clathrin-coated vesicles (CCVs), which will subsequently lose their coat to reform SVs. Neuron 2017 93, 854-866.e4DOI: (10.1016/j.neuron.2017.02.011) Copyright © 2017 Elsevier Inc. Terms and Conditions