Developmental time course of SNAP-25 isoform regulation of hippocampal long-term synaptic plasticity and hippocampus-dependent learning

SNAP-25 is essential in activity-dependent vesicle fusion and neurotransmitter release in the nervous system. During development and adulthood, SNAP-25 appears to have differential influences on long- and short-term synaptic plasticity in the hippocampus. The involvement of SNAP-25 in this process may be altered by two different splice variants expressed in adolescences versus adulthood in hippocampal neurons. This study suggests that the adolescent isoform, SNAP-25a can contribute to developmental regulation of the expression of LTD and LTP. In mice deficient in SNAP-25b, the adult isoform, Schaffer collateral-CA1 synapses showed slower release kinetics, reduced initial release probabilities, decreased LTP and enhanced LTD at 1 month. By 4 months of age, when mice have fully developed in the absence of SNAP-25b, the gene targeted mice appear to have compensated for the lack of the adult SNAP-25b isoform. Moreover, hippocampal-dependent training reversed reductions in LTP, but not LTD, seen at 1 month. In 4 month old adult mice, training prevented the compensatory reversal of LTD that had been observed prior to training. These findings support the hypothesis that immature SNAP-25a plays a strong role in the expression of plasticity at Schaffer collateral-CA1 synapses in adolescent mice, but compensatory mechanisms that reverse alterations in synaptic plasticity once mice reach adulthood.

Introduction displays its most significant increase coincides with a shift from expression of LTD to expression 120 of both LTD and LTP [7,13,14]. Since the switch between isoforms happens between postnatal 121 weeks 1-3, we compared hippocampal slices from one-month old and 4 month old mice, to 122 assess the early developmental and later adult impact of an absence of SNAP-25b on long-term 123 synaptic plasticity and learning. Improving our knowledge of presynaptic mechanisms of 124 plasticity, the molecules that mediate these changes and the outcomes they have on learning and 125 memory can help identify therapeutic targets for disorders associated with synaptic plasticity 126 dysfunction. 3   Technology, Rockingham, VT). After confirming the presence of Schaffer collateral-evoked 172 fEPSPs >1 mV in amplitude in CA1 stratum radiatum, and inducing LTP, 10 µM 6-cyano-7nitroquinoxaline-2,3-dione (CNQX) was bath-applied throughout the rest of the experiment to prevent synaptically-driven action potentials in CA3 pyramidal neurons from accelerating dye 175 release. Presynaptic boutons were loaded by bath-applying 5 µM FM1-43 (Molecular Probes) in 176 hypertonic ACSF supplemented with sucrose to 800 mOsm for 25 sec to selectively load the 177 rapidly-recycling pool (RRP) [17,18], then returned to normal ACSF. Stimulus-induced 178 destaining was measured after 30 min perfusion with dye-free ACSF, by bursts of 10 Hz bipolar 179 stimuli (150 µs DC pulses) for 2 sec applied once each 30 sec. We fitted a single exponential to 180 the first 6 fluorescence time course values, and decay time constants between groups compared 181 by two-tailed Student's t-test, as we have shown previously that the early release reflects 182 vesicular release from the RRP prior to recycling and reuse of vesicles [17,18].

198
The active place avoidance task used was described previously by Burghardt et al. 19 . In 199 this paradigm, mice are placed on a circular rotating platform that continuously turns clockwise 200 at a speed of 1 rpm. Over several days and multiple trials (Fig. 3a), mice (n = 12-17) learn to 201 identify the 60 shock zone guided by spatial markers on the walls surrounding the apparatus.

202
Entrance into the shock zone triggered a brief constant foot-shock (500ms, 60Hz, 0.2mA) with 203 an intershock interval of 1.5s that would cease upon leaving the shock zone. The middle point of 204 the animal was used as the reference point to determine the position of a rat and recorded using 205 behavioral software AnyMaze, Stoelting Co., Inc.). Using the same software, the number of 206 shock-zone entries was measured, where a decrease in shock-zone entries indicates learning.

207
During the initial pretraining trial (10 mins) when the shock was turned off, mice were allowed 208 to habituate to the apparatus and showed no preference for any area of the platform.

209
Subsequently, the shock was turned on and the mice completed 3 training sessions (10 mins 210 each) per day for 3 days, followed by an extinction trial (10 mins) on the next day when the 211 shock was turned off and the animals were allowed to ambulate freely into the zone previously 212 associated with the shock. After extinction, a conflict variant task was performed in order to test

220
Vesicular release probability is reduced in one month old SNAP-25b deficient 221 mice 222 SNAP-25a and SNAP-25b are functionally different in their ability to facilitate 223 exocytosis as they associate with the other core SNARE proteins in the SNARE complex [14].  The technique of loading presynaptic vesicles with a membrane impermeable dye permits 232 analysis of vesicle fusion dynamics in presynaptic terminals. FM1-43, preferentially loads into 233 presynaptic vesicles. Once the dye is loaded into synaptic vesicles, it can only be released when 234 vesicles fuse with the membrane and release their contents [17]. Two-photon analysis of FM1-43 235 makes it possible to directly image presynaptic rates of vesicular release [18]. To assess 236 neurotransmitter release probability directly, Schaffer collateral presynaptic terminal vesicles 237 were loaded with FM1-43 and the time course of fluorescent destaining in response to stimulus-238 evoked release of the dye was monitored using two-photon laser scanning microscopy [11,12].

239
Schaffer collateral axons were given a 2Hz stimulus train to evoke neurotransmitter release.

240
Schaffer collateral terminals in field CA1 of hippocampal slices from SNAP-25a mice showed significantly slower neurotransmitter release kinetics compared to wildtype controls (Fig. 1a), as 242 measured by a slower rate of fluorescence decay of FM1-43. in the developing brain, consistent with a previous study in which we showed that LTP at Schaffer collateral-CA1 synapses is reduced in both male and female SNAP-25b deficient mice 290 at 1 month of age [15]. In contrast, four-month old SNAP-25b deficient mice displayed sLTD that 291 did not differ from their littermate controls (Fig. 3b). These results also suggest compensatory 292 effects can restore synaptic plasticity to control levels and balance in adult mice lacking SNAP-  Place avoidance spatial learning 314 To evaluate the behavioral phenotypes associated with a lack of SNAP-25b throughout 315 the normal developmental period into adulthood, we used an active place avoidance assay 316 developed by Fenton and colleagues, in which rodents are placed on a turning metal grid 317 platform, and, using spatial cues, must learn to move to avoid a shock that is given when the 318 animal enters one quadrant of the circular field. Littermate controls and SNAP-25b deficient 319 mice were subject to multiday habituation, training trials, extinction and conflict discrimination 320 described in Burghardt et al. [19]. In contrast to deficits in this learning, we have observed 321 previously in one-month old mutant mice, adult mice lacking SNAP-25b showed no differences 322 in the initial learning phase in days 1-3 (Fig. 4b). However, after extinction (day 4), when the 323 shock zone was shifted 180° from its initial position, SNAP-25b deficient mice entered the new 324 shock zone fewer times than controls, suggesting a more rapid relearning of the new shock zone 325 location, a reflection of behavioral learning flexibility. No difference in anxiety-like behavior 326 was detected in four-month old SNAP-25b deficient mice (Fig. 4c), using an elevated plus maze.

327
Increased anxiety observed at one month in our previous study [15] was no longer detected in 328 older SNAP-25b deficient mice. Finally, motor function as assessed by total path length was not 329 different in SNAP-25b deficient mice and wildtype littermates (Fig. 4d). These data indicate that

369
Unlike adolescent mice lacking SNAP-25b, sLTD was altered in four-month old adult 370 SNAP-25b deficient mice compared to wildtype littermate controls only after passive avoidance 371 spatial learning task (Fig. 6b). Prior to training, four-month-old SNAP-25b deficient mice did not In mGluRII cLTD, lack of SNAP-25b did not alter the amplitude of mGluRII cLTD in 407 either one- (Fig. 7a) or four-month old (Fig. 7b)    NMDARs can be up regulated, resulting in enhanced LTP [29,30].

482
In contrast, levels of LTD in these older SNAP-25b deficient mice were similar to coordinate synchronous stimulus evoked release [28]. The absence of SNAP-25b may promote 496 enhanced functionality or increased recruitment of SNAP-25a or altered association of SNAP-that these residues alter protein-protein interactions [10]. Additionally, Munc 18-1, an accessory 500 SNARE protein which promotes SNARE complex formation, also interacts with SNAP-25 and 501 binds to SNAP-25b more easily than SNAP-25a [31]. To assess whether hippocampal dependent learning and memory is affected by lack of the 515 adult SNAP-25b isoform throughout development, hippocampal-dependent spatial learning was 516 evaluated in four month old SNAP-25b deficient mice using an active avoidance assay. Four 517 month old SNAP-25b deficient mice initially showed no differences in acquisition of learned 518 place avoidance, but they did show an enhanced conflict learning immediately after extinction, 519 i.e., a more rapid shift away from a previous shock zone to learn a new zone. On the second day 520 of conflict reversal learning, there was only a non-significant trend towards fewer shock zone 521 entries for the mice lacking SNAP-25b. The possibility exists that, since older SNAP-25b deficient mice did not experience the initial delay in learning we observed in younger mice, their 523 underlying memory deficit had largely disappeared as compensatory mechanisms were expressed 524 during development. Given that, at one month of age, this same mouse line showed enhanced 525 cognitive learning, we hypothesize that enhanced LTD in these adolescent animals accounted for 526 their performance during the relearning phase of the test, and both LTD and learning had 527 returned to normal in adulthood.

528
To address any underlying memory deficit not detected by the active avoidance assay, 529 additional behavioral learning and memory paradigms will be needed to test hippocampus- the initial learning phase [19].

536
Mice expressing only SNAP-25a exhibit a higher number of doublecortin positive 537 precursor cells [11], which implies a higher rate of neurogenesis in these mice. Increased 538 neurogenesis may lead to improved performance on cognitive flexibility tasks [19], and could  SNAP-25b implies a lower release probability following the induction of LTD. PPF is also an 580 indicator of presynaptic neurotransmitter release mechanisms. Larger PPF means that a smaller 581 number of vesicles released their content after the first pulse due to a lower release probability, 582 resulting in larger PPF in response to a second stimulus [17,32]. In a recent study [15], we found 583 that PPF in SNAP-25b deficient mice was significantly larger than in SNAP-25b expressing 584 wildtype mice, also suggesting that release probability is lower in mice expressing only SNAP-585 25a. In this earlier study [15], we also found that I/O ratios were unchanged, suggesting that 586 SNAP-25b deficient mice had homeostatically compensated for lower release probabilities by 587 upregulating postsynaptic sensitivity to glutamate.

589
The results described above regarding baseline synaptic transmission may put behavioral 590 findings associated with SNAP-25a into context. It is consistent with previous studies [20,28] showing that SNAP-25a containing SNARE complexes are slower to release FM1-43, indicating 592 lower release probability. Impaired transmitter release mechanisms may result in memories that 593 are not as persistent and stable as wildtypes, making them more susceptible to reversal in conflict 594 learning assays. Reduced transmitter release probability is suggested to impair the induction of 595 LTP, resulting in weaker synaptic strengthening and memory formation. Therefore, extinction in 596 adult SNAP-25b deficient mice may proceed faster than in control mice that express adult 597 SNAP-25b. It is noteworthy that the I/O relationship is similar to controls [15], suggesting that The absence of SNAP-25b did not alter the amplitude of mGluRII-dependent LTD 602 measured at Schaffer collateral-CA1 synapses when DCG-IV was bath applied, or NMDAR-603 dependent LTD when NMDA was bath applied, to hippocampal slices from these mice, 604 compared to littermate controls at four months of age. This suggests that SNAP-25a and Gβγ can 605 interact to induce cLTD in adult SNAP-25b-deficient mice with similar affinity to wildtype mice 606 during both adolescence and adulthood. These results are not consistent with recent 607 immunoprecipitation data showing that SNAP-25a-containing SNARE complexes associate with 608 Gβγ 50% less than SNAP-25b-containing complexes [25]. Since, even if Gβγ is less likely to 609 interact with SNAP-25a, a difference in chemically-induced NMDA or mGLuRII-dependent 610 LTD was not detected, it is possible that lower levels Gβγ and SNAP-25 interaction may still be 611 sufficient to permit expression of normal magnitude NMDAR and mGluRII cLTD. These results 612 also imply that the altered synaptic plasticity induced by stimulus in young and old mutant mice 613 (SNAP-25a) versus wild type (normal SNAP-25b) mice may have been in response to altered presynaptic activity, since we observed no difference in fEPSPs when mGluRII-LTD was 615 expressed in SNAP-25b deficient mice. Our previous work in rats has shown that presynaptic 616 infusion of Gβγ scavenging peptides, ct-SNAP-25 and structurally distinct mSIRK, can each 617 occlude mGluRII LTD [31]. Therefore, Gβγ binding to SNAP-25a in mice expressing only this 618 isoform may show larger sLTD compared to littermate controls, given evidence that SNAP-25a 619 favors LTD. An important next study will be to evaluate the vesicular release of FM1-43 from 620 SNAP-25a mutants after expressing each form of chemically-induced LTD.