Short Communication

Selective Perturbation of the BAR Domainof Endophilin Impairs Synaptic Vesicle

EndocytosisFREDRIK ANDERSSON, PETER LOW, AND LENNART BRODIN*

Department of Neuroscience, Karolinska Institutet, S-171 77 Stockholm, Sweden

KEY WORDS clathrin; neurotransmitter release; periactive zone; synapse

ABSTRACT The importance of the BAR domain of endophilin in synaptic vesicleendocytosis was tested in presynaptic microinjection experiments in the lamprey giantsynapse. Antibodies as well as Fab fragments directed to the BAR domain caused astimulus-dependent decrease in the number of synaptic vesicles along with an accu-mulation of shallow clathrin coated pits in the periactive zone. Moreover, the isolatedBAR domain protein also caused an accumulation of shallow-coated pits in the periac-tive zone, in addition to appearance of narrow tubules in synaptic regions. The BARdomain of endophilin is thus required for efficient progression of the synaptic vesiclecycle. Synapse 64:556–560, 2010. VVC 2010 Wiley-Liss, Inc.

INTRODUCTION

Clathrin-mediated synaptic vesicle endocytosisdepends critically on endophilin, a 40 kDa protein,which interacts with dynamin and synaptojanin (Gadet al., 2000; Ringstad et al., 1999). The N-terminalpart of endophilin consists of an N-BAR domain, aBAR domain with an N-terminal amphipatic helix(Gallop et al., 2006). BAR domains form curved phos-holipid-binding dimers that can sense and inducemembrane curvature. As yet, the specific roles ofBAR domains in nerve terminals are not well defined.Here, we test the effect of perturbing the N-BAR do-main of endophilin in the lamprey giant synapse.

MATERIALS AND METHODS

The BAR domain [amino acid (aa) 20–247], variableregion (aa 226–289), and the SH3 domain (aa 290–353) of lamprey endophilin (Ringstad et al., 1999)were cloned into a pGEX-6P-2 vector (GE Healthcare,Uppsala, Sweden) using BamH1 and EcoR1 (NewEngland Biolabs) and expressed as GST fusion pro-teins. The GST tag was cleaved, using PreScissionProtease (GE healthcare), before immunization inrabbits (Statens Veterinarmedicinska Anstalt, Upp-sala, Sweden). The serum was then purified on aHiTrap column (GE healthcare) with GST-taggedantigen. For generation of Fab fragments, total IgGwas purified on a protein A column (GE Healthcare).

The eluated IgG was thereafter cleaved using immobi-lized Papain (Pierce) over night at 378C. After cleavageof IgG, the Fab fragments were affinity purified using aNHS-activated HiTrap column coupled with GST-tagged antigen. Uncleaved IgG was removed using aprotein A column. Control injections were performedwith IgG from rabbit (085K7585, Sigma).

For the liposome-binding assay, Folch extract(Sigma) was dissolved in chloroform and methanol 2:1to a concentration of 10 mg/ml. The solvent was firstevaporated under Argon environment and thereafterin vacuum. The liposomes were dissolved in buffer(150 mM NaCl and 20 mM HEPES, pH 7.4), warmedat 378C for 7 min, and sonicated for 5 min, thereafterrun through a 0.4-mm filter (Whatman) using an ex-truder (Avanti Polar Lipids). N-BAR protein (5 mg)was used untreated or after preincubation with BARFab fragments or control IgG (25 mg). Ten micro-grams of liposomes were added, and buffer was addedto a final volume of 50 ml. The liposomes and proteinswere incubated at room temperature for 40 min andthereafter centrifuged on an airflow-driven centrifuge.

Contract grant sponsors: Swedish Research Council, K. & A. WallenbergsStiftelse

*Correspondence to: Lennart Brodin, Department of Neuroscience, Karolin-ska Institutet, S-171 77 Stockholm, Sweden. E-mail: lennart.brodin@ki.se

Received 2 November 2009; Accepted 1 December 2009

DOI 10.1002/syn.20772

Published online in Wiley InterScience (www.interscience.wiley.com).

VVC 2010 WILEY-LISS, INC.

SYNAPSE 64:556–560 (2010)

The pellet was eluated with 50 ml 2 � sample buffer(SB) and 10 ml loaded on a NuPage 4–12% bis tris gel(Invitrogen). The protein was transferred to a mem-brane (Millipore), and the relative amount of proteinin the different fractions were determined with aTyphoonTM (GE Healthcare) using an Alexa 488 (Invi-trogen) conjugated antibody against the BAR domainof lamprey endophilin.

Microinjections were performed as described (Ring-stad et al., 1999). Briefly, IgG, Fab fragments, andthe N-BAR domains of lamprey and rat (aa1-247; Gal-lop et al., 2006) endophilin were conjugated with anAlexa probe (Invitrogen) in carbonate buffer. Thebuffer was exchanged to injection buffer (10 mMHEPES, 250 mM potassium acetate), and the sub-stances were either pressure-injected (antibodies, Fabfragments) or injected with negative current (N-BARdomain) via sharp microelectrodes (resistance 70–100M�) while monitoring the fluorescence using a CCDcamera (Roper Scientific). Stimulation was appliedwith a suction electrode. The stimulation period wasended with fixation in 3% glutaraldehyde and 4% tan-nic acid, and thereafter the specimen was preparedfor electron microscopy (Gad et al., 2000).

To determine the concentration of injected BAR do-main, injected axons were scanned with a confocal

microscope (Nikon Eclipse C1) with a 40� 0.80 waterimmersion objective and the fluorescence intensitymeasured. Areas within micropipettes of similar di-ameter filled with known protein concentrations werescanned for comparison. The concentration of theinjected BAR domain was estimated to be in therange of 0.5 mM at the injection site.

Morphometry was performed on synapses with asingle active zone (Ringstad et al., 1999). Synapses inadjacent uninjected axons were used as controls.Axons were analyzed at several levels relative to theinjection site. Data were collected from the regionwithin 100 mm from the injection site. The lampreyand rat N-BAR proteins were used in one experimenteach with similar results. The BAR IgG was used inone experiment. In all other cases, two independentexperiments, each with several injected axons, wereperformed. In the bar graph in Figure 3, results fromrepresentative experiments are shown. In the amphi-physin SH3 domain microinjection experiment, thestimulation rate was reduced to 0.2 Hz to avoid dis-ruption of the synaptic structure (Shupliakov et al.,1997). Statistical analysis was carried out usingKaleidaGraph software with Student’s t-test or Wil-coxon Mann–Whitney rank sum test where appropri-ate (see legend of Fig. 3).

Fig. 1. A: Western blot of lamprey brain extract with IgG raisedagainst the BAR domain, the variable region, and the SH3 domainof lamprey endophilin, respectively, performed as described by Ring-stad et al. (1999). B: Confocal images from an axon injected withAlexa 488-labeled endophilin BAR IgG (green) and Alexa 546-la-beled synaptotagmin IgG (red; similar results were obtained withendophilin SH3 IgG). Scale bar 5 2 lm. C: Liposome sedimentation

assay testing the binding of the N-BAR domain of endophilin withFolch liposomes. The addition of Fab fragments lowered the BARdomain binding to 48% of the control (background subtracted). Add-ing unspecific IgG had no effect (97%). D: Electron micrograph of asynapse in an axon injected with control IgG and stimulated at 5Hz for 30 min. E: A synapse in an axon injected with BAR IgG (5Hz, 30 min). Arrows indicate coated pits. Scale bar 5 0.2 lm.

557A BAR DOMAIN IN SYNAPTIC VESICLE ENDOCYTOSIS

Synapse

RESULTS

Antibodies raised against the lamprey endophilinBAR domain reacted with a single band of 40 kDa inlamprey brain extract, which was also recognized byantibodies to the variable region and the SH3 do-main, respectively (Fig. 1A). When microinjected intoaxons endophilin antibodies accumulated at synapticrelease sites identified by labeling with comicroin-jected synaptotagmin antibodies (Fig. 1B). To studyeffects on synaptic vesicle recycling, injected axonswere stimulated at a physiological rate [5 Hz for 30min as used by Ringstad et al. (1999) and Gad et al.(2000)] followed by fixation. In axons injected withcontrol IgG (Fig. 1D), synapses contained large vesicleclusters (181.0 6 52.7 synaptic vesicles, mean 6 SD,n 5 7 synapses; uninjected control 212.8 6 50.8 syn-aptic vesicles, n 5 6 synapses; P > 0.05) and fewcoated pits (arrow in Fig. 1D; 1.34 6 0.47, n 5 7;uninjected control 1.68 6 0.88, n 5 6; P > 0.05). Incontrast, synapses in axons injected with BAR IgGhad a reduced number of synaptic vesicles (Figs. 1Eand 3A1; filled bars, injected; open bars, control).Coated pits, the majority of which were shallow, were

accumulated in the periactive zone (Figs. 2A and 3A1and 3A2). To rule out the possibility that the BARIgG acted by precipitating endophilin, BAR-directedFab fragments were prepared. (Fab fragments do notform complexes.) These Fab fragments inhibited bind-ing of the N-BAR domain to liposomes (Fig. 1C), andthey are not expected to interfer with the SH3domains, which are located at the distal tips of theBAR dimer (Wang et al., 2008). When injected, theFab fragments induced similar changes as the BARIgG (Figs. 2B and 3B1 and 3B2). Endosome-likeobjects were in some cases observed in synapticregions of IgG- and Fab fragment-injected axons.However, we did not detect coated pits on these struc-tures, making it uncertain whether they were endo-cytic intermediates (Andersson et al., 2008). The roleof the N-BAR domain was further probed by the injec-tion of the N-BAR domain protein (aa 1-247) to an ap-proximate concentration of 0.5 lM (see Materials andMethods section). The N-BAR domain indeed induceda reduction in the number of synaptic vesicles uponstimulation along with an accumulation of shallow-coated pits (Figs. 2C and 3C). The increase in the

Fig. 2. A–C: Electron micrographs showing the periactive zone (within 2 lm from a synapse) inaxons microinjected with: (A) BAR domain-directed IgG, (B) BAR domain-directed Fab fragments,and (C) the N-BAR domain. Stimulation was applied at 5 Hz for 30 min. Scale bar 5 0.2 lm.

558 F. ANDERSSON ET AL.

Synapse

number of coated pits was smaller in comparison withthat induced by the BAR antibodies and Fab frag-ments, but the distribution of stages was similar(Figs. 2C and 3C2). Additionally, in synaptic regionsof N-BAR domain-injected axons, we observed

tubules, often connected with endosome-like struc-tures (not shown). The narrowest tubules had a diame-ter of 30–40 nm. Presumably, they reflected tubulationof endosomelike objects (Andersson et al., 2008)exposed to a high-N-BAR concentration (Gallop et al.,2006). In axons injected with control IgG (Fig. 1D), thestages of coated intermediates (Fig. 3D) were similarto those in control axons, while in axons injected withthe SH3 domain of amphiphysin (Shupliakov et al.,1997) most coated pits were deeply invaginated(Fig. 3E).

Thus, three reagents expected to perturb N-BARdomain—membrane interactions cause accumulationof shallow-coated pits in stimulated synapses. A simi-lar phenotype was observed after microinjection ofIgG to the SH3 domain of rat endophilin (Ringstadet al., 1999). Endophilin has, however, been mostfirmly linked to late stages of endocytosis. In the SH3domain-directed antibody injection experiments ofRingstad et al. (1999), some free-coated vesicles werein fact detected, and microinjection of a SH3-bindingpeptide caused accumulation of deeply invaginatedcoated pits and of numerous free-coated vesicles (Gadet al., 2000). The present findings point to a role ofthe N-BAR domain in the invagination of coated pits,which, may, however, be indirect. Although it remainspossible that the N-BAR domain acts directly to bendthe membrane of the emerging coated pit, this actionwould probably be limited as little endophilin appearsto be present within the coat (Ringstad et al., 1999).An action at the rim of the pit, together with otheraccessory factors, is yet possible. Another possibilityis that the recruitment of synaptojanin had beenimpaired, resulting in elevated levels of PIP2 thatcould cause sequestration of coat components, preventreorganization of the growing coat (Mani et al., 2007),or influence actin dynamics (Shupliakov et al., 2002).Impaired recruitment of dynamin may also contributeas dynamin has been implicated in controllingthe rate of coated pit maturation (Loerke et al., 2009).In summary, the present experiments allow us toconclude that the BAR domain of endophilin isrequired for efficient progression of synaptic vesicleendocytosis.

ACKNOWLEDGMENTS

We thank Drs. Jenny Gallop and Harvey T. McMa-hon for providing endophilin N-BAR protein and Dr.Ole Kjaerulff for comments on the manuscript.

REFERENCES

Andersson F, Jakobsson J, Low P, Shupliakov O, Brodin L. 2008.Perturbation of syndapin/PACSIN impairs synaptic vesicle recy-cling evoked by intense stimulation. J Neurosci 28:3925–3933.

Gad H, Ringstad N, Low P, Kjaerulff O, Gustafsson J, Wenk M, DiPaolo G, Nemoto Y, Crun J, Ellisman MH, De Camilli P, Shuplia-kov O, Brodin L. 2000. Fission and uncoating of synaptic clathrin-

Fig. 3. A–C: Quantitative analysis of the microinjection experi-ments shown in Figure 2. Average number of synaptic vesicles (A1–C1, left panels; mean 6 SD) and number of clathrin-coated inter-mediates (A1–C1, middle panels; mean 6 SD). Relative distributionof clathrin-coated intermediates (A2–C2). Open bars, uninjectedcontrol axons; filled bars, axons injected with the indicated reagent.The lamprey N-BAR and rat N-BAR proteins were used with similarresults. The data in C were obtained with the lamprey N-BAR.Measurements were performed as described by Andersson et al.(2008). Specifications of n and statistics: A1 synaptic vesicles, n 5 6(control) and 6 (injected) synapses, respectively, t-test; A1-coatedintermediates: n 5 6 and 8 synapses, respectively, t-test; A2, n 5199 (control) and 138 (injected) coated intermediates, respectively;B1 synaptic vesicles, n 5 6 (control) and 7 (injected) synapses,respectively, Wilcoxon Mann–Whitney rank sum test (WMW); B1-coated intermediates: n 5 6 and 7 synapses, respectively, WMW;B2, n 5 199 (control) and 222 (injected) coated intermediates,respectively; C1 synaptic vesicles, n 5 5 (control) and 9 (injected)synapses respectively, t-test; C1-coated intermediates: n 5 5 and 9synapses, respectively, WMW; C2, n 5 227 (control) and 98(injected) coated intermediates, respectively; (D) n 5 115 (control)and 131 (injected) coated intermediates; (E) n 5 171 coated inter-mediates.

559A BAR DOMAIN IN SYNAPTIC VESICLE ENDOCYTOSIS

Synapse

coated vesicles are perturbed by disruption of interactions withthe SH3 domain of endophilin. Neuron 27:301–312.

Gallop JL, Jao CC, Kent HM, Butler PJ, Evans PR, Langen R,McMahon HT. 2006. Mechanism of endophilin N-BAR domain-mediated membrane curvature. EMBO J 25:2898–2910.

Loerke D, Mettlen M, Yarar D, Jaqaman K, Jaqaman H, DanuserG, Schmid SL. 2009. Cargo and dynamin regulate clathrin-coatedpit maturation. PLoS Biol 7:e57.

Mani M, Lee SY, Lucast L, Cremona O, Di Paolo G, De Camilli P,Ryan TA. 2007. The dual phosphatase activity of synaptojanin1 isrequired for both efficient synaptic vesicle endocytosis andreavailability at nerve terminals. Neuron 56:1004–1018.

Ringstad N, Gad H, Low P, Di Paolo G, Brodin L, Shupliakov O, DeCamilli P. 1999. Endophilin/SH3p4 is required for the transition

from early to late stages in clathrin-mediated synaptic vesicleendocytosis. Neuron 24:143–154.

Shupliakov O, Low P, Grabs D, Gad H, Chen H, David C, Takei K,De Camilli P, Brodin L. 1997. Synaptic vesicle endocytosisimpaired by disruption of dynamin-SH3 domain interactions. Sci-ence 276:259–263.

Shupliakov O, Bloom O, Gustafsson JS, Kjaerulff O, Low P,Tomilin N, Pieribone VA, Greengard P, Brodin L. 2002. Impairedrecycling of synaptic vesicles after acute perturbation of the pre-synaptic actin cytoskeleton. Proc Natl Acad Sci USA 99:14476–14481.

Wang Q, Kaan HY, Hooda RN, Goh SL, Sondermann H. 2008. Struc-ture and plasticity of endophilin and sorting nexin 9. Structure16:1574–1587.

560 F. ANDERSSON ET AL.

Synapse