Synaptic Activity Regulates the Abundance and Binding of Complexin

1318

Biophysical Journal

Volume 108

March 2015

1318–1329

Article Synaptic Activity Regulates the Abundance and Binding of Complexin Rachel T. Wragg,1 Ge´raldine Gouzer,1 Jihong Bai,2 Gianluca Arianna,1 Timothy A. Ryan,1 and Jeremy S. Dittman1,* 1

Department of Biochemistry, Weill Cornell Medical College, New York, New York; and 2Division of Basic Sciences, Fred Hutchinson Cancer Research Center, Seattle, Washington

ABSTRACT Nervous system function relies on precise chemical communication between neurons at specialized junctions known as synapses. Complexin (CPX) is one of a small number of cytoplasmic proteins that are indispensable in controlling neurotransmitter release through SNARE and synaptic vesicle interactions. However, the mechanisms that recruit and stabilize CPX are poorly understood. The mobility of CPX tagged with photoactivatable green fluorescent protein (pGFP) was quantified in vivo using Caenorhabditis elegans. Although pGFP escaped the synapse within seconds, CPX-pGFP displayed both fast and slow decay components, requiring minutes for complete exchange of the synaptic pool. The longer synaptic residence time of CPX arose from both synaptic vesicle and SNARE interactions, and surprisingly, CPX mobility depended on synaptic activity. Moreover, mouse CPX-GFP reversibly dispersed out of hippocampal presynaptic terminals during stimulation, and blockade of vesicle fusion prevented CPX dispersion. Hence, synaptic CPX can rapidly redistribute and this exchange is influenced by neuronal activity, potentially contributing to use-dependent plasticity.

INTRODUCTION Neurons communicate predominantly via chemical synaptic transmission between pre- and postsynaptic specializations where neurotransmitter release is choreographed by the interactions of hundreds of proteins (1). Complexin (CPX) is an essential cytoplasmic protein that plays a major role in the control of neurotransmitter release (2–8). CPX has both positive and negative regulatory activities that require direct binding to the SNARE complex as well as interactions with synaptic vesicles (SVs) (4,9–14). Suppression of spontaneous SV fusion requires both SNARE binding and SV interactions, whereas CPX can promote calcium-triggered fusion even in the absence of vesicle interactions (7,8,12,15). Additionally, SNARE complexes and SVs are repeatedly formed, consumed, and recycled during synaptic transmission, and thus CPX targeting may be directly affected by ongoing synaptic activity. Although the speed of CPX binding to SVs and assembling SNAREs is expected to have a major impact on synaptic strength, the processes that capture and stabilize synaptic CPX are not known. Although the mobility of intracellular calcium within and between synaptic structures has been studied extensively (16–19), protein mobility and the extent to which presynaptic proteins freely diffuse are less explored. Given that cytoplasmic synaptic proteins are not necessarily permanent residents of a particular synapse, several questions arise: How is CPX localized to the synaptic bouton? What is the rate of CPX diffusion in and out of the synapse? Does the ge-

ometry of the bouton affect apparent CPX mobility? What is the impact of transient binding interactions on sequestration of CPX? How do calcium influx and SV cycling affect the movement of CPX into and out of the synapse? To address these questions, presynaptic proteins were imaged in en passant synapses of living intact Caenorhabditis elegans and in cultured hippocampal neurons. Quantitative measurements of mobility of both photoactivatable green fluorescent protein (pGFP) and CPX fused to pGFP (CPXpG) revealed that small synaptic proteins exchange between synapses on a timescale of seconds with a diffusion rate set in part by synaptic morphology. Furthermore, SVand SNARE interactions significantly restricted the mobility of CPX while conferring a strong sensitivity to the SV cycle. Both chronic and acute changes in neuronal activity altered CPX mobility and presynaptic abundance. Dispersion of CPX required SV cycling, whereas depolarization and calcium influx were not sufficient to mobilize CPX. We developed a simple quantitative model of CPX mobility incorporating synapse morphology, restricted diffusion, chemical interactions, and synaptic activity. These studies clarify the processes underlying the regulation of CPX binding and provide a general quantitative framework for understanding the roles played by synapse geometry, binding interactions, and activity in shaping the dynamics of synaptic proteins. MATERIALS AND METHODS Strains

Submitted July 31, 2014, and accepted for publication December 29, 2014. *Correspondence: [email protected]

Animals were maintained on agar nematode growth media at 20
C and seeded with OP50 bacteria as previously described (20). Strains employed

Editor: Rong Li. Ó 2015 by the Biophysical Society 0006-3495/15/03/1318/12 $2.00

http://dx.doi.org/10.1016/j.bpj.2014.12.057

Complexin Dynamics at the Synapse in this study include: N2 Bristol, unc-13(s69), unc-64(e246), and tom1(ok285), JSD054: tauls27[pGFP þ mCherry-Rab3 under Punc-129], JSD079: tauls28[CPX-mpG þ mCherry-Rab3 under Punc-129], JSD084: tauIs33[CPX(KY/AA)-mpG þ mCherry-Rab3 under Punc-129], JSD130: tauIs27;unc-13(s69), JSD132: tauIs27;tom-1(ok285), JSD133: tauIs28; unc-13(s69), JSD135: tauIs28;tom-1(ok285), JSD223: tauls73[DeltaCT2Linker mpGFP under Punc-129], JSD230: tauls74[Soluble GFP under Punc-129], JSD240: tauIs79[DeltaCH-mpG þ mCherry Rab3 under Punc129], JSD278: tauIs28;unc-64(e246), JSD280: tauIs73;unc-64(e246), JSD298: tauls74;unc-13(s69), JSD299: tauls74;unc-64(e246), JSD305: tauIs73;tom-1(ok285), JSD393: tauIs119[CPX(DeltaCH,DeltaCT)-Linker mpGFP under Punc-129], JSD479: tauIs145[pGFP-L-SNG-1þ mCherry: Rab-3 under Punc-129], JSD740: tauIs27;unc-64(e246).

Calibration of the 405 nm photoactivation beam size The spatial extent of the 405 nm beam was estimated by systematically stepping the laser along a line containing a single 170 nm fluorescent bead and integrating the total measured fluorescence. An image was acquired at each position using only the parked laser to excite the bead fluorophores. This process was repeated several times on separate beads and the data were background subtracted, normalized, and averaged to create a plot of excitation efficiency versus laser position. The data were fit by a Gaussian curve with a standard deviation of 330 nm and a full-width at half maximum value of ~500 nm (see Fig. 1 E).

1319

A

CPX-pG mCherry-Rab3

B

axon C

D

pGFP imaging and quantification Animals were immobilized using a 30 mg/mL solution of 2,3-butanedione monoxime (J.T.Baker, Center Valley, PA) mounted on 2% agarose pads, and imaged on an inverted Olympus microscope (IX81), using a laser scanning confocal imaging system (Olympus Fluoview FV1000 with dual confocal scan heads), and an Olympus PlanApo 60X 1.42 NA objective. For the high time resolution pGFP experiments shown in Figs. 1, 2, and 3, pGFP was photoactivated with a 2 msec pulse from a stationary 405 nm laser beam centered on the synapse of interest based on mCherry::RAB-3 fluorescence. Line scans of 5 to 10 mm centered on the target synapse were collected every 1.5–2 msec, depending on the length of the line scan. Kymographs of sequential line scans were assembled and analyzed to generate time courses of fluorescence decay averaged over a 1 mm region centered on the target synapse. For the lower time resolution experiments in Figs. 3, 4, and 5, rectangular regions measuring 10  0.2 mm (400  8 pixels) centered on the target synapse were collected every 30 to 40 msec. Each image was collapsed to a line scan using maximal intensity projection. Consecutive projected line scans were then assembled into kymographs for the green (CPX-pG) and red (mCherry-RAB-3) channels. Experiments were rejected if mCherry::RAB-3 fluorescence indicated large drift or movement artifacts. Average fluorescence in a 1 mm region centered on the mCherry::RAB-3 punctum was computed for both channels and the green/red ratio was recorded over a 30 s period. The green channel fluorescence was corrected for the pGFP photoswitching artifact as described in Fig S2 in the Supporting Material). All analysis was performed in Igor using custom written software.

Hippocampal culture and live imaging Hippocampal CA3–CA1 regions were dissected from 1- to 2-day-old Sprague Dawley rats, dissociated (bovine pancreas trypsin, 5 min at room temperature) and plated as previously described (21). Cells were maintained in culture media consisting of minimal essential medium, 0.5% glucose, 2 mM Glutamax (Gibco, Grand Island, NY), 0.1 g/L transferrin (Calbiochem, Billerica, MA), 0.025 g/L insulin, 5% Fetal Bovine Serum (Atlanta Biologicals, Flowery Branch, GA), 2% B-27 (Gibco). After 48 h

E

F

G

FIGURE 1 Using pGFP to determine mobility of synaptic proteins. (A) Cartoon of an adult worm expressing pGFP in a dorsal motor neuron (DA/DB axon and neuromuscular junctions). The rectangle indicates the region imaged in this study. (B) Experimental scheme using pGFP expressed in a small number of motor neuron axons. Nonactivated pGFP (black) is distributed throughout the axon along with SVs marked by mCherry::RAB-3 (red circles). Photoactivation of a single synapse with a 405 nm laser pulse (purple) converts pGFP to a fluorescent state (green) followed by diffusion and reequilibration with axonal pGFP. (C) Kymographs of mCherry-Rab3 and pGFP line scans (~10 mm of axon) collected every 2 msec for 5 s. Three en passant synapses are apparent based on localized Rab3 fluorescence (gray arrows). About 1 s into the recording, a brief photoactivation of the middle synapse (#2, purple arrowhead) produced a transient increase in GFP fluorescence that rapidly dispersed throughout the axon. (D) Time course of 1-mm spatial averages for each of the three synapses as indicated. (E) The activated region was estimated to be ~500 nm in diameter based on scanning a 170 nm fluorescent bead (see Materials and Methods). (F) Six consecutive photoactivation trials at a single synapse show similar decay kinetics. Intertrial interval was 2 min. (G) Single trial fits to 1D (red, tD) and 3D diffusion (blue, Dapp). See Materials and Methods for model details.

Biophysical Journal 108(6) 1318–1329

1320

Wragg et al.

A

D

B

C

E

F

FIGURE 2 Effects of synaptic bouton geometry on protein mobility. (A) Example of bouton (blue) and axonal (black) decay kinetics for pGFP with fits to 1D diffusion. The cartoon inset indicates the locations of point photoactivation. (B) Average decay time courses (5SE) for axonal (black) and synaptic pGFP (blue). Number of experiments indicated on graph. Inset: Average decay time constants using 1D diffusion show a sixfold decrease in apparent mobility in synaptic boutons compared to axons. Data are mean 5 SE and **indicates p < 0.01 by Student’s t-test. (C) Individual 3D synaptic boutons were modeled as 1-mm long ellipsoids in Virtual Cell as described in the Materials and Methods using the indicated sizes. Seven boutons were placed evenly along a 24 mm length of axon with 3 mm spacing. The center bouton was used for photoactivation simulations (arrowhead). (D) Simulated normalized decay curves for values of bouton radius ranging from 100 nm (blue) to 500 nm (yellow). For each bouton radius, five values of diffusion coefficients ranging from 2 to 8 mm2/s were used (corresponding to the five curves for each radius). (E) Each decay curve from D was fit to a 1D decay to estimate the diffusion coefficient. The ratios of the estimated and actual diffusion coefficients used in the simulation were plotted as a function of bouton radius, illustrating the underestimation of diffusion with increasing bouton size. An exponential fit exhibited a space constant of 72 nm. The region shaded in yellow represents the typical range of bouton sizes. Note that the r ¼ 100 nm case did not produce a precise match to the true D value because of numerical error in the simulation using a voxel size of 20 nm. (F) Simulated decay curves (normalized to initial pGFP concentration) for pGFP photoactivation in the axon (black) and synapse (blue) using D ¼ 4 mm2/s and a bouton radius of 200 nm. of culture, this medium was supplemented with 2 mM cytosine b-D-arabinofuranoside hydrochloride. Cells were transfected using the calcium phosphate coprecipitation method after 7 days in vitro with complexin-GFP alone, or together with VAMP::mCherry or TeNT light chain (plasmid provided by M. Dong, Harvard Medical School). Experiments were performed at 14–21 days in vitro (7–14 days posttransfection). Coverslips were mounted in a laminar flow-perfusion and stimulation chamber on the stage of a custom-built laser-illuminated epifluorescence microscope. Cells were illuminated using solid-state diode pumped 488 nm and 561 nm laser illumination (100 mW, Coherent Saphire, Santa Clara, CA), shuttered with accoustooptic modulation. A 40X 1.3NA Fluor Zeiss objective was combined with a 1.6 Optivar tube lens, and appropriate dichroic and emission filters from Chroma (Bellows Falls, VT) (GFP, dichroic: zt488, emission: 510–560 nm; mcherry, dichroic: z561, emission: 580–630 nm). Live cell images were acquired with an Andor iXonþ (model DU-897ECS0-#BV) camera. Fluorescence dispersion recordings were performed at 2 Hz frame rate. During imaging, cells were maintained at 30
C and continuously perfused at ~200 mL/min with basal buffered saline (119 mM NaCl, 2.5 mM KCl, 2 mM CaCl2, 25 mM HEPES, 30 mM glucose at pH 7.4) containing 10 mM 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) and 50 mM D,L-2-amino-5-phosphonovaleric acid (AP5) to supBiophysical Journal 108(6) 1318–1329

press postsynaptic responses. Cells were depolarized with field potentials of ~10 V/cm for 1 msec via platinum-iridium electrodes. All reagents were from Sigma-Aldrich (Natick, MA) except when noted.

Hippocampal culture image analysis Images were analyzed in Image J (http://rsb.info.nih.gov/ij) by using custom written plugins (http://rsb.info.nih.gov/ij/plugins/time-series.html). 2 mm circular regions of interest (ROIs) were placed on visible varicosities based on the subtracted image of the 10 last images during stimulation from the 10 images before stimulus. Only boutons that did not split or merge, remained in focus and had similar responses throughout all trials that were selected for analysis (between 15 and 70 ROIs per cell). Fluorescence values were systematically corrected for background measured on surrounding nontransfected regions of the field. Changes in fluorescence (DF) were normalized to the starting fluorescence (F0) at individual ROIs. For each cell, measurements from three trials were averaged together to increase the signal/noise before fitting. Time-constants and amplitudes were determined by fitting the curves to a single exponential decay equation using OriginPro (OriginLab, Northampton, MA).

Complexin Dynamics at the Synapse

1321

A

C

B

D

E

H

I

F

G

J

FIGURE 3 Complexin mobility is restricted at synapses. (A) Kymographs of mCherry-Rab3 and CPX-1 fused to pGFP (CPX-pG) line scans collected every 2 msec for 6 s. Three en passant synapses along a 10 mm region of the axon are identified using Rab3 fluorescence (gray arrowheads). A brief photoactivation of the middle synapse (#2, purple arrowhead) produced a transient increase in CPX-pG fluorescence that dispersed throughout the axon and accumulated at neighboring synapses. (B) Time course of 1-mm spatial averages for each of the three synapses as indicated. (C) Synaptic fluorescence decay (uncorrected) for pGFP (gray) and CPX-pG (black) indicating the lower mobility of CPX-pG within the synapse, perhaps due to CPX binding to SV membrane as shown in the cartoon. (D) Average fluorescence decay (5SE) for axonal (black) and synaptic (blue) CPX-pG. For comparison, the average axonal pGFP decay is shown in purple. Based on the slower axonal mobility of CPX-pG compared to pGFP (diffusion coefficients of 2 and 4 mm2/s, respectively), simulations of synaptic CPX-pG generate the decay curve shown in red. Note that CPX-pG decays much more slowly than predicted (arrow). (E) Single trial decay curves for pGFP (gray) and CPX-pG (black) are overlaid with 1D diffusion fits (blue and red, respectively) sampled once every 2 msec over a 5 s interval. (F) Long timescale measurements made every 40 msec for pGFP (gray) and CPX-pG (black) together with 1D diffusion fits. (G) Average values of the decay time constants based on 1D diffusion using fast (black) and slow (gray) sampling rates. The number of synapses is indicated in the bars. (H) Schematic of an en passant synapse with immobile binding sites (orange). CPX (black) diffuses in and out of the bouton and binds to the synaptic sites with dissociation constant KD and concentration Btot. This scheme was used to develop a simple reaction-diffusion model in Virtual Cell. (I) Photoactivation in a single bouton using a binding model with five species. The fluorescent species (blue) diffuse out of the bouton, whereas nonfluorescent species (gray) diffuse into the bouton after photoactivation at t ¼ 0. The measured fluorescence was modeled as the sum of both fluorescent species (green) and compared to measured decay traces to estimate best fitting parameters. Unbound species (dotted lines) exchanged rapidly with the same time course as pGFP, whereas bound species (solid lines) exchanged slowly and were limited by the off-rate of binding. (J) The reaction-diffusion model adequately fits the example decay traces for both CPX-pG (black) and pGFP (gray) adjusting only the binding capacity (Btot/KB) and using a common Dapp of 4 mm2/s for both pGFP and CPX-pG. See Materials and Methods and Fig. S2 for details. Data are mean 5 SE.

Biophysical Journal 108(6) 1318–1329

1322

A

Wragg et al.

B

C

D

E

F

Correction of the pGFP photoswitching artifact For the fast sampling imaging in Fig. S2 A, each pG-Gyrin decay curve was fit to a normalized double exponential decay of the form: gðtÞ ¼ 1  a1  a3 þ a1 ea2 t þ a3 ea4 t : The fit parameters were averaged to create an average fit to the pG-Gyrin ensemble, and this curve is superimposed on several pG-Gyrin decay curves in Fig. S2 B (red curves). The average values from n ¼ 26 synapses were: a1 ¼ 0.13, a2 ¼ 0.62, a3 ¼ 0.3, and a4 ¼ 6.87. Because pG-Gyrin is relatively immobile on this timescale, g(t) represents the dynamics of pGFP photoswitching over several seconds following a 2 msec irradiation with 405 nm light. Individual pGFP traces were corrected by dividing the normalized fluorescence time course by the average double exponential decay to cancel out the photoswitching contribution. This correction removes a fast component of the decay kinetics from all traces, resulting in a Biophysical Journal 108(6) 1318–1329

FIGURE 4 Complexin is retained in synapses through multiple types of interactions. (A) Schematic of six optical probes used for these experiments. CH is the central helix and CTD is the C-terminal domain. KY/AA is a double point mutant within the CH domain that impairs SNARE binding. Probes are ordered from highest to lowest binding (top to bottom), as predicted from prior studies. (B) Cartoon of two binding modes for complexin at the synapse. The CTD (black) interacts with SVs, whereas the CH (gray) interacts with SNARE complexes. (C) Representative single trials of CPX-pG (gray), DCT-pG (blue), and pGFP (black) synaptic decay curves. pGFP/mCherry ratios were computed for each decay trace and then normalized to the initial value. Data are fit with a 1D diffusion decay curve with characteristic decay time constants as indicated. (D) Average tD values for each of the optical probes. (E) Residual fluorescence at 25 s normalized to initial fluorescence for each CPX variant. Note that pG-Gyrin is included for comparison (magenta). (F) Average values of the log of the dissociation constant (in micromolar) are plotted for each pGFP probe. Traces were fit to a 3D reaction diffusion model with immobile binding sites within the synaptic bouton and Dapp of either 2 or 4 mm2/s. Best-fit estimates of the dissociation constant were generated for each of the CPX probes, allowing only the binding site concentration and the dissociation constant to vary, (see Materials and Methods and Fig. S3). Data are mean 5 SE and number of synapses is indicated in (E). **differs from CPX-pG with p < 0.01, *differs from CPX-pG with p < 0.05. n.s., not significant. # differs from all other species with p < 0.01. Significance was determined using the Tukey Kramer test. Sample numbers indicated in (E).

flat profile for pG-Gyrin and a slower decay rate for pGFP. Using a slower sampling rate and recording fluorescence decays over longer timescales, the pG-Gyrin decay behaved like a single exponential process (Fig. S2 A, blue curve), presumably because the fast decay was undersampled. The single exponential correction factor was defined as: gðtÞ ¼ a1 þ ð1  a1 Þea2 t : The long timescale traces (25–40 s) were corrected using the average values of a1 and a2, 0.78 and 2.94, respectively, for n ¼ 18 synapses.

Pure diffusion in a three-dimensional geometry Three dimensional (3D) diffusion models were implemented in Virtual Cell (22) using a 50 mm long 200 nm diameter cylinder with sealed ends oriented along the y axis. Centered on this cylinder, a synaptic bouton was created using an ellipsoidal volume defined by ðx=rÞ2 þ ðy=lÞ2 þ ðz=rÞ2 %1;

Complexin Dynamics at the Synapse

1323

A

E

C

B

F

G

D

H

FIGURE 5 Chronic activity mutants bidirectionally alter complexin mobility. (A) Mean 5 SE decay curves for synaptic CPX-pG in WT (black), unc-64 Syntaxin mutant (red), and tom-1 Tomosyn mutant (blue) animals. (B) Individual traces were fit to 1D diffusion and average tD values are shown for each genotype. Both synaptic mutants significantly differ from WT. (C) Mean 5 SE synaptic decay curves for a C-terminal truncation of CPX (DCT-pG) are shown for each genotype. (D) Diffusion time constants for DCT-pG indicate that there is no significant change in mobility for unc-64 Syntaxin mutants and only a minor increase for tom-1 Tomosyn mutants. (E) Mean 5 SE decay curves for pGFP in each genotype. (F) Average time constant for pGFP is the same in both mutant backgrounds. The number of synapses is indicated in the bars for each genotype. **p < 0.01, *p < 0.05 by Student’s t-test. (G) Cartoon of the SV cycle (red arrows) depicting CPX (blue) binding and unbinding from SVs. If SV fusion drives the release of bound CPX (orange arrows), CPX mobility will be coupled to the SV cycle. (H) Individual CPX-pG decay traces from WT (black), unc-64 Syntaxin mutant (red), and tom-1 Tomosyn mutant (blue) animals are overlaid with simulated decays for three concentrations of binding sites using Dapp ¼ 2 mm2/s. Model binding site abundance is indicated on the right. Log(KD) values are 2.4, 2.7, and 2 for WT, unc-64, and tom-1, respectively.

where r is the radius of the bouton and l is the length of the bouton along the axon. For the simulations presented in this study, r was varied from 100 to 500 nm, whereas l was kept constant at 1 mm and the diffusion coefficient was varied over a range (1–8 mm2/s). The bouton volume was then copied to create six neighboring boutons of identical size, three on either side of the central bouton symmetrically spaced at 3 mm apart. Physiologically, en passant synapses are typically 1 mm long with a radius of 150–250 nm and spaced 3 mm apart on average (23,24). For synaptic photoactivation, the diffusing species was initially set to a value of 1.0 within the central bouton volume and zero elsewhere. The average concentration within the central bouton was computed as a function of time for each value of r and D. Decay curves were fit to a simple one-dimensional (1D) analytical solution used to estimate the diffusion time constant tD in Figs. 1–5:

CðtÞ ¼ erf


pffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffi t D =t þ t=pt D , ðexpðt D =tÞ  1Þ;

where tD ¼ w2 =D and w is the half-width of the rectangular region over which the average fluorescence was computed. This simplification ignores details of the 405 nm illumination beam profile or initial distribution of

pGFP within the bouton. The expression was normalized to its value at the first experimental sample point (t0): nðtÞ ¼ CðtÞ=Cðt0 Þ: During fast line scan acquisition, t0 was 2 msec, whereas for slow acquisition (longer timescale imaging), t0 was 50 msec. Simulations were run using a fully implicit finite volume regular grid with a uniform voxel size of 20 nm. In the case of a simple cylinder (r ¼ 100 nm), analytical solutions could be directly compared to the numerical simulations. A small numerical error produced a diffusion coefficient that was 3.8% smaller than the analytical value Dapp, providing an estimate for the numerical error present in all simulations presented in this study using a 20 nm mesh (Fig. 2 E). For comparisons with the data, the t ¼ 0 points were removed from the simulations and the traces were normalized to the value at t0.

Modeling the impact of CPX binding on its mobility For the immobile binding site model, five species were incorporated into the geometry described previously: free fluorescent CPX (CF), bound fluorescent CPX (CFB), free nonfluorescent CPX (C), bound nonfluorescent CPX (CB), and unoccupied binding sites (Bfree). The fluorescent and Biophysical Journal 108(6) 1318–1329

1324

Wragg et al.

nonfluorescent species participated in identical bimolecular reactions: CF þ Bfree # CFB and C þ Bfree # CB with forward rate constants kon and reverse rate constants koff. CFB, CB, and Bfree were all assigned a diffusion coefficient of zero and their location was restricted to the synaptic boutons. In contrast, the two unbound CPX species could freely diffuse throughout the axon. The apparent diffusion coefficient Dapp was fixed to either 4 mm2/s (based on pGFP) or 2 mm2/s (based on CPX-pG) and the bouton radius was fixed at 200 nm. These restrictions left a total of four free parameters: the total CPX concentration in the axon (Ctot), the total binding site concentration in a bouton (Btot), and the forward and reverse rates for the binding reaction (kon and koff). The dissociation constant was defined as KD ¼ koff =kon : After running simulations over a range of values for all four parameters, Ctot was fixed to 12 mM and kon set to 5 mM1 s1. Btot and KD were systematically varied to determine least square fits for all decays shown in Fig. 4 as well as Fig. S3 and Table S1. Average values for each CPX variant were computed and the average Log(KD) values are shown in Fig. 4 F with KD in units of mM. The middle bouton (position y ¼ 0) was chosen for photoactivation, so the initial C and CB concentrations were set to 0 within this ellipsoid because at time t ¼ 0, all species were fully converted to their fluorescent forms within this volume. CFB was computed using the bimolecular equilibrium formula where a is the synaptic volume fraction: total synapse volume/(total synapse volume þ axon volume):

CF B ¼

 1 aBtot þ Ctot þ KD 2a qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2  ðaBtot þ Ctot þ KD Þ  4aBtot Ctot ;

CF ¼ Ctot – aCFB and Bfree ¼ Btot – CFB were used to compute the other two species.

 1 a ¼ 1 þ 1:5 , ðL=N  1Þ , ða=rÞ2 ; where L is the total axon length, N is the total number of synaptic boutons, a is the axon radius, and r is the bouton radius (as above). For a 50 mm axon of radius 100 nm containing seven boutons, each of radius 200 nm, the synaptic volume fraction is ~0.3. In some simulations, a 24 mm axon was used to expedite computations. For L ¼ 24 mm, the volume fraction is ~0.52, but the decay kinetics were similar to the 50 mm case. In the nonphotoactivated boutons, the same formulas were used for C and CB, whereas CF and CFB were set to zero. As in the pure diffusion simulations, the t ¼ 0 point was removed from the decay traces and they were normalized to their values at t0 for comparison with the measured data.

RESULTS Assessing synaptic protein mobility in living animals using pGFP To examine the exchange of small cytoplasmic proteins between en passant synapses, pGFP was expressed in a small number of motor neurons in C. elegans (Fig. 1 A). GFP is a compact 27 kDa protein with a hydrodynamic radius of 2.3 nm (25) and thus serves as a model small protein for investigating mobility in the absence of specific binding interactions. pGFP is rapidly converted from a dim to a bright fluorescent state (using 488 nm excitation) upon illumination with 405 nm light (26,27). Individual synaptic boutons in a living intact animal were located using mCherry::RAB-3 fluorescence and briefly irradiated with Biophysical Journal 108(6) 1318–1329

a low intensity 405 nm light beam to create a fluorescent population of pGFP (Fig. 1, B–E). Synaptic pGFP fluorescence rapidly decayed with a half-decay time of a few hundred milliseconds, whereas neighboring synaptic boutons captured some of the mobile pGFP (Fig. 1, C and D). Photoactivation was restricted to a 1 mm region (Fig. 1 E) and the synaptic decay was reproducibly observed at single synapses upon repeated photoactivation indicating minimal photodamage (Fig. 1 F). Single trial fluorescence decay transients were fit to a simple 1D diffusion function with single fit parameter defined as tD (Fig. 1 G, see Materials and Methods for details). For 1D diffusion, tD is approximately the time required for 50% of the protein to escape the 1 mm region of axon. The mobility of pGFP was compared to relatively immobile proteins such as the SV transmembrane protein Synaptogyrin to control for bleaching and pGFP photoswitching dynamics (Fig. S1 A). Unexpectedly, Synaptogyrin-pGFP (Gyrin-pG) displayed a spontaneous decrease in fluorescence during the first few hundred milliseconds after photoactivation producing a decay component independent of its mobility, possibly arising from a transient bright fluorescent state (Fig. S1, B–E) (28). Correction of the mobility-independent photoswitching fluorescence transient To better characterize and correct the photoswitching transient associated with photoactivation, Gyrin-pG fluorescence decay traces were measured over longer time durations as well as at high temporal resolution (Fig. S2 A). Individual Gyrin-pG decay transients were normalized to the average decay, thereby eliminating the mobility-independent decay component (Fig. S2 B and Materials and Methods). As an independent verification of the correction strategy, conventional GFP was expressed in single axons and fluorescence recovery following GFP photobleaching was compared to either raw or corrected pGFP decay transients. Although the raw pGFP transient relaxed significantly faster than GFP recovery from photobleaching, the corrected pGFP transient closely matched the GFP transient (Fig. S2 C). Correction of pGFP transients resulted in a twoto threefold slower estimate of mobility due to the removal of a fast decay component (Fig. S2, D and E). Based on these results, all subsequent pGFP measurements in this study have been corrected for the photoswitching transient. Contributions of synaptic bouton morphology to apparent protein mobility The mobility of pGFP was measured either within synaptic boutons or in the axon between synapses (Fig. 2 A). Unexpectedly, pGFP escaped the synapse with a five- to sixfold lower mobility compared to an equivalent region along the axon between synaptic boutons (Fig. 2, A and B). The

Complexin Dynamics at the Synapse

apparent mobility difference between axons and boutons was independent of whether the photoswitching correction was applied (Fig. S2 F). Furthermore, fits to 1D diffusion generally described the axonal fluorescence decay better than the synaptic bouton fluorescence decay, as seen for example in Fig. 2 A. The 1D approximation fails to account for the significantly larger bouton volume (typically between 0.3 and 0.5 mm in diameter) relative to the surrounding axon (0.1–0.2 mm in diameter), so these differences could arise from a simple geometric effect. Using Virtual Cell (22), we developed a rudimentary but realistic 3D diffusion model of en passant synapses (Fig. 2 C). As the bouton radius was increased in the model, clearance of pGFP out of the synaptic bouton became progressively slower (Fig. 2, D and E), whereas clearance out of a 1 mm region of neighboring axon was unaffected (data not shown). Electron microscopy studies of presynaptic bouton morphology indicate that en passant synapses typically have a bouton radius of 200 nm in both C. elegans and in rat hippocampus (23,24,29). Using this radius value along an axon with a 100 nm radius, the synaptic and axonal decay transients were fit with an apparent diffusion coefficient of 3–5 mm2/s (Figs. 1 G and 2 F). The surprisingly small value for the diffusion coefficient of GFP, which is ~80–90 mm2/s in water, is consistent with a high degree of molecular crowding, viscosity, and tortuosity observed within small compartments such as synapses (30–34). Thus, protein mobility within the axonal compartment can be assessed quantitatively after accounting for pGFP photoswitching and bouton geometry. CPX is retained in synaptic boutons much longer than pGFP We next asked whether the 18 kDa cytoplasmic synaptic protein complexin (CPX-1) also rapidly exchanged between synapses. CPX-1 was tagged with pGFP (CPX-pG) and imaged using the same approaches described previously (Fig. 3, A and B). CPX-pG is a functional fusion protein and can fully replace endogenous CPX in null mutant animals (7,8,12). As shown in Fig. 3 C, CPX-pG exchanged more slowly compared to pGFP, perhaps due to interactions with SVs and SNARE proteins (9,10,12,35). Surprisingly, the mobility of CPX-pG in axons (where SV and SNARE binding should be minimal) was significantly lower than that of pGFP alone (Fig. 3 D) with an apparent diffusion coefficient of 2 rather than 4 mm2/s. The slower diffusion of CPX-pG in both compartments could arise from several contributions. First, the C-terminal domain of CPX-1 is known to bind membranes, suggesting that CPX-pG movement could be restricted by protein-lipid interactions with the axonal plasma membrane or other membrane-bound organelles within the axon. Second, CPX-pG is 67% larger (by mass) than pGFP and larger particles diffuse more slowly. For a globular protein, a 67% increase in mass is predicted to cause a 19% decrease in diffusion, so the observed

1325

50% decrease in axonal diffusion is too large to be accounted for by size considerations alone (25,36). Could the slower axonal diffusion of CPX-pG account for the slower synaptic exchange? Using the 3D bouton diffusion model developed previously with a diffusion coefficient of 2 mm2/s, the synaptic fluorescence decay of CPX-pG was simulated and compared to the measured synaptic decay (Fig. 3 D, red versus blue). The large discrepancy between predicted and measured mobility indicates that CPX-pG mobility is even further restricted within synaptic boutons. Fluorescence decay was measured over either a 5 s period with millisecond time resolution or a 30 s period with a slower sampling rate to examine the difference between pGFP and CPX-pG mobility on longer timescales (Fig. 3, E and F). 1D diffusion fits to the CPX-pG decay transient were systematically slower when measured over the longer time interval (Fig. 3 G), perhaps due to undersampling the fastest component of decay. Regardless of the measurement interval, synaptic CPX-pG exchanged almost 10-fold more slowly than pGFP alone (Fig. 3 G). Unlike GFP, synaptic proteins participate in specific biochemical processes at the synapse. A simple explanation for the lower mobility of synaptic CPX-pG is that binding interactions stabilize complexin within the synapse. The effects of CPX-pG binding to immobile sites within the synaptic bouton were explored using a modification of the 3D diffusion model described previously (Fig. 3 H). When the model was constrained to fit both short and long timescale measurements using an apparent diffusion coefficient of 4 mm2/s for pGFP and 2 mm2/s for CPX-pG, a simple bimolecular reaction was sufficient to account for the observed decay time course of CPX. According to this scheme, the rapid decay of CPX-pG over the first several seconds is dominated by unbound CPX-pG, whereas the slow decay is set largely by the low unbinding rate of CPX-pG from synaptic binding sites (Fig. 3 I). If the binding site concentration was lowered to zero, the decay kinetics matched pGFP alone (Fig. 3 J). Multiple interactions restrain CPX within the synapse To test the hypothesis that CPX-pG is retained in synaptic boutons via binding interactions with either SVs or SNAREs, mobility was measured for a series of mutated CPX-pG variants where single or multiple binding interactions were selectively disrupted (Fig. 4, A and B). Previous studies have characterized the binding of CPX to assembled SNARE complexes (11,14,37) and liposomes (10,12,35). Perturbing SV binding significantly increased CPX-pG mobility and decreased the residual fraction of synaptic protein (Fig. 4, C–E). Elimination of both SNARE and vesicle binding rendered the remaining peptide indistinguishable from pGFP alone by either measure. As described previously, the measured increases in mobility were much larger than those expected from probe size alterations (25). The Biophysical Journal 108(6) 1318–1329

1326

fluorescence measurements crudely separated the probes into two distinguishable groups: a slow group comprising full-length CPX, KY/AA, and DCH, and a fast group consisting of DCT, DCHDCT, and pGFP (Fig. 4, D and E). Using the 3D model with immobile binding sites, the apparent binding affinity and binding site concentration were estimated for each of the CPX probes assuming an apparent diffusion coefficient of either 2 or 4 mm2/s for the slow and fast groups, respectively (Figs. 4 F and S3). Estimated binding site concentration was largely unaffected by the domain mutations, whereas the apparent dissociation constant varied over four orders of magnitude as might be expected for perturbations of the ligand (Fig. S3, I–K). The apparent dissociation constants estimated here do not accurately reflect absolute binding affinities because the underlying model is highly simplified. For instance, the simulations ignore variability of synaptic morphology, binding site mobility, and variation in binding affinity between synapses. Nevertheless, these parameters provide a quantitative measure of relative changes in binding between CPX probes, demonstrating that this approach can detect binding interactions within the synapse. SV exocytosis modulates CPX mobility SVs and SNAREs continuously undergo cycles of exo- and endocytosis at living synapses. The binding substrate of CPX is therefore a dynamic variable in these imaging assays. Several proposed mechanisms of CPX action entail displacement of CPX before SV fusion (11,38,39). Such fusion-state dependent binding predicts that synaptic activity will alter the mobility and localization of synaptic CPX. To examine this issue, CPX-pG decay time courses were measured in a severe exocytosis mutant where both stimulus-evoked and spontaneous synaptic transmission are decreased by >90% (8,40,41). In unc-64 Syntaxin 1 mutants, CPX-pG decay was significantly slower than wildtype (WT) (Fig. 5, A and B). To determine if the effect of SV exocytosis on mobility is bidirectional, CPX-pG was imaged in tom-1 Tomosyn mutants where loss of an endogenous SNARE inhibitor causes a doubling of calcium-triggered SV fusion (42–44). In tom-1 animals, CPX-pG escaped the synapse at twice the WT rate (Fig. 5, A and B). The dependence of protein mobility on SV cycling might arise directly through interactions between CPX and SVs or indirectly through morphological or ultrastructural changes within the bouton involving cytoskeletal architecture and membrane-bound organelles (45–47). These possibilities were tested by examining the mobility of a truncated CPX-pG missing its C-terminal vesicle tether (DCT-pG) and pGFP alone. In the absence of SV binding, CPX diffusion no longer depended strongly on SV cycling (Fig. 5, C and D). Furthermore, pGFP escaped the synapse with similar kinetics in WT, unc-64 Syntaxin mutants, and tom-1 Tomosyn mutants (Fig. 5, E and F), suggesting that there were no major changes Biophysical Journal 108(6) 1318–1329

Wragg et al.

in synaptic ultrastructure or morphology in these synaptic mutants. Thus, CPX mobility is altered through direct interactions with SVs rather than through global changes in protein mobility. We modeled increased synaptic activity as a decrease in the effective number of binding sites using the diffusion and binding scheme developed in this study (Fig. 5, G and H). The basic features of CPX dynamics and the impact of synaptic activity were captured with a minimal number of assumptions and parameters. Synaptic stimulation acutely dispersed hippocampal CPX-GFP through SV cycling The pGFP kinetics measurements in chronic synaptic transmission mutants indicate that the activity state of the synapse contributes to the capture and retention of CPX through its binding interactions. One prediction of this model is that acute increases in synaptic activity will displace complexin and lower its synaptic abundance. To test this prediction, GFP-tagged mammalian complexin 1 (CPX-GFP) was expressed in cultured hippocampal neurons and its synaptic abundance was quantified before, during, and after stimulation. Rodent CPX-GFP and the presynaptic SV marker VAMP-mCherry showed strong colocalization in live neurons at rest (Fig. 6 A), consistent with the presynaptic localization of mammalian CPX (2,13). As predicted in the simple binding model, synaptic CPX-GFP fluorescence decreased, whereas interbouton axonal CPX-GFP concomitantly increased during a 10 Hz stimulus train (Fig. 6, B and C). This transient redistribution then recovered to resting levels on a timescale of 3 min. Similar activity-dependent dispersion has been described previously for the SV associated protein synapsin 1, and may be a general feature of cytoplasmic proteins that bind to SVs in an activity-dependent manner (48). These observations support the notion that CPX localization and mobility are regulated by synaptic activity. The role of SV cycling was examined in neurons coexpressing tetanus toxin light chain (TeNT-LC). TeNT eliminates SV fusion by cleaving the vSNARE VAMP2 (49,50). In TeNT-treated boutons, CPX-GFP no longer dispersed upon stimulation (Fig. 6, D and E). Because cleavage of VAMP2 does not significantly impact presynaptic calcium influx (51,52), CPX-GFP dispersion is not driven directly by elevated intracellular calcium. In fact, CPX accumulated at boutons to a small extent following stimulation in the absence of SV cycling. Thus, redistribution of synaptic CPX requires SV fusion in both worm and mammalian synapses. DISCUSSION Mechanisms underlying use-dependent complexin binding How does synaptic activity alter the localization and mobility of CPX? This study shows that CPX is sequestered within

Complexin Dynamics at the Synapse

Implications for use-dependence plasticity

A

B

D

1327

C

E

FIGURE 6 Dispersion of synaptic CPX-GFP during acute stimulation in cultured hippocampal neurons. (A) Colocalization of CPX-GFP (green) and the presynaptic marker VAMP-mCherry (red) in cultured hippocampal neurons. Scale bar ¼ 5 mm. (B) CPX-GFP fluorescence at rest (top) and during a train of 300 AP at 10 Hz (bottom). Note the decreased fluorescence in boutons and increased fluorescence in neighboring regions of axon upon stimulation. Scale bar ¼ 2 mm. (C) Time course of average fluorescence intensity (mean 5 SE) in synaptic boutons (red) and axonal regions (gray) from one neuron during a train of 300 AP at 10 Hz. (D) Time course of average synaptic bouton fluorescence (mean 5 SE) for neurons expressing the tetanus toxin light chain (TeNT). (E) Average change in bouton fluorescence following the 10 Hz stimulus train for control (red) and TeNT-expressing (blue) neurons.

synaptic boutons through a combination of geometry and binding in the absence of neuronal activity, whereas high levels of exocytosis release CPX from its binding sites. Furthermore, the mobility of a truncated CPX without its C-terminal domain is no longer highly sensitive to synaptic activity. Based on previous work, at least two mechanisms may underlie this phenomenon. First, the calcium-binding SV protein synaptotagmin has been suggested to displace CPX from the assembling SNARE complex during calcium-triggered exocytosis, and removal of the CPX C-terminal domain enhanced the efficacy of displacement (11,39). Second, the C-terminal domain of CPX was recently shown to bind with greater affinity to highly curved membranes (35). Collapse of the vesicle membrane during fusion could release CPX by decreasing its membrane binding affinity. In both scenarios, removal of the C-terminal domain is predicted to diminish CPX binding, and consequently, activity-dependent CPX mobility. Regardless of the mechanistic details, competition between activity and CPX binding would limit the impact of CPX on subsequent exocytosis.

Short-term synaptic plasticity is a ubiquitous feature of synaptic transmission, although the specific types and time courses of short-term synaptic plasticity differ widely between synapse types (53). Use-dependent increases in synaptic strength following a brief stimulus train are driven by elevated presynaptic calcium and several calcium-binding molecules including synaptotagmin (54), PKC (55), Munc13 (56,57), and calmodulin (57). Likewise, presynaptic short-term depression is thought to arise from SV depletion and a refractory state of release sites (53,58–60). Transient loss of an inhibitor might boost release, leading to a potentiated state. Alternatively, loss of a fusogenic molecule would momentarily weaken synaptic transmission leaving the synapse in a transiently depressed state. Phosphorylation of Munc18 by PKC has been suggested to underlie a form of synaptic potentiation by stabilizing a local pool of phospho-Munc18 thereby increasing release probability (61,62). In the case of complexin, acceleration or inhibition of SV cycling changes the effective binding of synaptic CPX. As CPX has both fusogenic and inhibitory roles in synaptic transmission, the impact of a brief decrease in CPX binding is difficult to predict. Fluctuations in synaptic protein abundance during activity may represent a novel presynaptic mechanism underlying use-dependent plasticity (63). The approaches described in this study provide, to our knowledge, a new set of tools for developing dynamic optical indicators of the molecular state of a synapse. For instance, the degree to which a critical protein is captured within a bouton is an indication of its binding affinity and the number of available binding sites. Alterations of either factor during stimulation may indicate a specific change in the state of the synapse. Although this study focused on CPX, a large number of other synaptic proteins are subject to diffusive redistribution along the axonal compartment over time. Not all synaptic proteins are anchored to the same extent and a wide variety of mobility values or immobile fractions have been reported on several presynaptic proteins (62,64–69). Further studies will be required to determine whether some of the unanchored proteins play a preferential role in use-dependent plasticity because they will have a larger capacity to collect or disperse from individual boutons during ongoing synaptic activity.

CONCLUSIONS Many proteins involved in the regulation of synaptic transmission are not permanently anchored to the lipid bilayer or cytoskeleton and are free to diffuse within and between synaptic boutons. Here, the rate of protein diffusion at en passant synapses was measured using high time resolution imaging of fluorescence redistribution following photoactivation. The studies presented here show that CPX is Biophysical Journal 108(6) 1318–1329

1328

localized to presynaptic boutons via SV and SNARE interactions. We further show that the amount of vesicle cycling determines the rate at which CPX escapes the synapse in both worms and mammals. This activity-dependent exchange of CPX was quantitatively modeled using a simple diffusion scheme with immobile binding sites restricted to the synaptic bouton. Finally, acute neuronal stimulation dispersed synaptic CPX, demonstrating that activity can rapidly alter the binding and dynamic exchange of synaptic CPX. Taken together, these experiments show that neuronal activity, synapse geometry, and binding interactions strongly influence the local exchange and abundance of this critical regulator of neurotransmitter release. SUPPORTING MATERIAL Three figures and one table are available at http://www.biophysj.org/ biophysj/supplemental/S0006-3495(15)00159-9.

ACKNOWLEDGMENTS We thank David DiGregorio and Bruce Bamber and the reviewers for advice and manuscript suggestions, Julia Marrs for help with the hippocampal cultures, and Les Loew, Boris Slepchenko, and Ronnie Yang for their expertise and assistance with Virtual Cell modeling. This work was supported by National Institutes of Health grants R01-GM095674 (J.S.D.), Basil O’Connor, and the Rita Allen Foundation Award 188423 (J.S.D.) and R01-MH085783 (T.A.R.).

REFERENCES

Wragg et al. 12. Wragg, R. T., D. Snead, ., J. S. Dittman. 2013. Synaptic vesicles position complexin to block spontaneous fusion. Neuron. 77:323–334. 13. Lin, M. Y., J. G. Rohan, ., R. H. Chow. 2013. Complexin facilitates exocytosis and synchronizes vesicle release in two secretory model systems. J. Physiol. 591:2463–2473. 14. Giraudo, C. G., A. Garcia-Diaz, ., J. E. Rothman. 2008. Distinct domains of complexins bind SNARE complexes and clamp fusion in vitro. J. Biol. Chem. 283:21211–21219. 15. Kaeser-Woo, Y. J., X. Yang, and T. C. Su¨dhof. 2012. C-terminal complexin sequence is selectively required for clamping and priming but not for Ca2þ triggering of synaptic exocytosis. J. Neurosci. 32: 2877–2885. 16. Regehr, W. G., and D. W. Tank. 1994. Dendritic calcium dynamics. Curr. Opin. Neurobiol. 4:373–382. 17. Tank, D. W., W. G. Regehr, and K. R. Delaney. 1995. A quantitative analysis of presynaptic calcium dynamics that contribute to shortterm enhancement. J. Neurosci. 15:7940–7952. 18. DiGregorio, D. A., A. Peskoff, and J. L. Vergara. 1999. Measurement of action potential-induced presynaptic calcium domains at a cultured neuromuscular junction. J. Neurosci. 19:7846–7859. 19. Sabatini, B. L., T. G. Oertner, and K. Svoboda. 2002. The life cycle of Ca(2þ) ions in dendritic spines. Neuron. 33:439–452. 20. Brenner, S. 1974. The genetics of Caenorhabditis elegans. Genetics. 77:71–94. 21. Ryan, T. A. 1999. Inhibitors of myosin light chain kinase block synaptic vesicle pool mobilization during action potential firing. J. Neurosci. 19:1317–1323. 22. Moraru, I. I., J. C. Schaff, ., L. M. Loew. 2008. Virtual Cell modelling and simulation software environment. IET Syst. Biol. 2:352–362. 23. Palay, S. L., and V. Chan-Palay. 1974. Cerebellar Cortex: Cytology and Organization. Springer Verlag, Berlin. 24. Shepherd, G. M., and K. M. Harris. 1998. Three-dimensional structure and composition of CA3—>CA1 axons in rat hippocampal slices: implications for presynaptic connectivity and compartmentalization. J. Neurosci. 18:8300–8310.

1. Su¨dhof, T. C., and J. Rizo. 2011. Synaptic vesicle exocytosis. Cold Spring Harb. Perspect. Biol. 3:3.

25. Hink, M. A., R. A. Griep, ., A. J. Visser. 2000. Structural dynamics of green fluorescent protein alone and fused with a single chain Fv protein. J. Biol. Chem. 275:17556–17560.

2. McMahon, H. T., M. Missler, ., T. C. Su¨dhof. 1995. Complexins: cytosolic proteins that regulate SNAP receptor function. Cell. 83: 111–119.

26. Patterson, G. H., and J. Lippincott-Schwartz. 2002. A photoactivatable GFP for selective photolabeling of proteins and cells. Science. 297:1873–1877.

3. Giraudo, C. G., W. S. Eng, ., J. E. Rothman. 2006. A clamping mechanism involved in SNARE-dependent exocytosis. Science. 313: 676–680.

27. Patterson, G. H., and J. Lippincott-Schwartz. 2004. Selective photolabeling of proteins using photoactivatable GFP. Methods. 32:445–450.

4. Xue, M., K. Reim, ., C. Rosenmund. 2007. Distinct domains of complexin I differentially regulate neurotransmitter release. Nat. Struct. Mol. Biol. 14:949–958. 5. Su¨dhof, T. C., and J. E. Rothman. 2009. Membrane fusion: grappling with SNARE and SM proteins. Science. 323:474–477. 6. Huntwork, S., and J. T. Littleton. 2007. A complexin fusion clamp regulates spontaneous neurotransmitter release and synaptic growth. Nat. Neurosci. 10:1235–1237. 7. Hobson, R. J., Q. Liu, ., E. M. Jorgensen. 2011. Complexin maintains vesicles in the primed state in C. elegans. Curr. Biol. 21:106–113. 8. Martin, J. A., Z. Hu, ., J. S. Dittman. 2011. Complexin has opposite effects on two modes of synaptic vesicle fusion. Curr. Biol. 21:97–105. 9. Chen, X., D. R. Tomchick, ., J. Rizo. 2002. Three-dimensional structure of the complexin/SNARE complex. Neuron. 33:397–409. 10. Seiler, F., J. Malsam, ., T. H. So¨llner. 2009. A role of complexin-lipid interactions in membrane fusion. FEBS Lett. 583:2343–2348. 11. Tang, J., A. Maximov, ., T. C. Su¨dhof. 2006. A complexin/synaptotagmin 1 switch controls fast synaptic vesicle exocytosis. Cell. 126: 1175–1187. Biophysical Journal 108(6) 1318–1329

28. Duan, C., V. Adam, ., D. Bourgeois. 2014. Structural basis of photoswitching in fluorescent proteins. Methods Mol. Biol. 1148:177–202. 29. Kittelmann, M., J. Hegermann, ., S. Eimer. 2013. Liprin-a/SYD-2 determines the size of dense projections in presynaptic active zones in C. elegans. J. Cell Biol. 203:849–863. 30. Terry, B. R., E. K. Matthews, and J. Haseloff. 1995. Molecular characterisation of recombinant green fluorescent protein by fluorescence correlation microscopy. Biochem. Biophys. Res. Commun. 217:21–27. 31. Brown, E. B., E. S. Wu, ., W. W. Webb. 1999. Measurement of molecular diffusion in solution by multiphoton fluorescence photobleaching recovery. Biophys. J. 77:2837–2849. 32. Ando, T., and J. Skolnick. 2010. Crowding and hydrodynamic interactions likely dominate in vivo macromolecular motion. Proc. Natl. Acad. Sci. USA. 107:18457–18462. 33. McGuffee, S. R., and A. H. Elcock. 2010. Diffusion, crowding & protein stability in a dynamic molecular model of the bacterial cytoplasm. PLOS Comput. Biol. 6:e1000694. 34. Luby-Phelps, K. 2013. The physical chemistry of cytoplasm and its influence on cell function: an update. Mol. Biol. Cell. 24:2593–2596. 35. Snead, D., R. T. Wragg, ., D. Eliezer. 2014. Membrane curvature sensing by the C-terminal domain of complexin. Nat. Commun. 5:4955.

Complexin Dynamics at the Synapse

1329

36. Berg, H. C., and E. M. Purcell. 1977. Physics of chemoreception. Biophys. J. 20:193–219.

T. C. Sudhof, and C. F. Stevens, editors. The Johns Hopkins University Press, Baltimore, MD.

37. Liu, J., T. Guo, ., S. F. Sui. 2006. Complexin is able to bind to SNARE core complexes in different assembled states with distinct affinity. Biochem. Biophys. Res. Commun. 347:413–419.

54. Sun, J., Z. P. Pang, ., T. C. Su¨dhof. 2007. A dual-Ca2þ-sensor model for neurotransmitter release in a central synapse. Nature. 450:676–682.

38. An, S. J., C. P. Grabner, and D. Zenisek. 2010. Real-time visualization of complexin during single exocytic events. Nat. Neurosci. 13: 577–583. 39. Xu, J., K. D. Brewer, ., J. Rizo. 2013. Subtle Interplay between synaptotagmin and complexin binding to the SNARE complex. J. Mol. Biol. 425:3461–3475. 40. Saifee, O., L. Wei, and M. L. Nonet. 1998. The Caenorhabditis elegans unc-64 locus encodes a syntaxin that interacts genetically with synaptobrevin. Mol. Biol. Cell. 9:1235–1252. 41. Wang, Z. W., O. Saifee, ., L. Salkoff. 2001. SLO-1 potassium channels control quantal content of neurotransmitter release at the C. elegans neuromuscular junction. Neuron. 32:867–881. 42. McEwen, J. M., J. M. Madison, ., J. M. Kaplan. 2006. Antagonistic regulation of synaptic vesicle priming by Tomosyn and UNC-13. Neuron. 51:303–315. 43. Gracheva, E. O., A. O. Burdina, ., J. E. Richmond. 2006. Tomosyn inhibits synaptic vesicle priming in Caenorhabditis elegans. PLoS Biol. 4:e261. 44. Gracheva, E. O., A. O. Burdina, ., J. E. Richmond. 2007. Tomosyn negatively regulates both synaptic transmitter and neuropeptide release at the C. elegans neuromuscular junction. J. Physiol. 585:705–709. 45. Phillips, G. R., J. K. Huang, ., D. R. Colman. 2001. The presynaptic particle web: ultrastructure, composition, dissolution, and reconstitution. Neuron. 32:63–77. 46. Triller, A., and H. Korn. 1985. Activity-dependent deformations of presynaptic grids at central synapses. J. Neurocytol. 14:177–192. 47. Denker, A., K. Kro¨hnert, ., S. O. Rizzoli. 2011. The reserve pool of synaptic vesicles acts as a buffer for proteins involved in synaptic vesicle recycling. Proc. Natl. Acad. Sci. USA. 108:17183–17188. 48. Chi, P., P. Greengard, and T. A. Ryan. 2001. Synapsin dispersion and reclustering during synaptic activity. Nat. Neurosci. 4:1187–1193. 49. Rangaraju, V., N. Calloway, and T. A. Ryan. 2014. Activity-driven local ATP synthesis is required for synaptic function. Cell. 156: 825–835. 50. Ahnert-Hilger, G., U. Weller, ., M. Gratzl. 1989. The tetanus toxin light chain inhibits exocytosis. FEBS Lett. 242:245–248. 51. Dreyer, F., A. Mallart, and J. L. Brigant. 1983. Botulinum A toxin and tetanus toxin do not affect presynaptic membrane currents in mammalian motor nerve endings. Brain Res. 270:373–375. 52. Llina´s, R., M. Sugimori, ., J. Marsal. 1994. Transmission at the squid giant synapse was blocked by tetanus toxin by affecting synaptobrevin, a vesicle-bound protein. J. Physiol. 477:129–133. 53. Regehr, W. G., and C. F. Stevens. 2001. Physiology of synaptic transmission and short-term plasticity. In Synapses. W. M. Cowan,

55. Fioravante, D., Y. Chu, ., W. G. Regehr. 2011. Calcium-dependent isoforms of protein kinase C mediate posttetanic potentiation at the calyx of Held. Neuron. 70:1005–1019. 56. Shin, O. H., J. Lu, ., T. C. Su¨dhof. 2010. Munc13 C2B domain is an activity-dependent Ca2þ regulator of synaptic exocytosis. Nat. Struct. Mol. Biol. 17:280–288. 57. Lipstein, N., T. Sakaba, ., N. Brose. 2013. Dynamic control of synaptic vesicle replenishment and short-term plasticity by Ca(2þ)-calmodulin-Munc13-1 signaling. Neuron. 79:82–96. 58. Hua, Y., A. Woehler, ., J. Klingauf. 2013. Blocking endocytosis enhances short-term synaptic depression under conditions of normal availability of vesicles. Neuron. 80:343–349. 59. Dittman, J. S., and W. G. Regehr. 1998. Calcium dependence and recovery kinetics of presynaptic depression at the climbing fiber to Purkinje cell synapse. J. Neurosci. 18:6147–6162. 60. Schneggenburger, R., T. Sakaba, and E. Neher. 2002. Vesicle pools and short-term synaptic depression: lessons from a large synapse. Trends Neurosci. 25:206–212. 61. Genc, O., O. Kochubey, ., R. Schneggenburger. 2014. Munc18-1 is a dynamically regulated PKC target during short-term enhancement of transmitter release. eLife. 3:e01715. 62. Cijsouw, T., J. P. Weber, ., R. F. Toonen. 2014. Munc18-1 redistributes in nerve terminals in an activity- and PKC-dependent manner. J. Cell Biol. 204:759–775. 63. Heine, M., L. Groc, ., D. Choquet. 2008. Surface mobility of postsynaptic AMPARs tunes synaptic transmission. Science. 320:201–205. 64. Kalla, S., M. Stern, ., N. Brose. 2006. Molecular dynamics of a presynaptic active zone protein studied in Munc13-1-enhanced yellow fluorescent protein knock-in mutant mice. J. Neurosci. 26:13054– 13066. 65. Tsuriel, S., R. Geva, ., N. E. Ziv. 2006. Local sharing as a predominant determinant of synaptic matrix molecular dynamics. PLoS Biol. 4:e271. 66. Tsuriel, S., A. Fisher, ., N. E. Ziv. 2009. Exchange and redistribution dynamics of the cytoskeleton of the active zone molecule bassoon. J. Neurosci. 29:351–358. 67. Bai, J., Z. Hu, ., J. M. Kaplan. 2010. Endophilin functions as a membrane-bending molecule and is delivered to endocytic zones by exocytosis. Cell. 143:430–441. 68. Ribrault, C., J. Reingruber, ., A. Triller. 2011. Syntaxin1A lateral diffusion reveals transient and local SNARE interactions. J. Neurosci. 31:17590–17602. 69. Spangler, S. A., S. K. Schmitz, ., C. C. Hoogenraad. 2013. Liprin-a2 promotes the presynaptic recruitment and turnover of RIM1/CASK to facilitate synaptic transmission. J. Cell Biol. 201:915–928.

Biophysical Journal 108(6) 1318–1329