Abstract
The release of neurotransmitters (NTs) at central synapses is dependent on a cascade of protein interactions, specific to the presynaptic compartment. Among those dedicated molecules, the cytosolic complexins play an incompletely defined role as synaptic transmission regulators. Complexins are multidomain proteins that bind soluble N-ethylmaleimide sensitive factor attachment protein receptor complexes, conferring both inhibitory and stimulatory functions. Using systematic mutagenesis and comparing reconstituted in vitro membrane fusion assays with electrophysiology in cultured neurons from mice of either sex, we deciphered the function of the N-terminus of complexin (Cpx) II. The N-terminus (amino acid 1–27) starts with a region enriched in hydrophobic amino acids (1–12), which binds lipids. Mutants maintaining this hydrophobic character retained the stimulatory function of Cpx, whereas exchanges introducing charged residues perturbed both spontaneous and evoked exocytosis. Mutants in the more distal region of the N-terminal domain (amino acid 11–18) showed a spectrum of effects. On the one hand, mutation of residue A12 increased spontaneous release without affecting evoked release. On the other hand, replacing D15 with amino acids of different shapes or hydrophobic properties (but not charge) not only increased spontaneous release but also impaired evoked release. Most surprising, this substitution reduced the size of the readily releasable pool, a novel function for Cpx at mammalian synapses. Thus, the exact amino acid composition of the Cpx N-terminus fine-tunes the degree of spontaneous and evoked NT release.
Significance Statement
We describe in this work the importance of the N-terminal domain of the small regulatory cytosolic protein complexin (Cpx) in spontaneous and evoked glutamatergic neurotransmitter release at hippocampal mouse neurons. We use biochemical assays to screen for amino acids of interest in the Cpx N-terminus and test these residues for functional relevance in spontaneous and Ca2+-triggered synaptic vesicle (SV) exocytosis using electrophysiology assays and site-directed mutagenesis. In addition to identifying crucial residues for clamping spontaneous release and promoting Ca2+-evoked transmission, we identify a single amino acid at Position D15 which determines SV priming, a function that was never before attributed to Cpx at vertebrate synapses.
Introduction
The transmission of information at synapses requires tightly controlled neurotransmitter (NT) release. Presynaptic proteins orchestrate the fusion of NT-filled synaptic vesicles (SVs) with the plasma membrane to release their contents onto the opposing postsynaptic site. The SNARE (soluble N-ethylmaleimide sensitive factor attachment protein receptor) proteins, namely, VAMP2, Syntaxin1a, and SNAP25, form the core SV fusion machinery (Sørensen et al., 2006; Zhou et al., 2015; Zhang et al., 2022). The calcium-sensor synaptotagmin 1, together with another protein modulating release, complexin (Cpx), clamp the complex together or stimulate membrane fusion in the absence or presence of Ca2+, respectively (DiAntonio and Schwarz, 1994; Fernández-Chacón et al., 2001; Reim et al., 2001; Huntwork and Littleton, 2007; Südhof, 2013; Rizo and Xu, 2015).
Cpxs are the best characterized SNARE complex binding proteins (Trimbuch and Rosenmund, 2016). Of the four Cpx isoforms in vertebrates, Cpx I, II, and III are expressed in the brain and the retina, whereas Cpx IV is found only in retinal ribbon synapses (Reim et al., 2005). In autaptic neurons, Cpxs inhibit synaptic transmission by arresting spontaneous SNARE complex assembly (Xue et al., 2010; Malsam et al., 2020), but paradoxically, they facilitate action potential-evoked exocytosis (Reim et al., 2001; Trimbuch et al., 2014). However, discrepancies in the role of Cpx in synaptic transmission at vertebrate synapses have been reported using different experimental approaches (Maximov et al., 2009; Yang et al., 2010, 2013; Kaeser-Woo et al., 2012). Across species, the role of Cpx apparently varies. In Caenorhabditis elegans and Drosophila, Cpx knock-out (KO) causes an increase in spontaneous release, suggesting a role of Cpx in inhibiting, or clamping, the fusion of release-ready vesicles, but KO also decreases the size of the readily releasable pool (RRP) and the evoked response (Huntwork and Littleton, 2007; Wragg et al., 2013; Cho et al., 2014). Similar results are obtained in neuroendocrine cells, where a Cpx II KO results in more spontaneous and less evoked release (Dhara et al., 2014). In mammalian neurons, Cpx predominantly stimulates NT release. For example, in neurons where Cpxs I–III are knocked out (Cpx-triple knock-out, Cpx-TKO), the evoked response is decreased but neither the spontaneous synaptic release nor the size of the RRP is changed (Xue et al., 2010; Trimbuch et al., 2014). Cross-species rescue experiments suggest that Cpx may have adapted to modulate release through suppression of spontaneous release or by boosting Ca2+-triggered release, depending on the needs of the synapse (Xue et al., 2009).
Cpx is made up of four main domains: the N-terminal domain (NTD), the accessory helix (AH), the central alpha-helix (CH), and the C-terminal domain (CTD). Each of those domains shows distinct interactions and physiological functions (Xue et al., 2007). The CH binds and stabilizes the partially assembled SNARE complex, whereas the AH arrests full SNARE complex assembly (Zhou et al., 2017; Malsam et al., 2020). The CTD of Cpx is involved in lipid interactions, stopping SNARE complex assembly, and has been linked to fusion pore formation and pore stabilization (Courtney et al., 2022; Hao et al., 2023). Finally, the NTD is crucial for the facilitatory function of Cpx in mammalian neurons, but its mode of action is still poorly understood (Xue et al., 2007, 2009, 2010). While no structural data are available so far, biochemical and nuclear magnetic resonance (NMR) spectroscopy data hint that the NTD may either interact with membranes (Zdanowicz et al., 2017) or fold back to interact with the SNARE complex (Xue et al., 2007). In addition, the NTD shows species-specific functions (Xue et al., 2008; Yang et al., 2013). Taken together, these results motivated us to perform a systematic examination of the function of the Cpx NTD. In this study, we aimed to dissect the distinct roles that the NTD of Cpx plays in spontaneous and Ca2+-evoked NT release using a structure–function approach. Amino acids of interest were identified with biochemical assays and tested for functional relevance in the synapse using site-directed mutagenesis of rescue constructs and electrophysiological recordings. Our results show that targeted mutations of single amino acids lead to distinct functional phenotypes that modulate synaptic neurotransmission and indicate an unexpectedly fine balance in the stimulatory role of Cpx on Ca2+-triggered release.
Materials and Methods
Animals maintenance and mouse lines
All mouse experiments were performed in accordance with the regulation of the animal welfare committee of the Charité–Universitätsmedizin Berlin. Time-pregnant females were anesthetized and killed at Embryonic Day (E)18 according to permission from the Landesamt für Gesundheit und Soziales Berlin under the license number G106/20.
Primary hippocampal cultures
Cpx I–III triple KO neurons were generated as previously described (Xue et al., 2008). Primary murine hippocampal neurons were prepared from E18 mice of either sex, as described previously (Arancillo et al., 2013). Briefly, hippocampi were dissected, and neurons dissociated by an enzymatic treatment using 25 units per milliliter of papain for 45 min at 37°C. For conventional microscopy and Western blot (WB) experiments on high-density cultures, 100 × 103 neurons/well (35 mm diameter) were plated on wild-type (WT) continental astrocyte feeder layers (Chang et al., 2018). For electrophysiology, low-density cultures of 3 × 103 neurons/well (35 mm diameter) were seeded on astrocyte microislands for autaptic cultures (Arancillo et al., 2013). Astrocyte feeder layers were prepared 1–2 weeks before neuronal seeding, as described previously (Arancillo et al., 2013). After plating, neurons were incubated in Neurobasal-A medium supplemented with 50 μg/ml streptomycin and 50 IU/ml penicillin at 37°C. For electron microscopy (EM), sapphire disks (6 × 0.12 mm, Wohlwend, art. #1292) were coated with PDL + collagen, and ∼30 × 103 cells/cm2 were seeded onto an astrocytic feeder layer. Electrophysiological, imaging, WB, or EM experiments were performed at DIV 14–20.
Lentiviral constructs and virus production
All lentiviral constructs were generated through the Gibson assembly method (New England Biolabs) with the corresponding cDNAs and with a human synapsin-1 promoter-driven lentiviral shuttle vector [f(syn), based on FUGW (Lois et al., 2002)] that contained a nuclear-localized signal (NLS) GFP that was fused N-terminally to a self-cleaving P2A peptide (Kim et al., 2011) to allow polycistronic translation. The Cpx II variants were cloned in frame upstream of the P2A sequence leading to the constructs f(syn)Cpx II-P2A-NLS.GFP-WPRE to allow expression of an unmodified M1 start amino acid of Cpx II. Lentiviral particles were prepared by the Charité Viral Core Facility as previously described (Lois et al., 2002; vcf.charite.de). Briefly, HEK293T cells were cotransfected with the shuttle vector f(syn)Cpx II-P2A-NLS.GFP-WPRE and helper plasmids, pCMVdR8.9 and pVSV.G with polyethylenimine. Virus containing supernatant was collected after 72 h, filtered, aliquoted, flash-frozen with liquid nitrogen, and stored at −80°C. For infection, ∼5 × 105–1 × 106 infectious virus units were pipetted onto DIV 1 hippocampal Cpx I–III triple KO neurons per 35-mm-diameter well.
Constructs for biochemical reconstitution assays
For reconstitution into small unilamellar vesicles (SUVs), protein constructs containing full-length VAMP2 fused to glutathione S-transferase (GST) and synaptotagmin 1-His6 lacking the luminal domain (Syt1, amino acids 57–421) are encoded by plasmids pSK28 (Kedar et al., 2015) and by pLM6 (Mahal et al., 2002), respectively. For reconstitution into giant unilamellar vesicles (GUVs), the t-SNARE complex consisting of Syntaxin1a and SNAP25B is encoded by pSK306. Synthetic genes encoding full-length rat Syntaxin1a (1–288) with an N-terminal His6-tag followed by a PreScission protease cleavage site and full-length mouse SNAP25 (1–206) with an N-terminal GST-tag followed by a PreScission protease cleavage site were subcloned into the bicistronic expression plasmid pETDuet1 resulting in pSK306. The cysteines in the SNAP25 linker region were replaced by serine residues. Short N-terminal extensions after removal of both tags with PreScission protease are highlighted in bold. Syntaxin1a, GPGMKDRTQELRTAKDSDDDDDVTVTVDRDRFMDEFFEQVEEIRGFIDKIAENVEEVKRKHSAILASPNPDEKTKEELEELMSDIKKTANKVRSKLKSIEQSIEQEEGLNRSSADLRIRKTQHSTLSRKFVEVMSEYNATQSDYRERSKGRIQRQLEITGRTTTSEELEDMLESGNPAIFASGIIMDSSISKQALSEIETRHSEIIKLENSIRELHDMFMDMAMLVESQGEMIDRIEYNVEHAVDYVERAVSDTKKAVKYQSKARRKKIMIIICCVILGIIIASTIGGIFG, and SNAP25, GPLGCGSSGMAEDADMRNELEEMQRRADQLADESLESTRRMLQLVEESKDAGIRTLVMLDEQGEQLERIEEGMDQINKDMKEAEKNLTDLGKFSGLSVSPSNKLKSSDAYKKAWGNNQDGVVASQPARVVDEREQMAISGGFIRRVTNDARENEMDENLEQVSGIIGNLRHMALDMGNEIDTQNRQIDRIMEKADSNKTRIDEANQRATKMLGSG.
To generate Cpx II point mutants, we used the QuikChange DNA mutagenesis kit (Qiagen; Ruiter et al., 2019) and the template encoding WT His6-Cpx II (pMDL80, Malsam et al., 2020). Thereby, the following Cpx II constructs were established: A12W (pJAC20), D15W (pSK258), D15N (pLB37), D15A (pSK255), and D15K (pSK259). In addition, native amino acids at the following Cpx II positions were replaced by 4-Benzoyl-L-Phenylalanine (BPA; Bachem) as described by Malsam et al. (2020): M1Bpa (pSK176), F3Bpa (pSK167), V4Bpa (pSK168), M5Bpa (pSK169), A8Bpa (pSK172), L9Bpa (pSK173), G10Bpa (pSK174), G11Bpa (pSK175), A12Bpa (pSK178), T13Bpa (pSK196), K14Bpa (pSK197), D15Bpa (pSK198), M16Bpa (pSK199), G17Bpa (pSK200), and K18Bpa (pSK201). The identity of all constructs was validated by DNA sequencing.
Protein expression and purification
In general, Escherichia coli (E. coli) BL21 (DE3; Stratagene) cells were transformed with expression vectors encoding the desired protein constructs. At an OD660 between 0.6 and 0.8, addition of 0.3 mM isopropyl-β-D-thiogalactopyranosid (IPTG)-induced protein expression. Afterward, cells were harvested by centrifugation (3,500 × g, 15 min, Sorvall H-1200) and lysed using a high-pressure pneumatic processor 110L (Microfluidizer). Cell fragments and insoluble material were removed by centrifugation at 60,000 rpm (70Ti rotor, Beckman Coulter) for 1 h, and the clarified supernatants were aliquoted and snap-frozen in liquid nitrogen.
Full-length VAMP2 (V2) was purified as described previously (Weber et al., 1998; Malsam et al., 2012) with the following modifications: cells were grown in ZYM media (Studier, 2005), and protein expression was induced with 0.3 mM IPTG for 3 h at 25°C.
Purification of Synaptotagmin-1-His6 (Syt1) was performed according to Malsam et al. (2012) with the following modifications: Ni-NTA beads with bound Syt1 were washed three times, 20 ml Mg-ATP wash buffer [25 mM HEPES-KOH, 400 mM KCl, 10% (w/v) glycerol, 20 mM imidazole, 1% (w/v) Triton X-100, 2 mM Mg-ATP, 3 mM β-mercaptoethanol (β-ME)], pH 7.5; 20 ml calcium wash buffer [25 mM HEPES-KOH, 400 mM KCl, 10% (w/v) glycerol, 20 mM imidazole, 20 mM CaCl2, 1% (w/v) Triton X-100, 3 mM β-ME], pH 7.5; and 20 ml EGTA wash buffer [25 mM HEPES-KOH, pH 7.5, 100 mM KCl, 10% (w/v) glycerol, 0.5 mM EGTA, 1% octyl-β-D-glucopyranoside (β-OG), 3 mM β-ME]. Syt1 was eluted with EGTA wash buffer containing 500 mM imidazole instead of EGTA, diluted twice to lower salt concentration to 50 mM and bound to a MonoS Sepharose column (GE Healthcare Biosciences). Finally, Syt1 was eluted with a gradient of 50 mM to 1 M KCl in 25 mM HEPES-KOH, pH 7.4, 1 mM Mg-ATP, 100 µM EGTA, 200 mM sucrose, 10% (w/v) glycerol, 1% (w/v) β-OG, and 1 mM DTT.
t-SNARE complexes consisting of full-length Syntaxin1a and SNAP25B were coexpressed in BL21-DE3 E. coli with 0.5 mM IPTG for 3 h at 25°C and lysed [50 mM HEPES-KOH, 400 mM KCl, 10% (w/v) glycerol, 200 mM sucrose], pH 7.4. Preassembled t-SNARE is bound to GST-beads in lysis buffer [1% (w/v) Triton X-100, 1.5% (w/v) cholate, 2 mM Mg-ATP, 5 mM TCEP] and washed with cleavage buffer [25 mM HEPES-KOH, 200 mM KCL, 10% glycerol, 200 mM sucrose, 1% (w/v) β-OG, 5 mM TCEP], pH 7.4. After an overnight cleavage with 200 U PreScission protease (Cytiva) and addition of fresh GST-beads to bind free GST and PreScission protease, the protein complex is purified by binding to a MonoQ Sepharose column (GE Healthcare Biosciences) at 90 mM KCl [sample dilution with 25 mM HEPES-KOH, 200 mM sucrose, 5% (w/v) glycerol and 1% (w/v) β-OG] and eluted with a gradient of 100 mM to 1 M KCl in 20 mM MOPS-KOH, pH 7.4, 200 mM sucrose, 5% (w/v) glycerol, and 1% (w/v) β-OG.
His6-Cpx II was expressed and purified as described by Malsam et al. (2012), but the expression was performed in BL21-DE3 codon+ E. coli for 2 h at 27°C, and no pefablock was used. Cpx II Bpa mutants were expressed as described in Malsam et al. (2020) and purified like the WT.
The concentrations of purified proteins were determined by SDS-PAGE and Coomassie blue staining using BSA as a standard and the Fiji software for quantification. For reconstitution assays, mutant and WT protein concentrations were directly compared on a single gel.
Protein reconstitution into liposomes
Atto488-DPPE (1,2-dipalmitoyl-SN-glycero-3-phosphoethanolamine) and Atto550-DPPE were purchased from ATTO-TEC, all other lipids were from Avanti Polar Lipids. For SUVs containing VAMP2 and Syt1, lipid mixes (3 µmol total lipid) with the following composition were prepared: 25 mol% 1-palmitoyl-2-oleoyl-SN-glycero-3-phosphoethanolamine (POPE), 29 mol% 1-palmitoyl-2-oleoyl-SN-glycero-3-phosphocholine (POPC), 25 mol% cholesterol (from ovine wool), 5 mol% PI (L-a-phosphatidylinositol), 15 mol% 1,2-dioleoyl-SN-glycero-3-phosphoserine (DOPS), 0.5 mol% Atto488-DPPE, and 0.5 mol% Atto550-DPPE. The t-SNARE GUV lipid mix (5 µmol total lipid) had the following composition: 34 mol% POPC, 15 mol% DOPS, 20 mol% POPE, 25 mol% cholesterol, 4 mol% PI, 1 mol% PI(4,5)P2 (L-a-phosphatidylinositol-4,5-bisphosphate), 0.05 mol% Atto647-DPPE, and 0.5 mol% tocopherol.
Synaptotagmin 1 and VAMP2 were reconstituted into SUVs at a protein-to-lipid ratio of 1:900 and 1:350 as described previously (Malsam et al., 2020) with the following buffer changes: 3 µmol dried lipid mix was resuspended with VAMP2 in 1.7% (w/v) octyl-β-D-glucopyranoside (β-OG) containing dilution buffer (25 mM HEPES-KOH, 550 mM KCl, 100 µM EGTA), pH 7.4. After addition of Syt1, small unilamellar Syt1/VAMP2 vesicles were formed by rapid detergent dilution by a threefold volume increase with β-OG free dilution buffer. Overnight dialysis (25 mM HEPES-KOH, 135 mM KCl, 100 µM EGTA, 1 mM DTT), pH 7.4, removed the detergent, followed by centrifugation via Nycodenz gradient to concentrate the liposomes. After a second overnight dialysis (25 mM HEPES-KOH, 135 mM KCl, 10 µM EGTA, 1 mM DTT), pH 7.4, SUVs were snap-frozen and stored at −80°C.
The reconstitution of t-SNAREs into SUVs and the generation of t-SNARE GUVs by electroswelling have been described previously (Malsam et al., 2012, 2020, respectively) and recent modifications of the protocol are specified in Kádková et al. (2024). The final lipid-to-protein ratios were determined by Atto647 fluorescence intensity measurements of the lipids and Coomassie blue staining of the proteins separated by SDS-PAGE.
Lipid mixing assay
Membrane fusion (lipid mixing) assays were performed as described previously (Malsam et al., 2012; Kádková et al., 2024). Briefly, t-SNARE GUVs (14 nmol lipid) were preincubated with VAMP2/Syt1-SUVs (2.5 nmol lipid) in the presence or absence of 6 µM Cpx II for 5 min on ice in 100 µl fusion buffer. After transferring samples into a prewarmed 96-well plate (37°C), the emitted fluorescence of Atto488 (ƛex = 485 nm; ƛem = 538 nm) was measured at intervals of 10 s. Ca2+ was added to a final free concentration of 100 µM after 2 min. Addition of 0.7% (w/v) SDS and 0.7% (w/v) n-Dodecyl-β-D-Maltosid after 4 min stopped the fusion reactions. The measured fusion-dependent fluorescent signals were normalized to the “maximum” fluorescent signal. As a negative control, SUVs were treated with botulinum neurotoxin type D to inactivate VAMP2, and the corresponding fluorescence signals were subtracted from individual measurement sets. For each mutant, three independent fusion experiments were performed. To determine the effect of the BPA mutagenesis, we normalized the fluorescence signals of the individual mutants to the WT values (Fig. 1A).
Functional Cpx II Bpa mutagenesis screen and lipid cross-links in the NTD. A, In vitro lipid mixing of Syt1/VAMP2 SUVs with t-SNARE GUVs in the absence/presence of Cpx-WT or Bpa mutants. Incubation scheme of the reconstituted assay (top-left panel). Representative fusion kinetics show fusion clamping by Cpx (before Ca2+ addition) and stimulation after Ca2+ addition at 2 min (bottom left, the fusion kinetic of Cpx II-D15K is shown). Positional Cpx mutants cause distinct effects before (right panel, red bars) and after Ca2+ addition (blue bars). All fusion values were normalized to the respective Cpx-WT value (100%). * Cpx Bpa variants D2Bpa, K6Bpa, and Q7Bpa could not be purified. Error bars indicate SEM with n = 3. B, Lipid-cross-links of positional Cpx Bpa mutants. Experimental incubation scheme is shown in the top-left panel. Representative cross-links of Cpx mutants to alkyne PC in Syt1/VAMP2 SUVs, as detected by click labeling of cross-linked alkyne PC with Alexa Fluor 647 and SDS-PAGE analysis (bottom-left panel). Quantification of positional lipid cross-links (right panel). Error bars indicate SEM with n = 3–6. Dotted line indicates background lipid cross-linking to Cpx-WT, lacking Bpa.
Detection of lipid interactions via cross-linking of Cpx II BPA mutants in an alkyne PC-containing bilayer
For detection of lipid cross-links (Hempelmann et al., 2023), 22 mol% POPC in the SUV bilayer was replaced by the corresponding amount of alkyne phosphocholine [alkyne PC; 16:0(alkyne)-18:1 PC, Avanti Polar Lipids]. vSNARE/Syt1-SUVs (5 nmol lipid) and t-SNARE GUVs (28 nmol lipid) were incubated with 6 µM Cpx-WT or BPA mutants in glutathione containing fusion buffer on ice for 30 min to obtain efficient docking and priming as described previously (Malsam et al., 2012). A cushion of 100 μl 70 mM sucrose containing fusion buffer was underlaid with 5 μl high sucrose buffer (240 mM sucrose, 1 mM EPPS, 1 mM DTT), pH 8.0, in a separate low-binding tube. After the incubation, samples were layered on top of the 105 μl cushion and spun at 10,000 × g for 20 min at 4°C in a swing out rotor (A-8–11 swing bucket rotor, Eppendorf) to isolate Cpx II bound to trans-SNARE complexes (SNAREpins) formed between the SUVs and GUVs. The supernatant containing free SUVs and soluble Cpx II was removed. The pellet was resuspended gently by short vortexing, and the suspension was exposed to UV light [365 nm, UV-LED lamp (Opsytec Dr. Gröbel) with 15 pulses of 1 s illumination (25 W/cm2), pausing for 2 s between pulses.] To fluorescently label the alkyne PC by click chemistry, final concentrations of the following reagents (in µM), 3,125 CuSO4, 125 Tris-[(1-benzyl-1H-1,2,3-triazol-4-yl)-methyl]-amin, 3,125 ascorbic acid, and 125 Picolyl-Alexa647 (Jena Bioscience), were added to the mix and incubated for 2 h at 37°C while gently shaking at 400 rpm. Proteins were precipitated by adding 1 ml precooled methanol (−20°C) and incubated overnight at −80°C. After centrifugation (20,000 × g, 30 min, 0°C), supernatants were removed, and the pellets were dried and resuspended in 12 µl 2× Laemmli buffer. Samples were analyzed by SD-PAGE and scanned for Picolyl-Alexa647 signal before protein staining with Coomassie blue. Fluorescence intensities corresponding to cross-linked lipids were normalized to respective Coomassie blue-stained Cpx II intensities using Fiji (Fig. 1B).
High-pressure freezing and freeze substitution
Sample preparation for EM using high-pressure freezing (HPF) and freeze substitution was performed as described previously (Weber-Boyvat et al., 2022). For HPF, neurons were cultured on sapphire glass disks and cryofixed at DIV15–16 using a Leica ICE HPF (p > 2,000 bar, cooling rate 10,000–12,000 K/s). After a brief wash in extracellular solution containing the following (in mM), 140 NaCl, 2.4 KCl, 10 HEPES, 2 CaCl2, 4 MgCl2, and 10 glucose, pH adjusted to 7.3 (with NaOH, 300 mOsm), sandwiches for HPF were assembled. The sandwiches consisted of the cultured sapphire glass disk, a 100 nm spacer ring, a second blank sapphire glass disk, and a 400 nm spacer ring; the cavity between the sapphire disks was filled with extracellular solution (without cryoprotectants). Directly after the assembly (∼2 min), the samples were high-pressure frozen. The table and chamber of the HPF, as well as the extracellular solution were kept at physiological temperature (37°C).
The freeze substitution was carried out using a Leica AFS2 freeze substitution device precooled to −90°C. After the freezing, the sandwiches were disassembled in anhydrous acetone, and the samples were subsequently transferred to cryotubes with a freeze substitution solution containing 1% (w/v) osmium tetroxide, 1% (v/v) glutaraldehyde, and 1% ddH2O in anhydrous acetone (both steps inside AFS2). The AFS2 was warmed up using the following stepwise heating protocol: −90°C (min. 5 h), −90 to −20°C (14 h; 5°C/h), −20°C (12 h), and −20 to 20°C (4 h; 10°C/h). At room temperature (RT), the sapphire glass disks were washed 4 × 15 min with anhydrous acetone, stained in 0.1% (w/v) uranyl acetate in acetone for 1 h, and again washed 4 × 15 min with anhydrous acetone. Epoxy resin [42.8% (w/w) epon, 31.2% (w/w) DDSA, 26% (w/w) MNA; Fluka] was used for infiltration in ascending concentration: 2 h in 30% epoxy/acetone, 2 h in 70% epoxy/acetone, and overnight pure epoxy. The following day, the sapphire glass disks were transferred to embedding molds containing epon/araldite [26.3% (w/w) epon, 18.7% (w/w) araldite, 51.9% (w/w) DDSA, 2.98% (w/w) BDMA; EMS] for polymerization (60°C, 48 h).
EM
Sapphire glass disks were removed from epon blocks using thermal shock. Ultrathin sections (50 nm) were prepared with an Ultracut UCT ultramicrotome (Leica Microsystems) and collected on formvar-coated 200 mesh copper or nickel grids. The sections were poststained for contrast enhancement using 2% uranyl acetate (w/v) in ddH2O for ∼3 min and Reynold's lead citrate for ∼1 min. For image acquisition, a FEI Tecnai G2 20 transmission electron microscope with an accelerating voltage of 200 kV and equipped with a Veleta 2 × 2 K CCD camera (Olympus) was used. Hippocampal synapses were preselected at low magnification to avoid bias and imaged subsequently with a pixel size of 0.71 nm.
Image analysis
Image analysis and quantification were performed using custom ImageJ macros and Python scripts (Weber-Boyvat et al., 2022; https://github.com/janakroll/synapse-analysis). In ImageJ, electron micrographs of all groups per culture were pooled as image stacks, and the order of images was mixed randomly. Only following this randomization were images excluded if their quality was too poor for quantification, if the active zone (AZ) membrane was cut transversally, or if the synapse showed signs of necrosis (swollen or washed out boutons). About 70 images per group and culture were acquired, of which ∼60 images were analyzed.
In ImageJ, the presynaptic AZ membrane, docked SVs and cytosolic SVs were marked. The AZ membrane was defined as the part of the presynaptic membrane that is opposed to the postsynaptic density. SVs were counted as docked SVs if no cleft was visible between outer SV membrane and AZ membrane. Using Python, the distribution of SVs was assessed by calculating the shortest distance between SV membranes and AZ membrane.
Electrophysiology
Whole-cell patch–clamp recordings were performed on autaptic cultures at RT at DIV 14–21. Synaptic currents were recorded using a Multiclamp 700B amplifier (Axon Instruments) controlled by Clampex 9 software (Molecular Devices). A fast perfusion system (SF-77B; Warner Instruments) continuously perfused the neurons with the extracellular solution containing the following (in mM): 140 NaCl, 2.4 KCl, 10 HEPES (Merck Millipore), 10 glucose (Carl Roth), 2 CaCl2 (Sigma-Aldrich), and 4 MgCl2 (Carl Roth; ∼300 mOsm), pH 7.4. Somatic whole-cell recordings were obtained using borosilicate glass pipettes, with a tip resistance of 2–4 MΩ and filled with the following internal solution (in mM): 136 KCl, 17.8 HEPES, 1 EGTA, 4.6 MgCl2, 4 Na2ATP, 0.3 Na2GTP, 12 creatine phosphate, and 50 U/ml phosphocreatine kinase (∼300 mOsm), pH 7.4. Membrane capacitance and series resistance were compensated by 70% und data filtered by low-pass Bessel filter at 3 kHz and sampled at 10 kHz using an Axon Digidata 1322A digitizer (Molecular Devices).
Neurons were clamped at −70 mV and action potentials triggered by a 2 ms depolarization to 0 mV to measure EPSCs (excitatory postsynaptic currents). Afterward, EPSCs were induced in the presence of the competitive AMPA receptor antagonist NBQX (3 μM, Tocris Bioscience). Paired-pulse stimulation for excitatory autapses was assessed by the induction of two action potentials with an interstimulus interval of 20 ms. Spontaneous events (miniature EPSCs, mEPSCs) were detected for 40 s. Electrophysiological traces were filtered at 1 kHz, and the range of parameters for inclusion of selected events using a conventionally defined template algorithm in AxoGraph X (AxoGraph Scientific) were 0.15–1.5 ms rise time and 0.5–5 ms half-width, and false-positive mEPSC events obtained in NBQX (3 µM) were subtracted to calculate the frequency of spontaneous events. The RRP was determined by applying a hypertonic 500 mM sucrose solution for 5 s and integrating the transient inward response component (Rosenmund and Stevens, 1996). The probability of vesicular release (Pvr) was calculated by dividing the average charge of the EPSC by the RRP charge. Spontaneous release rate was calculated by dividing the mEPSC frequency by the number of SVs in the RRP. The number of SVs in the RRP was calculated by dividing the RRP size by the mEPSC charge.
In order to estimate SV fusogenicity, we applied on each neuron in paired experiments 500 mM for 5 s or 250 mM sucrose solution for 10 s. The charge transfer of the transient synaptic current was measured and divided by the RRP size from the same neuron to obtain the fraction of RRP released by 250 mM sucrose solution. The onset of sucrose response was identified. We measured the time from the onset of the sucrose solution application to the onset of the response manually (T1). We used an “open tip” response configuration to calculate the time needed for the sucrose solution to reach the neuron (T2). Subsequently, the sucrose response onset latency was defined as the interval between the time when sucrose solution reaches the neuron and the onset time of sucrose response (T1–T2). To determine the peak release rate, we first digitally filtered the synaptic current traces at 10 Hz. The peak amplitude of sucrose response was then divided by the RRP size from the same neuron to obtain the peak release rate. Data were analyzed off-line using the AxoGraph X software (AxoGraph Scientific).
Immunostaining and quantification
Primary hippocampal neuronal mass cultures were rinsed in phosphate-buffered saline (PBS) and then fixed for 10 min in 4% paraformaldehyde, permeabilized with 0.1% PBS-Tween 20 solution, and blocked for an hour with 5% normal donkey serum. Primary antibodies were used to immune stain overnight at 4°C the following proteins: Cpx (1:1,000; Synaptic Systems; 122 002) and VGlut1 (1:4,000; Synaptic Systems; 135 304). Subsequently, secondary antibodies labeled with Alexa Fluor 647 anti-guinea pig IgG, Alexa Fluor 488 anti-rabbit, and Alexa Fluor 405 anti-chicken raised in donkey serum (each 1:500; Jackson ImmunoResearch Laboratories) were applied for 1 h at RT, respectively. After washing in PBS, glass coverslips were mounted on glass slides in Mowiol (Polysciences). Images were acquired with an Olympus IX81 epifluorescent microscope equipped with a MicroMax 1300YHS camera using the MetaMorph software (Molecular Devices). The analysis was performed off-line with ImageJ.
WB
For detection of Cpx protein levels by WB, protein lysates were obtained from mass cultures of Cpx-TKO hippocampal neurons (DIV 14–16) grown on WT astrocyte feeder layers. Briefly, cells were lysed using 50 mM Tris/HCl, pH 7.9, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, 1% Nonidet P-40, 1% sodium deoxycholate, and protease inhibitors (complete protease inhibitor cocktail tablet, Roche Diagnostics). Proteins were separated by SDS-PAGE and transferred overnight at 4°C to nitrocellulose membranes. After blocking with 5% milk powder (Carl Roth) for 1 h at RT, membranes were incubated with rabbit anti-Cpx I/II (1:1,000; Synaptic System) and mouse anti-tubulinIII (1:750; Sigma-Aldrich) antibodies for 1 h at RT. The membranes were washed several times with PBS-Tween before being incubated with the corresponding horseradish peroxidase-conjugated goat secondary antibodies (all from Jackson ImmunoResearch Laboratories). Protein expression levels were visualized with ECL Plus Western Blotting Detection Reagents (GE Healthcare Biosciences).
Structural analysis of Cpx mutants
Structures of Cpx were visualized in ChimeraX (Pettersen et al., 2021). The sequence of the mutated amino acid was imported from AlphaFold CoLab (Jumper et al., 2021; Mirdita et al., 2022). To generate illustrative representations of Cpx NTD mutants, we moved the NTD into place by hand.
Experimental design and statistical analysis
For our electrophysiological and immunocytochemical experiments, we recorded or imaged the same number of neurons per day to minimize variability. Neurons that were analyzed are represented on our bar graph as a single dot. The number of cells recorded or imaged is then explicitly written in the figure legend as well as the number of independent cultures recorded. Statistical tests were performed with Prism 7 (GraphPad Software). For bar plots data are represented as mean ± SEM. First, all data were tested for normality using the D’Agostino and Pearson’s test. If they pass the parametric assumption, the Kruskal–Wallis test was performed followed by Dunn's test. All experiments in the manuscript are repeated at least three times unless explicitly notified. Significance and p values were calculated and reported in the corresponding figure.
Data and materials availability
Data and materials of this study are available from the corresponding authors upon reasonable request.
Results
Biochemical identification of residues within the N-terminus of Cpx that modify membrane fusion
To assess the role of the NTD of Cpx in regulating spontaneous and Ca2+-triggered NT release, we performed a systematic mutagenesis screen. Individual amino acids were mutated, and the impact on vesicle fusion was assessed in a lipid mixing assay on reconstituted vesicles (Malsam et al., 2012, 2020). We incubated SUVs containing the SV-residing- (v-) SNARE VAMP2, the calcium-sensor synaptotagmin-1 (Syt1), and a quenching pair of fluorescence-labeled lipids with GUVs containing the plasma membrane target- (t-) SNAREs, Syntaxin1, and SNAP25, in the presence or absence of the Cpx isoform II, Cpx II (from here on referred to as Cpx; Fig. 1A, top left, liposome fusion assay). After a 5 min incubation at 0°C, lipid mixing was assessed at 37°C by measuring fluorescence before and after the addition of Ca2+. The presence of Cpx-WT strongly reduced the spontaneous fusion of reconstituted vesicles before Ca2+ was added. This reduction was observed from an approximately fourfold lower fluorescence signal from lipid mixing (dequenching) compared with signal in the absence of Cpx (Fig. 1A). The post-Ca2+ lipid mixing, reflecting Ca2+-triggered fusion, showed a slight tendency to increase in the presence of Cpx-WT (Fig. 1A). These results suggest that lipid fusion assays can recapitulate the reported role of Cpx in clamping spontaneous fusion of SVs (Xue et al., 2010) but may lack the sensitivity or complex native environment to reflect the reported role of Cpx in facilitating Ca2+-triggered release (Xue et al., 2007, 2010).
We next investigated how single residues in the NTD of Cpx affect lipid mixing in this reconstituted vesicle fusion assay. Each of the first 18 NTD amino acids of Cpx was exchanged for the hydrophobic, photo-activatable, unnatural amino acid benzoyl-phenylalanine (Bpa; Young et al., 2010; Malsam et al., 2020). Mutants with aspartate at Position 2, lysine at Position 6, and glutamine at Position 7 mutated to Bpa (D2Bpa, K6Bpa, and Q7Bpa, respectively) were insoluble using our E. coli expression system and could therefore not be purified for use in this assay. Analogous to a standard mutagenesis approach, we measured the functional contribution of each amino acid with lipid fusion assays in vitro without photoactivation before and after the addition of Ca2+. With the exception of an inhibition by L9Bpa, none of the NTD Cpx mutants affected Ca2+-triggered fusion compared with Cpx-WT. However, Ca2+-independent, spontaneous fusion varied by a factor of 4 for the distinct mutants, ranging from 80 to 400% fluorescence signal when normalized to Cpx-WT (Fig. 1A). Generally, the introduction of the bulky and hydrophobic Bpa at amino acids 1–10 had only minor effects on the fusion kinetics. Bpa at positions 11–18 resulted in an increase in spontaneous fusion, reaching a maximal increase with mutation of the aspartate at Position 15 (Fig. 1A, right, red bars; before Ca2+, 414 ± 57%; after Ca2+, 120 ± 7%).
Since the NTD of Cpx has also been implicated in lipid interactions (Zdanowicz et al., 2017), we assessed putative lipid interaction sites using a lipid cross-linking assay. Using a setup similar to the lipid mixing assay, the fusion reaction was arrested in the prefusion state, and Syt1/VAMP2 SUVs docked to the t-SNARE GUVs were isolated (Fig. 1B). We exposed the Bpa introduced at individual sites to UV light, leading to photoactivation, to cross-link Bpa to nearby lipids. Cross-linked Cpx II–lipid adducts were detected by employing alkyne PC-containing SUV bilayers. To visualize cross-links, we labeled alkyne PC with a fluorophore (Alexa Fluor 647), using click chemistry, and the cross-linked products were analyzed by SDS-PAGE and fluorescence scanning of the Alexa Fluor 647 signal (Fig. 1B; Hempelmann et al., 2023). UV-dependent alkyne PC cross-links were quantified by normalizing the Alexa Fluor 647 fluorescence signal to the corresponding Cpx amount detected by Coomassie blue staining (Fig. 1B). Exemplary SDS-PAGE gels reveal prominent lipid interactions of M5Bpa and A12Bpa (Fig. 1B, left panel). The Cpx-WT control (lacking Bpa) and D15Bpa and K18Bpa do not show UV-dependent lipid cross-links (Fig. 1B, left panel). In a complete mutagenesis screen of the NTD, cross-linking between Cpx Bpa mutants and the SUV bilayer was observed throughout most of the N-terminus up to Position 13 with prominent cross-linking with Positions 4 and 8 (Fig. 1B, right panel), suggesting extensive lipid interactions within the Cpx NTD.
Cpx-A12W and Cpx-D15W have distinct functional effects on NT release
Since our in vitro fusion assays suggested a strong role of D15 in regulating spontaneous fusion (Fig. 1A) and a role of A12 in both spontaneous fusion (Fig. 1A) and lipid binding (Fig. 1B), we first investigated these residues for their role in regulating SV fusion in the more complex, native environment of the synapse. For consistency with the biochemical experiments, we mutated these sites to tryptophan, closely mimicking the chemical properties of Bpa substitution (A12W and D15W; Fig. 2B,C). In lipid mixing assays, indeed the A12W and D15W mutants exhibited an increase in spontaneous fusion (Cpx-WT, 5.81 ± 0.62%; A12W, 11.51 ± 0.57%; and D15W, 18.38 ± 0.93%; Fig. 3A). Therefore, we went on to perform structure–function experiments on these single-point mutants in hippocampal glutamatergic neurons from mice lacking Cpx isoforms I, II, and II (Cpx-TKO).
Single mutations of the NTD of Cpx II. A, AlphaFold structure of Cpx II whose four different domains are color coded, NTD, N-terminal domain (gray); AH, alpha-accessory helix (green); CH, alpha-central helix (light pink); and CTD, C-terminal domain (dark blue). B, enlargement of the AlphaFold prediction of the structure of Cpx II WT NTD highlighting the amino acids of interest for this study. C, Sequence alignment of Cpx II WT NTD and the various mutants used in this study. D, Prediction of the structure of the different mutants of Cpx II NTD.
Cpx-A12W and Cpx-D15W have distinct functional effects on NT release. A, In vitro reconstitution assay shows that Cpx-A12W and Cpx-D15W stimulated Ca2+-independent lipid mixing (for experimental setup, see Fig. 1A). B, Examples traces (left) and quantification of the EPSC amplitude (right) recorded for autaptic hippocampal neurons obtained from Cpx-TKO mice expressing no rescue or rescued with Cpx-WT, Cpx-A12W, or Cpx-D15W. C, Example traces (left) and quantification of the mEPSC amplitude and frequency (right) obtained from the same neurons as in B. D, Quantification of the spontaneous release rate obtained from the same neurons as in B. E, Examples traces (left) and quantification of the RRP induced by 500 mM sucrose application (right) obtained from the same neurons as in (B). F, Examples traces (top) and quantification of STP determined by 50 Hz stimulation and normalized to the first EPSC (bottom) obtained from the same neurons as in B. The artefacts are blanked in example traces in B and F. The example traces in C were filtered at 1 kHz. In B–E, data points represent a single recorded neuron. Data are expressed as mean ± SEM; asterisks on the graph show the significance comparisons to Cpx-rescue (***p ≤ 0.001, ****p ≤ 0.0001, nonparametric Kruskal–Wallis test). N = 3 independent cultures. For WB and ICC showing protein and synaptic expression see also Extended Data Figure 3-1.
Figure 3-1
Protein and synaptic expression of Cpx-A12 W and Cpx-D15 W mutants in mice hippocampal neuron cultures corresponding to Figure 3. A, Example image (left) of a Western blot against complexin (top) and Tubulin (bottom) for Cpx-WT rescue and Cpx mutants in Cpx-TKO hippocampal neuron cultures. Bar graph (right) corresponds to the intensity of the complexin bands in each condition normalized to WT (n = 3 independent cultures). B, Example images (left) and quantification (right) of immunofluorescence labeling for Complexin and V-Glut1 for continental hippocampal cultures of Cpx-TKO neurons or infected with Cpx-rescue, Cpx-A12 W or Cpx-D15 W. Scale bar, 10 μm. The quantification of the immunofluorescence intensity of Complexin is normalized to the immunofluorescence intensity of V-Glut1. The values are normalized to the one obtained for Cpx-WT rescue (red). In (B), data points represent a field of view. Data are expressed as mean ± SEM, asterisks on the graph show the significance comparisons to Cpx-rescue (*** p ≤ 0.001, nonparametric Kruskal-Wallis test). N = 3 independent cultures. Download Figure 3-1, TIF file.
To characterize the effects of the A12W and D15W mutations on synaptic transmission, we expressed Cpx-A12W, Cpx-D15W, or Cpx-WT in Cpx-TKO autaptic mouse culture (Fig. 3B). Cpx-A12W and -D15W mutants were expressed using lentiviral transduction, and expression levels of reexpressed WT and mutant Cpx were monitored by WB (Extended Data Fig. 3-1A) analysis and immunocytochemistry (ICC; Extended Data Fig. 3-1B). Although total protein expression levels for Cpx-A12W and Cpx-D15W were lower than for Cpx-WT (Extended Data Fig. 3-1A), as measured by WB analysis, we observed with ICC that the three variants showed a similar signal intensity when normalized to VGlut1 puncta intensity. Therefore, they were properly localized at putative synaptic sites and were locally expressed at comparable levels (Extended Data Fig. 3-1B). The synaptic transmission phenotypes of each Cpx mutant were compared with the physiological properties of Cpx-TKO and Cpx-WT rescue in autaptic neurons (Fig. 3). As previously reported, electrophysiological recordings of cultured, autaptic Cpx-TKO glutamatergic neurons displayed a significantly reduced EPSC amplitude (Fig. 3B), reduced Pvr (Fig. 3E), and increased facilitation during trains of high-frequency stimulation compared with Cpx-WT rescue (Fig. 3; Xue et al., 2010; Trimbuch et al., 2014). Also consistent with previous work, loss of Cpx affected neither the hyperosmotic sucrose-evoked pool of fusion competent vesicles, the RRP (Fig. 3E), nor the frequency of mEPSC (Xue et al., 2010; Trimbuch et al., 2014).
The Cpx-A12W and Cpx-D15W mutants showed differential effects on EPSC amplitude rescue as compared with those on Cpx-WT rescue (Fig. 3B). Cpx-D15W expressing cells displayed reduced EPSC amplitudes at a level similar to Cpx-deficient neurons (D15W, 0.88 ± 0.1 nA; n = 62; TKO, 0.42 ± 0.07 nA; n = 67) compared with that to neurons rescued by Cpx-WT (2.19 ± 0.26 nA; n = 83), whereas Cpx-A12W rescued to similar levels as Cpx-WT. This suggests that D15 is an important amino acid for the synchronous NT release. Interestingly, analysis of spontaneous release showed that neurons expressing Cpx-A12W and Cpx-D15W displayed approximately a twofold increase in mEPSC frequency compared with Cpx-WT rescue (Cpx-WT rescue, 3.89 ± 0.49 Hz; n = 65; A12W, 7.42 ± 0.90 Hz; n = 48; D15W, 7.87 ± 0.67 Hz; n = 48; Fig. 3C), whereas the mEPSC amplitude was similar in all groups (Fig. 3C). This increase in spontaneous release mimics the observed increases in spontaneous lipid mixing seen in the in vitro vesicle fusion assay (Fig. 1A,B) and suggests that the A12 and D15 residues in the NTD of Cpx may play a role in clamping spontaneous fusion.
To further examine the role of Cpx's NTD in spontaneous fusion, we calculated the spontaneous release rate (Fig. 3D), a measure of spontaneous fusion frequency with respect to the number of fusion competent RRP vesicles per cell. Application of hypertonic sucrose revealed that the Cpx-A12W mutant had a similar RRP size compared with Cpx-WT rescue, while Cpx-D15W displayed a drastic reduction in RRP size (A12W, 0.321 ± 0.12 nC; n = 42; D15W, 0.095 ± 0.012 nC; n = 50; Cpx-WT rescue, 0.423 ± 0.100 nC; n = 70; Fig. 3E). The D15W phenotype on RRP size was surprising, as loss of Cpx leads to changes in release efficacy (Fig. 3B; Xue et al., 2010; Trimbuch et al., 2014) with no effect on RRP size (Fig. 3E; Xue et al., 2010; Trimbuch et al., 2014). The spontaneous vesicular release rate, calculated by dividing absolute mEPSC frequency by the number of vesicles in the RRP, showed that spontaneous release was increased for Cpx-A12W and even more robustly for Cpx-D15W (Cpx-WT rescue, 0.00196 ± 0.00038 s−1; n = 65 vs D15W, 0.00961 ± 0.00132 s−1 n = 54; Fig. 3D), suggesting an active role in unclamping, promoting spontaneous vesicle fusion, with both mutations.
To probe whether the observed changes in spontaneous release are related to changes in release efficacy, we calculated synaptic Pvr (Fig. 3E) and examined the short-term plasticity (STP) pattern as revealed by a 50 Hz stimulation (Fig. 3F). Pvr, calculated by dividing the charge of the EPSC by the charge of the RRP (Rosenmund and Stevens, 1996), was not significantly different between the Cpx-A12W and Cpx-D15W mutants and Cpx-WT rescue (Cpx-WT rescue, 5.7 ± 0.7%; n = 70; A12W, 8.2 ± 1.3%; n = 40; D15W, 7.1 ± 1.3%; n = 47; Fig. 3E). Additionally, unlike Cpx-TKO neurons, neither Cpx-A12W nor Cpx-D15W showed a facilitation pattern but rather a depression, similar to Cpx-WT rescue neurons (Fig. 3F). These results demonstrate that the unclamping and associated spontaneous release related to mutating sites A12 and D15 is not related to altered release efficacy.
The unique phenotype of Cpx-D15W on spontaneous NT release depends on the shape of the amino acid
Because the phenotype of the single-point mutation at position D15W of Cpx reduced vesicle priming, revealing a novel phenotype in mammalian synapses, we investigated whether the decrease in RRP was due to the loss of the negatively charged amino acid, aspartate (D), or the addition of a hydrophobic amino acid, namely, tryptophan (W). To test this, we created three additional single amino acid substitutions for Cpx-D15 (Fig. 2B,C), introducing either a neutral and small amino acid (Cpx-D15A), a positively charged amino acid (Cpx-D15K), or a polar uncharged amino acid (Cpx-D15N). First, we determined the effects of these novel Cpx-D15 mutants on fusion clamping by analyzing spontaneous and triggered release in the reconstituted liposome fusion assay (Fig. 4A). The mutants affected Ca2+-independent membrane fusion in a gradient manner ranging from no clamping (−Cpx) to maximum inhibition Cpx-WT: D15W > D15A > D15K > D15N (−Cpx, 20.68 ± 0.81%; D15W, 18.31 ± 1.80%; D15A, 16.09 ± 1.62%; D15K, 13.32 ± 1.43%; D15N, 9.91 ± 2.28%; and Cpx-WT, 6.69 ± 2.00%; Fig. 4A). The bulkier or more hydrophobic the amino acid substitution was, the more spontaneous membrane fusion was increased. As the phenotype of Cpx-D15N in in vitro experiments is comparable to Cpx-WT, we suggest that the actual charge of the amino acid at Position D15 is less critical.
The unique phenotype of Cpx-D15W on spontaneous NT release depends on the shape of the amino acid. A, In vitro reconstitution assay in the absence of Cpx (−Cpx) or in presence of Cpx-WT or Cpx-D15 mutants with increasing hydrophobicity (D15K > D15N > D15A > D15W) shows lipid mixing kinetics before and after addition of Ca2+. Error bars indicate SEM with n = 3 replicates. B, Examples traces (left) and quantification of the EPSC amplitude (right) recorded for autaptic hippocampal neurons obtained from Cpx-TKO mice expressing no rescue or rescued with Cpx-rescue, Cpx-D15W, Cpx-D15K, Cpx-D15A, or Cpx-D15N. C, Examples traces of mEPSC. D–F, Quantification of the mEPSC frequency (D), mEPSC amplitude (E), and spontaneous release rate (F) obtained from the same neurons as in B. G, Correlative plot of the pre-calcium mean obtained in our biochemical liposome fusion assays against the mEPSC frequency, both normalized to Cpx-rescue values. H, Examples traces (left) and quantification of the RRP induced by 500 mM sucrose application (right) obtained from the same neurons as in B. I, Correlative plot of the pre-calcium mean obtained in our biochemical liposome fusion assays against the RRP, both normalized to Cpx-rescue values. The artifacts are blanked in example traces in B. The example traces in C were filtered at 1 kHz. In B, D–H, data points represent a single recorded neuron. Data are expressed as mean ± SEM; asterisks on the graph show the significance comparisons with Cpx-rescue (*p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001; nonparametric Kruskal–Wallis test). N = 3 independent cultures. For WB and ICC showing protein and synaptic expression, see also Extended Data Figure 4-1.
Figure 4-1
Protein and synaptic expression of Cpx-D15 mutants in mice hippocampal neuron cultures corresponding to Figure 4. A, Example image (left) of a Western blot against complexin (top) and Tubulin (bottom) for Cpx-WT rescue and Cpx mutants in Cpx-TKO hippocampal neuron cultures. Bar graph (right) corresponds to the intensity of the complexin bands in each condition normalized to WT (n = 4 - 5 independent cultures). B, Example images (left) and quantification (right) of immunofluorescence labeling for Complexin (green) and V-Glut1 (red) for continental hippocampal cultures of Cpx-TKO neurons or infected with CPX-rescue, Cpx-D15 K, Cpx-D15N, Cpx-D15A or Cpx-D15 W. Scale bar, 10 μm. The quantification of the immunofluorescence intensity of Complexin is normalized to the immunofluorescence intensity of V-Glut1. The values are normalized to the one obtained for Cpx-WT rescue. In (B), data points represent a field of view. Data are expressed as mean ± SEM, asterisks on the graph show the significance comparisons to Cpx-rescue (**** p ≤ 0.0001, nonparametric Kruskal-Wallis test). N = 3 - 4 independent cultures. Download Figure 4-1, TIF file.
We next examined the Cpx-D15 mutants on NT release in cultured glutamatergic neurons. Expression levels of Cpx-D15 mutants induced by lentivirus in continental hippocampal neuronal cultures were verified by WB (Extended Data Fig. 4-1A) or ICC (Extended Data Fig. 4-1B). Although we noticed a weaker overall protein expression of our Cpx mutants in comparison with the Cpx-WT rescue construct (Extended Data Fig. 4-1A), we also show that the synapse-specific expression level of Cpx mutants was significantly increased compared with Cpx-TKO and comparable with Cpx-WT rescue expression in VGlut1 puncta (Extended Data Fig. 4-1B). In autaptic glutamatergic neurons, evoked EPSCs recorded from Cpx-D15W were decreased to a similar extent as neurons lacking Cpx altogether (Cpx-TKO) when compared with Cpx-WT rescue (Cpx-WT rescue, 2.66 ± 0.2 nA; n = 97; Cpx-TKO, 0.77 ± 0.07 nA; n = 94; D15W, 1.18 ± 0.11 nA; n = 49; Fig. 4B). However, none of the mutants Cpx-D15K, Cpx-D15A, or Cpx-D15N showed the same decrease in EPSC amplitude (D15K, 2.1 ± 0.2 nA; n = 69; D15A, 1.93 ± 0.23 nA; n = 42; and D15N, 3.11 ± 0.45 nA; n = 42; Fig. 4B). Interestingly, the size of the RRP for Cpx-D15K and Cpx-D15A was still decreased (Cpx-WT rescue, 0.34 ± 0.03 nA; n = 70; D15K, 0.22 ± 0.03; n = 54; and D15A, 0.16 ± 0.03 nA; n = 30; Fig. 4H), and the mEPSC frequency was still increased [Cpx-WT rescue, 2.53 ± 0.28 nA (n = 94); D15K, 6.47 ± 0.52 (n = 68); and D15A, 5.95 ± 0.78 nA (n = 41); Fig. 4C–F] compared with Cpx-WT rescue. Hence, the observed increase in mEPSC frequency or decrease in RRP size was not due only to the loss of a negative charge at Position 15 or the inclusion of a hydrophobic amino acid like a tryptophan.
In contrast, mutating D15 into an asparagine (D15N) rescued all the electrophysiological parameters to Cpx-WT rescue levels (Fig. 4B,F,H). The RRP size between Cpx-WT rescue and Cpx-D15N were comparable (Cpx-rescue, 0.34 ± 0.03 nA; n = 70; D15N, 0.24 0.04; n = 36; Fig. 4H), and the mEPSC frequency in Cpx-D15N neurons was restored to that of the Cpx-WT rescue (Cpx-WT rescue, 2.53 ± 0.28 nA; n = 94; D15N, 4.26 ± 0.76 nA; n = 38; Fig. 4D). The calculated spontaneous release rate per vesicle was also at Cpx-WT rescue level (Cpx-WT rescue, 0.0012 ± 0.0002 s−1; n = 78 vs D15N, 0.0019 ± 0.0003 s−1; n = 38; Fig. 4F). These results are in accordance with the in vitro experiments, which showed that Cpx-D15N has an only minor effect compared with Cpx-WT rescue (Fig. 4A).
To gain further insight to these findings altogether, we examined correlations between our biochemical and functional experiments (Fig. 4G,I). We observed remarkably strong correlations between the Ca2+-independent spontaneous vesicle fusion activity obtained through our liposome fusion assay and both the mEPSC frequency and the RRP measured in our electrophysiological experiments. When these behaviors were plotted against one another for the four D15 mutants and Cpx-WT rescue, the mEPSC rate was positively correlated (R2 = 0.89; Fig. 4G), and the RRP size was negatively correlated (R2 = 0.92; Fig. 4I) to the extent of spontaneous liposome fusion. This suggests that the D15 residue in Cpx plays a crucial role in clamping SNARE complexes to arrest spontaneous SV fusion. Additionally, these results strengthen the underlying assumption that pre-Ca2+ fusion rates in the reconstitution assay reflect spontaneous release in mammalian synapses.
Cpx-D15W partially increases SV fusogenicity
So far, we have shown that the shape of the amino acid at Position 15 in Cpx NTD affects spontaneous NT release and the size of the pool of SVs ready to be released (RRP) in opposite directions. To further dissect the origin of these regulatory functions, we probed the role of Cpx in vesicle fusogenicity. The fusogenicity of SVs is a measure of the activation energy required for vesicles to transition from the primed to the fused state and can be investigated by examining the release kinetics of sucrose-triggered responses in a sucrose-concentration–dependent manner. A 5 s application of 500 mM sucrose represents a saturating stimulus, releasing the entire RRP. In WT excitatory neurons, an intermediate dose of sucrose (250 mM) releases about a third of the RRP, and changes in the RRP fraction released in 250/500 mM sucrose indicates altered vesicle fusogenicity (Basu et al., 2007; Xue et al., 2010; Fig. 5A). The fraction of RRP released in Cpx-TKO neurons is significantly decreased compared with Cpx-WT rescue (consistent with previous observations; Xue et al., 2010; Fig. 5B). In contrast, Cpx-D15W showed a slightly, but not significantly (p = 0.29), increased fraction of RRP released with 250 mM sucrose as compared with Cpx-WT rescue (Cpx-WT rescue, 30.74 ± 3.57%; n = 39; D15W, 38.01 ± 5%; n = 48; Cpx-TKO, 19.09 ± 3.67%; n = 36; Fig. 5B). Another measure by which vesicle fusogenicity is reflected is the kinetics of the charge evoked by hypertonic sucrose application (Xue et al., 2010). To examine vesicle fusogenicity by analysis of release kinetics, we quantified the response onset latency of the 250 mM sucrose stimuli, which is slowed down when fusogenicity is lowered. On the one hand, the onset response latency for Cpx-D15W was slightly faster compared with Cpx-WT rescue (Cpx-WT rescue, 0.64 ± 0.05 s; n = 39; vs D15W, 0.5 ± 0.04 s; n = 48; p = 0.033; Fig. 5D) consistent with an increase in fusogenicity. On the other hand, the peak release rate which corresponds to the amplitude of the peak response divided by the RRP size is not significantly changed between Cpx-D15W and Cpx-WT rescue (Cpx-WT rescue, 0.64 ± 0.07 s−1; n = 39; D15W, 0.42 ± 0.05 s−1; n = 48; p = 0.074; Fig. 5E). Taken together, these results suggest that while the D15W mutation may have some effects on the fusion kinetics of SVs, it is not as clear-cut as the effects of Cpx-WT versus Cpx-TKO on vesicle fusogenicity.
Cpx-D15W partially increases SVs fusogenicity. A, Example traces of the RRP obtained from the application of 500 and 250 mM sucrose from Cpx-TKO, Cpx-rescue, and Cpx-D15W neurons. B, Quantification of the fraction of RRP released, determined as the ratio from the charge transfer for 500 mM sucrose and the charge transfer for 250 mM sucrose obtained from Cpx-TKO, Cpx-rescue, and Cpx-D15W neurons. C, Quantification of Pvr determined as the percentage of the RRP released upon one AP obtained from the same neurons as in B. D, Quantification of the response onset latency determined from the application of 250 mM sucrose and obtained from the same neurons as in B. E, Quantification of the peak release rate determined from the application of 250 mM sucrose and obtained from the same neurons as in B. F, Quantification of the number of vesicles released obtained by dividing the RRP charge measured by 500 mM sucrose application by the mEPSC charge for the same neuron. G, Example of HPF fixation followed by EM (HPF-EM) images of neurons from high-density cultures of Cpx-TKO, Cpx-rescue, and Cpx-D15W neurons. Scale bar, 200 nm. H, Quantification of number of docked vesicles, PSD length, number of SV within 50 nM of AZ obtained for Cpx-TKO, Cpx-rescue, and Cpx-D15W. I, SV distribution within 100 nm of AZ obtained from the same neurons as in H. In B–F, data points represent a single recorded neuron. Data are expressed as mean ± SEM; asterisks on the graph show the significant comparisons with Cpx-rescue (*p ≤ 0.05; ****p ≤ 0.0001; nonparametric Kruskal–Wallis test). N = 2–3 independent cultures.
The decreased RRP phenotype of Cpx-D15W does not exhibit a morphological correlate
We observed that the Cpx-D15W mutant has an impaired RRP, as shown by a decreased charge evoked by 500 mM sucrose, compared with either Cpx-WT rescue or Cpx-TKO (Fig. 3E). To translate the impairment of SV priming to the number of fusion competent SVs, we divided the RRP charge by the mean charge of mEPSC response (Fig. 3C,E), and confirmed that Cpx-D15W expressing neurons have a threefold smaller number of RRP SVs compared with Cpx-WT rescue or Cpx-TKO neurons (Cpx-WT rescue, 3,455 ± 302; n = 78; Cpx-TKO, 3,682 ± 414; n = 78; and D15W, 1,130 ± 114; n = 38; Fig. 5F). To investigate whether the vesicle priming defect in the D15W mutant translates to an ultrastructural defect, we obtained electron micrographs from synapses and analyzed vesicle docking (Watanabe et al., 2013; Imig et al., 2014). Neurons were plated at high density and fixed using HPF followed by freeze substitution, epon embedding, and sectioning to 50 nm thickness. Electron microscopic transmission images were analyzed as shown previously (Weber-Boyvat et al., 2022), and the number of docked SVs, considered as those in direct contact with the plasma membrane, was quantified as a function of AZ length. All three groups showed normal postsynaptic density length (Fig. 5G,H). Moreover, the density of docked vesicles was not different between Cpx-WT rescue, Cpx-TKO, and Cpx-D15W groups (Cpx-WT rescue, 1.1 ± 0.1; n = 120; Cpx-TKO, 1.05 ± 0.1; n = 132; and D15W, 1.19 ± 0.05; n = 129; Fig. 5H). Similarly, the SV distribution profiles within 100 nm of the AZ were comparable between Cpx-WT rescue, Cpx-TKO, and Cpx-D15W (Fig. 5I). These results suggest that the general synaptic organization of neurons lacking Cpx or neurons expressing the Cpx-D15W mutant is unaltered. The discrepancy between the low number of readily releasable vesicles measured with electrophysiology and the absence of a docking phenotype measured via EM suggests that Cpx-D15W impairs SV priming downstream of SV docking.
Cpx-D15W unclamps spontaneous vesicular release in WT neurons
While vesicle priming in Cpx-D15W mutant is impaired (Figs. 3E, 4H), spontaneous release (Figs. 3D, 4F) and, to some extent, vesicle fusogenicity (Fig. 5D) are enhanced in comparison with Cpx-WT rescue. To test whether these phenotypes are loss- or gain-of-function effects, we overexpressed Cpx-WT and Cpx-D15W mutants using lentiviral transduction in WT neurons. Overexpression of Cpx-WT in WT neuronal cultures had no measurable impact on any of the electrophysiological parameters we studied in comparison with WT neurons expressing GFP (Fig. 6). Cpx-D15W overexpression also had no effect on evoked EPSC responses (GFP, 3.51 ± 0.57 nA; n = 35; Cpx-WT, 3.66 ± 0.49 nA; n = 35; Cpx-D15W, 4.12 ± 0.62 nA; n = 37; Fig. 6A). However, the frequency of mEPSC events increased only when Cpx-D15W was overexpressed (GFP, 5.14 ± 1.18 Hz; n = 34; Cpx-WT, 5.28 ± 1 Hz; n = 34; and Cpx-D15W, 7.95 ± 1.19 Hz; n = 34; Fig. 6D) suggesting that Cpx-D15W has a dominant gain-of-function effect on spontaneous synaptic transmission. However, neither the RRP size, the SV release probably, nor the STP dynamics during a 10 Hz action potential train were modified by the expression of exogenous Cpx-D15W (Fig. 6B,C,E) suggesting that lentiviral-induced D15W expression is probably too mild to overcome the function of the endogenous Cpx. Taken together, the analysis shows that the unusual D15W mutant phenotype is only dominant in unclamping to promote spontaneous release.
Cpx-D15W has a dominant positive effect on spontaneous vesicular release in WT neurons. A, Examples traces (left) and quantification of the EPSC amplitude or EPSC charge (right) recorded for autaptic hippocampal neurons obtained from Cpx-WT mice overexpressing GFP, Cpx-WT, or Cpx-D15W. B, Example traces (left) and quantification of the RRP induced by 500 mM sucrose application and Pvr (right) obtained from the same neurons as in A. C, Example traces (left) and quantification of the spontaneous release rate, mEPSC amplitude, and frequency (right) obtained from the same neurons as in A. D, Examples traces (left) and quantification of STP determined by high-frequency stimulation at 10 Hz and normalized to the first EPSC (right) obtained from the same neurons as in A. The artifacts are blanked in example traces in A and D. The example traces in C were filtered at 1 kHz. In A–C, data points represent a single recorded neuron. Data are expressed as mean ± SEM; asterisks on the graph show the significance comparisons to Cpx-rescue (*p ≤ 0.05, nonparametric Kruskal–Wallis test). N = 3 independent cultures.
Cpx-M5E mutant disrupts the binding of Cpx N-terminus to the SNARE complex in a charge-dependent manner
Our initial screen with lipid fusion assays predicted the effects of Cpx mutants on regulating spontaneous SV fusion but suggested no effect of Cpx on Ca2+-triggered fusion. This is in contrast to many studies demonstrating the facilitatory effect of Cpx on Ca2+-triggered SV fusion in the synapse (Reim et al., 2001; Trimbuch et al., 2014). For instance, our previous work reported that mutating the N-terminal residues Cpx-M5E, K6E disrupts the positive effect of Cpx on SV fusogenicity and synchronous release (Xue et al., 2010). This phenotype was accompanied by a disruption of binding of Cpx NTD to the SNARE complex in nuclear resonance measurements (Xue et al., 2010). Our liposome fusion assay did not show a phenotype in the Cpx-M5-Bpa mutant (Fig. 1A). As the mutation of M5 into a negatively charged glutamic acid was crucial to disrupt the interaction of Cpx with the C-terminus of SNARE complex (Xue et al., 2010), we utilized Cpx-M5E mutants in parallel with Cpx-M5W that sterically best mimic the insertion of the unnatural amino acid Bpa.
The Cpx-M5E mutant (Fig. 2B,C) was sufficiently expressed and colocalized with the presynaptic glutamatergic marker VGlut1 (Extended Data Fig. 7-1A,B). In our functional experiments, Cpx-M5E largely mimics neurons lacking Cpx (Cpx-TKO). The amplitude of the evoked EPSC response is significantly decreased compared with Cpx-WT rescue (Cpx-WT rescue, 5.37 ± 0.57 nA; n = 48; Cpx-TKO, 0.87 ± 0.12 nA; n = 47; and M5E, 1.49 ± 0.2 nA; n = 49; Fig. 7A). Moreover, Cpx-M5E showed paired-pulse facilitation comparable with Cpx-TKO (Cpx-WT rescue, 1.06 ± 0.13; n = 46; Cpx-TKO, 1.59 ± 0.12; n = 46; and M5E, 1.65 ± 0.11; n = 46; Fig. 7A) and an impairment of SV spontaneous release as shown by the mEPSC frequency decrease compared with Cpx-WT rescue and Cpx-TKO (Cpx-WT rescue, 3.4 ± 0.6 Hz; n = 41; Cpx-TKO, 3.09 ± 0.44 Hz; n = 42; and M5E, 1.9 ± 0.8 nA; n = 27; Fig. 7B). On the other hand, the RRP of Cpx-M5E was not different from the RRP for Cpx-WT rescue or Cpx-TKO neurons. In contrast to the phenotype of the Cpx-M5E mutant, when we mutated M5 into a tryptophan (Cpx-M5W; Fig. 2B,C), all measured electrophysiological parameters (Pvr, RRP, mEPSC amplitude, and STP) behaved like WT (Fig. 7A–D). These results are in full accordance with our biochemical experiments, in which Cpx-M5Bpa had no effect on fusion kinetics (Fig. 1A). Taken together, these results support the previously published idea that Cpx NTD residue M5 is crucial for the facilitatory function of Cpx (Xue et al., 2010) but additionally suggest that the unmasking of this role depends on the charge of the mutated amino acid.
Insertion of a negative charge at Position 5 (Cpx-M5E) disrupts the stimulatory function of the Cpx N-terminus. A, Example traces (left) and quantification of the EPSC amplitude (right) and paired-pulse ratio (PPR) at 20 ms intervals recorded for autaptic hippocampal neurons obtained from Cpx-TKO mice expressing no rescue or rescued with Cpx-WT, Cpx-M5E, or Cpx-M5W. B, Example traces (left) and quantification of the mEPSC amplitude and frequency (right) obtained from the same neurons as in A. C, Example traces (left) and quantification of the RRP induced by 500 mM sucrose application (right) obtained from the same neurons as in A. D, Examples traces (left) and quantification of STP determined by 50 Hz stimulation and normalized to the first EPSC (bottom) obtained from the same neurons as in A. The artifacts are blanked in example traces in A and D. The example traces in B were filtered at 1 kHz. In A–C, data points represent a single recorded neuron. Data are expressed as mean ± SEM; asterisks on the graph show the significance comparisons to Cpx-rescue (*p ≤ 0.005; ***p ≤ 0.001; nonparametric Kruskal–Wallis test). N = 3 independent cultures. For WB and ICC showing protein and synaptic expression, see also Extended Data Figure 7-1.
Figure 7-1
Protein and synaptic expression of Cpx-M5 mutants in mice hippocampal neuron cultures corresponding to Figure 7. A, Example image (left) of a Western blot against complexin (top) and Tubulin (bottom) for Cpx-WT rescue and Cpx mutants in Cpx-TKO hippocampal neuron cultures. Bar graph (right) corresponds to the intensity of the complexin bands in each condition normalized to WT (n = 3 independent cultures). B, Example images (left) and quantification (right) of immunofluorescence labeling for Complexin and V-Glut1 for continental hippocampal cultures of Cpx-TKO neurons or infected with Cpx-rescue, Cpx-M5E or Cpx-M5 W. Scale bar, 10 μm. The quantification of the immunofluorescence intensity of Complexin is normalized to the immunofluorescence intensity of V-Glut1. The values are normalized to the one obtained for Cpx-WT rescue. In (B), data points represent a field of view. Data are expressed as mean ± SEM, asterisks on the graph show the significance comparisons to Cpx-rescue (* p ≤ 0.05, **** p ≤ 0.0001, nonparametric Kruskal-Wallis test). N = 3 independent cultures. Download Figure 7-1, TIF file.
Discussion
The SNARE complex is at the core of the canonical release machinery common for all regulated release paths. At the synapse, additional modulators are needed to accelerate and tune release properties to adjust the release function to the need of synaptic communication. Cpx is the best characterized SNARE complex binding protein (Ishizuka et al., 1995; McMahon et al., 1995; Takahashi et al., 1995). It can perform both release-accelerating and clamping functions, depending on species and synapse (Trimbuch and Rosenmund, 2016 for review). All structural and mutagenesis studies show that its accelerating and clamping functions are embedded in the NTDs and CTDs (Xue et al., 2009, 2010; Buhl et al., 2013). The NTD is so far the only region known to be essential for the accelerating function of Cpx (Xue et al., 2009, 2010), but how the NTD regulates release is poorly understood. In this study, we combined Cpx mutagenesis with in vitro reconstitution assays and functional analysis at central synapses to define functionally critical regions within the NTD as well as the putative role of lipid binding of the assumed amphipathic helical secondary structure. We show that (1) the proximal region of the N-terminus of Cpx from Position 1–12 is a lipid interacting region (Fig. 1), (2) residues between 11 and 16 are important for reducing spontaneous release, (3) mutating the aspartate at Position 15 results in a drastically decreased RRP (Figs. 3–6), and (4) Position M5 is functionally relevant in promoting NT release in synapses (Fig. 7) and sensitive to lipid interaction (Fig. 1). Our detailed analysis of the NTD of Cpx shows that Cpx function is even more complicated than initially thought, because the facilitatory NTD can also modulate vesicle priming and spontaneous release.
Overall, our data add significant knowledge to our understanding of the mechanisms of action of Cpx, especially on its facilitatory role. Based on previous studies, the N-terminus, formed by the residues 1–27, is essential for the positive action of Cpx on vesicle release probability (Xue et al., 2010). From previous work, two possible interaction modes for the N-terminus emerge, binding to the SNARE complex itself or interacting with lipid membranes. NMR experiments using a paramagnetic probe at Position 12 demonstrated that the N-terminus can interact with the C-terminus of all three SNARE complex forming proteins (Xue et al., 2010; Choi et al., 2018). Assuming an alpha-helical structure of the N-terminus, the helix would contain an amphipathic surface that could also favor membrane binding, where the hydrophobic site is formed by evolutionary conserved residues 1, 5, 8, 9, 12, and 16 (Xue et al., 2010). However, we did not observe the expected pattern of alternating effects on either fusion or on lipid interactions within the NTD that would be indicative of this helical structure. Our systematic assessment of lipid interactions using single-residue Bpa–mediated cross-linking (Fig. 1) showed that the strongest interactions with lipids take place between residues 1 and 12, but the assay was not specifically selective for hydrophobic residues. Our in vitro fusion assays with single-residue Bpa mutant of Cpx did not display any signs of activation of fusion for the first 10 residues (Fig. 1). Thus, on the one hand, changing already hydrophobic residues in this region to Bpa does not affect Cpx function. In contrast, single Bpa mutations at D2, K6, and Q7 rendered Cpx insoluble in the heterologous E. coli expression system, indicating that the position of hydrophobic amino acids is critical. However, this is likely due to the property of the fusion assay, as deletion of the entire NTD only impairs Ca2+-triggered release in mammalian synapses, but not in in vitro fusion assays (Bera et al., 2022). On the other hand, mutagenesis of the hydrophobic M5 residue to a hydrophilic glutamate strongly reduces both spontaneous and evoked release suggesting that lipid interactions in this region could have a stimulatory function and do not contribute to clamping. This finding is consistent with a previous study, which shows that the Cpx NTD can be functionally replaced by a general membrane fusion peptide from an unrelated protein (Lai et al., 2016).
Our in vitro fusion assay revealed a novel role for the N-terminus of Cpx in fusion clamping, as mutating residues 11–17 showed an enhanced pre-Ca2+ fusion activity (Fig. 1). We could confirm that this behavior in pre-Ca2+ fusion activity ascertained in our biochemical analysis indeed reflects a clamping function of spontaneous release, as residues 12 and 15 when mutated and expressed in Cpx-TKO neurons also showed an increase in spontaneous SV fusion (increase in mEPSC frequency; Figs. 3, 4). Specifically, we observed that spontaneous release was enhanced when residues 12 and 15 were made bulky and hydrophobic and the capability to form hydrogen bonds was impaired. Shon et al. (2018) postulate a new “linker-opened” stage in which Cpx I NTD interacts with the unfolded linker region of VAMP2. Both A12 and D15 are positioned in Cpx NTD's linker region involved in this postulated linker repulsing arrangement, but D15 is even closer to the SNARE motifs, which results in stronger impairment by mutagenesis compared with A12. Another important observation linking the behavior of proteins driving fusion within the biochemical assays and within the native habitat of the synapse was the strong correlation between pre-Ca2+ fusion activity in lipid mixing assays and the degree of spontaneous release recorded with electrophysiology in mutants with graded effects on the phenotype (Fig. 4G). However, due to technical limitations, our in vitro assay could not resolve the Ca2+-triggered phase in correlation with the functional evoked response. Overall, these results strongly suggest the fundamental nature of the role of Cpx in SNARE-mediated membrane fusion, as the phenotype is reflected both in a situation containing the minimal players involved in the process (reconstituted liposome fusion assays) and in an environment containing far more players than necessary for fusion (native synapse). In turn, taken together these results emphasize the impact of the Cpx NTD on clamping spontaneous release.
One striking observation we made through our detailed analysis of the Cpx NTD was the novel, single-residue–dependent role of Cpx in mammalian cells in determining the pool of primed, fusion competent vesicles. Previous papers have shown that in the Drosophila model, abolishing Cpx (Cpx−/−) can lead to a decrease in the size of the RRP (Jorquera et al., 2012). In mammalian neurons, deletion of Cpx (Cpx-TKO) does not modify the size of the RRP (Xue et al., 2008, 2010). Nevertheless, we found that altering the acidic nature of the aspartic acid at Position 15 resulted in a strong reduction of the RRP size (Figs. 3, 4). The surprising impairment in RRP size in D15W mutant compared with Cpx-TKO or Cpx-WT rescue suggests D15W has a dominant negative phenotype that selectively impairs vesicle priming function. The effect was not accompanied by a defect in SV docking (Fig. 5), a phenotype typically associated with priming deficiency, arguing that the impairment in vesicle priming occurs at a step downstream of vesicle docking. Comparison of our functional experiments to our data from the in vitro assay also showed that the impacts on mEPSC frequency and RRP were inversely correlated (Fig. 4G,I). Such a behavior points toward a destabilized primed vesicle state, where primed vesicles transition more easily into the unprimed or fused state. Interestingly, we observed a similar behavior in Syntaxin1 mutants (Salazar-Lázaro et al., 2024) that putatively interfere with the “primary interface” interaction between synaptotagmin1 (Syt1) and the SNARE complex as defined by protein crystallography (Zhou et al., 2015). Additionally, a recent publication using optical tweezers shows that removal of the NTD from Cpx changes the disassembly of Cpx and the SNARE complex (Hao et al., 2023), emphasizing the role of the NTD in stabilizing the SNARE complex. Why does the Cpx-D15W mutation interfere with SV priming, while the Cpx-A12W does not? While our current approaches cannot provide a detailed mechanistic explanation for this discrepancy, we can speculate that Cpx-D15W is in a privileged position to modify the interaction of Cpx with either Syt1 or the SNARE complex through ionic repulsion or steric hindrance.
In summary, our analysis of the Cpx N-terminus showed a diverse and complex regulation of release function where single-residue mutations in close proximity have discrete and distinct effects on vesicle priming, spontaneous release, and the efficacy of Ca2+-triggered release. A possible explanation for such behavior is that Cpx, along with synaptotagmin, the SNARE complex, and the membranes interact at a closely confined space. The spatial position of the NTD in the tripartite interface, for example, in relation to the SNARE complex and synaptotagmin, still needs to be structurally determined. However, our study contributes to a better understanding of the role of the small cytosolic regulator, Cpx, in the fine-tuning of vesicle release and ultimately in regulation of synaptic transmission.
Footnotes
This work was funded by the Deutsche Forschungsgemeinschaft (DFG; German Research Foundation) project 278001972-TRR186 (to T.H.S. and C.R.). We are grateful to Berit Söhl-Kielczynski, Bettina Brokowski, Katja Pötschke, Ursula Göbel, Lara Braun, Susanne Kreye, and Heike Lerch for their excellent technical assistance. We also thank Denisa Jamecna and Doris Höglinger for sharing their expertise in lipid cross-linking using click chemistry and Daniela Schweinfurth and Britta Brügger for sharing their knowledge in protein–lipid extraction. We are grateful to Andrew Plested for his comments on the manuscript. We thank the services of the Charité viral core facility for virus production and the electron microscopy core facility for the technical support. Molecular graphics were performed with UCSF ChimeraX, developed by the Resource for Biocomputing, Visualization, and Informatics at the University of California with support from the National Institutes of Health R01-GM129325 and the Office of Cyber Infrastructure and Computational Biology, National Institute of Allergy and Infectious Diseases.
The authors declare no competing financial interests.
- Correspondence should be addressed to Thomas H. Söllner at thomas.soellner{at}bzh.uni-heidelberg.de or Christian Rosenmund at christian.rosenmund{at}charite.de.
