Regulation of Synaptic Vesicle Recycling by Complex Formation Between Intersectin 1 and the Clathrin Adaptor Complex AP2

Regulation of synaptic vesicle recycling by complex formation between intersectin 1 and the clathrin adaptor complex AP2

Arndt Pechsteina,b, Jelena Bacetica,1, Ardeschir Vahedi-Faridia,1, Kira Gromovaa,1, Anna Sundborgerb,1, Nikolay Tomlinb, Georg Krainerc, Olga Vorontsovab, Johannes G. Schäfera, Simen G. Owed, Michael A. Cousine, Wolfram Saengera, Oleg Shupliakovb,2, and Volker Hauckea,c,2

aInstitute of Chemistry and Biochemistry, Freie Universität and Charité Universitätsmedizin Berlin, 14195 Berlin, Germany; cLeibniz-Institut für Molekulare Pharmakologie, 13125 Berlin, Germany; bDepartment of Neuroscience, Linné Center in Developmental Biology and Regenerative Medicine, Karolinska Institutet, 171 77 Stockholm, Sweden; dDepartment of Anatomy, University of Oslo, N-0317 Oslo, Norway; and eMembrane Biology Group, Centre for Integrative Physiology, University of Edinburgh, EH8 9XD Edinburgh, Scotland, United Kingdom

Edited* by Pietro De Camilli, Yale University and the Howard Hughes Medical Institute, New Haven, CT, and approved January 21, 2010 (received for review September 29, 2009) Clathrin-mediated synaptic vesicle (SV) recycling involves the spa- over, association of the SH3A-B linker region of intersectin with tiotemporally controlled assembly of clathrin coat components at AP2 inhibits binding of the inositol phosphatase synaptojanin 1. phosphatidylinositiol (4, 5)-bisphosphate [PI(4,5)P2]-enriched mem- These data identify the intersectin-AP2 complex as an important brane sites within the periactive zone. Such spatiotemporal control regulator of clathrin-mediated SV recycling in synapses. is needed to coordinate SV cargo sorting with clathrin/AP2 recruitment and to restrain membrane fission and synaptojanin-mediated Results uncoating until membrane deformation and clathrin coat assembly Antibodies Targeting the Linker Region Between SH3 Domains A and are completed. The molecular events underlying these control mech- B of Lamprey Intersectin 1 Disrupt SV Endocytosis at Early Stages. anisms are unknown. Here we show that the endocytic SH3 domain- Acute perurbation of intersectin interactions in living lamprey containing accessory protein intersectin 1 scaffolds the endocytic synapses with antibodies raised against the SH3 domain-containing process by directly associating with the clathrin adaptor AP2. Acute module revealed multiple effects on SV recycling (13). Endocytosis perturbation of the intersectin 1-AP2 interaction in lamprey synapses was inhibited at the stage of constricted CCPs, consistent with a in situ inhibits the onset of SV recycling. Structurally, complex for- defect in dynamin-mediated fission. However, microinjection of mation can be attributed to the direct association of hydrophobic “ antibodies targeting SH3 domains A to C, including the connecting peptides within the intersectin 1 SH3A-B linker region with the side linker regions of intersectin 1, also led to an accumulation of large sites” of the AP2 α-andβ-appendage domains. AP2 appendage membrane expansions, indicative of endocytosis inhibition at early association of the SH3A-B linker region inhibits binding of the inosi- stages (13). To further explore the underlying mechanisms of the tol phosphatase synaptojanin 1 to intersectin 1. These data identify early endocytic phenotype associated with intersectin perturbation the intersectin-AP2 complex as an important regulator of clathrin- mediated SV recycling in synapses. at vertebrate synapses, we raised antibodies against different parts of the SH3 module, including the linker region between SH3 do- fi fi endocytosis | synapse | scaffolding proteins | appendage | synaptojanin mains A and B of lamprey intersectin (LIS-linker). Af nity-puri ed LIS-linker IgG specifically recognized two bands of expected molecular weights of about 200 and 170 kDa (13) (lamprey inter- ynaptic vesicles (SVs), following their activity-dependent exo- sectin 1 short and long isoforms) in detergent-extracted lamprey cytic fusion with the presynaptic plasma membrane, are recy- S brain homogenates, and this reactivity was lost upon preincubation cled by compensatory endocytosis at the periactive zone (1–3), with the antigenic peptide (Fig. S1A). Similar results were seen if largely via clathrin-mediated reinternalization of fully fused SV antibody specificity was assessed by immunoprecipitation of native membrane (4). Clathrin-coated pit (CCP) formation (5) proceeds endogenous intersectin 1 (Fig. S1D). LIS-linker IgG also recog- through the assembly of endocytic proteins at phosphatidylinositiol (4, 5)-bisphosphate [PI(4,5)P ]-enriched membrane sites (6, 7). A nized endogenous intersectin 1 accumulated within SV clusters in 2 reticulospinal axons, as seen previously (13); again, this labeling was key factor in the assembly pathway is the heterotetrameric adaptor B complex AP2, whose α-andβ2-appendage domains act as major completely abolished by preincubation with the antigen (Fig. S1 and C). A similar accumulation of intersectin 1 at SV clusters is also recruitment platforms for accessory proteins (6, 7), regulating dis- fi tinct steps within the pathway. Despite our extensive knowledge observed in mammalian synapses of hippocampal mossy bers from regarding the endocytic interactome, we know comparably little the CA3 region and synapses established by Schaffer recurrent about the structural components within the periactive zone that colaterals from CA1 and at corresponding postsynaptic elements scaffold the endocytic process, thereby allowing the high fidelity of SV recycling. Such spatiotemporal control is needed to coordinate SV cargo protein sorting with coat recruitment (8) and to restrain Author contributions: O.S. and V.H. designed research; A.P., J.B., A.V.-F., K.G., A.S., N.T., fi J.G.S., S.G.O., and O.S. performed research; K.G., A.S., N.T., G.K., O.V., and M.A.C. con- membrane ssion and uncoating until membrane deformation and tributed new reagents/analytic tools; A.P., A.V.-F., K.G., G.K., M.A.C., W.S., and V.H. an- CCP assembly are completed. Moreover, stabilizing scaffolds may alyzed data; and A.P., O.S., and V.H. wrote the paper. aid coupling of SV exo- and endocytosis (1, 3). The Drosophila The authors declare no conflict of interest. multidomain protein Dap160, an ortholog of mammalian inter- *This Direct Submission article had a prearranged editor. sectin, has been postulated to act as an endocytic scaffold of the Data deposition: Protein structures are accessible under PDB ID codes 3HS8 and 3HS9. – periactive zone (9 11), although its precise role in SV recycling in 1J.B., A.V.-F., K.G., and A.S. contributed equally to this work. mammalian nerve terminals remains largely unclear (12). 2To whom correspondence may be addressed. E-mail: [email protected] or oleg. Here we show that intersectin 1 scaffolds the endocytic process [email protected]. by directly associating with AP2. Acute perturbation of intersectin- This article contains supporting information online at www.pnas.org/cgi/content/full/ AP2 complex formation blocks the onset of SV recycling. More- 0911073107/DCSupplemental.

4206–4211 | PNAS | March 2, 2010 | vol. 107 | no. 9 www.pnas.org/cgi/doi/10.1073/pnas.0911073107 Downloaded by guest on September 27, 2021 (Fig. 1A and Fig. S2 A–E). Thus, intersectin 1 is accumulated at observed instead in these synapses (Fig. 1 E, F, and I). These presynaptic sites both in lamprey and in mammals. functional data indicate that the linker region between SH3 Having established that LIS-linker IgGs specifically recognize domains A and B of intersectin plays an important role at early intersectin 1 in vitro and in situ, we microinjected these IgGs into stages of clathrin-mediated SV endocytosis, perhaps via direct lamprey reticulospinal axons and stimulated at 5 Hz for 20 min to interactions with one of the major coat components. induce SV recycling at a physiological rate. Analysis of ultrathin sections revealed a profound loss of SVs at active zones in syn- Intersectin 1 via its SH3A-B Linker Region Directly Associates with the α β apses from these axons (Fig. 1 B, C, and H), but not from control - and -Appendage Domains of AP2. To directly address the possibility that intersectin 1 interacts with clathrin/AP2, we per- axons microinjected with nonspecific IgGs (Fig. 1 G and H) (13). formed immunoprecipitations from rat (Fig. 2A) and lamprey No morphological alterations were observed in axons micro- B fi D brain extracts (Fig. 2 ). Antibodies speci c for intersectin 1 (Fig. injected with LIS-linker IgGs kept at rest (Fig. 1 ). This change S1E)efficiently coprecipitated not only the established intersectin was accompanied by an accumulation of large membrane pockets 1 binding partner dynamin 1, but also the plasmalemmal clathrin and invaginations and a concomitant increase in the number of adaptor AP2 (Fig. 2 A and B). Hsc70 or AP1 were absent from the B C CCPs, predominantly at early stages of endocytosis (Fig. 1 , , immunoprecipitates. Conversely, intersectin was found in immu- H–J ), indicative of an inhibition of clathrin-mediated SV recy- noprecipitated material using monoclonal antibodies against AP2 cling. This phenotype was clearly distinct from that observed (see below), indicating that intersectin and AP2 are present in a following microinjection of antibodies against SH3 domains C to tight complex in vivo in both lamprey and in mammals. This E of intersectin, which did not induce large membrane expansions conclusion is supported by the near perfect colocalization of around the active zone (Fig. 1 E, F, and J). A significant accu- intersectin 1 with AP2, but not with AP1 in astrocytes (Fig. S2 F mulation of coated intermediates in the periactive zones was and G) or primary neurons in culture (Fig. S2H). We also con- CELL BIOLOGY

Fig. 1. Antibodies targeting the linker region between SH3 domains A and B of intersectin 1, a pre- and postsynaptic endocytic protein, inhibit SV recycling. (A) Intersectin 1 accumulates at pre- and postsynaptic membrane sites. Depicted are quantifications of immunogold particles accumulated over nerve terminals and dendrites in mossy fiber terminals and recurrent Schaffer collateral synapses within the mouse hippocampus labeled with intersectin 1-specific antibodies. The difference between densities of gold particles in each synaptic compartment is not significantly different for both types of synapses (P > 0.05; n = 20 in each case; Student’s t test). Bars depict mean densities of gold particles ± SEM. Background is below two particles per square micrometers in both cases. (B and C) Electron micrographs of reticulospinal synapses in an axon stimulated at 5 Hz for 20 min after injection of antibodies against the linker region between SH3 domains A and B of lamprey intersectin (LIS-linker). The vesicle cluster (SVC) is depleted, large membrane invaginations (curved arrows)and clathrin-coated intermediates are present. Small arrow, active zone. a, axoplasm. (D) Electron micrograph of an intact reticulospinal synapse after injection of LIS-linker antibodies kept at rest. Note the presence of a densely packed SV cluster at the active zone. (E) Electron micrograph of a reticulospinal synapse in an axon stimulated at 5 Hz for 20 min after injection of antibodies against the SH3 domains C to E of lamprey intersectin (LIS-CE). Note the accumulation of CCPs and the absence of membrane invaginations. (F) Electron micrograph of an intact reticulospinal synapse stimulated at 5 Hz for 20 min from a reticulospinal axon from the same preparation as in E.(G) A synapse from an axon microinjected with nonspecific IgG and stimulated at 5 Hz. (H) Bar graph showing a reduction in the number of SVs in LIS antibody injected synapses. LIS-CE antibodies cause a strong increase in the number of CCPs (I), whereas LIS-linker antibody-injected synapses display numerous membrane pockets and invaginations (see curvature index, CI) (J). Nonspecific IgGs did not induce significant changes in synaptic morphology when compared to noninjected control synapses (H–J). Depicted are the means ± SD (*, P < 0.05; **, P < 0.01; ***, P < 0.001; Student’s t test). (Scale bar, A and B,1μm.)

Pechstein et al. PNAS | March 2, 2010 | vol. 107 | no. 9 | 4207 Downloaded by guest on September 27, 2021 firmed that intersectin 1 associates with CCPs in synapses by immuno-electron microscopy. In lamprey reticulospinal synapses stimulated at 5 Hz for 30 min, intersectin immunogold labeling on clathrin/AP2-coated intermediates at sites of endocytosis was predominantly associated with coats (Fig. 2C). Given the strong early endocytic phenotype elicited by microinjection of LIS-linker IgG, we suspected that the intersectin 1 SH3A-B linker region is involved in complex formation with AP2. Analysis of the primary sequences of the linker region from various species indeed revealed the presence of two evolutionary conserved sequences that resemble AP2 binding motifs found in other endocytic proteins. Although none of the two strictly con- forms to any of the known AP2 binding consensus sequences, the strong conservation of large aromatic residues (W, F) flanked by negative charges suggested that the WADF and WDxW peptides might bind to AP2 (Fig. 2D). To directly test this possibility, we prepared GST fusion proteins comprising the WT linker or mutants, in which one or both putative AP2-binding peptides had been mutationally inactivated or deleted (Fig. 2F). Native AP2 was efficiently retained by immobilized GST-intersectin 1 linker but not by a mutant lacking the conserved aromatic residues within the putative AP2 binding motifs. A GST-fused WADF peptide also retained AP2 from brain extracts, albeit less efficiently, whereas GST-WDAW was ineffective (Fig. 2E). As most other endocytic proteins target the appendage domains of AP2 (6, 7), we incubated GST-linker region fusion proteins with purified His6-tagged α-orβ2 appendage domains. We observed potent binding of GST-linker to both ear domains and this was dependent on conserved aromatic amino acids within both peptides. Under these conditions, avid binding of GST-WADF and a much weaker association of GST-WDAW with both appendages was observed (Fig. S3). Isothermal titration calorimetry revealed a thermodynamic preference of the intersectin 1 linker for the AP2- β- over the AP2-α-appendage domain with binding constants of 490 nM and 5 μM, respectively (Fig. 2G). To verify that the intersectin 1 SH3A-B linker region indeed can associate functionally with AP2, we coexpressed intersectin 1-mCer linker (WT or AxxA mutant) with the pH-sensitive reporter synaptopHluorin (14) in primary cultures of cerebellar granule neurons (CGNs). CGNs showed a robust increase in fluorescence after stimulation by either KCl or a train of action potentials, reflecting fusion of SVs with the cell surface (Fig. S4 A and B). This fluorescence remained elevated in neurons expressing intersectin 1-mCer (WT) after stimulation, indicative of a block in SV endocytosis. By contrast, fluorescence in neurons expressing the AP2 binding-defective mutant decreased Fig. 2. Intersectin associates with AP2 via direct binding to the α- and β2- A B A back to its initial value (Fig. S4 and ). The slight difference in appendage domains of AP2. ( ) Immunoprecipitation from detergent- fl extracted rat brain lysates using control or anti-intersectin 1 (ITSN1) anti- uorescence increase in neurons expressing intersectin 1-mCer bodies coupled to beads. Samples were analyzed by SDS/PAGE and immu- (WT) compared to mutant-expressing cells during the late phase noblotting. 1% input, 1% of the total amount of rat brain extract (RBE) of stimulation likely reflects inhibition of compensatory SV en- added to the assay. (B) Same as in A but using detergent extracted lamprey docytosis accompanying exocytic fusion events. In agreement brains instead. 1% input, 1% of the total amount of lamprey brain extract with this, kinetic rise times of pHluorin de-quenching are iden- (LBE) added to the assay. (C) Electron micrograph of CCPs from the periactive tical in neurons expressing intersectin 1-mCer (WT) or mutant zone of a lamprey reticulospinal synapse stimulated at 5 Hz for 30 min and (Fig. S5A). This interpretation is corroborated by experiments labeled with intersectin LIS-AC antibodies. Note the accumulation of gold μ D using FM1-43, which reveal a selective inhibition of SV recycling, particles on CCPs. (Scale bar, 0.2 m.) ( ) Multiple sequence alignment of B–D intersectin 1 SH3A-SH3B. Putative AP2-binding peptides are highlighted in but not exocytosis (Fig. S5 ). A similar block of SV endo- purple. (E) GST or GST-intersectin 1 fusion proteins were immobilized and cytosis at early stages was observed in reticulospinal axons of the incubated with RBE in the presence or absence of ATP. Samples were ana- lamprey following microinjection of a lamprey intersectin-1 lyzed by SDS/PAGE and immunoblotting. (F) Schematic representation of the linker region-derived AP2 binding peptide. Analysis of ultrathin proteins used for pull downs. Putative AP2 binding motifs and correspond- sections revealed a partial loss of SVs at active zones, an accu- ing mutants are labeled (compare also to D). (G) High-affinity binding of mulation of large membrane pockets and invaginations, and a β intersectin 1 linker to the AP2 appendage as determined by isothermal concomitant increase in the number of CCPs in intersectin 1- titration calorimetry (ITC) analysis. Intersectin 1-derived linker peptide was α β peptide (Fig. S4 C, D, and F–H), but not in non-injected control injected into -or 2-appendage. Depicted are integrated, normalized, and E–H dilution-corrected heats of reaction, Q, versus molar peptide/protein ratio, R. synapses (Fig. S4 ). Thus, overexpression or microinjection Heats of reaction obtained by injecting peptide into α- (circles) or β2- of an intersectin 1-derived AP2 binding peptide impairs SV en- appendage (squares) were fitted by a one-site binding model (solid lines), docytosis, likely via sequestration of endogenous AP2, confirm- K yielding the given D values. ing our biochemical data.

4208 | www.pnas.org/cgi/doi/10.1073/pnas.0911073107 Pechstein et al. Downloaded by guest on September 27, 2021 PNNWADFSSTWP) and used it for cocrystallization experiments with purified α-orβ2-appendage. We obtained crystals for both α- and β2-appendages that diffracted to a maximal resolution of 1.9 Å and 2.15 Å, respectively. The structure model refined with excellent B-factors for α-appendage and peptide ligand. Somewhat higher B-factors were obtained for the β2- appendage complex (Table S1). The intersectin 1 WADF peptide binds to an extensive surface on the β-sandwich subdomain of the α-appendage (also termed the side site) (Fig. 3A). W843 and F846 (within mouse intersectin 1L), which form the core consensus motif, are seen in shallow pockets. F846 is π-stacked to F740 of AP2α and W843 is hydrogen-bonded to G780. Addi- tional hydrogen bonds involve E702, E718, Q723, and Q782, which interact with the peptide backbone or with the side chain of S848 (Fig. 3B). The same intersectin-1 peptide can also accommodate the sandwich subdomain of the β2-appendage (Fig. 4C), a rare, if not unique feature among the known endocytic proteins. Of note, the side site of the β2-sandwich lies on the opposite face of the subdomain compared to the α-appendage (compare Fig. 3 A and C). The WADF peptide largely binds via aromatic amino acids W843, and F846.W843 is surrounded by Y815 and N758. F846 projects into a shallow pocket formed by F757 and Q756 (Fig. 3D). Additional major contacts involve hydrogen bonds between Y815 and Q756 with the peptide backbone (Fig. 3D and Table S2). The conformation of the WADF peptide is similar to that previously reported for a short peptide fragment derived from Eps15 (Table S3) (15, 16). Intersectin 1 binding to the side site of the β2-appendage was confirmed by site-directed mutagenesis. GST-fused β2-appendage avidly bound to intersectin 1, the FxDxF motif-containing endocytic protein AP180, and epsin 1, an endocytic adaptor that predominantly associates with the top site of β2 via DPF motifs. Mutation of Y815 to A815 selectively abolished association with intersectin 1 and AP180, identifying both proteins as β2 side-site specific ligands, whereas binding to epsin 1 was unaffected. Conversely, the GST-β2 appendage Y888V mutant had selectively lost the ability to pulldown epsin 1 although retaining complex formation with intersectin 1 and AP180 (Fig. 3E). Selective association of intersectin 1 with the crystallographically determined side site of the β2-appendage was finally demonstrated using purified His6-tagged intersectin 1 linker (Fig. 3F). We conclude that intersectin 1 contacts the side sites of both AP2 appendage domains, a feature that may allow CELL BIOLOGY Fig. 3. The intersectin 1-derived WADF motif peptide accommodates the the protein privileged access to the AP2 hub during early stages sandwich subdomain of the AP2α or AP2β appendages. (A) Ribbon diagram of CCP formation. These structural and biochemical data fit well showing the binding site for the mouse intersectin 1-derived peptide (amino with the early endocytic defects observed upon perturbation of acids 840–851, blue) in complex with the α-appendage (gray). The peptide binds intersectin-AP2 complex formation in situ. to anelongated pocket atthesidesite of theα-appendagesandwichsubdomain. (B) Close-up view showing direct molecular contacts between the intersectin 1- Association of AP2 with Intersectin 1 Prevents Synaptojanin 1 Binding. derived peptide (blue) and the side site of the α-appendage. See text for details. Based on our data reported here and on previous studies, inter- C ( ) Ribbon diagram showing the binding site for the mouse intersectin 1-derived sectin 1 acts as a major endocytic scaffold and recruitment platform peptide (blue) in complex with the β2-appendage (gray). The peptide binds to an β D that may coordinate AP2-based CCP assembly with dynamin- elongated pocket at the side site of the 2-appendage sandwich subdomain. ( ) fi Close-up view showing direct molecular contacts between the intersectin 1- dependent membrane ssion and synaptojanin 1-mediated PI(4,5) derived peptide (blue) and the side site of the β2-appendage. See text for details. P2-hydrolysis. Intersectin 1 association with dynamin 1 has been (E) One-hundred microgram GST-fusion protein or GST immobilized on gluta- shown to involve its SH3 domains A, C, and E, whereas synapto- thione beads were incubated with 2 mg RBE. Samples were analyzed by SDS/ janin 1 preferentially binds to SH3A (13). To obtain insights into PAGE and immunoblotting. ITSN1 binding was abolished in the side site mutant possible mechanisms for coordinating these various activities of Y815A but unaffected by mutation of the top site of the β2-appendage. (F)In intersectin 1, we tested whether intersectin 1 can simultaneously vitro binding assay: 50 μg GST-fusion protein or GST immobilized on glutathione bind to dynamin 1, synaptojanin 1, and AP2. To this aim, we first beads were incubated with 10 μgHis6-tagged ITSN1 linker. A side site mutant β employed mouse monoclonal antibodies that selectively associate (Y815A) of the 2-appendage fails to bind to the ITSN1 linker. with the platform subdomain (the so-called top site) of the AP2 α-appendage. When immobilized on beads, these antibodies pre- Structural Basis for the Association of Intersectin 1 with the Sandwich cipitated native AP2 from rat brain together with substantial Subdomains of AP2α and β2. The observation that antibody- amounts of intersectin 1. Synaptojanin 1 and amphiphysin, an mediated perturbation of intersectin 1 association with AP2 endocytic protein that associates predominantly with the platform potently inhibits SV endocytosis at early stages (Fig. 1) promp- subdomain of AP2α, were both absent from these immunopreci- ted us to investigate the structural basis of complex formation. pitates (Fig. 4A). To further explore the relationship between re- We synthesized the major AP2 binding peptide (compare Fig. 2; cruitment of AP2 and synaptojanin 1 to intersectin 1, we carried out

Pechstein et al. PNAS | March 2, 2010 | vol. 107 | no. 9 | 4209 Downloaded by guest on September 27, 2021 affinity chromatography experiments from lamprey spinal cord extracts using immobilized GST-SH3A-C of lamprey intersectin 1 in the presence of IgG against the AP2-binding linker region (LIS- linker IgG). LIS-linker IgGs effectively competed recruitment not only of native AP2, but also of synaptojanin 1-p145 (the brain- specific short isoform of the enzyme). By contrast, GST-SH3A-C binding to dynamin 1 was much less affected (Fig. 4B). Similar results were obtained if the purified AP2α or β2 appendage domains were taken as competitors (Fig. 4 C and D). Conversely, purified recombinant synaptojanin 1-p145 tail domain did not in- terfere with AP2 binding to GST-SH3A-C (Fig. S6A), suggesting that synaptojanin 1 is unable to disrupt intersectin 1-AP2 complex formation directly. Collectively, our data suggest that association of intersectin 1 with AP2 may prevent it from binding to synaptojanin 1-p145. To test this prediction more directly, we carried out affinity chromatography experiments using GST-tagged SH3A domain fusion proteins either containing or lacking the AP2-binding linker peptide. As seen in Fig. 4F, GST-SH3A, a fusion protein lacking the ability to associate with AP2, displayed a greatly improved ability to pull down native endogenous synaptojanin 1-p145 from rat brain extracts when compared to GST-SH3A-linker. By contrast, both versions of GST-tagged SH3A avidly bound to purified synaptojanin 1-tail domains in the absence of AP2 (Fig. S6B). To investigate whether native endogenous endocytic protein complexes may be subject to a similar regulatory mechanism, we carried out immunoprecipation experiments from detergent-extracted rat brains using mouse monoclonal antibodies directed against synaptojanin 1. In agreement with previous data (13, 17), synaptojanin 1-p145 was found to be associated with the SH3 domain-containing proteins endophilin A1 and intersectin 1. Addition of AP2α-orβ2-appendage domains to these extracts selectively impaired association of synaptojanin 1-p145 with endogenous intersectin 1, whereas complex formation with endophilin A1 was unaffected (Fig. 4E). These data further support our conclusion that intersectin complex formation with AP2 impairs its association with synaptojanin 1. Discussion In this work we show that the endocytic scaffold intersectin 1 regulates early steps in clathrin-mediated SV recycling by directly associating with the clathrin adaptor complex AP2. Biochemical and crystallographic analyses identify an evolutionary conserved peptide motif that accommodates the side sites of the sandwich subdomains of the AP2 α-andβ-appendages by a two-pin-plug mechanism. Blocking complex formation between AP2 and intersectin 1 by specific antibodies functionally arrests SV recycling at early stages, resulting in the accumulation of numerous plasma membrane-derived membrane pockes and invaginations. Thus, our data argue in favor of intersectin-1 complex formation with AP2 during early steps of CCP assembly. We predict similar phenotypes to be observed in mammalian neurons derived from knockout mice lacking both intersectins 1 and 2. In this context we note the presence of putative binding motifs for both AP2α (i.e., 1690WVxFD) Fig. 4. Association of intersectin 1 with AP2 impairs its binding to synapto- and clathrin in intersectin 2, an intersectin paralog expressed in the janin 1. (A) Top site-selective AP2 α-appendage antibodies (AP.6) coimmuno- precipitate intersectin but not synaptojanin, AP180, or amphiphysin from RBE. Samples were analyzed by SDS/PAGE and immunoblotting. 1% input, 1% of the total amount of RBE added to the assay. (B) Anti-intersectin linker domain immunoblot signals for synaptojanin 1 and dynamin 1 seen in C (mean ± SEM; antibodies inhibit complex formation of intersectin SH3A-C with AP2 and n = 3). (E) AP2 α-orβ2-appendage domains inhibit complex formation synaptojanin 1, but not with dynamin. GST-SH3A-C (LIS) or GST (100 μg) were between synaptojanin 1 and intersectin 1, but not endophilin A1. RBEs were used for affinity chromatography from lamprey spinal cord extracts (LCE) in subjected to immunoprecipitations with anti-synaptojanin 1-p145 antibodies the presence of the indicated molar ratios of IgGs targeting the intersectin 1 in the presence or absence of AP2 α-orβ2-appendage domains (500 μg). linker region or control IgGs. Samples were analyzed by SDS/PAGE and Depicted are the relative levels of intersectin 1 or endophilin A1 found in the immunoblotting. 1% input, 1% of the total amount of LCE added to the assay. immunoprecipitate normalized to the control condition (mean ± SEM; n =2). (C) Same as in B except that 1.5 mg purified α-orβ2-appendage domains were Representative immunoblots for intersectin and endophilin are shown below used as competitors. Samples were analyzed by SDS/PAGE, staining with the bar diagram. (F) GST-SH3A-linker, GST-SH3A, or GST (50 μg) were used for Ponceaus S (to visually illustrate dynamin 1 association), and immunoblotting affinity chromatography from RBEs. Samples were analyzed by SDS/PAGE and for synaptojanin 1, dynamin 1, and clathrin heavy chain as a control. 3% input, immunoblotting for synaptojanin 1, AP2α,orβ-actin. 1.5% input, 1.5% of the 3% of the total amount of LCE added to the assay. (D) Quantification of the total amount of RBE added to the assay.

4210 | www.pnas.org/cgi/doi/10.1073/pnas.0911073107 Pechstein et al. Downloaded by guest on September 27, 2021 brain of intersectin-1 knockout mice (18). It is likely that the scaf- Materials and Methods folding function of intersectin is aided by its association with the Afﬁnity Chromatography and Immunoprecipitation Experiments. Detergent AP2-binding endocytic protein Eps15/Eps15R. The resulting mul- extracts were prepared from rat or lamprey brain using established pro- tiprotein complex might then serve as a template to scaffold cedures (13, 23). For immunoprecipitation experiments, antibodies were coupled to protein A/G PLUS Agarose (Santa Cruz Biotechnology) and assembly of AP2-containing coat complexes (19) at PI(4,5)P2- incubated with 4 mg rat or lamprey brain Triton X-100 extract in buffer A

enriched membrane sites within the periactive zone. (20 mM Hepes buffer, pH 7.4, containing 50 mM KCl, 2 mM MgCl2,and1% Our data also indicate that AP2 binding to intersectin 1 might Triton X-100) plus protease inhibitors (Sigma) in a total volume of 1 mL for occlude recruitment of the PI(4,5)P2-phosphatase synaptojanin 1. 2 h at 4 °C. Following extensive washes, samples were eluted with sample These results suggest a potential, although at this stage clearly hy- buffer and analyzed by SDS/PAGE and immunoblotting. For afﬁnity chormatography or direct binding experiments, 100 or 50 μgGST-fusion pothetical, mechanism whereby direct complex formation between proteins were coupled to GST-Bind resin (Novagen) and incubated with

AP2 and intersectin inhibits premature recruitment of synaptojanin 1 2mgratbrainextractor10μgHis6-tagged recombinant protein, respec- to CCPs at early stages, thereby preventing premature PI(4,5)P2- tively, in a total volume of 1 mL for 1 h at 4 °C on a rotating wheel. Fol- hydrolysis. How can such competition be explained in molecular lowing extensive washes, samples were eluted with sample buffer and terms? Earlier data have shown that synaptojanin 1 preferentially analyzed by SDS/PAGE and immunoblotting. associates with the SH3A domain of intersectin 1 (13), a site in close Microinjection of Compounds into Lamprey Reticulospinal Axons and Analysis proximity to the major AP2 binding sequence within the SH3A-B by Electron Microscopy. Spinal cord preparations, microinjection experiments, linker region. It is thus possible that intersectin 1-bound AP2 by steric and electron microscopic analysis were essentially done as described before hindrance selectively impairs synaptojanin 1-p145 recruitment to (13). See SI Materials and Methods for details. intersectin 1-containing endocytic sites during the initial stages of CCP assembly. As clathrin recruitment and assembly proceeds, Protein Crystallography. Detailed procedures of protein crystallography are available as SI Materials and Methods. Briefly, crystals were grown at 18 °C engagement of the clathrin terminal domain with the side site of the using the sitting-drop vapor-diffusion method. X-ray data were collected at AP2 β-appendage (20) might release the AP2 clamp from the beamline BL2 at BESSY-II, Berlin and processed using HKL2000 and scalepack. intersectin linker, allowing for the recruitment of synaptojanin 1 to The phase problem was solved by molecular replacement with CCP4 pro- intersectin 1 within the assembled coat. Likely, other endocytic SH3- gram molrep using the 1.60-Å resolution structures of the α- and β2- domain containing proteins, most notably endophilin, are involved in appendage domains (PDB code 1KYF) as models. The statistics of the structures are reported in Table S1. the stable tethering of synaptojanin 1 to late-stage endocytic intermediates (17), thereby initiating PI(4,5)P2 hydrolysis and uncoating ACKNOWLEDGMENTS. We thank Drs. Thomas Südhof (Stanford University, (19). This hypothetical scenario fits well with recent TIRF-based CA), Pietro De Camilli (Yale University, CT), and Gero Miesenbock (Oxford University) for the kind gift of reagents. This work was supported by the imaging data showing that neuronal synaptojanin 1-p145 is recruited Deutsche Forschungsgemeinschaft Grant HA2686/ 3-1/ FOR 806 (to V.H.), to CCPs at late stages preceding membrane fission, but is excluded Swedish Research Council (Projects 13473, 20587), the Wallenbergs Founda- from early endocytic intermediates (21) and is compatible with the tion and Linné Center DBRM (O.S.), Grant SFB 449/ Z3 (to W.S.), (Exc 257- Neurocure), and the European Science Foundation (O.S. and V.H.). A.P. and J. metastability of newly assembling CCPs observed in living cells (22). B. are the recipients of scholarships from the Deutsche Forschungsgemein- Future studies will need to address these possibilities further. schaft (GRK 1123).

1. Murthy VN, De Camilli P (2003) Cell biology of the presynaptic terminal. Annu Rev 13. Evergren E, et al. (2007) Intersectin is a negative regulator of dynamin recruitment to Neurosci 26:701–728. the synaptic endocytic zone in the central synapse. J Neurosci 27:379–390. 2. Sudhof TC (2004) The synaptic vesicle cycle. Annu Rev Neurosci 27:509–547. 14. Anggono V, et al. (2006) Syndapin I is the phosphorylation-regulated dynamin I 3. Shupliakov O (2009) The synaptic vesicle cluster: a source of endocytic proteins during partner in synaptic vesicle endocytosis. Nat Neurosci 9:752–760. neurotransmitter release. Neuroscience 158:204–210. 15. Edeling MA, et al. (2006) Molecular switches involving the AP-2 beta2 appendage 4. Granseth B, Odermatt B, Royle SJ, Lagnado L (2006) Clathrin-mediated endocytosis is regulate endocytic cargo selection and clathrin coat assembly. Dev Cell 10:329–342. Neuron the dominant mechanism of vesicle retrieval at hippocampal synapses. 51: 16. Schmid EM, et al. (2006) Role of the AP2 beta-appendage hub in recruiting partners – CELL BIOLOGY 773 786. for clathrin-coated vesicle assembly. PLoS Biol 4:e262. 5. Bonifacino JS, Lippincott-Schwartz J (2003) Coat proteins: shaping membrane transport. 17. Ringstad N, et al. (1999) Endophilin/SH3p4 is required for the transition from early to Nat Rev Mol Cell Biol 4:409–414. late stages in clathrin-mediated synaptic vesicle endocytosis. Neuron 24:143–154. 6. Schmid EM, McMahon HT (2007) Integrating molecular and network biology to 18. Yu Y, et al. (2008) Mice deficient for the chromosome 21 ortholog Itsn1 exhibit decode endocytosis. Nature 448:883–888. vesicle-trafficking abnormalities. Hum Mol Genet 17:3281–3290. 7. Wieffer M, Maritzen T, Haucke V (2009) SnapShot: endocytic trafficking. Cell 137:382. 19. Cremona O, et al. (1999) Essential role of phosphoinositide metabolism in synaptic e1–3. vesicle recycling. Cell 99:179–188. 8. Jung N, Haucke V (2007) Clathrin-mediated endocytosis at synapses. Traffic 8: 20. Thieman JR, et al. (2009) Clathrin regulates the association of PIPKIgamma661 with 1129–1136. the AP-2 adaptor beta2 appendage. J Biol Chem 284:13924–13939. 9. Koh TW, et al. (2007) Eps15 and Dap160 control synaptic vesicle membrane retrieval 21. Perera RM, Zoncu R, Lucast L, De Camilli P, Toomre D (2006) Two synaptojanin 1 and synapse development. J Cell Biol 178:309–322. Proc Natl Acad Sci 10. Koh TW, Verstreken P, Bellen HJ (2004) Dap160/intersectin acts as a stabilizing scaffold isoforms are recruited to clathrin-coated pits at different stages. USA – required for synaptic development and vesicle endocytosis. Neuron 43:193–205. 103:19332 19337. 11. Marie B, et al. (2004) Dap160/intersectin scaffolds the periactive zone to achieve high- 22. Loerke D, et al. (2009) Cargo and dynamin regulate clathrin-coated pit maturation. fidelity endocytosis and normal synaptic growth. Neuron 43:207–219. PLoS Biol 7:e57. 12. Thomas S, et al. (2009) Intersectin regulates dendritic spine development and somato 23. Diril MK, Wienisch M, Jung N, Klingauf J, Haucke V (2006) Stonin 2 is an AP-2-dependent dendritic endocytosis but not synaptic vesicle recycling in hippocampal neurons. J Biol endocytic sorting adaptor for synaptotagmin internalization and recycling. Dev Cell 10: Chem 284:12410–12419. 233–244.

Pechstein et al. PNAS | March 2, 2010 | vol. 107 | no. 9 | 4211 Downloaded by guest on September 27, 2021