Q-SNARE Syntaxin 7 confers actin-dependent rapidly replenishing synaptic vesicles upon high activity

Replenishment of readily releasable synaptic vesicles (SVs) with vesicles in the recycling pool is important for sustained transmitter release during repetitive stimulation. Kinetics of replenishment and available pool size define synaptic performance. However, whether all SVs in the recycling pool are recruited for release with equal probability is unknown. Here, using comprehensive optical imaging for various presynaptic endosomal SNARE proteins in cultured hippocampal neurons, we demonstrate that part of the recycling pool bearing the endosomal Q–SNARE Syntaxin 7 (Stx7) is preferentially mobilized for release during high–frequency repetitive stimulation. Recruitment of the SV pool marked with the Stx7–reporter requires high intra–terminal Ca2+ concentrations and actin polymerization. Furthermore, disruption of Stx7 function by overexpressing the N–terminal domain selectively abolished this pool. Thus, our data indicate that endosomal membrane fusion involving Stx7 is essential for adaptation of synapses to respond high-frequency repetitive stimulation.

7 cluster of SVs at presynaptic terminals. To examine detailed localization of Stx7 within 157 the terminals, we next performed immunoelectron microscopy. Since our initial trials to 158 detect endogenous Stx7 with the same antibody did not produce reliable signals, we 159 expressed Stx7-SEP as performed for SEP imaging, and proceeded to immunostaining 160 using anti-se antibody. Whereas SypHy signals with the same detection method were 161 evenly spread all over SV clusters in the terminals, Stx7-SEP signals were relatively 162 sparse and restricted to an area within SV clusters ( Fig 2B). Notably, in many instances, 163 Stx7-SEP signals were located distant from active zone structures, which is consistent 164 with a previous observation under STED microscopy (Wilhelm et al, 2014). 165 Stx7 content in the synaptosomal fraction isolated from rat brains was reported to 166 be 78.6 copies per synapse 7 . However, the synaptosomal fraction also contains 167 postsynaptic compartments in which Stx7 might be present (Appendix Fig S4). To 168 estimate a fractional contribution of presynaptic Stx7 in the synaptosomal fraction, we 169 quantified the Stx7 content in purified SVs where Stx7 was shown to be highly enriched 170 by western blot analysis (Takamori et al, 2006). We estimated the copy number of Stx7 171 molecules per SV as 0.14 ± 0.03 (Figs 2C, D and Appendix Fig S5). Assuming that an 172 average synaptosome contains ~380 SVs, total Stx7 molecules located on SVs is 173

Stx7-SEP is sorted to SV subpopulation that requires HFS for exocytosis 178
One of the most striking features revealed by our comprehensive SEP imaging analysis 179 (Figs 1E,F) was that Stx7-SEP responded preferentially to high-frequency stimulation 180 (HFS) at 40 Hz, but rarely responded to 10-Hz stimulation. To confirm this 181 phenomenon and to avoid any bias possibly due to heterogeneity intrinsic to each 182 bouton or culture preparation, we continuously monitored changes in SEP fluorescence 183 at individual boutons with various stimulation frequencies ranging from 5 Hz to 40 Hz 184 at 5 min intervals (Fig 3A). Unlike SypHy, which reliably exhibited a robust exocytotic 185 fluorescence increase irrespective of stimulation frequency (Fig 3A, top), Stx7-SEP 186 hardly responded during 5-Hz or 10-Hz stimulation, while it showed robust 187 fluorescence increases at 20 Hz and 40 Hz in the same boutons (Fig 3A, bottom). 188 8 We then wondered if overexpression of Stx7-SEP simply attenuated SV exocytosis 189 generally, for instance by inactivating Ca 2+ channels, or if Stx7-SEP localized at 190 non-synaptic areas where non-SV type secretory vesicles that responded only to HFS 191 were present. To exclude these possibilities, we co-expressed Stx7-SEP and 192 in which a pH-sensitive orange fluorescent protein, mOrange2, was fused to the luminal 193 region of synaptophysin, instead of SEP (Egashira et al, 2015(Egashira et al, , 2016, and monitored 194 their fluorescence changes simultaneously in the same presynaptic boutons (Appendix 195 Fig S6). When analyses were restricted in bouton-like structures where clear 196 fluorescence increases of Syp-mOr were detected, Stx7-SEP rarely responded to 10-Hz 197 stimulation (Appendix Fig S6), but exhibited drastic increase in response to 40-Hz 198 stimulation (Appendix Fig S6). Hz and 10 Hz (Fig 3C). Comparisons between SypHy and Stx7-SEP by replotting as a 212 function of stimulus numbers clearly showed that rise kinetics of Stx7-SEP were 213 identical to those of SypHy during 20-Hz and 40-Hz stimulation, whereas those of 214 Stx7-SEP were much slower than SypHy at 5 Hz and 10 Hz (Fig 3D). The fact that Stx7-SEP responded preferentially to high repetitive stimulation (> 20 Hz) 219 suggested that exocytosis of Stx7-SEP-bearing SVs would require a high concentration 220 of Ca 2+ . To test this hypothesis, we raised external Ca 2+ from 2 to 8 mM, and examined 221 9 whether even 10-Hz stimulation would cause robust responses. The response of SypHy 222 was facilitated in the presence of 8 mM (Fig 4A, left), as observed previously 223 (Chanaday & Kavalali, 2018). Notably, Stx7-SEP also exhibited a robust exocytic 224 response to 10-Hz stimulation when external Ca 2+ was raised to 8 mM ( Fig 4A). Ca 2+ /calmodulin-dependent fast SV replenishment after RRP depletion is also mediated 229 by an actin-dependent process in calyx of Held synapses (Sakaba & Neher, 2001, 2003. 230 We therefore asked whether actin dynamics are involved in recruitment of Stx7-SEP-231 bearing vesicles for release during HFS. In accordance with a previous report (Hua et al, 232 2011), actin depolymerization with latrunculin A (Lat-A: 5 µM) did not show 233 remarkable effects on SypHy recycling during 10-Hz stimulation ( Fig 4B). However, 234 Lat-A significantly retarded the SypHy response upon 40-Hz stimulation (Fig 4B,  235 upper right), indicating that there exists a subset of SVs for which mobilization for 236 release is actin-dependent during intense stimulation. This result is consistent with 237 previous results reported using neurons in which actin isoforms were conditionally 238 deleted (Wu et al, 2016). Notably, more pronounced reduction by Lat-A was observed 239 for Stx7-SEP responses upon 40-Hz stimulation (Fig 4B), indicating that Stx7-SEP 240 localizes to a subpopulation of recycling SVs for which recruitment for release is actin-241 dependent. Lat-B, another actin polymerization inhibitor, also decreased the SypHy and 242 Stx7-SEP responses to similar extent at 40 Hz, but not at 10 Hz (Appendix Fig S7). 243 Finally, when Stx7-SEP and SypHy were monitored during 10-Hz stimulation in the 244 presence of 8 mM Ca 2+ , Lat-A significantly reduced the responses of both (Fig 4C). 245 To further determine whether Lat-A treatment reduced the size of the total recycling 246 pool or whether it simply slowed SV recruitment for release, we extended stimulation 247 lengths in the presence of Baf to measure the size of the total recycling pool of SypHy-248 laden vesicles ( Fig 4D). As expected, kinetics measured as  exo and the total recycling 249 pool size measured as the plateau at the end of stimulation, of SypHy responses during Taken together, these results indicate that a portion of the SV recycling pool is 256 recruited through a pathway requiring Ca 2+ -dependent actin polymerization, and that 257 Stx7-SEP preferentially localizes to that pool.  To understand the molecular mechanism by which Stx7 is selectively sorted to a 263 subpopulation of the SV recycling pool, we constructed two deletion mutants either 264 lacking the NTD (Stx7-NTD-SEP) or the SNARE motif (Stx7-SNARE-SEP) (Fig  265   5A). When these mutants were expressed in cultured neurons, Stx7-NTD-SEP was 266 properly sorted to Syb2-positive bouton-like puncta ( Fig 5B) in which it preferentially 267 localized to the intracellular acidic compartments, with albeit higher luminal pH 268 compared to genuine SVs (Figs 5C,D and Appendix Fig S8). However, unlike Stx7-269 SEP, Stx7-NTD-SEP responded to 10-Hz stimulation as it did to 40-Hz stimulation 270 ( Fig 5E) and TeNT treatment did not completely abolish the responses (Fig 5F). 271 Furthermore, responses of Stx7-NTD-SEP at 40 Hz were no longer sensitive to Lat-A 272 treatment ( Fig 5G). Thus, the NTD is essential for proper sorting of Stx7 to the actin-273 dependent SV subpool that is preferentially recruited during HFS. By contrast, Stx7- Although results described above demonstrate that Stx7-SEP is directed to a 281 subpopulation of the SV recycling pool for which recruitment for release requires high 282 Ca 2+ and actin polymerization, whether Stx7 is necessary for this subpopulation of SVs 283 remains unclear, especially given that only a small population of SVs carries Stx7 (Fig  284   2). To address this question, we first examined whether silencing of Stx7 with a specific 285 shRNA affects the SV recycling sub-pool. However, chronic knockdown of Stx7 286 expression severely reduced SypHy responses at 10 Hz and 40 Hz, indicating that Stx7 287 is necessary to establish SV pool per se during synapse development and maturation 288 processes (Appendix Fig S9). As an alternative approach to inactivate Stx7, we 289 explored specific dominant-negative effects by overexpressing the N-terminal domain 290 of Stx7. The rationale for this is that the NTD of Stx7 interacts with its own SNARE 291 motif, thereby inhibiting SNARE complex formation with cognate SNAREs (Antonin et 292 al, 2002). In addition, the results above (Fig 5) clearly demonstrate that the NTD of 293 Stx7 is essential for its proper sorting to the actin-dependent SV recycling pool. To this 294 end, we placed a P2A self-cleaving peptide between the SypHy and Stx7-NTD  Treatment with Lat-A in addition to Stx7-NTD overexpression did not further slow 302 release kinetics (Fig 6D), indicating direct involvement of Stx7 in the actin-dependent 303 SV recycling pool that rapidly and preferentially replenishes RRP upon HFS. 304 305 306

Animals 610
Pregnant ICR mice were purchased from SLC, Japan. All mice were given food and 611 water ad libitum. Animals were kept in a local animal facility with a 12-h light and 12-612 h dark cycle. Ambient temperature was maintained around 21˚C with a relative 613 humidity of 50%. All animal experiments were approved by the Institutional Animal 614 Care and Use Committee of Doshisha University. 615 616

Statistics 617
All data are shown as the mean ± standard error of the mean (SEM). Unpaired t-tests 618 were applied to compare means of two experimental groups. All statistical tests were 619 two-tailed, and the level of statistical significance is indicated by asterisks: *p < 0.05, 620 **p < 0.01, ***p < 0.001. n.s.; not significant. 621 622

Data availability 623
All relevant data that support the findings of this study are available from the 624 corresponding authors upon request. The detailed data for Fig. 1E-H, 3D, E, 4