VAMP4 directs synaptic vesicles to a pool that selectively maintains asynchronous neurotransmission

Abstract

Synaptic vesicles in the brain harbor several soluble N-ethylmaleimide-sensitive-factor attachment protein receptor (SNARE) proteins. With the exception of synaptobrevin2, or VAMP2 (syb2), which is directly involved in vesicle fusion, the role of these SNAREs in neurotransmission is unclear. Here we show that in mice syb2 drives rapid Ca(2+)-dependent synchronous neurotransmission, whereas the structurally homologous SNARE protein VAMP4 selectively maintains bulk Ca(2+)-dependent asynchronous release. At inhibitory nerve terminals, up- or downregulation of VAMP4 causes a correlated change in asynchronous release. Biochemically, VAMP4 forms a stable complex with SNAREs syntaxin-1 and SNAP-25 that does not interact with complexins or synaptotagmin-1, proteins essential for synchronous neurotransmission. Optical imaging of individual synapses indicates that trafficking of VAMP4 and syb2 show minimal overlap. Taken together, these findings suggest that VAMP4 and syb2 diverge functionally, traffic independently and support distinct forms of neurotransmission. These results provide molecular insight into how synapses diversify their release properties by taking advantage of distinct synaptic vesicle-associated SNAREs.

Figure 1. Synaptic localization of VAMP4

(a) Expression of VAMP4 in area CA1 of the hippocampus. Strong immunofluorescence is observed in the cell body layer (s.p. stratum pyramidale), consistent with VAMP4 localization in the Golgi apparatus. In addition, punctate staining is observed in every layer of the hippocampus including stratum radiatum (s.r.) and stratum oriens (s.o.). Calibration bar: 50 µm (b–d) Double immunostaining with CCK demonstrates that VAMP4 is expressed in CCK-positive terminals in the str. radiatum (arrows). Calibration bar: 15 µm. (e, f) Expression of VAMP4 in presynaptic terminals (t) is confirmed at the electron microscopic level using silver-enhanced immunogold particles. The absence of a clear postsynaptic density is consistent with this example being an inhibitory synapse Calibration bar: 200 nm.

Figure 2. VAMP4 mediates evoked asynchronous neurotransmitter release

(a) Traces (left) and average peak values (right) of evoked IPSCs in uninfected syb2 deficient neurons (n=16), or infected with VAMP4 (n=31) or syb2 (n=35). Here and in all subsequent figures bars represent mean ± standard error of the mean (s.e.m.) and "*" denotes statistical significance between the groups assessed by one-way ANOVA-Fisher test at p<0.05. (b) (left) Response of the same cells shown in (a) to 50 action potentials (APs) triggered at 10 Hz. (Right) Average cumulative charge transfer of responses. Dotted line represents the time point at which the three groups begin to be statistically different. (c) Normalized traces (gray) and average normalized traces (black) of the 1st, 10th, 20th, 30th 40th and 50th IPSC in a 10 Hz train of stimulation. The left graph displays the variance of the normalized current calculated for 0.5 ms, starting at 40 ms. Levene’s test was used to determine the significant differences between the variances using a F value <0.05 (n=24). The current values averaged between 40 and 40.5 ms are different only at the first IPSC, where the variance is not different. (d) Representative traces (left) and mean charge transfer values integrated during the first 2 s of 30 s hypertonic stimulation (+0.5 M sucrose) (right) in non-infected syb2 deficient neurons (n=4 cells) or those infected with VAMP4 (n=6) and syb2 (n= 3). (e) Representative traces (left) and average frequency (right) of mIPSCs in non-infected syb2 deficient neurons (n=5 cells) or those infected with VAMP4 (n=12) and syb2 (n=8).

Figure 3. VAMP4 mediated evoked neurotransmitter release is susceptible to slow Ca²⁺ buffering

(a) Representative traces for IPSCs evoked by a single AP in syb2- or VAMP4-infected syb2 knockout neurons after pre-incubation in DMSO (vehicle) or EGTA-AM (10 µM) (n= 8 for each group) for 10 minutes. N.S. denotes lack of significance. (b) (Left) Representative traces for IPSCs evoked by 50 APs triggered at 10 Hz in syb2- or VAMP4-infected neurons after pre-incubation in DMSO (vehicle) or EGTA-AM (n= 8 for each group). (Right) Average cumulative charge transfer of responses. The gradual increase is neurotransmission seen after EGTA-AM treatment is attributable to Ca²⁺ buffer saturation. Dotted line represents the time from which all groups are statistically different. (c) (left) Representative traces of IPSCs evoked by 50 APs applied at 10 Hz in wild type hippocampal neurons overexpressing syb2 and pre-incubated with DMSO or EGTA-AM. (right) Average cumulative charge transfer values of IPSCs in neurons overexpressing syb2 and pre-incubated with DMSO or EGTA-AM (n=9 for each group). The difference between the two groups is not statistically significant at any time point. (d) (left) IPSCs in neurons overexpressing VAMP4 under the same conditions as in c. (right) Average cumulative charge transfer values of IPSCs evoked by 50 APs at 10 Hz in neurons overexpressing VAMP4 and pre-incubated with DMSO (n=16) or EGTA-AM (n=14). (e, f) Traces depict a progressive increase in asynchronicity of individual evoked IPSCs after VAMP4 overexpression compared to syb2 overexpression. Thick lines denote the mean of all traces depicted.

Figure 4. VAMP4 loss-of-function attenuates the extent of asynchronous release

(a) Immunoblot analysis of neurons pooled from 4 coverslips infected with L-307 virus (control) or VAMP4 knockdown viruses (KD1 and KD2). Syx1=Syntaxin 1; Syt1=Synaptotagmin 1; Syp1=Synaptophysin 1. (b) Traces and average cumulative charge histograms (obtained from individual IPSC traces) depicting the synchronicity of synaptic responses during 1 Hz stimulation (in 2 mM Ca²⁺) after VAMP4 knockdown (control: n=11, KD1: n=15, KD2: n=12). (c) The same analysis as in b performed on 1 Hz synaptic responses recorded after 600 APs at 20 Hz (control: n=15, KD1: n=11, KD2: n=12). Dashed line indicates statistical significance between VAMP4 knockdown (KD2) and control groups (p< 0.05). The difference between the KD1 group and controls was not significant (p=0.2). (d) Late asynchronous release in the presence of 8 mM Ca²⁺ evoked by 50 APs applied at 10 Hz in rat hippocampal neurons infected with L-307 virus (control) or VAMP4 knockdown virus (KD1). Normalized traces depict the kinetic difference between the two groups. (e) Average charge transfer values of the late asynchronous release measured in 8 mM Ca²⁺ within 5 s after the cessation of 10 Hz stimulation in neurons infected with control (n=14), KD1 (n=12) or KD2 (n=11) viruses. (f) Average cumulative charge transfer for IPSCs in neurons infected with control (n=13), KD1 (n=11) or KD2 (n= 11) viruses. Dotted line represents the time point when the KD1 and KD2 groups are statistically different from control. (g) Average values of initial IPSC peaks measured in neurons infected with control (n=13), KD1 (n=11), KD2 (n= 11) constructs.

Figure 5. The ternary complex of VAMP4 with syntaxin 1/SNAP-25 does not engage complexins or synaptotagmin 1

(a) VAMP4 forms a stable SDS-resistant complex with Syntaxin 1 and SNAP-25 similar to syb2. Purified recombinant proteins (~5 µM) were mixed and incubated for 30 min at 20°C and complex formation was visualized by PAGE and Coomassie staining. (b) Complexin 2 binds syb2-, but not VAMP4-, containing ternary complex. GST-complexin 2 glutathione beads were incubated with the two preassembled ternary complexes and the bound material was visualized by PAGE and Coomassie staining. All samples were boiled prior to loading to disrupt the ternary complexes and reveal individual proteins. Asterisk indicates the GST tag alone present on the GST-complexin 2 beads. Note that recombinant SNAP-25 used in this study contained a minor proteolytic degradation product (second band slightly below 25 kDa). (c) VAMP4 containing ternary complex does not bind endogenous complexins and synaptotagmin 1. GST-SNAP-25 (control) and VAMP4- or syb2-containing complexes were immobilized on glutathione beads via the GST-SNAP-25 protein (upper panel). Rat brain detergent extract was incubated with the beads for 1 hour at 4°C and, following washing, the bound material was visualized by immunoblotting (lower panel).

Figure 6. Trafficking of VAMP4 at central synapses

(a) Schematic structures of VAMP4-pHluorin and syb2-mOrange constructs. (b) (Left) Co-localization (arrows) of syb2-pHluorin and syb2-mOrange in hippocampal synapses after co-infection. (Right) The correlation between syb2-mOrange and syb2-pHluorin fluorescence in the presence of 50 mM NH₄Cl (n= 529 synapses, 7 experiments). (c) (Left) Co-localization of VAMP4-pHluorin and syb2-mOrange in hippocampal synapses after co-infection. (Right) The positive correlation between syb2-mOrange and VAMP4-pHluorin fluorescence in the presence of 50 mM NH₄Cl (n= 412 synapses, 6 experiments). (d) Average traces of syb2-mOrange and syb2-pHluorin fluorescence change at 282 individual boutons in response to 400 APs at 20 Hz followed by NH₄Cl treatment (n=5 experiments, p>0.05 difference between the groups is not significant). (e) Average traces of syb2-mOrange and VAMP4-pHluorin fluorescence change at 798 individual boutons in response to 400 APs at 20 Hz followed by NH₄Cl treatment (n=11 experiments, p<0.05). Inset: VAMP4-pHluorin individual traces. VAMP4-pHluorin trafficking in response to 400 APs at 20Hz at 5 different synapses. These traces demonstrate the spectrum of stimulation-induced VAMP4-pHluorin trafficking from exo-endocytic or positive to endo-exocytic or negative.

Figure 7. VAMP4 traffics independently of syb2

(a) VAMP4-pHluorin trafficking under stimulation (20 Hz for 20s) in wild type (green), syb2-deficient (black) and SNAP-25-deficient (gray) neurons. The bar graph shows the fraction of VAMP4 internalization in wild type (n=4), syb2-deficient (n= 6), and SNAP-25-deficient (n= 3) neurons. n denotes number of experiments. (b) VAMP4 L25A structure and its exo-endocytic trafficking under stimulation (20 s, 20 Hz) in syb2-deficient neurons. The bar graph shows the comparison of syb2-pHluorin and VAMP4 L25A-pHluorin exocytosis (as fluorescence increase) in wild type and syb2-deficient neurons (syb2: n= 12 wild type, n=7 syb2 deficient; VAMP4: n=3 wild type, n=7 for syb2 deficient). n denotes number of experiments. (c) Average traces of fluorescence change evoked by 400 APs applied at 20 Hz for boutons exhibiting VAMP4 trafficking before and after application of folimycin (n=4 experiments). (d) Dual color imaging experiments in folimycin indicate that VAMP4 can support low levels of exocytosis in the same boutons that show robust syb2-mediated fusion (n=40 boutons).

Figure 8. VAMP4 trafficking enables asynchronous release during intense activity

(a) Recovery of synaptic responses during 1 Hz stimulation after 600 action potentials applied at 20 Hz in control (n=15), VAMP4 overexpressing (n=12) and VAMP4 L25A (n=10) expressing neurons. These experiments followed the same setting as in Fig. 3. Shaded area denotes the region where synaptic responses between control and VAMP4 overexpressing neurons are significantly different (p<0.05). In contrast, the same comparison between control and VAMP4 L25A expressing neurons did not reveal statistical significance (p>0.05). (b) Representative traces depicting desynchronized synaptic responses triggered after intense stimulation in synapses overexpressing VAMP4. This desynchronization of release was attenuated after expression of VAMP4 L25A mutant. Bar graphs 200 ms vs. 50 pA. (c) Cumulative charge averaged from multiple experiments depicting the desynchronization of synaptic responses after intense activity in synapses expressing VAMP4. Dashed line indicates statistical significance betweenVAMP4 and control groups. In contrast, VAMP4 L25A expressing synapses did not show statistical difference with respect to controls.