Molecular determinants of complexin clamping and activation function

  1. Manindra Bera
  2. Sathish Ramakrishnan
  3. Jeff Coleman
  4. Shyam S Krishnakumar  Is a corresponding author
  5. James E Rothman  Is a corresponding author
  1. Yale Nanobiology Institute, United States
  2. Department of Cell Biology, Yale University School of Medicine, United States
  3. Department of Pathology, Yale University School of Medicine, United States
  4. Departments of Neurology, Yale University School of Medicine, United States

Abstract

Previously we reported that Synaptotagmin-1 and Complexin synergistically clamp the SNARE assembly process to generate and maintain a pool of docked vesicles that fuse rapidly and synchronously upon Ca2+ influx (Ramakrishnan et al., 2020). Here, using the same in vitro single-vesicle fusion assay, we determine the molecular details of the Complexin-mediated fusion clamp and its role in Ca2+-activation. We find that a delay in fusion kinetics, likely imparted by Synaptotagmin-1, is needed for Complexin to block fusion. Systematic truncation/mutational analyses reveal that continuous alpha-helical accessory-central domains of Complexin are essential for its inhibitory function and specific interaction of the accessory helix with the SNAREpins enhances this functionality. The C-terminal domain promotes clamping by locally elevating Complexin concentration through interactions with the membrane. Independent of their clamping functions, the accessory-central helical domains of Complexin also contribute to rapid Ca2+-synchronized vesicle release by increasing the probability of fusion from the clamped state.

Editor's evaluation

Bera and colleagues revisit several mechanistic questions mainly centered on the accessory helix of mouse complexin (mCpx) and its contribution to the 'fusion clamp' property of mCpx whereby mCpx-SNARE interactions prevent full assembly and subsequent membrane fusion. This clamping function is believed to help generate a metastable pool of release-ready vesicles at the synapse, and it has been studied in a wide variety of systems including mouse, fly, worm, squid, fish, and diverse in vitro biochemical preps over the past ~ 20 years. The authors derive several conclusions from their efforts, but most relevant is a reiteration of a previous proposal that the accessory helix region of mCpx stabilizes a pre-fusion clamped state via interactions with SNAREs.

https://doi.org/10.7554/eLife.71938.sa0

Introduction

Neurons communicate with each other at synaptic contacts by releasing neurotransmitters from synaptic vesicles (SVs). This process is tightly controlled by activity-dependent changes in the presynaptic Ca2+ concentration and can occur in less than a millisecond after the neuronal spike (Südhof, 2013; Kaeser and Regehr, 2014). SV fusion is catalyzed by presynaptic SNARE proteins. The SNAREs on the opposing membranes (VAMP2 on the synaptic vesicle membrane; Syntaxin and SNAP25 on the presynaptic plasma membrane) assemble into a four-helix bundle that catalyzes fusion by forcing the two membranes together (Söllner et al., 1993; Weber et al., 1998). Related SNARE proteins are universally involved in intracellular transport pathways and by themselves can constitutively catalyze fusion (Weber et al., 1998; McNew et al., 2000). As such, Ca2+-evoked neurotransmitter release occurs from the readily releasable pool (RRP) of vesicles docked/primed at the presynaptic active zone (Südhof, 2013; Kaeser and Regehr, 2014). The current view is that at a single RRP vesicle, the SNARE complexes are firmly held (‘clamped’) in a partially assembled state (SNAREpins) close to the point of triggering fusion. Upon Ca2+ influx, multiple SNAREpins are synchronously activated to drive ultrafast SV fusion and neurotransmitter release (Südhof and Rothman, 2009; Südhof, 2013; Rizo and Xu, 2015; Rothman et al., 2017; Brunger et al., 2019).

It is well-established that the late stages of SV fusion are tightly regulated by two synaptic proteins – the presynaptic Ca2+ release sensor Synaptotagmin-1 (Syt1) and Complexin (CPX) (Südhof, 2013; Südhof and Rothman, 2009; Rizo and Xu, 2015; Brunger et al., 2019). CPX is an evolutionarily conserved cytosolic protein that bind and regulate synaptic SNARE complex assembly (McMahon et al., 1995; Huntwork and Littleton, 2007; Martin et al., 2011; Trimbuch and Rosenmund, 2016; Mohrmann et al., 2015). Biochemical and biophysical analyses show that CPX promotes the initial stages of SNARE assembly but then blocks complete assembly (Li et al., 2011; Kümmel et al., 2011; Lai et al., 2014; Krishnakumar et al., 2015). Thus, it can both facilitate and subsequently inhibit SV fusion. CPX contain distinct domains that mediate the dual clamp/activator function (Xue et al., 2007; Giraudo et al., 2008; Trimbuch and Rosenmund, 2016; Mohrmann et al., 2015). The largely unstructured N-terminal domain (residues 1–26 of mammalian CPX1) activates Ca2+-regulated vesicular release (Xue et al., 2010; Lai et al., 2016) while the α-helical accessory domain (CPXacc, residues 26–48) serves as the primary clamping domain (Xue et al., 2007; Giraudo et al., 2008; Maximov et al., 2009; Yang et al., 2010; Kümmel et al., 2011; Cho et al., 2014). A central helical sequence within CPX (CPXcen, residues 48–70) binds the groove between pre-assembled Syntaxin and VAMP2 and is essential for both function (Chen et al., 2002; Xue et al., 2007; Giraudo et al., 2008; Maximov et al., 2009). The remainder c-terminal portion (residues 71–134) has been shown to preferentially associate with curved lipid membrane via an amphipathic helical region and promotes the clamping function (Kaeser-Woo et al., 2012; Wragg et al., 2013; Gong et al., 2016).

The relative strength of CPX facilitatory vs inhibitory activities differs across species (Yang et al., 2013; Trimbuch and Rosenmund, 2016; Mohrmann et al., 2015; Xue et al., 2009). As a result of this intricate balance, genetic perturbations of CPX can produce apparently contradictory effects in different systems. For example, knockout (KO) of CPX in neuromuscular synapses of C. elegans and Drosophila results in increased spontaneous release, decreased evoked release with overall reduction in the RPP size (Huntwork and Littleton, 2007; Cho et al., 2014; Martin et al., 2011; Hobson et al., 2011; Wragg et al., 2013). In model mammalian synapses, CPX KO abates both spontaneous and evoked release with no significant change in the RRP size (Reim et al., 2001; Xue et al., 2008; López-Murcia et al., 2019) but acute CPX knockdown (KD) reduces synaptic strength, but also increases spontaneous release with a concomitant reduction in the number of primed vesicles (Maximov et al., 2009; Yang et al., 2010; Kaeser-Woo et al., 2012; Yang et al., 2013). Some of the apparent discrepancies might be related to the perturbation method used (Yang et al., 2013); nonetheless, the physiological role of mammalian CPX in regulating SV fusion and the underlying mechanisms remains in the center of debate (Mohrmann et al., 2015; Trimbuch and Rosenmund, 2016).

The interpretation of the physiological experiments can be limited by presence of the different CPX isoforms and possible compensatory homeostatic mechanisms. As such, the experiments in live synapses need to be complemented with a reductionist approach where the variables are limited, and the components can be rigorously controlled or altered. It is our hypothesis that the most direct mechanistic insight can be obtained from fully controlled cell-free systems. We have described a biochemically defined fusion setup based on a pore-spanning lipid bilayer setup that is well-suited for this purpose (Ramakrishnan et al., 2018; Ramakrishnan et al., 2019; Ramakrishnan et al., 2020).

Using this in vitro setup, which allows for precision study of the single-vesicle fusion kinetics, we recently demonstrated that mammalian CPX (mCPX), along with Syt1 and SNAREs, are essential and sufficient to achieve Ca2+-regulated fusion under physiologically relevant conditions (Ramakrishnan et al., 2020). Our data revealed that mCPX and Syt1 act co-operatively to clamp the SNARE assembly process and produce a pool of docked vesicles. The study also revealed that there are at least two types of clamped SNAREpins under a docked vesicle – a small subset that are reversibly clamped by binding to Syt1 (which we termed ‘central’) and a larger population that are thought to be free of Syt1 and require mCPX for clamping (termed ‘peripheral’). We further established that Syt1s’ ability to oligomerize and bind SNAREpins via the ‘primary’ binding site on SNAP25 is key to its ability to clamp central SNAREpins and that the activation of these Syt1-associated SNAREpins is sufficient to elicit rapid, Ca2+-synchronized vesicle fusion (Ramakrishnan et al., 2020).

Building on this work, here we use a systematic in vitro reconstitution strategy to obtain new and direct insights into the molecular basis of mCPX clamping function and its role in establishing Ca2+-regulated release. We report that mCPX inhibitory function requires a delay in overall fusion kinetics and involves well-defined interaction of the accessory-central helical fragments with the SNAREpins. The accessory-central helical domains also stimulate Ca2+-triggered vesicle fusion from the clamped state. Overall, we find that under physiologically-relevant conditions, mCPX is essential to generate/maintain a pool of docked vesicles and to promote Ca2+-triggered rapid ( < 10ms) and synchronous fusion of the docked vesicles.

Results

To dissect the mCPX clamping functionality, we used physiologically relevant reconstitution conditions similar to our previous work (Ramakrishnan et al., 2020). Typically, we used small unilamellar vesicles (SUV) with ~70 copies (outward facing) of VAMP2 (vSUV) without or with ~25 copies Syt1 (Syt1-vSUV) (Figure 1—figure supplement 1). We employed pre-formed t-SNAREs (1:1 complex of Syntaxin1 and SNAP-25) in the planar bilayers (containing 15% PS and 3% PIP2) to both simplify the experimental approach and to bypass the requirement of SNARE-assembling chaperones, Munc18 and Munc13 (Baker and Hughson, 2016). Mammalian CPX1 (wild type or variants) was included in solution, typically at 2 μM unless noted otherwise (Figure 1—figure supplement 2). We used fluorescently labeled lipid (2% ATTO647N-PE) to track docking, clamping and spontaneous fusion of individual vesicles and a content dye (sulforhodamine B) to study Ca2+-triggered fusion of docked vesicles from the clamped state.

To focus on the ‘clamping’ of constitutive fusion events, we monitored large ensembles of vesicles to determine the percent remaining unfused as a function of time elapsed after docking and quantified as ‘survival percentages’ (Ramakrishnan et al., 2019; Ramakrishnan et al., 2020; Ramakrishnan et al., 2018). Docked immobile vesicles that remained un-fused during the initial 10-min observation period were defined as ‘clamped’ and the ‘docking-to-fusion’ delay enabled us to quantify the strength of the fusion clamp (Ramakrishnan et al., 2019; Ramakrishnan et al., 2020; Ramakrishnan et al., 2018). Since we track the fate of single vesicles, this analysis allowed us to examine the ‘clamping’ mechanism, independent of any alteration in the preceding docking sub-step.

Our earlier results showed that Syt1 alone can meaningfully delay but not stably clamp SNARE-mediated fusion. Similarly, mCPX, on its own, is ineffective in clamping SNARE-driven vesicle fusion. In fact, both Syt1 and mCPX are needed to produce a stably ‘clamped’ state which can then be reversed by Ca2+ (Ramakrishnan et al., 2020). It is possible that Syt1 and mCPX1 either act jointly to generate a new intermediate state in the SNARE assembly pathway or operate sequentially, with the kinetic delay introduced by Syt1 enabling mCPX to arrest SNARE assembly. To distinguish between these possibilities, we developed a mimic for the Syt1 clamp – a lipid-conjugated ssDNA that is capable of regulating SNARE-driven fusion in situ. Without directly interacting with the SNAREs, the specific base-pair hybridization of the complementary ssDNA reconstituted into the SUVs and the planar bilayer introduces a steric barrier which is expected to, and indeed does delay fusion (Figure 1, Figure 1—figure supplement 3). Moreover, this docking-to-fusion delay could be varied by adjusting the number of ssDNA molecules (Figure 1—figure supplement 3).

Figure 1 with 4 supplements see all
Syt1 and mCPX act sequentially to arrest SNARE-driven fusion.

(A) Schematic of the programmable DNA-based mimetic used to simulate the Syt1 clamp on the SNARE-driven fusion. Annealing of the complementary ssDNA reconstituted into the SUV and the bilayer in dsDNA sterically counters the polarized SNARE assembly process and introduces a docking-to-fusion delay reminiscent of Syt1 (B) Survival analysis (Kaplan-Meier plot) curve shows that a nominal dock-to-fusion delay introduced by 20 copies of ssDNA (purple) allows mCPX to arrest spontaneous fusion of vSUVs. In contrast, no clamping was observed with 5 copies of ssDNA (yellow) which created no appreciable delay in the fusion kinetics. This suggests a sequentially mode of action for Syt1 and mCPX, wherein the kinetic delay introduced by Syt1 enables mCPX to block SNARE-driven fusion. Data was obtained from a minimum of three independent experiments, with at least 100 vesicles analyzed for each condition. A representative survival curve is shown for clarity.

Figure 1—source data 1

Data and summary statistics for DNA-regulated fusion assay.

https://cdn.elifesciences.org/articles/71938/elife-71938-fig1-data1-v1.xlsx

We then assessed the effect of mCPX on ssDNA-regulated fusion of vSUV in the absence of Ca2+ (Figure 1). mCPX was able to near-completely arrest spontaneous fusion of vSUV to generate stably docked vesicles, provided that the rate of SNARE-mediated fusion was sterically delayed by ~20 copies of ssDNA (Figure 1, Figure 1—figure supplement 4). The majority of the vSUVs were immobile following docking to the t-SNARE-containing suspended bilayer (Figure 1, Figure 1—figure supplement 4), and they rarely fused over the initial observation period. In contrast, little or no inhibition was observed in control experiments with ~5 copies of ssDNA that did not introduce a detectable delay in the fusion process, as all docked vesicles proceeded to fuse spontaneously typically within 1–2 s (Figure 1, Figure 1—figure supplement 4). This suggests that it is the delay in fusion per se that is necessary for the mCPX inhibitory function, and importantly that the mCPX clamp is not dependent or influenced by the ssDNA molecules (Figure 1—figure supplement 4). Thus, our data indicates that Syt1 and mCPX likely act sequentially to produce a synergistic clamp, with the delay introduced by Syt1 meta-stable clamp enabling CPX to bind and block the full assembly of the SNARE complex.

Next, we investigated the role of the distinct domains of mCPX in establishing the fusion block using Syt1 containing vSUV (Syt1-vSUVs). On their own, a majority (~80%) Syt1-vSUVs that docked to the t-SNARE containing bilayer surface were mobile and fused on an average 5–6 s after docking, while a small fraction (~20%) were immobile and stably clamped (Figure 2A and B). Inclusion of 2 μM wild-type mCPX (mCPXWT) enhanced the vesicle docking rate, with an ~ three-fold increase in the total number of stably docked vesicles and >95% of Syt1-vSUVs remaining immobile post-docking (Figure 2A and B). This is consistent with our earlier findings (Ramakrishnan et al., 2020). A truncation mutant (mCPX26-134) lacking the unstructured N-terminal domain had very little or no effect on the vesicle docking rate or the fusion clamp, with vesicle behavior near identical to CPXWT (Figure 2A and B). Deletion of the CPXacc in addition to N-terminal domain (mCPX48-134) increased the number of docked vesicles (~ two-fold) but abrogated the inhibitory function with majority of the docked vesicles proceeding to fuse spontaneously (Figure 2A and B). Targeted mutations in CPXcen (R48A Y52A K69A Y70A; mCPX4A) that disrupt its interaction with the SNAREpins completely abolished both the stimulatory effect on vesicle docking and the fusion clamp (Figure 2A and B). In fact, both the CPXacc deletion (mCPX48-134) and CPXcen modifications (mCPX4A) resulted in complete loss of mCPX inhibitory function and could not be rescued even at highest concentration (20 μM) tested (Figure 2C, Figure 2—figure supplement 1). Deletion of the c-terminal domain (mCPX26-83) lowered the clamping efficiency (Figure 2A and B) with ~50% vesicles clamped under the standard experimental conditions (2 μM mCPX26-83). However, the inhibitory function was rescued simply by raising the concentration and was completely restored at 20 μM mCPX26-83 (Figure 2C, Figure 2—figure supplement 1).

Figure 2 with 2 supplements see all
Molecular determinants of Complexin clamping function.

The effect of mCPX mutants on docking and clamping of spontaneous fusion was assessed using a single-vesicle analysis with a pore-spanning bilayer setup. (A) Inclusion of mCPX increases the number of docked Syt1-vSUVs and this stimulatory effect is greatly reduced when the interaction of the CPXcen to the SNAREpins is disrupted targeted mutations (mCPX4A). In contrast, deletion of the N-terminal domain (CPX26-134) or accessory helix (CPX48-134) or the c-terminal portion (CPX26-83) exhibit limited effect of the vesicle docking. In all cases, a mutant form of VAMP2 (VAMP24X) which eliminated fusion was used to unambiguously estimate the number of docked vesicles after the 10 min interaction phase. (B) The time between docking and fusion was measured for each docked vesicle and the results for the whole population are presented as a survival curve (Kaplan-Meier plots). Syt1-vSUVs (black curve) are diffusively mobile upon docking and fuse spontaneous with a half-time of ~5 s. Addition of soluble mCPX (red curve) fully arrest fusion to produce stably docked SUVs that attach and remain in place during the entire period of observation. CPX mutants with impaired SNARE interaction (mCPX4A, green curve) or lacking the accessory helical domain (mCPX48-134, yellow curve) fail to clamp fusion whilst the removal of c-terminal portion (mCPX26-83, purple curve) produces a partial clamping phenotype. The N-terminal domain is not involved in establishing the fusion clamp (C) End-point analysis at 10 s post-docking shows that the both the accessory helix deletion (mCPX48-134) and CPXcen modifications (mCPX4A) result in complete loss of inhibitory function and cannot be rescued even at 20 μM concentration. In contrast, the clamping function of the c-terminal deletion mutant (mCPX26-83) is fully restored at high CPX concentration. The average values and standard deviations from three independent experiments (with ~300 vesicles in total) are shown. **p < 0.01; *** p < 0.001 using the Student’s t-test.

Figure 2—source data 1

Data and summary statistics of docking and survival analysis for CPX mutants.

https://cdn.elifesciences.org/articles/71938/elife-71938-fig2-data1-v1.xlsx

Altogether, we conclude that the CPXcen-SNAREpin interaction promotes vesicle docking, and this interaction along with CPXacc are critical for mCPX mediated clamping under physiologically relevant experimental conditions. The c-terminal domain plays an auxiliary role and contributes to the mCPX inhibitory function likely by concentrating it on vesicle surfaces due to its curvature-binding region. Supporting this, a CPX mutant (CPXL117W) that enhances the curved membrane association of the c-terminal domain (Seiler et al., 2009) increased the clamping efficiency as compared to CPXWT (Figure 2—figure supplement 2).

Biophysical and structural studies have demonstrated that binding of the CPXcen to the SNAREpins positions the CPXacc to effectively block complete SNARE assembly (Kümmel et al., 2011; Giraudo et al., 2008; Krishnakumar et al., 2015). While the precise mode of action is under debate, there is evidence that this involves specific interactions of CPXacc with the c-terminal region of the SNAREpins (Kümmel et al., 2011; Malsam et al., 2020). Critical information about these inter-molecular interactions was provided by the X-ray structure of mCPX bound to a mimetic of a pre-fusion half-zippered SNAREpins (Kümmel et al., 2011). It revealed that the CPXcen is anchored to one SNARE complex, while its CPXacc extends away and binds to the t-SNARE in a second SNARE complex in a site normally occupied by the C-terminus of the VAMP2 helix (Kümmel et al., 2011; Krishnakumar et al., 2015). This trans-insertion model suggest a straightforward mechanism by which CPXacc can block the complete assembly of the SNARE complex (Kümmel et al., 2011; Krishnakumar et al., 2015).

To ascertain if the hydrophobic CPXacc-t-SNARE binding interfaces observed in the crystal structure are involved in clamping in our in vitro system, we tested known CPX mutants designed to either enhance (D27L E34F R37A, ‘super-clamp’ mutant mCPXSC) or weaken (A30E A31E L41 A44E, ‘non-clamp’ mutant 1 mCPXNC1) this interaction (Giraudo et al., 2009; Kümmel et al., 2011). Survival analysis of Syt1-vSUVs showed that the binding interface mutants indeed alter the inhibitory activity of CPX as predicted (Figure 3A, Figure 3—figure supplement 1). The mCPXNC1 abrogated the fusion clamp and was inactive even at higher (20 μM) concentration (Figure 3A, Figure 3—figure supplement 1). In contrast, mCPXSC increased the clamping efficiency and produced stably docked vesicles at lower concentrations (IC50 ~0.5 μM) compared to the mCPXWT (IC50 ~1 μM) (Figure 3B, Figure 3—figure supplement 1). These findings strongly support the notion that the CPXacc-t-SNARE interactions observed in the pre-fusion mCPX-SNAREpin crystal is relevant for the CPX clamping function and is physiologically relevant.

Figure 3 with 4 supplements see all
Specific interaction of mCPX accessory helix with SNAREs enhances its clamping function.

(A) End-point survival analysis (measured at 10 s post docking) using Syt1-vSUVs demonstrates that disrupting the binding of the CPXacc to either the t-SNAREs (CPXNC1) or the VAMP2 (CPXNC2) abrogates the clamping function, and that a helix breaking mutation (CPXGP) introduced between CPXcen and CPXacc also abrogates the fusion clamp. (B) In contrast, mutations designed to enhance the binding of CPXacc to t-SNAREs (CPXSC) increase the potency of the CPX clamp. This indicates efficient clamping by CPX requires a continuous rigid helix along with specific interaction of the CPXacc with the assembling SNARE complex. (C) Supporting this notion, survival analysis (Kaplan-Meier plots) shows that both Drosophila and C. elegans CPXs, which have very low sequence identity with the mCPX accessory domain, and a CPX mutant with a randomized accessory helical sequence (CPXEAAK) have poor clamping efficiency under standard (2 μM) experimental conditions and only partial clamping at higher (20 μM) concentration. The average values and standard deviations from three to four independent experiments (with ~250 vesicles in total) are shown. *** indicates p < 0.001 using the Student’s t-test.

Figure 3—source data 1

Data and summary statistics of survival analysis for ceCPX, dmCPX and mCPX mutants.

https://cdn.elifesciences.org/articles/71938/elife-71938-fig3-data1-v1.xlsx

Another key feature of the pre-fusion crystal structure is that the mCPX helix (CPXcen +CPXacc) forms a rigid bridge between two SNARE complexes (Kümmel et al., 2011; Krishnakumar et al., 2015). To test whether the rigidity of mCPX is important for clamping, we used a mCPX mutant (mCPXGP) having a helix-breaking linker (GPGP) inserted between CPXcen and CPXacc. We found that disrupting the continuous helix indeed reduced the clamping efficiency (Figure 3A, Figure 3—figure supplement 1) indicating that the continuity and rigidity of the CPX helix is mechanistically important for its inhibitory function. This is also consistent with other previous studies (Chen et al., 2002; Xue et al., 2007; Cho et al., 2014; Radoff et al., 2014).

Recently, site-specific photo-crosslinking studies in a reconstituted fusion assay revealed that CPXacc (of closely related mammalian isoform CPXII) binds to the c-terminal portions of SNAP25 and VAMP2 and both interactions are important for the mCPX inhibitory function (Malsam et al., 2020). The binding interface for SNAP25 was nearly identical to CPXacc-t-SNARE interface observed in the crystal structure while the opposite side of the CPXacc was found to interact with VAMP2 (Malsam et al., 2020). Note that this portion of VAMP2 was missing in the pre-fusion SNAREpin mimetic used for in the crystal structural analysis (Kümmel et al., 2011). To understand if the aforementioned CPXacc-VAMP2 interaction is also part of the clamping mechanism in our cell-free system, we used a mCPX mutant (K33E R37E A40K A44E; non-clamp mutant 2, mCPXNC2) that reverses the charge on key binding residues and is thus expected to disrupt this interaction (Malsam et al., 2020). mCPXNC2 also failed to clamp spontaneous fusion of Syt1-vSUVs in our in vitro assay (Figure 3A, Figure 3—figure supplement 1) and was phenotypically analogous to the t-SNARE non-binding mutant (mCPXNC1). This indicated the CPXacc interacts with both t- and v-SNAREs to block full-zippering. As expected, because their central helix is unaltered, the majority of CPXacc mutants tested retained the ability to promote vesicle docking process albeit lower than mCPXWT (Figure 3—figure supplement 2).

CPXcen is broadly conserved with ~75% amino acid sequence identity across diverse species, whereas CPXacc is highly divergent with ~25% sequence identity (Figure 3—figure supplement 3). Nonetheless, cross-species rescue experiments have been largely successful, and in fact, CPXacc could be exchanged without impairing function in mammalian, fly and nematode synapses (Xue et al., 2009; Cho et al., 2014; Radoff et al., 2014). This raises the question whether the distinct CPXacc-SNARE interactions that are vital for mCPX inhibitory functionality in our in vitro assays are physiologically relevant. To address this, we examined the clamping ability of the C. elegans (ceCPX) and Drosophila (dmCPX) orthologs of mCPX in our in vitro reconstituted assay. Under standard experimental conditions (2 μM CPX), both ceCPX and dmCPX were able to promote vesicle docking (Figure 3—figure supplement 2) but were considerably less efficient (~15% and ~ 30%, respectively) in preventing spontaneous fusion of Syt1-vSUV (Figure 3C) as compared near-complete ( > 95%) fusion clamp observed with mCPX (Figure 2B and C). Interestingly, simply increasing the concentrations improved the clamping efficacy of both dmCPX and ceCPX, with ~60–70% of docked vesicles stably-clamped at 20 μM concentration (Figure 3C) and remained Ca2+-sensitive (Figure 3—figure supplement 4).

This suggests that specific molecular interactions of CPXacc with SNAREs likely increase the potency of the mCPX inhibitory function and that this effect may be occluded at high concentrations of CPX. To verify this, we examined the effect of the mCPX mutant wherein the endogenous CPXacc domain (residues 26–48) is replaced with an artificial alpha helix based on a Glu-Ala-Ala-Lys (EAAK) motif repeated seven times (Radoff et al., 2014). Noteworthy, this construct (mCPXEAAK) was able to fully-restore CPX inhibitory functionality in C. elegans neuromuscular synapses (Radoff et al., 2014). In our in vitro assay, CPXEAAK enhanced initial docking (Figure 3—figure supplement 2) but failed to clamp spontaneous fusion (~10% efficiency) under standard experimental conditions (2 μM CPX) and was moderately effective (~50% efficiency) at higher (20 μM CPX) concentration (Figure 3C, Figure 3—figure supplement 4). We note that the accessory helix of mCPXEAAK is more hydrophobic in nature and interestingly resembles couple of the gain-of-function ‘super-clamp’ mutations with residue Asp-27 and Glu-34 replaced with Ala (Figure 3—figure supplement 3). This could potentially explain mCPXEAAK ability to partially clamp vesicle fusion at high (20 μM) CPX concentration. Overall, our data supports the notion the specific CPXacc-SNARE interaction is functionally relevant and likely enhances CPX inhibitory function.

Finally, we evaluated the probability and rate of Ca2+-triggered fusion from the clamped state in the presence and absence of mCPX. We used Syt1-vSUV loaded with Sulforhodamine B (fluorescent content marker) to track full-fusion events and lipid-conjugated Ca2+ indicator (Calcium green C24) attached to the planar bilayer to estimate the time of arrival of Ca2+ at/near the docked vesicles (Figure 4A). Consistent with our previous study, the influx of free Ca2+ (100 μM) triggered simultaneous fusion of >90% of the Syt1/mCPX-clamped vesicles (Figure 4B). These vesicles fused rapidly and synchronously, with a characteristic time-constant (τ) of ~11 msec following the arrival of Ca2+ locally (Figure 4C). Considering that the majority of Ca2+-triggered fusion occurs within a single frame (13 ms), we suspect that the true Ca2+-driven fusion rate is likely <10 ms.

Complexin increases the probability of Ca2+-triggered vesicular release.

(A) The effect of mCPX on Ca2+-triggered fusion was assessed using a content-release assay with Sulforhodamine-B loaded vesicles. Sulforhodamine-B is largely self-quenched when encapsulated inside an SUV. Fusion of the vesicle results in dilution of the probe, which is accompanied by increasing fluorescence. The Ca2+-sensor dye, Calcium Green, introduced in the suspended bilayer (via a lipophilic 24-carbon alkyl chain) was used to monitor the arrival of Ca2+ at/near the docked vesicles. A representative fluorescence trace before and after the addition of 100 μM Ca2+ shows that the rise in Sulforhodamine-B (red curve) fluorescence intensity occurs within a single frame (13 ms) of Ca2+ binding to local Calcium green (green curve) (B) End-point analysis at 1 min post Ca2+-addition shows that >90% of all Syt1/mCPX-clamped vesicles (~70 copies of VAMP2 and ~25 copies of Syt1) fuse following Ca2+ addition as compared to ~70% of Syt1-clamped vesicles (~13 copies of VAMP2 and ~25 copies of Syt1). Inclusion of mCPX enhances the fusion probability even under the low-VAMP2 condition suggesting that mCPX promote Ca2+-triggered fusion independent of its clamping function. (C) Kinetic analysis shows that the clamped vesicles with or without mCPX fuse rapidly following Ca2+-addition with near identical time constant of ~11 ms. This represents the temporal resolution limit of our recordings (13 ms frame rate) and the true Ca2+-triggered fusion rate may well be below 10 ms. (D) Deletion and mutational analysis under low-VAMP2 conditions (SUVs with ~13 copies of VAMP2 and ~25 copies of Syt1) show that the deletion of CPXacc (CPX48-134, blue bar) or disruption of CPXcen-SNARE interaction (CPX4A, green bar) abrogate the stimulatory function, but deletion of the N-terminal portion (CPX26-134, yellow bar) or the c-terminal domain (CPX26-83, purple bar) has no effect. The stimulatory function does not require rigid CPXacc-CPXcen helix (mCPXGP, orange bar) nor clamping specific CPXacc-SNARE interaction as non-clamping CPXEAAK mutant (cyan bar) and C. elegans ortholog (ceCPX, brown bar) retain stimulatory function. The average values and standard deviations from three independent experiments (with ~100 vesicles in total) are shown. ** p < 0.01, ***p < 0.001 using the Student’s t-test.

Figure 4—source data 1

Data and summary statistics of effect of mCPX mutants on calcium activation of fusion.

https://cdn.elifesciences.org/articles/71938/elife-71938-fig4-data1-v1.xlsx

In absence of mCPX, we observed a relatively small number of docked vesicles prior to Ca2+ influx and this precluded any meaningful quantitative analysis. Hence, to obtain stably docked vesicles without mCPX, we used low VAMP2 conditions that is, SUVs containing ~13 copies of VAMP2 and ~25 copies of Syt1 (Figure 1—figure supplement 1). We have previously demonstrated that under these conditions, Syt1 alone is sufficient to produce stably-clamped vesicles (Ramakrishnan et al., 2019) and that is what we observe, with >95% of docked vesicles immobile post-docking. Addition of Ca2+ (100 μM) triggered rapid and synchronous fusion of ~70% of these Syt1-clamped vesicles (with τ ~ 11 msec) as compared to >90% fusion of Syt1/mCPX-clamped vesicles (Figure 4B and C). Besides mCPX, the number of SNAREpins available on a given vesicle is also different between the two conditions (~13 VAMP2 in Syt1-alone vs. ~70 VAMP2 in Syt1/CPX). Hence, to verify that the observed effect is directly attributable to mCPX, we tested and confirmed that the inclusion of mCPX under low VAMP2 conditions increased the Ca2+-triggered fusion probability (~90%) from the clamped state (Figure 4B). This indicated that besides clamping vesicle fusion, mCPX also promotes Ca2+-triggered vesicle fusion from the clamped state.

To identify the molecular aspects underlying mCPX stimulatory function, we examined the effect of mCPX mutants on Ca2+-triggered release under low VAMP2 conditions (Figure 4D). Deletion of the N-terminal alone (mCPX26-134) or the N- and C-terminal domains (mCPX26-83) had very little or no effect on the mCPX stimulatory function (Figure 4D). However, deletion of the CPXacc in addition to N-terminal domain (mCPX48-134) or disrupting the CPXcen-SNARE interaction (mCPX4A) abrogated the mCPX activation function (Figure 4D) suggesting that the CPXcen and CPXacc domains are crucial for mCPX’s stimulatory function. In contrast to their clamping function, disrupting the rigidity and continuity of the CPXcen-CPXacc helix with the GPGP insert (mCPXGP) had no effect on the activation function (Figure 4D). Furthermore, the mCPX mutant with a randomized CPXacc (CPXEAAK) and the C. elegans ortholog (ceCPX), both of which lack the clamping functionality under the experimental conditions (Figure 3), retained the ability to promote Ca2+-triggered fusion of the docked vesicles (Figure 4D). Taken together, our data suggest that specific interactions of CPXcen with SNAREpins are required for the mCPX stimulatory function and the CPXacc can act independently of CPXcen via a mechanism different from that involved in clamping vesicle fusion.

Discussion

Our data indicates the mCPX is critical to produce the ‘clamped’ state and also contribute towards synchronizing fusion to Ca2+ influx. In addition, we find that the stimulatory and clamping functionality of mCPX are mechanistically separable. There is a long-standing debate over the role of CPX in establishing a fusion clamp and perhaps the best evidence in support has come from biochemical analyses (Giraudo et al., 2006; Giraudo et al., 2008; Kümmel et al., 2011; Lai et al., 2014) and physiological studies in invertebrate synapses (Huntwork and Littleton, 2007; Cho et al., 2014; Martin et al., 2011; Hobson et al., 2011). In the case of mammalian synapses, a role for CPX in blocking spontaneous release events remains controversial because KD/KO manipulations yield seemingly contradictory results and show neuron-specific differences (Xue et al., 2008; López-Murcia et al., 2019; Maximov et al., 2009; Yang et al., 2013). Here, using a fully defined albeit simplified cell-free system we provide compelling evidence that mCPX is an integral part of the overall clamping mechanism and delineate the molecular mechanism of mCPX inhibitory function. The distinct effects of different CPX truncation and targeted mutations match with data obtained from other reductionist or even physiological systems (Giraudo et al., 2006; Giraudo et al., 2008; Kümmel et al., 2011; Cho et al., 2014; Lai et al., 2014; Gong et al., 2016) forcefully arguing for the physiological relevance of results obtained from our in vitro reconstituted assay.

Our experiments indicate that mCPX inhibitory function entails distinct and specific interactions of the CPXcen and CPXacc domains with assembling SNAREpins, and that the c-terminal domain augments clamping function by increasing the local concentration and/or by proper orientation of CPX via interactions with the vesicle membrane (Figure 2). Our results indicate that CPXcen binds in the groove between assembling Syntaxin and VAMP2 helices at the early stages of vesicle docking to stabilize the partially-zippered SNAREpins, consequently promote vesicle docking. This in turn positions CPXacc to block further zippering of SNARE complex both by directly capturing the VAMP2 c-terminus and by simultaneously occupying its binding pocket on the t-SNARE. In line with earlier reports (Chen et al., 2002; Xue et al., 2007; Radoff et al., 2014; Cho et al., 2014; Kümmel et al., 2011), we find that a continuous, rigid CPX helix is essential for a stable fusion clamp. However, the precise configuration of this clamped state under the docked vesicles has been unclear. This is in large part due to the observed variability in the positioning of the CPXacc (Choi et al., 2016; Malsam et al., 2020; Zhou et al., 2017; Kümmel et al., 2011). CPXacc has been proposed to interact with c-terminal portion of the t-SNARE and VAMP2, both in a cis configuration that is CPXcen and CPXacc bound to the same SNAREpin (Choi et al., 2016; Malsam et al., 2020) or in a trans configuration that is CPXcen and CPXacc interacting with neighboring SNAREpins (Choi et al., 2016; Kümmel et al., 2011; Krishnakumar et al., 2015; Cho et al., 2014). We favor the trans insertion clamping model as this arrangement would enable CPX to regulate the distinct central and peripheral SNAREpin populations (Figure 5, see below).

Figure 5 with 2 supplements see all
Synergistic regulation of SNARE-mediated fusion by CPX and Syt1.

(A) Model of pre-fusion CPX-Syt-SNARE complex containing the central and peripheral SNAREpins connected via CPX trans-clamping interaction. The central SNAREpins, which are responsible for the Ca2+-triggered fusion, are bound to and sterically clamped by two Syt molecules - one independently at the ‘primary’ interface and other in the conjunction with CPXcen (red) at the ‘tripartite’ interface. The CPXacc (yellow) emanating from the central SNAREs reaches out to bind and clamp the peripheral SNAREpin (dark gray). This molecular model was generated using the X-ray crystal structures 5W5C (Zhou et al., 2017) and 3RL0 (Kümmel et al., 2011) (see Figure 5—figure supplement 1). Noteworthy, the positioning of peripheral SNAREpins in this model is likely to be flexible considering the inherent variability in the localization of CPXacc (B) Organization of pre-fusion CPX-Syt-SNARE complex at the synaptic vesicle-plasma membrane interface. In addition to the ‘bridging interaction’, the primary C2B domain (gray) also self-assembles to an oligomeric structure which strengthens the Syt1 clamp on the central SNAREpins. The SNAREpins are multi-colored, CPX is cyan and tripartite C2B is pink. Only a single cross-linked SNAREpins is shown, but multiple SNARE complexes are likely involved in driving rapid SV fusion (see Figure 5—figure supplement 2). We have omitted the transmembrane domains of SNAREs/Syt and the Syt C2A domains for clarity.

Noteworthy, we observe that the specific interactions of the CPXacc with the synaptic SNARE proteins increase the potency of the clamp, and in accordance mCPX is ~2–3 fold more efficient in establishing the fusion clamp as compared to dmCPX or ceCPX under the same experimental conditions (Figure 3). However, the divergence in clamping ability among the mammalian, fly, and nematode CPXs is diminished at higher concentrations of CPX. This might explain the puzzling observation that in physiological analyses, when CPX is over-expressed, cross-species rescue experiments are largely successful yet CPXacc-SNARE disrupting mutants’ exhibit limited effect on the CPX clamping ability (Yang et al., 2010; Cho et al., 2014; Radoff et al., 2014). Considering that the CPXacc is highly divergent across different species, it is conceivable that CPXacc has distinctively evolved to optimally bind and clamp the species-specific SNARE partners. Additional biochemical/structural studies are needed to address this question.

Overall, our data strongly argues that mCPX has an intrinsic capacity to inhibit SNARE-dependent fusion and under minimal conditions is required (along with Syt1) to generate and maintain a pool of release-ready vesicles. Indeed, functionality of mCPX observed in our in vitro system perfectly matches with physiological studies in model invertebrate systems (Martin et al., 2011; Hobson et al., 2011; Huntwork and Littleton, 2007; Cho et al., 2014). However, recent physiological studies in mammalian synapses reported that acute CPX loss reduces SV fusion probability but does not unclamp spontaneous fusion. Hence, they conclude that CPX is dispensable for ‘fusion clamping’ in mammalian neurons (López-Murcia et al., 2019). It is worth noting that under these conditions CPX removal abates both spontaneous and evoked neurotransmitter release without changing the number of docked vesicles (López-Murcia et al., 2019). This suggests that acute CPX loss likely affects the late-stage vesicle priming process, and it is possible this ‘loss-of-fusion’ phenotype occludes CPX role in regulating spontaneous fusion events. Indeed, rescue experiments in CPX1/2/3 triple-knockout mouse background show that the CPXacc mutants enhances the spontaneous fusion events without altering evoked release, revealing that mCPX has a strong suppressive clamping function (Malsam et al., 2020). It is feasible mCPX also plays a more specialized role in mammalian synapses and is primarily involved in stabilizing newly primed synaptic vesicles and prevents their premature fusion (Dhara et al., 2014; Chang et al., 2015). In doing so, mCPX may function as a fusion clamp in an activity-dependent manner and is critical to blocking spontaneous/tonic and asynchronous vesicular release (Dhara et al., 2014; Chang et al., 2015; Yang et al., 2010) and indirectly promoting synchronous SV exocytosis.

mCPX on its own is ineffective in clamping SNARE-driven vesicle fusion, as the c-terminal portion of VAMP2 assembles into the SNARE complex far faster than free CPX can bind to prevent its zippering (Gao et al., 2012). As such, a delay in SNARE zippering is required for the CPX to bind and thereby block fusion. The fact that sufficient delay can be artificially provided by ~20 copies of DNA duplexes (Figure 1) suggest that under physiological conditions, Syt1 (and perhaps other proteins on the SV) might hinder the SNARE assembly by a simple steric mechanism, enabling mCPX to function as a fusion clamp. This is supported by the observation that the Syt1 clamp or the formation of the central SNAREpins are not strictly required for mCPX clamping function (Ramakrishnan et al., 2020).

Ca2+-activation studies (Figure 4) show that mCPX also contributes to Ca2+-triggered vesicle fusion from the clamped state. Reinforcing our earlier reports (Ramakrishnan et al., 2019; Ramakrishnan et al., 2020), we find that Syt1 and a small number of SNAREs are largely sufficient to get Ca2+-evoked fusion with ~70% of vesicles fusing in response to 100 μM Ca2+. Inclusion of mCPX increases the fusion probability with >90% Ca2+-triggered fusion from the clamped state (Figure 4). We do not observe any change in the fusion kinetics (τ~11ms) without or with mCPX (Figure 4), at least with our current time resolution of ~13 ms and persistent high Ca2+ levels as opposed to Ca2+-transients in the synapse.

Deletion/mutational analyses reveal that the α-helical CPXcen and CPXacc are the minimal domain required for the activation function (Figure 4). Specifically, the well-defined CPXcen-SNARE interactions (Chen et al., 2002) was found to be critical for the stimulatory function and this effect is observed even low VAMP2 conditions that is with vesicles containing Syt1-clamped central SNAREpins only (Figure 4). This is in line with our previous finding that CPXcen interaction with the SNAREs, independent of the clamping functionality, is important for Ca2+-evoked release in Drosophila neuromuscular junctions (Cho et al., 2014). Our data shows that CPXacc also contributes to the activation function, but the underlying mechanism is unclear. CPXacc could act indirectly by promoting CPXcen binding (Radoff et al., 2014) or directly by interacting with the SNARE complex albeit in a manner different from the clamping interactions.

The data presented here, taken together with our earlier report (Ramakrishnan et al., 2020), suggests a parsimonious model of how Syt1 and CPX could regulate SNARE-mediated fusion (Figure 5, Figure 5—figure supplement 1, Figure 5—figure supplement 2). We posit that under every docked vesicle, there are two types of SNAREpins – the central SNAREpins which are bound to Syt1 and are responsible for Ca2+-triggered release and peripheral SNAREpins which are not bound to Syt1 and thus, not directly regulated by Ca2+. We further suggest that the central and peripheral SNAREpins are equal in number and are assembled as a pair via a common, bridging molecule of CPX (Figure 5A). At the early stages of SV docking, Syt1 oligomers bind and clamp sub-set of central SNAREpins via the ‘primary’ interface (Ramakrishnan et al., 2020). CPX bind the Syt1-associated central SNAREpins via the CPXcen which positions the CPXacc helix to bind the t-SNAREs an oppositely-oriented SNAREpin occupying the space where the C-terminal half of VAMP2 would ordinarily zipper to drive fusion. In this way, CPXacc acts to clamp the peripheral SNAREpin. This ‘bridging model’ (Figure 5, Figure 5—figure supplement 1) is based on the known ‘trans-clamping’ interaction observed in the pre-fusion CPX/SNAREpin crystal structure (Kümmel et al., 2011) and is validated by biochemical and functional analyses both previously (Cho et al., 2014; Krishnakumar et al., 2015; Krishnakumar et al., 2011; Kümmel et al., 2011) and in the current work. In addition, CPXacc might also directly interact with the peripheral VAMP2 c-terminus to prevent its assembly (not shown in Figure 5).

As evidenced in the recent crystal structure (Zhou et al., 2017), CPXcen binding to the central SNAREpins likely creates a new binding interface for second Syt1 to bind the same SNAREpins. Thus, mCPX could regulate Ca2+ triggered vesicle fusion via the ‘tripartite’ interface (Figure 5, Figure 5—figure supplement 1, Figure 5—figure supplement 2). Supporting this proposition, we have previously shown the ‘tripartite’ interface is not necessary to produce stably docked vesicles but is required for efficient Ca2+-triggered fusion from the clamped state (Ramakrishnan et al., 2020). In fact, disrupting binding of Syt1 to the tripartite interface lowers the fusion probability (~25%) similar to that observed with the removal of mCPX (Ramakrishnan et al., 2020). Furthermore, as the tripartite binding motif is largely conserved among different Synaptotagmin isoforms, so it is possible that mCPX binding could enable synergistically regulation of vesicular release by different calcium sensors (Volynski and Krishnakumar, 2018; Zhou et al., 2017). In addition to creating the ‘tripartite’ interface, mCPX binding might also promote vesicle fusion by stabilizing the full zippering SNARE complex. Obviously, this model is highly speculative and further functional studies (with higher temporal resolution, physiological Ca2+ dynamics and different calcium sensors) as well as high-resolution structural data of vesicle-membrane junctions are needed to dissect the precise role of mCPX and its synergistic action with Syt1 in regulating Ca2+-triggered vesicular fusion from the clamped state.

Materials and methods

Proteins and materials

Request a detailed protocol

The following cDNA constructs, which have been previously described (Krishnakumar et al., 2013; Ramakrishnan et al., 2019; Ramakrishnan et al., 2020), were used in this study: full-length VAMP2 (VAMP2-His6, residues 1–116); full-length VAMP24X (VAMP2-His6, residues 1–116 with L70D, A74R, A81D, L84D mutations), full-length t-SNARE complex (mouse His6-SNAP25B, residues 1–206 and rat Syntaxin1A, residues 1–288); Synaptotagmin (rat Synaptotagmin1-His6, residues 57–421); Complexins (human His6-Complexin 1, residues 1–134; C. elegans His6-Complexin, residues 1–143; Drosophila His6-Complexin1, residues 1–139). All mCPX mutants (truncations/point-mutations) were generated in the same background. All proteins were expressed and purified as described previously (Krishnakumar et al., 2013; Ramakrishnan et al., 2019; Ramakrishnan et al., 2020). All the lipids used in this study were purchased from Avanti Polar Lipids (Alabaster, AL). ATTO647N-DOPE was purchased from ATTO-TEC, GmbH (Siegen, Germany) and Calcium Green conjugated to a lipophilic 24-carbon alkyl chain (Calcium Green C24) was custom synthesized by Marker Gene Technologies (Eugene, OR). HPLC-purified DNA sequences (5’-ATCTCAATTATCCTATTAACC-3’ and 5’-GGTTAATAGGATAATTGAGAT-3’) conjugated to cholesterol with a 15 atom triethylene glycol spacer (DNA-TEG-Chol) were synthesized at Yale Keck DNA sequencing facility.

Liposome preparation

Request a detailed protocol

VAMP2 ( ± Syt1) were reconstituted into small unilamellar vesicles (SUV) were using rapid detergent (1% Octylglucoside) dilution and dialysis method as described previously (Ramakrishnan et al., 2019; Ramakrishnan et al., 2020). The proteo-SUVs were further purified via float-up using discontinuous Nycodenz gradient. The lipid composition was 88 (mole) % DOPC, 10% PS and 2% ATTO647-PE for VAMP2 ( ± Syt1) SUVs and we used protein: lipid (input) ratio of 1:100 for VAMP2 for physiological density, 1: 500 for VAMP2 at low copy number, and 1: 250 for Syt1. Based on the densitometry analysis of Coomassie-stained SDS gels and assuming the standard reconstitution efficiency, we estimated the vesicles contain 73 ± 6 (normal physiological-density) or 13 ± 3 (low-density) and 25 ± 4 copies of outward-facing VAMP2 and Syt1 respectively (Figure 1—figure supplement 1).

Single-vesicle fusion assay

Request a detailed protocol

All the single-vesicle fusion measurements were carried out with suspended lipid bilayers as previously described (Ramakrishnan et al., 2018; Ramakrishnan et al., 2019; Ramakrishnan et al., 2020). Briefly, t-SNARE-containing giant unilamellar vesicles (80% DOPC, 15% DOPS, 3% PIP2 and 2% NBD-PE) were prepared using the osmotic shock protocol and busted onto Si/SiO2 chips containing 5 µm diameter holes in presence of HEPES buffer (25 mM HEPES, 140 mM KCl, 1 mM DTT) supplemented with 5 mM MgCl2. The free-standing lipid bilayers were extensively washed with HEPES buffer containing 1 mM MgCl2 and the fluidity of the t-SNARE containing bilayers was verified using fluorescence recovery after photo-bleaching using the NBD fluorescence.

Vesicles (100 nM lipids) were added from the top and allowed to interact with the bilayer for 10 min. The ATTO647N-PE fluorescence introduced in the vesicles were used to track vesicle docking, post-docking diffusion, docking-to-fusion delays and spontaneous fusion events. The time between docking and fusion corresponded to the fusion clamp and was quantified using a ‘survival curve’ whereby delays are pooled together, and their distribution is plotted in the form of a survival function (Kaplan-Meier plots). For the end-point analysis, the number of un-fused vesicles (survival percentage) was estimated ~10 s post-docking. After the initial 10 min, the excess vesicles were removed by buffer exchange (3 x buffer wash) and 1 mM CaCl2 was added from the top to monitor the effect of Ca2+ on the docked vesicles. The number of fused (and the remaining un-fused) vesicles was estimated (end-point analysis) ~ 1 min after Ca2+-addition. CPX protein (at the indicated final concentration) were added to the experimental chamber and incubated for 5 min prior to the addition of the vesicles. Note: Pre-incubation with either the bilayer or the vesicle does not affect the clamping ability of mCPX and we chose to use pre-incubation with the bilayer (prior to adding SUVs) for the sake of convenience (Figure 1—figure supplement 2). All experiments were carried out at 37 °C using an inverted laser scanning confocal microscope (Leica-SP5) and the movies were acquired at a speed of 150ms per frame, unless noted otherwise. Fate of each vesicles were analyzed using our custom written MATLAB script described previously (Ramakrishnan et al., 2018). The files can be downloaded at: https://www.mathworks.com/matlabcentral/fileexchange/66521-fusion-analyzer-fas.

Single-vesicle docking analysis

Request a detailed protocol

To get an accurate count of the docked vesicles, we used VAMP2 mutant protein (L70D, A74R, A81D, and L84D; VAMP24X) that eliminates fusion without impeding the docking process (Krishnakumar et al., 2013). For the docking analyses, 100 nM VAMP24X containing SUVs (vSUV4X) were introduced into the chamber and allowed to interact with the t-SNARE bilayer for 10 min. The bilayer was then thoroughly washed with the running buffer (3 x minimum) and the number of docked vesicles were counted, using Image J software.

DNA-regulated single vesicle fusion assay

Request a detailed protocol

To prepare ssDNA containing vesicles, dialyzed VAMP2 or t-SNARE containing SUVs were incubated with complementary DNA-TEG-Chol for 2 hr at room temperature with mild-shaking. The v-SUVs were further purified using the Nycodenz gradient. We used the lipid: DNA-TEG-Chol input ratios of 1:2000, 1:1000, 1:500, and 1: 200 produce vSUVs with approximately 5, 10, 20, 50 copies of ssDNA per vesicles respectively. To identify the optimal condition for the single-vesicle fusion assays, we first tested the fusogenicity of ssDNA containing vesicles using bulk-fusion assay (Figure 1—figure supplement 2). Fusion of vSUV with t-SNARE liposomes were un-affected up to 20 copies of ssDNA, but we observed some reduction in fusion levels with 50 copies of ssDNA (Figure 1—figure supplement 2). Correspondingly, in the single-vesicle fusion setup, vSUV with 5, 10, and 20 copies of ssDNA docked and fused spontaneously with progressive docking-to-fusion delays, but the majority of 50 ssDNA-vSUV remained docked and un-fused (Data not shown). So, we chose to test the effect of Cpx on 20 ssDNA-vSUV, with 5 ssDNA-vSUV as the control.

Calcium dynamics

Request a detailed protocol

We used a high-affinity Ca2+-sensor dye, Calcium Green (Kd of ~75 nM) conjugated to a lipophilic 24-carbon alkyl chain (Calcium Green C24) introduced in bilayer to monitor the arrival of Ca2+ (100 μM). To estimate the arrival of Ca2+ at or near the docked vesicle precisely, as indicated by increased in Calcium green fluorescence at 532 nm, we used resonant scanner to acquire movies at a speed of up to 13 ms per frame with 512 × 32 resolution. For each vesicle fusion kinetics, calcium arrival was monitored over area of an individual hole (5 μm diameter) to get the high signal-to-noise ratio and vesicle fusion was monitored with 0.5 μm ROI around the docked vesicle. In these experiments, we used Sulforhodamine-B loaded Syt1-vSUV and tracked full-fusion events using increase in fluorescence signal due to dequenching of Sulforhodamine-B.

Data availability

All data generated or analyzed during this study are included in the manuscript and supporting files. Source data files for Figures 1, 2, 3, 4, and gel blots associated supplements are provided.

References

Decision letter

  1. Axel T Brunger
    Reviewing Editor; Stanford University School of Medicine, Howard Hughes Medical Institute, United States
  2. Vivek Malhotra
    Senior Editor; The Barcelona Institute of Science and Technology, Spain

Our editorial process produces two outputs: i) public reviews designed to be posted alongside the preprint for the benefit of readers; ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your Research Advance "Molecular Determinants of Complexin Clamping in Reconstituted Single-Vesicle Fusion" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Vivek Malhotra as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

While we find the work potentially interesting, there are 3 concerns (see below) that should be addressed .

Essential revisions:

1. The authors report that Cpx can clamp fusion in the presence of 20 ssDNAs, while no clamping was observed with 5 ssDNAs. Since ssDNA is an synthetic factor, conclusions about the copy number of ssDNAs should be drawn with caution. Please perform titration experiments with various ratios of SNAREs and Cpx to demonstrate the clamping kinetics at a variety of protein ratios and concentrations.

2. The calcium concentration used in the experiments presented in Figure 4 is very high (1 millimolar). Thus, the "activating" effect of Cpx (established in several previous studies in neuronal cultures and in reconstitution experiments) could may have been masked by the high calcium concentration. Please repeat these experiments at a more physiological relevant calcium concentration (e.g., 10-100 μM).

3. For adding Cpx, the authors used the protocol listed as: "CPX protein (at the indicated final concentration) were added to the experimental chamber and incubated for 5 minutes prior to the addition of the vesicles." Thus, Cpx was incubated with plasma membrane mimicking bilayers. However, previous work has shown that Cpx also interacts with synaptic vesicles and that this interaction may be essential for its function. Moreover, since Cpx is a curvature sensor, as washed by the injection flow of the vesicles, the effective Cpx concentration used in the present experiments may be much lower than claimed. Moreover, as shown in their previous publication [Ramakrishnan 2020 eLife], Cpx can be washed off. Please co-incubate Cpx with vesicles, and then inject calcium to trigger fusion.

4. The authors utilize several past lines of evidence to support the fusion clamp view of Cpx function including older studies of RNAi knock-down of mouse Cpx1/2, studies from model invertebrates such as worm and fly, as well as the authors' own earlier in vitro work. In general, work on Cpx in cultured mammalian neurons has failed to observe large increases in spontaneous SV fusion rates over the past ~8 years and the field has moved away from relying on RNAi knock down of Cpx1/2 due to artefacts associated with Cpx3 expression. Thus, the authors are advised to be cautious about using the older siRNA KD data as an argument supporting an inhibitory role for Cpx of spontaneous release in neurons. Regardless, the inhibitory role of Cpx is well established in reconstituted systems where calcium-independent fusion is monitored (which is potentially distinct from spontaneous release in neurons that actually depends on low levels of calcium and may depend on other fusion sensors than synaptotagmin1/2). Please comment and discuss.

5. Conclusions from the lack of an observed Cpx effect on the speed of calcium-triggered fusion may be a bit overstated. There could be changes that are not being detected but are nonetheless physiologically relevant. Synapses display a variety of kinetics with distinct speeds ranging from 100 seconds of microseconds to milliseconds, so there is room for interesting kinetics well below the current 13 milliseconds cutoff of these experiments. Perhaps there are interesting Cpx effects in that physiological range. Please comment and discuss.

6. The temporal resolution for fast fusion experiments is 13 milliseconds. Thus, it is questionable to use a 7 millisecond time interval for the plots shown in Figure 4C. Moreover, in Figure 4A, the dotted lines to indicate the half-maximum of Sulforhodamine B jump between two time points. However, the signal of Sulforhodamine B reaches its maximum at the next time point. Therefore, the definition of "fusion time" is unclear. The single vesicle fusion assay used in this manuscript is based on the previous eLife article where the authors stated that "We typically observed the fluorescence signal increase at the bilayer surface between about three frames (~ 100 milliseconds) after calcium addition (Figure 2—figure supplement 1). We therefore used 100 milliseconds as the benchmark to accurately estimate the time-constants for the ca2+-triggered fusion reaction." Why the authors used the much shorter 7 millisecond time interval in this study? Please clarify and discuss.

7. For the purpose of delaying fusion long enough to allow the complexin clamp to form, the authors replace synaptotagmin with duplex DNA. Unlike the complexin/synaptotagmin machine, however, this system once clamped cannot proceed to fusion, so it is unclear if the duplex DNA clamped state resembles the physiologically clamped state. Since no physiological experiments are performed, the claims drawn by the present studied should be qualified.

Reviewer #1 (Recommendations for the authors):

1. The authors utilize several past lines of evidence to support the fusion clamp view of Cpx function including older studies of RNAi knock-down of mouse Cpx1/2, studies from model invertebrates such as worm and fly, as well as the authors' own earlier in vitro work. In general, work on Cpx in cultured mammalian neurons has failed to observe large increases in spontaneous SV fusion rates over the past ~8 years and the field has moved away from relying on RNAi knock down of Cpx1/2 due to artefacts associated with Cpx3 expression. I believe that continued use of the older siRNA KD data as an argument supporting an inhibitory role for Cpx doesn't seem like a constructive way forward when aiming for the mammalian Cpx research community to digest and assimilate this new in vitro data.

2. There were no mentions of fly or worm Cpx constructs in the Methods section. For instance, was fly Cpx a CAAX box variant? Could that explain its poor clamping? And there was no description of the Cpx(EAAK) construct. The current lack of relevant Materials documentation is not acceptable for publication.

3. Conclusions from the lack of an observed Cpx effect on the speed of calcium-triggered fusion may be a bit overstated. There could be changes that are not being detected but are nonetheless physiologically relevant. Synapses display a variety of kinetics with distinct speeds ranging from 100s of microseconds to milliseconds, so there's room for interesting kinetics well below the current 13 msec cutoff of these experiments. Perhaps there are interesting Cpx effects in that physiological range.

4. For the helix-breaking experiments performed in this manuscript using the GPGP insertion (Results section lines 194-200), there is a long history of past work on precisely this concept for Cpx clamping, but these were not directly mentioned or referenced. I would suggest including references to Chen 2002, Xue 2007, Cho 2014 and Radoff 2014 for their past work on destabilizing the propagation of secondary structure from the accessory to central helix and its requirement for Cpx inhibitory function.

Reviewer #2 (Recommendations for the authors):

1. The authors report that Cpx can clamp fusion in the presence of 20 ssDNAs, while no clamping was observed with 5 ssDNAs. Since ssDNA is an artificial factor, one cannot get any useful information from the copy number ssDNAs. Therefore, a titration experiment with various ratios of SNAREs and Cpx should be performed to demonstrate clamping kinetics.

2. The Cpx truncations and 4A mutant used in Figure 2 have been shown to reduce spontaneous release before Ca++. What hasn't been shown is the difference on clamping kinetics for these Cpx variants. Therefore, a titration experiment is necessary.

3. The Ca++ concentration in Figure 4 is too high to test the effect of Cpx. At that high Ca++ concentration, the activation effect of Cpx could be overridden by Ca++. Since effective Ca++-triggered fusion has been observed with 10-100uM Ca++ by many in vitro proteoliposome assays, they have to test 10-100uM Ca++ with much more physiological relevance.

4. For adding Cpx, the authors used the protocol listed as: "CPX protein (at the indicated final concentration) were added to the experimental chamber and incubated for 5 min prior to the addition of the vesicles." Basically, Cpx was incubated with plasma membrane mimicking bilayers. However, many papers have shown that Cpx should interact with synaptic vesicles for its function. Moreover, since Cpx is a curvature sensor, as washed by the injection flow of the vesicles, their effective Cpx concentration would be much lower than what has been claimed. As shown in their previous publication [Ramakrishnan 2020 eLife], Cpx can be washed off. Therefore, the authors should co-incubate Cpx with vesicles, and then inject Ca++ to trigger fusion.

5. The temporal resolution for fast fusion experiments is 13msec. The foundation of using a 7 msec time interval for their plots shown in Figure 4C is not clear. No explanation was provided. There is another unclear definition about fusion time. In the Figure 4A, the authors tried to use dotted lines to indicate the half-maximum of Sulforhodamine B jump between two time points. However, the signal of Sulforhodamine B reaches its maximum at the next time point. Therefore, it's not clear about their definition on "fusion time". The single vesicle fusion assay used in this manuscript is based on their previous publications [Ramakrishnan 2019 Langmuir; Ramakrishnan 2020 eLife]. In the method section of their 2020 eLife paper, the authors stated that "We typically observed the fluorescence signal increase at the bilayer surface between about three frames (~100 msec) after ca2+ addition (Figure 2—figure supplement 1). We therefore used 100 msec as the benchmark to accurately estimate the time-constants for the ca2+-triggered fusion reaction." It would be good to learn why the authors start to use 7 msec to estimate fusion in this paper.

Reviewer # 3 (Recommendations for the authors):

There are many typographical and/or grammatical errors that the authors should fix. In Figure 2, the color-coding scheme should be consistent. That is, mCPX(48-134) should not be blue in panels A and C and yellow in panel B. In the bar graphs in panel C, the survival percentage 0 should be on, not above, the x-axis.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Molecular Determinants of Complexin Clamping and Activation Function" for further consideration by eLife. Your revised article has been evaluated by Vivek Malhotra (Senior Editor) and a Reviewing Editor.

The manuscript has been much improved but there are some remaining issues that need to be addressed before acceptance, as outlined below:

It is gratifying that the activating effect of Cpx becomes apparent at 100 μm ca2+. At that concentration, Cpx does not alter the fusion kinetics. However, it is possible that there is an effect on the fusion kinetics at even lower calcium concentrations. Please perform an additional experiment at a lower ca2+ concentration, e.g., 25 μm ca2+.

Lines 150-169: how were some of the numbers presented in this paragraph (~80%, ~20%, ~5-fold, etc.) derived from the figures cited?

Lines 239-250: the discussion of the mCPX(EAAK) construct is unclear. How does this explain the ability of this construct to rescue clamping (50%, similar to dmCPX and ceCPX) at higher concentrations?

Line 269: How many VAMP2 are on a vesicle? Is it ~70 (text), 60 (as suggested by the notation Syt1-VAMP60 in Figure 4B/C), or ~74 (Figure 4 legend)?

Lines 275-276: The phrase: "Deletion of the N-terminal (mCPX26-134) or the C-terminal domain (mCPX26-83)…" is not quite right. The second of these constructs lacks the N-terminal and the C-terminal domains.

Lines 279-281: "In fact, CPXacc deletion or CPXcen modification lowered fusion probability even below that observed under no mCPX condition (Figure 4D) suggesting that the CPXcen and CPXacc domains are crucial for mCPX's stimulatory function." How does the second part of the sentence follow from the first?

Lines 311-313: how CPXacc could be "capable of directly capturing the VAMP2 C-terminus and simultaneously occupying its binding pocket on the t-SNARE." This would seem to be a notable feat for any sequence, to say nothing of a sequence that can be swapped out for distantly related ones like (EAAK)7. Please clarify.

Lines 239-250: The results with mCPX(EAAK) are intriguing/puzzling. The authors seem to suggest that (EAAK)7 has 30% sequence identify with mCPXacc – is that really true? And if not, how should we think about the ability of this construct to rescue clamping (50%, similar to dmCPX and ceCPX) at higher concentrations?

https://doi.org/10.7554/eLife.71938.sa1

Author response

Essential revisions:

1. The authors report that Cpx can clamp fusion in the presence of 20 ssDNAs, while no clamping was observed with 5 ssDNAs. Since ssDNA is an synthetic factor, conclusions about the copy number of ssDNAs should be drawn with caution. Please perform titration experiments with various ratios of SNAREs and Cpx to demonstrate the clamping kinetics at a variety of protein ratios and concentrations.

We have examined the clamping function of mCPX under different reconstitution conditions in the DNA-regulated fusion experiments. Specifically, we assessed the effect of varying the mCPX concentration, VAMP2 density, ssDNA numbers and select mCPX mutants (both individually and in combination). This data, included as Figure 1 Supplement 4 in the revised manuscript, is consistent with our conclusions regarding the need for delay in fusion kinetics for mCPX to arrest SNARE-driven fusion and the underlying molecular mechanisms.

2. The calcium concentration used in the experiments presented in Figure 4 is very high (1 millimolar). Thus, the "activating" effect of Cpx (established in several previous studies in neuronal cultures and in reconstitution experiments) could may have been masked by the high calcium concentration. Please repeat these experiments at a more physiological relevant calcium concentration (e.g., 10-100 μM).

As recommended, we have repeated the ca2+-triggered fusion experiments with 100 µM ca2+ and find that mCPX increases the probability of fusion from the clamped state (~90% with mCPX vs. ~70% without mCPX) but does not alter the kinetics of release (t~11 ms under all conditions). We further carried out deletion/mutation analysis to identify that the CPXcen + CPXacc constitute the minimal domain for the stimulatory function. Interestingly, we find that neither the continuous, rigid helical domain nor the CPXacc-SNARE clamping interaction are required for the activation function. This data is included as Figure 4 in the revised manuscript and the Title/Abstract/Result/Discussion section have been revised accordingly.

3. For adding Cpx, the authors used the protocol listed as: "CPX protein (at the indicated final concentration) were added to the experimental chamber and incubated for 5 minutes prior to the addition of the vesicles." Thus, Cpx was incubated with plasma membrane mimicking bilayers. However, previous work has shown that Cpx also interacts with synaptic vesicles and that this interaction may be essential for its function. Moreover, since Cpx is a curvature sensor, as washed by the injection flow of the vesicles, the effective Cpx concentration used in the present experiments may be much lower than claimed. Moreover, as shown in their previous publication [Ramakrishnan 2020 eLife], Cpx can be washed off. Please co-incubate Cpx with vesicles, and then inject calcium to trigger fusion.

In the development of the in vitro assay, we systematically tested the effect of pre-incubation of CPX with the bilayer or SUVs and found no difference in the kinetics or clamping ability. We chose to use pre-incubation with the bilayer (prior to adding SUVs) for the sake of convenience. We have additionally tested and confirmed this proposition by using mCPX mutants lacking the c-terminal domain (mCPX26-83) or carrying L117W mutation that enhances its interaction with lipid bilayer (mCPXL117W). This data is included as Figure 1 Supplement 2 in the revised manuscript and explicated in the Methods section.

4. The authors utilize several past lines of evidence to support the fusion clamp view of Cpx function including older studies of RNAi knock-down of mouse Cpx1/2, studies from model invertebrates such as worm and fly, as well as the authors' own earlier in vitro work. In general, work on Cpx in cultured mammalian neurons has failed to observe large increases in spontaneous SV fusion rates over the past ~8 years and the field has moved away from relying on RNAi knock down of Cpx1/2 due to artefacts associated with Cpx3 expression. Thus, the authors are advised to be cautious about using the older siRNA KD data as an argument supporting an inhibitory role for Cpx of spontaneous release in neurons. Regardless, the inhibitory role of Cpx is well established in reconstituted systems where calcium-independent fusion is monitored (which is potentially distinct from spontaneous release in neurons that actually depends on low levels of calcium and may depend on other fusion sensors than synaptotagmin1/2). Please comment and discuss.

Our intent is to provide necessary background and highlight the ongoing debate on the clamping function of CPX, particularly in mammalian synapses. Thus, emphasizing the need for an in vitro reconstituted system to investigate the mechanistic details and complement the physiological studies. Indeed, we have clearly noted (in the same paragraph) that the observed differences in physiological studies could be due to perturbation method used and homeostatic compensatory mechanisms. We believe that we have adequately addressed this issue within the manuscript.

5. Conclusions from the lack of an observed Cpx effect on the speed of calcium-triggered fusion may be a bit overstated. There could be changes that are not being detected but are nonetheless physiologically relevant. Synapses display a variety of kinetics with distinct speeds ranging from 100 seconds of microseconds to milliseconds, so there is room for interesting kinetics well below the current 13 milliseconds cutoff of these experiments. Perhaps there are interesting Cpx effects in that physiological range. Please comment and discuss.

We have now repeated the experiment with physiologically-relevant ca2+ concentration (100 µM) and find that CPX increases the probability, but not the kinetics of fusion (See response to Q2 for details). It is possible that effect of CPX on kinetics of ca2+-activated fusion might be occluded by the time-resolution (13 ms) of our current experimental setup. We have mentioned this caveat in the Discussion section of the revised manuscript.

6. The temporal resolution for fast fusion experiments is 13 milliseconds. Thus, it is questionable to use a 7 millisecond time interval for the plots shown in Figure 4C. Moreover, in Figure 4A, the dotted lines to indicate the half-maximum of Sulforhodamine B jump between two time points. However, the signal of Sulforhodamine B reaches its maximum at the next time point. Therefore, the definition of "fusion time" is unclear.

We use the initial jump in Sulforhodamine-B and Calcium-green signal (marked by dashed lines) rather the peak fluorescence to define the fusion time as this likely correlate to earliest instance of fusion pore opening. Since this occurs within a single frame of recording, we used t1/2 (~7 msec) of our single frame (13 msec) for the plots in Figure 4C. We strongly believe that the fusion occurs <10 msec in these conditions.

7. The single vesicle fusion assay used in this manuscript is based on the previous eLife article where the authors stated that "We typically observed the fluorescence signal increase at the bilayer surface between about three frames (~ 100 milliseconds) after calcium addition (Figure 2—figure supplement 1). We therefore used 100 milliseconds as the benchmark to accurately estimate the time-constants for the ca2+-triggered fusion reaction." Why the authors used the much shorter 7 millisecond time interval in this study? Please clarify and discuss.

Consistent with our previous report, we observe the ca2+ fluorescence signal at the bilayer surface ~100 msec after addition. The ca2+-addition to the top of the chamber is denoted by arrow in Figure 4A in the revised manuscript (Note: This arrow was inadvertently placed at the 100 msec mark in our older version of the Figure 4A. We apologize for the confusion)

8. For the purpose of delaying fusion long enough to allow the complexin clamp to form, the authors replace synaptotagmin with duplex DNA. Unlike the complexin/synaptotagmin machine, however, this system once clamped cannot proceed to fusion, so it is unclear if the duplex DNA clamped state resembles the physiologically clamped state. Since no physiological experiments are performed, the claims drawn by the present studied should be qualified.

We used DNA-regulated fusion assay to assess and confirm that the delay in fusion is necessary for mCPX to arrest SNARE assembly. We have now carried out systematic titration of VAMP2, ssDNA and CPX concentration (including CPX mutants) to establish that the clamping mechanisms observed under DNA-regulated fusion reaction is identical to those observed in the presence of Syt1 (see response to Q1 for details). This allows us to directly infer that under physiologically-relevant conditions delay in fusion kinetics, likely imparted by Syt1 or other SV proteins, is required for CPX to clamp SNARE-mediated fusion.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

The manuscript has been much improved but there are some remaining issues that need to be addressed before acceptance, as outlined below:

It is gratifying that the activating effect of Cpx becomes apparent at 100 μm ca2+. At that concentration, Cpx does not alter the fusion kinetics. However, it is possible that there is an effect on the fusion kinetics at even lower calcium concentrations. Please perform an additional experiment at a lower ca2+ concentration, e.g., 25 μm ca2+.

We find that the vast majority of the ca2+-triggered vesicle fusion occurs within a single frame irrespective of the ca2+-concentration (0.1 mM or 1 mM) or acquisition rate (13 msec or 36 msec). Furthermore, we observe that the mutations in Syt1 or CPX alter the probability of fusion, but not the fusion kinetics. This suggests that the fusion rate is likely much faster than the temporal resolution (13 msec) of our experimental setup. As such, any alterations in fusion kinetics, even at lower (25 mM) ca2+ concentration, might not be resolved in the current setup. We are working to increase the temporal resolution (1-2 ms) of our fusion system and to include better ca2+ control. Our future work is focused on using the upgraded fusion setup to dissect the role of Syt1/CPX in modulating vesicle fusion kinetics and as such, is beyond the scope of the current work.

Lines 150-169: how were some of the numbers presented in this paragraph (~80%, ~20%, ~5-fold, etc.) derived from the figures cited?

The fusion data for Syt1-vSUVs (black curve) is derived from the survival analysis shown in Figure 2B. Consistent with our previous reports (Ramakrishnan et al. 2019 and 2020), we observe that ~80% of Syt1-vSUVs fuse within 4-5 sec and remainder (~20%) stay ‘clamped’ and remain ca2+-sensitive up to 2-3 hours.

The fold changes in the number of docked vesicles are derived from the docking analysis shown in Figure 2A. We inadvertently included the wrong data for the no CPX condition in Figure 2A. We have corrected this mistake and have revised the fold-increase values in the manuscript. For example, we observe ~2-3 docked vesicles per 5 mm hole for Syt1-vSUVs (grey bar) and the inclusion of mCPXWT (red bar) increases the number of docked vesicles to ~7-8 vesicles per 5 mm hole, i.e. ~3-fold increase in the number of docked vesicles.

Lines 239-250: the discussion of the mCPX(EAAK) construct is unclear. How does this explain the ability of this construct to rescue clamping (50%, similar to dmCPX and ceCPX) at higher concentrations?

Based on the sequence alignment of the accessory helical region alone (mCPX residues 26-48) we stated that mCPXEAAK has 30% sequence similarity to mCPX. We acknowledge that this might be over-simplification as the accessory helix alignment is likely shaped by the central helix also. Indeed, if the central helix region is included (mCPX residues 26-70) in the sequence alignment, then the mCPXEAAK accessory helix region has low-degree sequence similarity with the mCPXacc (Figure 3 Supplement 3). Nonetheless, we find that the mCPXEAAK accessory helix is more hydrophobic in nature and in fact, resemble couple of gain-of-function ‘super-clamp’ mutations (residues Asp-27 and Glu-34 replaced by Ala) that has been shown to increase the overall clamping efficiency (Figure 3 Supplement 3). This could possibly explain mCPXEAAK ability to partially rescue clamping at 20 mM, similar to dmCPX and ceCPX, even though it was an artificial accessory helix sequence.

We have included the sequence alignment data as Figure 3 Supplement 3 in the revised manuscript. We have also revised the relevant section (Page 8) to remove the reference to 30% sequence identity and included the following language “We note that the accessory helix of mCPXEAAK is more hydrophobic in nature and indeed, resembles couple of the gain-of-function ‘super-clamp’ mutations, with residue Asp-27 and Glu-34 replaced with Ala (Figure 3 Supplement 3). This could potentially explain how a randomized accessory helix sequence could partially rescue clamping at high (20 mM CPX) concentration.”

Line 269: How many VAMP2 are on a vesicle? Is it ~70 (text), 60 (as suggested by the notation Syt1-VAMP60 in Figure 4B/C), or ~74 (Figure 4 legend)?

We sought to reconstitute 60 copies of VAMP2 per vesicles, but based on the densitometry analysis, we estimate that each vesicle contains 73 ± 6 of outward facing VAMP2. To be consistent, we now denote as ~70 VAMP2 (text) or as VAMP70 (Figure 4 and Figure 1 Supplement).

Lines 275-276: The phrase: "Deletion of the N-terminal (mCPX26-134) or the C-terminal domain (mCPX26-83)…" is not quite right. The second of these constructs lacks the N-terminal and the C-terminal domains.

We apologize for this error and have fixed the language in the revised manuscript.

Lines 279-281: "In fact, CPXacc deletion or CPXcen modification lowered fusion probability even below that observed under no mCPX condition (Figure 4D) suggesting that the CPXcen and CPXacc domains are crucial for mCPX's stimulatory function." How does the second part of the sentence follow from the first?

We agree. This sentence is a confusing and we have revised this section (Page 9) to read as follows: “Deletion of the CPXacc in addition to N-terminal domain (mCPX48-134) or disrupting the CPXcen-SNARE interaction (mCPX4A) abrogated the mCPX activation function (Figure 4D) suggesting that the CPXcen and CPXacc domains are crucial for mCPX’s stimulatory function”.

Lines 311-313: how CPXacc could be "capable of directly capturing the VAMP2 C-terminus and simultaneously occupying its binding pocket on the t-SNARE." This would seem to be a notable feat for any sequence, to say nothing of a sequence that can be swapped out for distantly related ones like (EAAK)7. Please clarify.

Our data, taken together with the previous studies (Kummel et al. 2011, Krishnakumar et al. 2011, Malsam et al. 2020) strongly indicate that the CPXacc binds to c-terminal end of both VAMP2, and t-SNAREs and these interactions are part of the clamping mechanism. This dual-binding modality might be feasible considering that the c-terminal portion of the SNARE proteins, esp. VAMP2 is likely unstructured and flexible in the pre-fusion clamped state.

https://doi.org/10.7554/eLife.71938.sa2

Article and author information

Author details

  1. Manindra Bera

    1. Yale Nanobiology Institute, New Haven, United States
    2. Department of Cell Biology, Yale University School of Medicine, New Haven, United States
    Contribution
    Conceptualization, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-9297-8126
  2. Sathish Ramakrishnan

    1. Yale Nanobiology Institute, New Haven, United States
    2. Department of Pathology, Yale University School of Medicine, New Haven, United States
    Contribution
    Conceptualization, Formal analysis, Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-7844-2234
  3. Jeff Coleman

    1. Yale Nanobiology Institute, New Haven, United States
    2. Department of Cell Biology, Yale University School of Medicine, New Haven, United States
    Contribution
    Methodology, Resources
    Competing interests
    No competing interests declared
  4. Shyam S Krishnakumar

    1. Yale Nanobiology Institute, New Haven, United States
    2. Departments of Neurology, Yale University School of Medicine, New Haven, United States
    Contribution
    Conceptualization, Formal analysis, Investigation, Supervision, Writing – original draft, Writing – review and editing
    For correspondence
    shyam.krishnakumar@yale.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-6148-3251
  5. James E Rothman

    1. Yale Nanobiology Institute, New Haven, United States
    2. Department of Cell Biology, Yale University School of Medicine, New Haven, United States
    Contribution
    Conceptualization, Investigation, Supervision, Writing – review and editing
    For correspondence
    james.rothman@yale.edu
    Competing interests
    No competing interests declared

Funding

National Institute of Diabetes and Digestive and Kidney Diseases (DK027044)

  • Shyam S Krishnakumar
  • James E Rothman

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

This work was supported by National Institute of Health (NIH) grant DK027044. We thank Dr. Kirill Grushin for help with the structural models.

Senior Editor

  1. Vivek Malhotra, The Barcelona Institute of Science and Technology, Spain

Reviewing Editor

  1. Axel T Brunger, Stanford University School of Medicine, Howard Hughes Medical Institute, United States

Publication history

  1. Preprint posted: July 5, 2021 (view preprint)
  2. Received: July 12, 2021
  3. Accepted: March 15, 2022
  4. Version of Record published: April 20, 2022 (version 1)

Copyright

© 2022, Bera et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 295
    Page views
  • 46
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Manindra Bera
  2. Sathish Ramakrishnan
  3. Jeff Coleman
  4. Shyam S Krishnakumar
  5. James E Rothman
(2022)
Molecular determinants of complexin clamping and activation function
eLife 11:e71938.
https://doi.org/10.7554/eLife.71938

Further reading

    1. Neuroscience
    Camille S Wang et al.
    Research Article Updated

    Synapses maintain both action potential-evoked and spontaneous neurotransmitter release; however, organization of these two forms of release within an individual synapse remains unclear. Here, we used photobleaching properties of iGluSnFR, a fluorescent probe that detects glutamate, to investigate the subsynaptic organization of evoked and spontaneous release in primary hippocampal cultures. In nonneuronal cells and neuronal dendrites, iGluSnFR fluorescence is intensely photobleached and recovers via diffusion of nonphotobleached probes with a time constant of ~10 s. After photobleaching, while evoked iGluSnFR events could be rapidly suppressed, their recovery required several hours. In contrast, iGluSnFR responses to spontaneous release were comparatively resilient to photobleaching, unless the complete pool of iGluSnFR was activated by glutamate perfusion. This differential effect of photobleaching on different modes of neurotransmission is consistent with a subsynaptic organization where sites of evoked glutamate release are clustered and corresponding iGluSnFR probes are diffusion restricted, while spontaneous release sites are broadly spread across a synapse with readily diffusible iGluSnFR probes.

    1. Neuroscience
    Nikoloz Sirmpilatze et al.
    Research Article

    During deep anesthesia, the electroencephalographic (EEG) signal of the brain alternates between bursts of activity and periods of relative silence (suppressions). The origin of burst-suppression and its distribution across the brain remain matters of debate. In this work, we used functional magnetic resonance imaging (fMRI) to map the brain areas involved in anesthesia-induced burst-suppression across four mammalian species: humans, long-tailed macaques, common marmosets, and rats. At first, we determined the fMRI signatures of burst-suppression in human EEG-fMRI data. Applying this method to animal fMRI datasets, we found distinct burst-suppression signatures in all species. The burst-suppression maps revealed a marked inter-species difference: in rats, the entire neocortex engaged in burst-suppression, while in primates most sensory areas were excluded—predominantly the primary visual cortex. We anticipate that the identified species-specific fMRI signatures and whole-brain maps will guide future targeted studies investigating the cellular and molecular mechanisms of burst-suppression in unconscious states.