Presynaptic c-Jun N-terminal Kinase 2 regulates NMDA receptor-dependent glutamate release

Activation of c-Jun N-terminal kinase (JNK) signaling pathway is a critical step for neuronal death occurring in several neurological conditions. JNKs can be activated via receptor tyrosine kinases, cytokine receptors, G-protein coupled receptors and ligand-gated ion channels, including the NMDA glutamate receptors. While JNK has been generally associated with postsynaptic NMDA receptors, its presynaptic role remains largely unexplored. Here, by means of biochemical, morphological and functional approaches, we demonstrate that JNK and its scaffold protein JIP1 are also expressed at the presynaptic level and that the NMDA-evoked glutamate release is controlled by presynaptic JNK-JIP1 interaction. Moreover, using knockout mice for single JNK isoforms, we proved that JNK2 is the essential isoform in mediating this presynaptic event. Overall the present findings unveil a novel JNK2 localization and function, which is likely to play a role in different physiological and pathological conditions.


Results
c-Jun N-terminal kinase (JNK) regulates presynaptic glutamate release. Glutamate release experiments, monitored as release of preloaded [3H]D-aspartate ([3H]D-ASP), were performed to investigate the role of JNK kinase in the presynaptic compartment. The radiactive tracer allows to measure glutamate exocytosis from isolated nerve terminals avoiding the problems related to the enzymatic degradation of glutamate 24,25 . From here on, throughout the manuscript, the release of preloaded [3H]D-ASP will be referred to as glutamate release. Mouse cortical terminals were tested for glutamate release using different stimuli up-down superfused on a thin layer of synaptosomes, according to a standard procedure 26 . Transient (90 s) exposure of cortical nerve terminals to a mild depolarizing stimulus (8 mM KCl) induced a significant exocytotic-like [ 3 H]D-Asp overflow (158 6 37% vs. control. **p , 0.01), which is comparable to that obtained by 10 min exposure to 100 mM NMDA (202 6 21% vs. control. **p , 0.01). NMDA stimulus was applied in Mg 21 -free medium to ensure the operability of the presynaptic NMDA receptor 24 . Similarly, a milder 10 min stimulus of 50 mM (s)-AMPA was able to give a significant [ 3 H]D-Asp overflow (98 6 11% vs. control. **p , 0.01) ( fig. 1A). Both stimuli are known to trigger Ca 21 -dependent neurotransmitter release 27 .
Biochemical analysis of c-Jun N-terminal Kinase (JNK) activation upon different stimuli in cortical mouse synaptosomes. Among the many intracellular partners activated by the above used stimuli, the involvement of the JNK protein has been investigated in cortical mouse synaptosomes. Synaptic JNK activation, measured as phosphorylated protein induced by different stimuli, was measured by biochemical experiments (fig. 1B-C-D). Only 10 min exposure to NMDA (100 mM) 1 glycine (1 mM) in Mg 21 -free medium stimulus induced a significant increase in JNK phosphorylation (163 6 18% vs. control 5 100%. **p , 0.01). KCl and AMPA left unchanged the level of pJNK respect to the control. The NMDA-evoked release experiments were performed in Mg 21 -free medium solution to maintain the NMDA receptors active without the Mg 21 'block' 26 .
Notably the NMDA blocker D-AP5 (50 mM) was able to reverse the NMDA-mediated increase in JNK phosphorylation (Fig. 1E). An additional study was carried out to exclude the possibility that the basal release of glutamate reaches a sufficient concentration able to induce NMDA activation in absence of any exogenous stimulus and leading to JNK phosphorylation. The NMDAR antagonist D-AP5 was applied to synaptosomes for 10 min and no changes in phosphorylation of JNK (110.0 6 8.9% vs Ctrl 100%) were detected ( fig. 1F).
It was demonstrated that postsynaptic NMDAR induces JNK activation through the interaction of JNK scaffolding protein JIP1. In line with this, it was observed that the NMDAR activity in JIP1 KO mice is reduced 28 . Here in order to assess whether also presynaptic NMDAR-signaling pathway might involve the signaling formed by NMDA/JIP1/JNK proteins, we tested for the presence of JNK and JIP1 at the presynaptic level.
For this, we used an ultrastructural fractionation of synaptosomes 9 and purity of the fractions were analyzed before use. PSD-95 immunoreactivity was observed only at the post-synaptic fraction (Post) while Syntaxin 1a (STX1a) resulted enriched in the presynaptic fraction (Pre) and absent in the Post. The non-synaptic soluble proteins (NSSP) fraction was only contaminated by STX1a which it could be attributed to docking vesicles at the presynaptic membrane ( fig. 1G). As expected JNK and its scaffolding protein JIP1 were present at both NSSP and Post fractions. Most importantly they were also clearly present at the presynaptic site ( fig. 1G).
To demonstrate that, in synaptosomes, upon NMDA stimulus the activation of JNK is mainly presynaptic, cortical synaptosomes were stimulated and later subjected to ultrasynaptic fractionation. We found that JNK phosphorylation was strongly augmented at the presynaptic site ( fig. 1H).
To confirm that JNK activation triggered by NMDA could pass through JNK-JIP1 interaction, we used a specific JNK inhibitor (L-JNKi1) which is able to disrupt the interaction of JNK with its JNKbinding domain (JBD) targets 29 . It is a cell permeable peptide (CPP) composed the JBD amino acid sequence linked to 10-amino acid carrier peptide derived from the HIV-Tat 48-57 sequence 29 . Cortical synaptosomes were treated for 30 min with 2 mM of L-JNKi1 after which the same stimuli above described were applied. In accordance with already published data the time frame and concentration used were enough to let the inhibitor diffuse and inhibit JNK 29,30 . In line with the evidences that L-JNKi1 does not affect JNK phosphorylation 31 , pre-incubation of synaptosomes with L-JNKi1 resulted in increased JNK activation when NMDA stimulus was applied (149 6 10% vs control 5 100%. *p , 0.05). In contrast, basal JNK phosphorylation was unaltered when L-JNKi1 was applied alone ( fig. 1B). In the case of KCl and AMPA the application of L-JNKi1 left JNK phosphorylation unaffected ( fig. 1C-D). Altogether these results indicate that NMDA stimulus specifically modulates JNK activity.
L-JNKi1 does not induce changes in vesicle machinery proteins and in NMDA or AMPA receptor subunit levels. We next monitored the expression levels of some presynaptic proteins such as syntaxin 1a (STX1a), synapsin (SYN), MUNC18-1 and synaptotagmin 1 (SYT1) since their alteration compromises the neurotransmitter release efficiency 32,33 . In our experimental conditions NMDA and L-JNKi1, when applied alone or in combination, did not modify the expression of STX1a, SYN, MUNC18-1 and SYT1 in synaptosomes ( Supplementary Fig. 1A). Similarly, NMDA and AMPA receptor subunits expression ( Supplementary Fig. 1B) were unchanged, thus excluding the possibility that NMDA-mediated effect on presynaptic JNK might depend on specific modifications of receptors subunits.
The capability of JNK inhibition to reduce the effect of presynaptic NMDA was also investigated by measuring the vesicle exocytosis. The fluorescent dye (Synaptogreen), which is incorporated into presynaptic vesicles, is released into the extracellular space losing its fluorescence. The fluorescence decrease reflects the exocytosis extent. The NMDA-evoked exocytosis was monitored by spectrofluorophotometer measurement. Synaptosomes, pre-loaded with the fluorescent dye, were exposed to NMDA and glycine stimulus, as control, or with a pretreatment for 30 min with L-JNKi1 or D-AP5. The decrease in fluorescence intensity due to the released dye was registered for 6 min in the two preparations ( fig. 2C). As expected, NMDA (100 mM) 1 glycine (1 mM) stimulus induced a reduction in fluorescence (258.7 6 0.9%) which it was reduced by L-JNKi1 application (228.1 6 1.1% vs. NMDA 1 glycine. t-test **p , 0.01) and more stronger with D-AP5 that was used as control since is an NMDA channel blocker (29.    JNK mediates presynaptic NMDA-dependent glutamate release. Glutamate spontaneously released in the synaptic cleft can in turn activate the NMDARs located at the presynaptic terminals thus allowing a further tonic facilitation of glutamate release 34 . Here we performed electrophysiological recordings to investigate the functional role of JNK in glutamate release. We studied the effect of L-JNKi1 on NMDA-dependent presynaptic glutamate release measuring miniature excitatory postsynaptic currents (mEPSCs), according to a protocol previously described 34 . The application of D-AP5 in the presence of extracellular tetrodotoxin (1 mM) and picrotoxin (100 mM), and with MK-801 (10 mM) in the patchpipette (in order to block NMDA channels in the cell we were recording from), induced a reversible increase of the cumulative inter-event interval (iei) distribution (***p , 0.001, K-S test, C57BL6/J vs. D-AP5, n 5 11; fig. 4A) and of their median values (**p , 0.01, Student's t-test, ctrl vs. D-AP5, n 5 11; fig. 4B), whilst the cumulative and median amplitude of the same events were unaffected (p . 0.05, K-S test, C57BL6/J vs. D-AP5, n 5 11; fig. 4A; p . 0.05 Student t-test, ctrl vs. D-AP5, n 5 11; fig. 4B). The reduction in frequency by D-AP5 is due to the blockade of presynaptic NMDA receptors which facilitate tonic glutamate release. In slices pre-incubated with L-JNKi1 the effect of D-AP5 was lost. In fact we found that perfusion with D-AP5 did not lead to significant changes of inter-event interval (p .  inhibition reduces glutamate release mediated by presynaptic NMDA receptors, suggesting that JNK plays a functional role in NMDA-dependent glutamate release in acute brain slices. NMDA does not facilitate glutamate release and JNK phosphorylation in JNK2 knock-out mice. To identify which JNK isoform is activated by NMDA stimulus in presynaptic compartment, glutamate release experiments were conducted on cortical synaptosomes obtained from JNK knock-out (KO) mice. Glutamate release, measured as [ 3 H]D-Asp efflux, was evoked by NMDA for 10 min. Interestingly, NMDA evoked release was significantly lower (28 6 19% over basal overflow; n 5 4 Newman-Keuls, **p , 0.01) only in JNK2 KO mice compared to JNK1 and JNK3 KO and wild-type (WT) mice ( fig. 5A). This result suggests that presynaptic JNK2 specifically controls NMDAevoked release.
In order to corroborate the notions indicating a specific involvement of JNK2 in the presynaptic NMDA-evoked glutamate release, we investigated whether the absence of JNK2 reduced JNK phosphorylation. The JNK KO mice used present a deletion of a specific JNK gene, therefore the other remaining JNK proteins are still detectable. To appreciate the different protein expression, JNK immuno-  reactivity in JNK KO mice has then been analyzed. As already reported 35 JNK proteins are present as double isoforms (54 kDa and 46 kDa bands) and at around 46 kDa JNK2 shows a faint band up to the JNK1 one. In JNK1 KO mice the JNK1 46 kDa band disappeared, while the JNK2 46 kDa band is lacking in JNK2 KO. In both JNK2 and JNK3 KO mice the 56 kDa bands was found to be reduced ( fig. S1C).
Synaptosomes prepared from cortex of JNK-KO mouse were subjected to NMDA stimulus and, as expected, the phosphorylation of JNK induced by NMDA application was totally abolished in JNK2 KO mice as compared with control. In contrast, it was still maintained in JNK1 KO and in JNK3 KO mice where NMDA induced an increase of pJNK immunoreactivity (165 6 13% over control 5 100%; n 5 5 t-test **p , 0.01) (176 6 10% over control 5 100%; n 5 4 t-test **p , 0.01) ( fig. 5B) which is totally comparable with the one obtained in WT mice ( fig. 1B). Interestingly NMDA stimulus increased both bands (46 and 54 kDa) of JNK phosphorylation in JNK1 and JNK3 KO mice. This might be due to the fact that JNK2, which is present in both JNK1 and JNK3 KO animals, has two splicing variants at 46 and 54 kDa 36 . In all cases, application of L-JNKi1 alone did not affect JNK phosphorylation, suggesting that JNK functionality in both KO mice is not perturbed in the presence of JNK inhibitor, as also reported for wild-type mice ( fig. 5A). The presence of JNK2 at the presynaptic site was then detected by WB on an ultrasynaptic fractionation of mice cortical synaptosomes ( fig. 5C).
Ultrastructural characterization of synaptosomes and localization of JNK2 protein. The localization of JNK2 at the presynaptic site was also confirmed by electron microscopy technique. We evaluated the ultrastructure of synaptosomes by analyzing corticals from mice. The typical morphology of synaptosomes was clearly observed by electron microscopy ( fig. 6A) and show mainly vesicles of 0.2-0.8 mm diameter that contain the complete presynaptic terminal (further distinguished by the conspicuous presence of neurotransmitter-filled vesicles, mitochondria, smooth endoplasmic reticulum, microtubules and neurofilaments); a synaptic cleft (about 20-30 nm); a postsynaptic component is distinguishable by presence of amorphous electrondense material containing black granules along the corresponding plasma membranes so-called 'postsynaptic density' (PSD) 9 .
Immunocytochemical analysis on synaptosomes shows the presence of JNK 2 protein both in pre-and post-synaptic structure ( fig. 6B).
JNK2 colocalizes with presynaptic markers in mice cortical synaptosomes. Confocal microscopy experiments were performed to assess the colocalization of JNK2 with two well-known markers of the presynaptic machinery. Synaptotagmin-1 (SYT 1) is localized in synaptic vesicles and it triggers calcium-induced exocytosis 37 . Synaptosomal-associated protein 25 (SNAP-25) is a membrane protein with a cytosolic domain that contributes in the formation of the exocytotic fusion complex (SNARE complex), where it assembles with syntaxin-1 and synaptobrevin 38  showed a high proportion of JNK2 immunoreactive synaptic puncta, indicating a widespread distribution of JNK2 while a complex colocalization pattern was characterized by different degrees of alternating single and double labeling in synaptosomal puncta. For quantitative analysis, to avoid interference from fluorescent dots aspecifically bound on the glass, image analysis was selectively performed by analyzing only the synaptosomal strucures identified through the transmitted light channel. Protein colocalization, including the evaluation of the colocalized threshold area, were analyzed by using two colocalization coefficients (Pearson's and Manders), and visual inspection of the spatial relations (Plot Profile) between the two immunoreactive fluorescence signal intensities. JNK2 colocalizes with both SYT 1 (13.00 6 1.35%) and SNAP-25 (15.00 6 2.21%) with respect to the total synaptosomal area ( fig. 7G). The percentage difference (17%) was not significant suggesting that JNK2 does not preferentially colocalize with SYT 1 or SNAP-25. The Pearson's coefficient was 0.62 6 0.07 and 0.69 6 0.08 respectively, indicating that in all the synaptosomal structures JNK2 immunoreactivity colocalizes with both synaptosomal proteins (colocalization is considered positive for values ranging 0.5-1) ( fig. 7H). Also in this case, the percentage difference was not significative (about 10%) suggesting that JNK2 does not correlate preferentially with SYT 1 or SNAP-25. Although Manders's coefficients are less efficient compared to Pearson's coefficient, for colocalization analysis, they can be used to evaluate the reciprocal association ratio between fluorescence markers. All Manders's coefficients were above 0.5 (association is considered positive for values ranging 0.5-1), with the exception of the JNK2 presence in the SYT 1 signal which was lower (0.36 6 2.21) ( fig. 7P-R). Interestingly, the percentual difference for both Manders coefficients were significatively higher for JNK2/SNAP-25 immunofluorescence respect to JNK2/SYT 1, (34% and 12%, respectively, t-test ***p , 0.001) suggesting a preferential interaction of JNK2 with SNAP-25. Plot profiles traces showed a rather heterogeneous fluorescence distribution for both JNK2/SYT 1 and JNK2/SNAP-25 double immunofluorescence ( fig. 7I-S). Traces displaying associated spatial increases and decreases of fluorescence intensity (indicating colocalization; fig. 7 yellow arrows pointing up) were alternated to traces characterized by a disjoint trend (indicating lack of colocalization; fig. 7 yellow arrows pointing down). Associated and disjoint traces homogenously could be followed for about 0.5 micron which can be a reasonable estimation for a single presynaptic size (0.5-1.5 micron). This suggests that presynaptic structures display only colocalization or single staining for the analyzed markers.
JNK2 isoform modulates presynaptic NMDA-dependent glutamate release. Electrophysiological recordings were also performed in transgenic mice in order to confirm which JNK isoform was involved in NMDA-dependent glutamate release. For this, we  fig. 8B). Notably, the decrease of glutamate release due to D-AP5 was specifically prevented in JNK2 KO mice. Indeed, perfusion with D-AP5 did not produce significant changes neither in mEPSC iei (p . 0.05, K-S test,     fig. 8B). Overall these electrophysiological data suggest that only the JNK2 isoform mediates presynaptic release of glutamate under our experimental conditions.

Discussion
Here we report that JNK2 protein is implicated in the control of the presynaptic NMDA-evoked glutamate release in cortex. JNK has been widely demonstrated to be downstream target of postsynaptic NMDA receptors 39 . Importantly, the NMDAR activity is linked to the presence of JNK interacting protein 1 (JIP1) protein 28 . Therefore this intracellular signaling pathway involves the combined interaction of three partners: NMDAR, JIP1 and JNK. Although there is general consensus that NMDA receptors are located only at the postsynaptic site, few are the evidences that NMDAR are also present at the presynaptic site where they control neurotransmitter release 26,40 . However it has never been directly demonstrated, at the presynaptic site, that the NMDAR activity is mediated by the interaction with JIP1 and JNK. To date, only few works have proposed a presynaptic role for JNK. In PC12 cells it has been shown that JNK, activated by nerve growth factor (NGF), controls Ca 21 -evoked release due to direct phosphorylation of synaptotagmin-4 (SYN4) 41 . However, this protein is known to be located either pre or postsynaptically. Recent findings suggested that JNK can be stimulated by NGF via presynaptic p75 NTR receptor 16 . Furthermore, JNK3 has been shown to be involved in the molecular organization of the presynapse 42 . Using presynaptic release assay, together with biochemical, imaging and electrophysiological experiments, here we identify for the first time that JNK is also present at presynaptic sites. Specifically we defined that JNK plays an essential and specific role in controlling glutamate release mediated by presynaptic NMDAR in cortical nerve endings. Release experiments were performed on synaptosomes in a superfusion system 13 where each event affecting glutamate release is attributable only to the functional presynaptic compartment. To inhibit JNK protein activity we have used a commercial cell-permeable peptide (L-JNKi1) containing an effector part, JNK binding domain (JBD), linked to a cargo sequence, i.e. the 10 amino acid long arginine rich domain of the HIV-1 Tat protein (48)(49)(50)(51)(52)(53)(54)(55)(56)(57). The effect of L-JNKi1 at the presynaptic compartment could be justified since we have demonstrated, for the first time, the presynaptic expression of both JIP1 and JNK by using a synaptic fractionation protocol.
Our biochemical experiments on isolated cortical presynaptic terminals show that, among the applied stimuli, only NMDA induces JNK phosphorylation. This effect was slightly but not significantly reduced by L-JNKi1 since, as already published for D-JNKi1 43,44 , L-JNKi1 does not block JNK activation but its downstream effect. In addition, using synaptosomes fractionation analysis, we showed that NMDA stimulus increases JNK phosphorylation at the presynaptic fraction, confirming the presynaptic activation of JNK.
An additional experiment to evaluate the exocytosis in cortical synaptosomes has been done using time-lapse imaging. Among the stimuli used only NMDA-evoked Synaptogreen exocytosis was reduced by L-JNKi1 application which corroborate our previous results. In addition, also our electrophysiological findings conducted in mouse cortical slices revealed a functional role of JNK in mediating the NMDA-induced glutamate release.
In support of our data a more stable form of L-JNKi1, D-JNKi1, reduced excitotoxicity and protected against cerebral ischemia 44,50 , and albeit only postsynaptic activity was studied, a presynaptic effect cannot be ruled out.
In order to identify which JNK isoform is involved in neurotransmitter release during NMDA stimulation we also employed JNK transgenic knock-out mice 21 . JNK3 deletion leads to protection from excitotoxicity 51 and hypoxia 52 , while JNK1 and JNK2 seem to be important for apoptosis during brain development 19 . Using knockout mice for single JNK isoforms, we report that JNK2 is the critical player involved in this presynaptic phenomenon. Notably JNK2 KO mice display a dramatic reduction of NMDA-evoked release, being associated with a decreased NMDA-elicited JNK activation, as demonstrated by western blot experiments.
Electron microscopy experiments were performed on cortical synaptosomes and, as previously reported 9 , the synaptosomal morphology was well preserved. Whereas JNK1 53 and JNK3 42 have been associated with a synaptic presence and function, on the other hand JNK2 was only reported to be localized between the cytoplasm and the nucleus 54 . Here, by immunogold staining, we have clearly identified the presence of JNK2 at both post-and pre-synaptic site. These results were confirmed and extended in confocal microscopy where a quantitative approach allowed to get more information of JNK 2 expression pattern. Colocalization experiments indicate that JNK2 is expressed in synaptic puncta in high proportion. This observation supports our functional data demonstrating that JNK2 can interfere with the SNARE complex machinery thereby controlling glutamate release. Finally, also our electrophysiological recordings highlight that JNK2 KO mice display reduced miniature EPSCs attributable to activation of presynaptic NMDA receptors.
Altogether our data led us to conclude that JNK is also localized at presynaptic site being able to control glutamate release evoked by NMDA stimulus.
We here describe that specifically the isoform 2 of JNK cooperates with NMDA receptors in regulating glutamate release. In the experiment in which the synaptic exocytosis was measured, the inhibition of JNK activity reduces neurotransmitter release when NMDA is applied. This leads to speculate that the mechanism by which NMDA controls glutamate overflow might pass through an interaction between JNK2, presynaptic JIP1 and other proteins of the release machinery. This aspect needs further investigation in future work.
In light of the mounting evidence indicating the involvement of JNK signaling in different diseases 55 , the present findings highlight a presynaptic localization and function of JNK2. This protein will represent a novel and selective target to modulate release-regulating NMDA autoreceptors, whose activation reinforces glutamate release from the presynaptic component of asymmetric synapses then controlling excitatory transmission in CNS.

com/scientificreports
Health and approved by the Ethical Committee on animal experiments of EBRI ''Rita Levi-Montalcini'' Foundation (Rome, Italy). Brains were rapidly dissected out on ice in order to collect cortical tissues. All the procedures necessary to prepare the brain fractions were done at 0-4uC. Synaptosome preparation. Purified synaptosomes were prepared as previously described 11,13,57 . Briefly, the tissue was homogenized in 10 volumes of 0.32 M sucrose, buffered to pH 7.4 with Tris-(hydroxymethyl)-amino methane [Tris, final concentration (f.c.) 0.01 M] using a glass/Teflon tissue grinder (clearance 0.25 mm); the homogenate was centrifuged at 1,000 3 g for 5 min to remove nuclei and debris and the supernatant was gently stratified on a discontinuous Percoll (Sigma Aldrich, Italy) gradient (2%, 6%, 10% and 20% v/v in Tris-buffered sucrose) and centrifuged at 33,500 3 g for 5 min. The layer between 10% and 20% Percoll (synaptosomal fraction) was collected and washed by centrifugation. The synaptosomal pellets were resuspended in a physiological solution (Standard Medium) with the following composition (mM): NaCl, 140; KCl, 3; MgSO 4 , 1.2; CaCl 2 , 1.2; NaH 2 PO 4 , 1. Tat 31-51 (2 nM) were applied at the end of the first fraction collected (b1) and maintained in the superfusion medium till the end of the superfusion period. In NMDA experiments the medium was replaced, at t 5 20 min, with a Mg 21 -free ions solution to permit NMDAR activation. In KCl experiments synaptosomes were transiently (90 s) exposed, at t 5 39 min, to high K 1 containing medium (8 mM). In these experiments, fractions were collected according to the following scheme: two 3 min fractions (basal release), one before (t 5 36-39 min) and one after (t 5 45-48 min) a 6 min fraction (t 5 39-45 min; evoked release). Fractions collected and superfused synaptosomes were counted for radioactivity. The amount of radioactivity released into each superfusate fraction was expressed as a percentage of the total synaptosomal tritium content at the start of the fraction collected (fractional efflux). Agonist and high K 1 -induced effects were estimated by subtracting the neurotransmitter content into the fractions corresponding to the basal release from those corresponding to the evoked release.
Measurements of synaptic vesicle exocytosis. Spectrophotometer measurements. Synaptogreen (SIGMA-ALDRICH, Italy), also known as FM1-43, was used to measure synaptic vesicle fusion with the plasma membrane as previously described 45,46 . In brief, synaptosomes (0.3 mg/mL) were incubated in 2 ml of Standard Medium for 2 min at 37uC in mild agitation. Synaptogreen (50 mM) was added and after 1 min stimulation with NMDA (100 mM) and glycine (1 mM) was applied to load the fluorescent dye. After 3 min, synaptosomes were washed twice to remove non-internalized Synaptogreen in Standard Medium containing 1 mg/mL BSA. Synaptosomes (200 mg of proteins) were then resuspended in 2 mL of Mg 21free Standard Medium, and incubated in a stirred and thermostated cuvette maintained at 37uC in a RF-5301PC spectrofluorophotometer (Shimadzu, Japan). Release of accumulated Synaptogreen was induced by the addition of NMDA (100 mM) and glycine (1 mM) after 10 min of re-equilibration, and measured as the decrease in fluorescence upon release of the dye into solution (excitation 510 nm, emission 625 nm) for 6 min after stimulus. Data points were obtained at 2 s intervals, and data presented as the Ca 21 -dependent decrease in FM1-43 fluorescence. The synaptosomes were pre-incubated, when indicated, with L-JNKi1(Sigma-Aldrich, USA) for 30 min before depolarization with NMDA and glycine. The average of the fluorescence of the last 20 registered points are compared to the one at the stimulus point. The value is expressed as percentage of decreased fluorescence after stimulus for each trace.
Microscopy measurements. Synaptosomes were seeded on coverslips in Standard Medium for staining procedures. FM1-43 loading into synaptic vesicles was performed as described above for spectrophotometer measurements and placed on a time-lapse system mounted on an inverted fluorescence microscope (TiE; Nikon, Japan), equipped with an incubation chamber (Okolab), a cooled CCD camera (Clara; Andor), Perfect Focus System to avoid z-axis focus fluctuations and a Niss Elements imaging software (Nikon). Video recordings were performed with a 40 3 oil objective (N.A 1.4) for 6 (30 mM KCl) or 30 min (NMDA 100 mM 1 glycine 1 mM) by taking 14 bit images every 2 s at room temperature (25uC). Fluorescence was recorded and video images were analyzed by using the Spot module of the ImarisSuite 7.6H software (Bitplane A.G., Zurich, Switzerland) by selecting a 1.5 mm radius diameter of fluor-escent puncta. Fluorescence intensity was measured in 15-20 randomly selected synaptosomes within the microscopic field. The criteria for fluorescence puncta inclusion in the data analysis were the spherical shape ranging between 0.5-1.5 micron and the stimulation-dependent destaining response. Destaining time courses were generated by normalization of each fluorescence spot trace. Quantification of FM1-43 responses was accomplished by calculating the average percentage of fluorescence loss. A group of control experiments was run to test the specificity of Synaptogreen loading and destaining. Ca 21 free medium drastically reduced the dye loading and inhibited the stimulus-induced dye release. The Synaptogreen photobleacing rate was also measured in absence of the stimulus and compared to the stimulated Synaptogreen destaining traces for each experimental group.
Stimulation of synaptosomes in biochemical experiments. Stimulation of synaptosomes was performed in batch as previously reported 27 . Re-suspended synaptosomes were divided into 0.2 ml aliquots and gently agitated at 37uC until stimulation. Following 20 min mild agitation at 37uC, JNK inhibitor L-JNKi1 (2 mM) or NMDA receptor inhibitor D-AP5 (50 mM) were added as indicated and left for 30 min at 37uC before stimulus and till the end of the experiments. Synaptosomal suspensions were subjected to differential stimulation (see Results for details) with either NMDA (100 mM) 1 glycine (1 mM) or AMPA (100 mM) for 10 min at 37uC. KCl (8 mM) stimulus was added for 90 seconds after which 0.2 ml of warm medium containing inhibitors, where needed, was added.
The experiments were terminated by adding 0.6 ml of cold HEPES buffer and immediate centrifugation at 12,000 3 g to pellet the synaptosomes. In NMDA stimulation experiments, external Mg 21 ions were omitted to avoid the functional inactivation of the receptor.
Protein concentrations were then determined by Bradford assay and the samples were directly analyzed by immunoblotting following resuspension in Laemmli buffer.
Ultrastructural fractionation. As reported in Ref. 9, after stimuli or just after preparation, synaptosomes were pelleted and resuspended in 300 mL HEPES buffer (4uC; pH 7.4) and 50 mL was removed and kept as a total. Synaptosomes were pelleted by centrifugation (16 000 g; 5 min; 4uC) and resuspended in 300 mL 0.32 m sucrose and 0.1 mm CaCl 2 . Protease inhibitors were used in all extraction steps. Synaptosomal solution was further diluted 1510 in ice-cold 0.1 mm CaCl 2 and then mixed with an equal volume of 2 3 solubilisation buffer (2% Triton X-100, 40 mm Tris, pH 6.0; 4uC). Following 30 min incubation at 4uC the insoluble material (synaptic junctions) was pelleted by centrifugation (40 000 g, 30 min, 4uC). The supernatant (NSSP) was decanted and proteins precipitated with 6 volumes of acetone at 220uC and centrifuged (18 000 g; 30 min; -15uC). The synaptic junction pellet was resuspended in 10 volumes of 1 3 solubilisation buffer (1% Triton X-100, 20 mm Tris, pH 8.0; 4uC) and incubated for 30 min at 4uC and then centrifuged (40 000 g, 30 min, 4uC). The pellet contained the insoluble postsynaptic density and the supernatant contained the presynaptic active zone. The protein in the supernatant (presynaptic fraction) was acetone precipitated and collected as above. Proteins were then treated for Western Blot (WB) analysis.
Immunofluorescence. Immediately after preparation synaptosomes were seeded on coverslip coated with Poly-L-Lysine (50 mg/ml), washed twice with PBS and fixed in 2% (w/v) freshly prepared paraformaldehyde in PBS for 10 min at room temperature. After four washes synaptosomes were permeabilized with 0.05% (v/v) Triton X-100/ PBS for 5 min, washed and incubated overnight in a mix solution of the following primary antibodies: rabbit anti-JNK2 (Santa Cruz, USA) and mouse anti-Synaptotagmin I (Santa Cruz, USA), or rabbit anti-JNK2 and mouse anti-SNAP25 (SMI 81) (Covance, USA). All primary antibodies were diluted 15200 in 0.05% (v/v) Triton X-100/PBS at 4uC. Unbound antibody was removed by four washes 10 min each with PBS at room temperature. Bound antibody was detected by incubation with donkey anti-rabbit Alexa 488 (15500) and donkey anti-mouse Alexa 555 (15600) conjugated secondary antibodies (Invitrogen, USA) at 4uC for 2 h. After three washes with PBS at room temperature, coverslips were mounted on slides with Hydromount TM mounting gel. Controls were performed by omitting the primary antibodies.
Confocal microscopy and image analysis. Slides were examined under a confocal laser scanning microscope (Leica SP5, Leica Microsystems, Wetzlar, Germany) equipped with 4 laser lines and a transmitted light detector for differential interference contrast (DIC; Nomarski) imaging. Confocal acquisition setting was identical between slides. Bright field images of the synaptosomes were corrected for homogeneities of the brightness across the image by using Adobe Photoshop highpass filter. Contrast of synaptosome edges was enhanced with ImageJ CLAHE (Contrast Limited Adaptive Histogram Equalization; http://rsbweb.nih.gov/ij/ plugins/clahe/index.html) filter plugin. For production of figures, brightness and contrast of images were globally enhanced by using linear histogram correction and images were slightly oversaturated. Image analysis was performed by using Imaris Suite 7.4H or Image J 1.4 (imagej.nih.gov/ij/) softwares on seven different images derived from each group. Image analysis was performed under visual control to determine thresholds that subtracts background noise and takes into account www.nature.com/scientificreports synaptosomes structures. For all image-processing steps, the images were compared with the original raw data to make sure that no structures were introduced that were not seen in the original data series or that structures present in the original data series were not removed. To determine fluorescent signal colocalization between different channels the ImarisColoc module was used. Before colocalization analysis, a median filter was applied to the images to reduce the background noise and source channels threshold were established by automatic thresholding of the P Value. The degree of channels colocalization was analyzed by considering the following indexes: the % of dataset colocalized (colocalized area), thresholded Manders's coefficient A and B, Pearson's coefficient in colocalized area. To evaluate the amount of the total synaptosomoal area several masks, by using Imaris Surface module, were manually drawn enclosing the entire synaptosomes visualized through the light transmitted channel. Inside these masks, to evaluate the relative immunofluorescent areas a mask for each channel was automatically drawn using the Imaris Surface module. To evaluate the spatial relations between channels intensity we used the ImageJ tool Plot Profile to create a plot of fluorescence intensity values across a line drawn on the image.
Samples preparation for electron microscopy. Synaptosomal fraction was fixed in 2.5% glutaraldehyde (Electron Microscopy Science-EMS) in 0.05 M phosphate buffer pH 7.4 for 3 hours at room temperature, post-fixed in 1% osmium tetroxide (Electron Microscopy Science-EMS), dehydrated in graded acetone and embedded in Spurr (Epoxy resin, EMS). For a morphological analysis, semithin sections (1 mm thick) were cut on a Reichert-Jung Ultracut ultramicrotome and stained with toluidine blue. Ultrathin sections were cut and stained with uranyl acetate and lead citrate, then examined with a Zeiss electron microscope.
For post-embedding immunolabeling, sections (80 nm) were placed on 200-mesh formvar carbon nickel grids (EMS), etched in 1% sodium periodate for 60 min and in H2O2 1% for 15 min. Residual aldehyde groups were quenched with 0.05 M glycine in PBS, pH 4 for 20 min. After 30 minutes incubation in blocking buffer (phosphate buffer solution pH 7,4 added with 5% bovine serum albumin), the samples were incubated with polyclonal anti-JNK2 antibody (1520; SC-827 NH2-terminal S. Cruz) in phosphate-buffered saline added with 1% bovine serum albumin (incubation buffer) overnight at room temperature. After washing, samples were incubated in incubation buffer with goat anti-rabbit secondary antibody conjugated to 10-nm gold particles (1510; AURION Immuno Gold Reagents & Accessories, Wageningen, the Netherlands) for 3 hours at room temperature. Samples were washed in incubation buffer, then in distilled water and counterstained with saturated uranyl acetate solution in H 2 O for 1 hour. Grids were air-dried and viewed under the electron microscope.
Layer II neurons of the EC were recorded in whole-cell patch-clamp configuration using 1.5 mm borosilicate glass electrodes (3-4 MV) pulled with a vertical puller (PP-83 Narishige, Japan) in voltage-clamp mode 59 . Filling solution containing (in mM): CsMeSO 4 , 130; HEPES, 5; EGTA, 0.5; MgCl 2 , 1; NaCl, 1; CaCl 2 , 0.34; QX-314, 5 and MK-801, 1; adjusted to pH 7.3 with CsOH. Unless otherwise stated, neurons were held at 270 mV. Current signals were filtered at 2 kHz and digitized at 10 kHz using a Multiclamp 700 B differential amplifier linked to a Digidata 1440A, operated by the pClamp10 software (Molecular Devices, USA). No series resistance compensation was implemented, in order to keep a low signal-to-noise ratio, however, series resistance and whole-cell capacitance were monitored continuously during the experiment and recordings were discarded if series resistance changed by more than 15% from control conditions. Spontaneous synaptic events were analyzed offline using the pClamp10 software package (Molecular Devices, USA). The spontaneous events were detected from 3 min trace records, through an algorithm based on the minimization of the sum of squared errors between data and a template function. Single template waveforms were created by averaging a number of spontaneous events following a visual inspection of a representative trace in each cell. The template-matching threshold was set between 5 and 5.5, providing a good balance to avoid detection of false events.
After 50 min of wash in t-TBS, the blots were incubated for 1 h at room temperature with peroxidase-conjugated goat anti-rabbit or anti-mouse IgG secondary antibodies (UCS Diagnostic).
After 50 min of washes in t-TBS bands immunoreactivity was detected by enhanced chemiluminescence (ECL; WESTAR, Cyanagen, Italy). In all experiments where an evaluation of phosphorylated proteins was required, pan antibodies were applied on the same membranes after stripping procedure (stripping buffer from SignaGen, USA).
Statistical Analysis. In release experiments, analysis of variance was performed by ANOVA followed by Newman-Keuls multiple-comparison test, when comparing only a single treatment group a Student's two-tailed t-test was used. Data were considered significant for p , 0.05 at least. Appropriate controls with enzyme inhibitors were always run in parallel.
Statistical analysis for biochemical experiments was performed with GraphPad PRISM 4 (GraphPad Software, USA). Values shown represent the mean 6 SEM of a different number of separate experiments (as indicated in figure legends). One-way or Two-ways ANOVA followed by Tukey's test was carried out when intergroup comparisons were required.
Changes in miniatures events amplitude or inter-event interval was determined by comparison of their cumulative distributions, with the Kolmogorov-Smirnov (K-S) test, or by comparison of their median values, using the 2-tailed Student t-test. The threshold for statistical significance was set at p , 0.05.