Evidence of Presynaptic Localization and Function of the c-Jun N-Terminal Kinase

The c-Jun N-terminal kinase (JNK) is part of a stress signalling pathway strongly activated by NMDA-stimulation and involved in synaptic plasticity. Many studies have been focused on the post-synaptic mechanism of JNK action, and less is known about JNK presynaptic localization and its physiological role at this site. Here we examined whether JNK is present at the presynaptic site and its activity after presynaptic NMDA receptors stimulation. By using N-SIM Structured Super Resolution Microscopy as well as biochemical approaches, we demonstrated that presynaptic fractions contained significant amount of JNK protein and its activated form. By means of modelling design, we found that JNK, via the JBD domain, acts as a physiological effector on T-SNARE proteins; then using biochemical approaches we demonstrated the interaction between Syntaxin-1-JNK, Syntaxin-2-JNK, and Snap25-JNK. In addition, taking advance of the specific JNK inhibitor peptide, D-JNKI1, we defined JNK action on the SNARE complex formation. Finally, electrophysiological recordings confirmed the role of JNK in the presynaptic modulation of vesicle release. These data suggest that JNK-dependent phosphorylation of T-SNARE proteins may have an important functional role in synaptic plasticity.


Introduction
The c-Jun NH 2 -terminal protein kinase (JNKs) is part of the mitogen-activated protein kinase (MAPK) family [1,2] and is represented in mammals by three genes coding for three isoforms: JNK1, JNK2, and JNK3. JNK1 and JNK2 expression is essentially ubiquitous, while JNK3 presence is restricted to brain and testis [3]. JNKs activation is led by several types of extracellular stress stimuli such as heat shock, UV irradiation, inflammatory cytokines, A oligomers-induced toxicity, and excitotoxicity, primarily leading to cell death.
JNK phosphorylates numerous important substrates, including transcription factors AP-1, c-Jun, and Fos, leading to activation of nuclear response [3][4][5]. Furthermore, JNK regulates intracellular signalling pathways phosphorylating many cytosolic and cell-membrane targets. Many data pointed out that the JNK activation in the brain is higher than in any other mammalian tissue [6][7][8], suggesting that kinase might play an important role in the central nervous system (CNS), regulating both physiological and pathological brain processes.
All these evidences point out that JNK pathway in the CNS is particularly intricate, since it is not only subjected to different types of stimulations, which modulate JNK response, but it also plays different roles depending on its localization, having JNK pool distinct functions in the various cellular districts. Since neurons are exquisitely polarized cells, it is plausible that JNK activation may have different roles and functions in diverse subcellular localizations, from nuclear to cytoplasmic compartments as well as axons, dendrites, and dendritic spines. This effect is not only due to compartmentalization, but also due to the huge variety of JNK's targets. For instance c-Jun, the JNK elective target, is localized in the nucleus and regulates gene expression, while microtubule associated proteins (MAPs) are in the cytoplasm [23][24][25][26] and control axon elongation and dendrite branching [22]. More recently, JNK has been also identified at synaptic sites [27][28][29]; however its role in this district as well as its activation level is not completely clear yet.
Different works suggested that JNK is preferentially located in postsynaptic regions, where it interacts with key synaptic scaffold proteins (PSD-95 and PSD-93) as well as with GluR2 AMPAR subunit (GluR2L) [30]. We demonstrated that JNK has an important role in regulating glutamate receptors expression at postsynaptic density level after A -induced toxicity in hippocampal neurons [17,31]. Furthermore, many lines of evidence demonstrated the importance of NMDA receptors (NMDARs) as JNK activators in the postsynaptic compartment [9,32], suggesting that JNK might have a crucial role in synaptic plasticity as well as in excitotoxicity. NMDA-induced excitotoxicity in fact, mediated mainly by postsynaptic NMDARs, powerfully activates JNK, and JNK specific inhibition prevents NMDARmediated neuronal death in cortical neurons [9,10,25] as well as in organotypic hippocampal slices culture [9]. However, less is known about the biological role of presynaptic NMDARs (pre-NMDARs) and their downstream pathways. The pre-NMDARs have been extensively characterized in the last years as new key-players during development and memory consolidation phases. They are mainly involved in synaptic strength modulation, through mechanisms ranging from spontaneous neurotransmitter release facilitation to LTP/LTD [33][34][35][36][37][38][39]. The pre-NMDARs are abundantly expressed in the developing nervous system [40][41][42] although their expression decreases in the adult, persisting only in most plastic brain areas (such as neocortex, cerebellum, and hippocampus) [43]. Importantly, their aberrant activation has been also associated to pathological events [44,45]. Although JNKs presence in postsynaptic compartment has been characterized, there is lack of data for their presence and physiological role at presynaptic sites. Here we investigate JNK localization and role in presynaptic compartment. JNK identification at the presynaptic site and its activation after NMDARs stimulation may be important not only to clarify this issue, but also for the identification of new targets contributing to synaptic plasticity. Moreover, JNK downstream pathway definition could represent a new therapeutic tool to prevent synaptic dysfunction in neurodegenerative diseases.

Crude Synaptosomes Extraction.
Crude synaptosomes were used for calcium currents evaluation. Brains were homogenated with a glass-Teflon grinder in SHE buffer (Hepes 4 mM pH 7.3, sucrose 320 mM, EGTA 1 mM, Sigma) with protease inhibitors and phosphatase inhibitors. After a first centrifuge (10.000 10 min 4 ∘ C), supernatant was collected and again centrifuged (12.500 , 20 min 4 ∘ C). Pellet was resuspended in SHE buffer and subjected to a third centrifuge (12.500 , 20 min, 4 ∘ C) in order to obtain crude synaptosomal fraction pellet, which was resuspended in KRH buffer (NaCl 125 mM, KCl 5 mM, MgSO 4 1.2 mM, KH 2 PO 4 1.2 mM, Hepes NaOH pH 7.4 25 mM, glucose 6 mM, CaCl 2 2 mM, Sigma) and kept at 37 ∘ C 5% CO 2 for 45 min, for functional reactivation. Crude reactivated synaptosomes were then resuspended at a final concentration of 1 mg-500 g/ml in KRH. After resuspension, synaptosomes were kept at 37 ∘ C 5% CO 2 for 25 min for an habituation step.

Calcium Currents Evaluation.
Calcium currents were measured on crude synaptosomes. Briefly, after reactivation step, synaptosomes were pelleted as described before, resuspended in 300 l 1 : 1 mixture using KRH and calcium sensitive cell-permeating fluorophore Fluo-4 (Fluo-4 Direct6 Calcium Assay, Life Technologies). Samples were placed in a whole black 96-well (Corning Costar5 96-Well Black Polystyrene Plate) for 25 min 37 ∘ C 5% CO 2 to allow Fluo-4 to permeate cell membranes. After habituation fluorescence was measured with a Tecan F500 spectrophotometer for 180 cycles (corresponding to approximately 7 minutes) at 37 ∘ C. NMDA (100 M, with glycine coagonist 1 M final concentration) or KCl (50 mM final concentration) were injected 100 l/sec at cycle 20 (45 sec). Fluorescence was recorded by a top mode reading optimizing Z-position.

Synaptosomes
Stimulation. 25 minutes before stimulation D-JNKI1 peptide (Xigen) is added to the medium at a final concentration of 2 M. Synaptosomes were treated with NMDA 100 M (Abcam) with glycine coagonist 1 M (Sigma) or KCl 50 mM (Sigma) for 2 , 5 , 7 , or 10 at 37 ∘ C 5% CO 2 . For NMDA stimulation MgSO 4 was omitted from the buffer. At the end of the stimulation synaptosomes were resuspended in lysis buffer. For western blotting analysis Bonny lysis buffer was used [25] while for SNARE complex evaluation SNARE buffer was used [37].
2.9. Electrophysiology. After a recovery period of at least 1 h, an individual slice was transferred to a recording chamber and continuously perfused with extracellular solution at room temperature (22/25 ∘ C). Whole-cell voltage-clamp recordings were made from cortical neurons using an EPC-8 patch-clamp amplifier (HEKA Elektronik, Lambrecht/Pfalz, Germany). Pipettes of borosilicate glass, with a tip resistance between 2.0 and 3 MΩ, were used for patch-clamp recordings [42]. The intracellular solution had the following composition (mM): K-gluconate 117, KCL 13, MgCl 2 ⋅6H 2 0 2, Hepes 10, CaCl 2 1, EGTA 11, Na 2 ATP 2, Na 3 GTP 0.5, pH 7.3. Voltageclamp recordings were accepted only if the series resistance was less than 10 MΩ. Data were filtered at 3 kHz and digitized at 10 kHz using the filter and analog/digital converter of the amplifier. Digitized data were stored on computer disk using the Pulse software (HEKA Elektronik).

Statistical Analysis.
At least three independent replicates have been performed for each experiment. Statistical analysis was done using Graph Pad Prism 6 program. Data were analysed using Student's t-test and one-or two-way ANOVA, followed by Dunnet/Tukey's post hoc test. All data were shown as mean ± SEM with statistical significance given at < 0.05.

Computational
Analysis. The pairwise alignment between Syntaxin-1 (Uniprot ID Q16623) and Syntaxin-2 (Uniprot ID P32856) was performed using Needle software available in the European Molecular Biology Open Software Suit (EMBOSS) (PMID: 10827456). The structure of Syntaxin-1 for the docking simulations with JNK1 was extracted from the Syntaxin-1-Munc18A complex (10.1038/emboj.2008.37). The structure of JNK1 complexed with a small peptide belonging to JIP scaffold protein [43] was selected for the docking simulations. The structure of Snap25 was extracted from the 20S supercomplex [44] while the same JNK1 structure selected for the docking with Syntaxin-1 was also used. The docking simulation was performed using HADDOCK 2.2 [45] program that is one of the most suitable software to correctly predict the proteinprotein complexes. The images have been obtained using the software UCSF Chimera v.1.10.1.

JNK Is Located in the Presynaptic Compartment.
To investigate JNK localization in presynaptic compartment, we used N-SIM Structured Super Resolution Microscopy on isolated cortical synaptosomes (Figure 1(a)). Synaptosomes were double stained for Synaptophysin, a constitutive and specific presynaptic protein (labeled in red) and for the JNK active form p-JNK (labeled in green, Figure 1 in Figure 1(a) (Merge panel), the JNK active form p-JNK was located in presynaptic compartment and colocalized with Synaptophysin. This evidence is also confirmed by Pearson's coefficient value, corresponding to 0.54 ± 0.08. JNK presynaptic localization was further investigated comparing the presynaptic to the postsynaptic fraction and to total brain homogenate. JNK and activated form p-JNK were evaluated by western blotting on total cortical homogenate, cortical synaptosomal lysate (presynaptic compartment), and TIF fractionation (postsynaptic compartment; Figure 1(b)). The P-JNK activated form, as well as the JNK total form, was detectable in the presynaptic compartment (Figure 1(b)) confirming the Super Resolution Microscopy data (Figure 1(a)).
In the postsynaptic TIF fractionation, identified by PSD95 scaffold protein, JNK was present as expected. The purity of the synaptosomes was confirmed by the low presence of PSD-95 in the presynaptic fraction, which as described by Evans [46] is accountable to the nonfunctional PSD structure bound to active zone. Furthermore TIF fraction purity was proved by the low presence of Synaptophysin in postsynaptic fractions.

JNK Is Rapidly Activated after Pre-NMDARs Stimulation Both in Young and in Adult Mice.
To investigate pre-NMDAR-mediated JNK activation, we evaluated p-JNK/JNK ratio by western blotting on cortical synaptosomes purified from young (p14) and adult mice after 2 , 5 , and 10 minutes of treatment with NMDA (100 M) and glycine (1 M).
As previously described, pre-NMDARs expression decreases during neurodevelopment [43]: to verify if pre-NMDAR-mediated JNK activation was conserved in adult mice, synaptosomes were extracted from aged mice. To test if residual pre-NMDARs expression was enough to trigger depolarization, calcium currents were measured on synaptosomes by pre-ncubation with Fluo4 cell-permeant calcium sensitive dye. Synaptosomes were stimulated with NMDA (100 M) and glycine (1 M) or KCl 50 mM as positive control (Figure 2(f)). Recording of calcium influx was performed with a spectrophotometer for 7 minutes, corresponding to 180 fluorescence-reading cycles, with stimulus injection at 45 sec (cycle 20, Figure 2(f) black arrow). This time-point has been chosen as a compromise to avoid aberrant fluorescence measurement due to homeostasis loss caused by lack of CO 2 and humidity control in the spectrophotometer chamber. As expected KCl injection led to a rapid and significant increase of intrasynaptosomal calcium levels lasting over time. Similarly, NMDA administration led to a comparable significant increase in fluorescence if compared to controls injected with vehicle, starting 45 sec (20 cycles) after injection and remaining stable over time. Unexpectedly we observed a drop in fluorescence level in vehicle injected synaptosomes (CTR); however, in line with previous data, NMDA+ Gly stimulation led to rapid JNK activation at 2 min (50% increase * * * < 0.001), which increases at 5 min (60% increase, * * * < 0.001) and persisted at 10 min (50% increase, * * < 0.01; Figure 2(g)). These data suggested that pre-NMDAR-JNK axis was conserved during adult life. In fact, pre-NMDARs retained the ability to trigger excitatory response. The stimulation with NMDA (100 M) and glycine (1 M) induced JNK activation in both young (p14) and adult synaptosomes, and the consequent p-JNK/JNK ratio was comparable. As for young synaptosomes, also in this case, no significant protein degradation events were observed: no significant change of presynaptic proteins levels (Syntaxin-1 and Syntaxin-2, Snap25, and Vamp) at the considered timepoints (2 5 and 10 ) was found (Figures 2(h)-2(k)).

Investigation of JNK Possible Targets.
To get more insight into JNK modulation of vesicle release, JNK binding domain (JBD) was searched among presynaptic machinery proteins amino acidic sequences. In recent years, the extensive characterization of the JBDs in several JNK substrates [5] allowed identification of a canonical amino acid pattern for the JNK-substrate interaction. This canonical pattern is composed of two positively charged and two hydrophobic residues separated by two spacer regions with different  [47]. Among the critical steps involved in neurotransmitter release we decided to consider vesicle docking and priming steps, since previously reported data showed that JNK specific inhibition is not associated to any modification in calcium influx after NMDAR stimulation [25]. Thus it is unlikely that JNK could act on modulating this process in the presynaptic compartment. As reported in Figure 3(a), there are two possible JBDs in Syntaxin-1 and three in Syntaxin-2, respectively. Two out of three possible JBDs are conserved between Syntaxin-1 and Syntaxin-2 (Figure 3(a)). In particular, the JBD among residues 148-155 is fully conserved between the two isoforms ( Figure 3(a)) making this region a very good candidate to be involved in the interaction with JNK. Analyzing also the sequence of Snap25, a unique putative JBD has been found (Figure 3(b)). The presence of putative JBDs pattern in three presynaptic machinery proteins is a strong indication of possible physical interaction of those proteins with JNK.

JNK Interacts with SNARE Proteins.
The presence of JBDs sequences in three members (Syntaxin-1, Syntaxin-2, and Snap25) of the SNARE protein family allowed us to predict the possible three-dimensional (3D) organization of the JNK-SNARE protein complexes using computational approach, the protein-protein docking (Figures 3(c)-3(d)). The JBD sequence, fully conserved in both isoforms of Syntaxin (residues 148-155), has been mapped on the structure of Syntaxin-1 [48], which we have chosen as a representative structure also for Syntaxin-2, because they share a sequence identity of 64.4% and sequence similarity of 80% (Figure 3(c)). The unique JBD sequence located between residues 30 and 35 in Snap25 was also mapped on the 3D structure (Figure 3(d)).
The JBD present on the SNARE proteins and the region of JNK1 contacting the JIP peptide was used to drive the docking simulation performed with HADDOCK 2.2 software [49]. The best Syntaxin-1-JNK and Snap25-JNK complexes were reported in Figures 3(c) and 3(d) with residues belonging to the JBDs underlined. Both complexes have a favourable HADDOCK score that is a sum of different energy contribution terms (i.e., electrostatic energy, Van der Waals energy, desolvation energy, and restraints violation energies). In detail, the Syntaxin-1-JNK best complex reports an HADDOCK score of −24.9 ± 6.2 while the Snap25-JNK1 reports an HADDOCK score of −115.3 ± 0.8, underlining that the second complex formation could be more favoured. In both complexes the JBDs are involved in a dense network of hydrophobic and electrostatic interaction with JNK that are crucial for the stability of the complexes.
To prove these interactions, SNARE proteins were evaluated with western blotting, after immunoprecipitation of Syntaxin-1, Syntaxin-2, and Snap25 on purified cortical synaptosomal lysates. JNK was found in the coprecipitate; in particular the JNK protein band was more detectable with Syntaxin-2 and Syntaxin-1 and was less intensely visible with Snap25 (Figures 4(a), 4(b), and 4(c)). To confirm the interaction of JNK with Syntaxin-1, Syntaxin-2, and Snap25, the reverse immunoprecipitation on purified cortical synaptosomal lysates was performed. According to data previously found mostly Syntaxin-2 and Syntaxin-1 were traceable in JNK's coprecipitate (Figure 4(d)). However, a precise quantification is not possible due to the different efficiency of the antibodies used.

SNARE Proteins Interact Both with JNK2 and with JNK3.
To investigate if a specific JNK isoform was involved in the interaction with SNARE protein complex, single JNK isoform precipitation was performed on cortical synaptosomal lysates. After precipitation of JNK3 [7,50], an isoform known to be more sensitive to NMDA stimulation, and of JNK2, an isoform known to be important for neural plasticity, SNARE protein expression was evaluated by western blotting in the coprecipitate. Presence of both Syntaxin-2 and Snap25 was traceable in JNK2 coprecipitate ( Figure 5  both SNAREs were present in JNK3 coprecipitate ( Figure 5, right panel). These results indicated that both JNK isoforms, JNK3 and JNK2, interact with Syntaxin and Snap25, without having a preferential interaction with a specific isoform.

JNK Inhibition Reduces the Frequency but Not the Amplitude of mEPSCs in Cortical Slices.
To determine the site of synaptic modifications induced by JNK, the miniature excitatory postsynaptic currents (mEPSCs) were examined in the presence of JNK inhibitor, D-JNKI1. Miniature synaptic events are due to the spontaneous release of glutamate vesicles, which in turn activates postsynaptic glutamate receptors. According to the quantal hypothesis, changes in the amplitudes of mEPSCs reflect a change in postsynaptic responsiveness, whereas presynaptic mechanisms result in no change in the amplitude of the mEPSCs but may cause a change in the frequency of mEPSCs recorded in the postsynaptic cell [51,52]. The mEPSCs were, therefore, recorded in cortical slices, pretreated for 1 hr with D-JNKI1 (2 M), in the presence of TTX (1 M) to prevent action-potential firing, and bicuculline (20 M) to block GABAergic currents. As shown in Figure 6(a) (sample traces of mEPSCs) and in Figures 6(b)-6(c) (quantification of mEPSCs frequency and amplitude), D-JNKI1 induced a significant decrease in the frequency of mEPSCs if compared to control conditions ( * < 0.05, = 8) but had no statistically significant effect on mEPSCs amplitude. These results support the hypothesis that JNK can act presynaptically to regulate transmitter release.

JNK Inhibition Reduces SNARE Complex Formation after NMDA Stimulation.
Having demonstrated JNK interaction with Syntaxin-1/Syntaxin-2 as well as with Snap25 and that the specific inhibition of JNK, with D-JNKI1, reduced spontaneous release in an in vitro slice preparation. We here studied the effect of D-JNKI1 on JNK-SNARE complex formation in control condition and after NMDA stimulation. The SNARE complexes (Syntaxin + Snap25 + Vamp) are detergent resistant structures and can be visualized by western blotting on unboiled samples [53]. Anti-Syntaxin-1 antibody was used to visualize SNARE complex at approximately 75-100 KDa and used for densitometric quantification. Higher MW signals, ascribable to more structured complexes, enriched with other release-machinery factors, were detected too.  there was a significant 50% increase of complex formation ( * < 0.05) after 7 minutes of treatment, due to the increased vesicle release triggered by the pre-NMDARs activation (Figure 7(a)). When synaptosomes were firstly pretreated with D-JNKI1 (2 M) for 25 minutes and then stimulated with NMDA, SNARE complex assembly was significantly prevented ( # < 0.05) and complex levels were restored to nonstimulated control levels (see Figure 7(a)). These results showed that JNK inhibition prevented SNARE proteins complexation, according to electrophysiology data. Finally, the levels of SNARE complex proteins were evaluated in boiled samples for all the conditions, control synaptosomes with and without D-JNKI1 as well as synaptosomes stimulated by NMDA with and without D-JNKI1. No change in protein levels was observed under these conditions (Figures 7(b)-7(d)). However we were not able to detect any reduction of noncomplexed Syntaxin-1 in unboiled NMDA treated samples, even though a significant reduction could have been expected due to Syntaxin-1 assembly in SNARE complex. This might probably be due to the huge enrichment of Syntaxin-1 in synaptosomal fraction, whose alterations in the protein levels cannot be conclusively evaluated by western blotting.

Discussion
JNK kinase plays an important role in regulating several functions in the central nervous system and in synaptic plasticity processes [54,55]. In particular, a specific JNK pool in the postsynaptic site is able to modulate synaptic strength regulating both AMPA and NMDA glutamate receptors localization in the postsynaptic density region (PSD). This feature is partially due to JNK interaction with PSD scaffold protein PSD95 [56], but also by its direct interaction with NR2A/B [57,58]. Importantly, in pathological conditions, such as Alzheimer's disease (AD) synaptopathy, JNK induces a massive removal of glutamate receptors (AMPA and NMDA) from PSD region, leading to LTP impairment, LTD increase [18,31], and dendritic spines loss [17,31]. JNK's role in the regulation of glutamatergic synapse dysfunction and dysmorphogenesis is intriguing and not completely established. The JNK postsynaptic function is well recognized. In fact, JNK is rapidly activated after glutamate receptors stimulation [59], and in addition plays a key role in NMDA-induced neuronal-death (excitotoxicity; see [10,60,61]). Despite the great amount of data supporting JNK function after NMDA activation, nothing is known by now about JNK involvement in presynaptic NMDARs (pre-NMDARs) downstream pathway. The pre-NMDARs are present at the presynaptic site in the more plastic brain regions in adult, such as cortex and hippocampus [62], where they control neurotransmitter release [63,64]. The activation of JNK signalling after NMDA stimulation at the presynaptic site is not proven yet, and only few studies have proposed a presynaptic role for JNK [65]. Many findings in literature show that synaptosomes preparation is a valid tool to value presynaptic compartment and related receptors, as previously described. By studying pre-NMDARs stimulation in synaptosomes, we showed for the first time JNK activation in the presynaptic site. Thanks to both microscopic and biochemical approaches, in fact, we detected JNK and its activated form, p-JNK, in cortical presynaptic terminals. Notably its levels in presynaptic site seem comparable to those of postsynaptic site, previously characterized ( [17], see Figures 1 and 2). We proved and examined JNK activation after pre-NMDA receptors stimulation directly at pre-synaptic site (synaptosomes) in both young and adult cortex preparations. Interestingly, JNK activation rates were comparable for young and adult preparations, indicating that pre-NMDARs stimulation induced JNK activation in the presynaptic site at both ages. In addition, this process is conserved after neurodevelopment although pre-NMDARs expression is widely decreased and persisted only in cortex and hippocampus in adult [62]. No alteration of presynaptic proteins levels was observed in synaptosomes, suggesting that JNK is not involved in protein degradation, as what occurs in the postsynaptic compartment [17,18,30].
Literature evidences suggest that presynaptic kinases regulate neurotransmitter release, mainly affecting three steps: stored vesicle mobilization [66], axonal ending depolarization due to calcium influx [67], and presynaptic vesicles fusion [68]. In this context the role of JNK in regulating presynaptic tasks is poorly understood: there are only few data reporting vesicle mobilization as an important step in spontaneous release [69], while JNK inhibition had already been proven not to change calcium intracellular levels [25] suggesting that it is unlikely that the kinase could modulate calcium channels or NMDA receptors permeability. Thus thanks to our modelling design finding, we examined whether JNK acts as a physiological effector of SNARE complex formation regulating vesicle release. The biochemical data confirmed the prediction of the computational analysis, suggesting the interaction of JNK with Syntaxin-1, Syntaxin-2, and Snap25 as well as with the SNARE complex formation. We have demonstrated that Syntaxin-1 and Syntaxin-2 share around 64% of sequence identity and there is only one JBD motif fully conserved in both isoforms, while we found only one complete JBD motif into the Snap25 sequence. Once we got these data from the sequence alignment, we were able to map the JBD on the Syntaxin-1/Syntaxin-2 and Snap25 threedimensional (3D) structures in order to perform the docking simulation. The results of the docking simulations can show the possible conformations of the Syntaxin-1/Syntaxin-2-JNK and Snap25-JNK complexes. However, the electrophysiological and biochemical experiments proved the rationale of the docking hypothesis.
To get more insight into JNK presynaptic function, we examined the effects of the specific JNK inhibitor D-JNKI1 on mEPSCs recorded from cortical slices. By preventing JNK action, the frequency, but not the amplitude, of mEPSCs was strongly reduced (50%) thus confirming a modulation of presynaptic release probability by JNK. These results combined with the biochemical findings suggest that the interaction of JNK with Syntaxin-1 and Syntaxin-2 as well as Snap25 at the presynaptic level could represent an important mechanism for regulating vesicle release.
Whether the JNK binding to Snap25 is direct or mediated by Syntaxin is difficult to establish since there are many interactions among the T-SNARE proteins that form the complex, including Syntaxin and Snap25 which, in resting conditions, are partially assembled in a primed structure [70][71][72]. For this reason, we evaluated NMDA-mediated SNARE complex assembly in presence of specific D-JNKI1 inhibitor. As expected, JNK inhibition reduces SNAREs complexation, indicating JNK as a modulator of vesicle docking-priming, via Syntaxin/Snap25 direct interactions and probably phosphorylation. Such effect may result in a fine regulation of neurotransmitter release by JNK signalling pathway.
We then tried to clarify which JNK isoforms are more prone to interact with Syntaxin-1/ Syntaxin-2 and Snap25. The brain specific JNK3 isoform, majorly related to NMDARs activation [7,50], is the most intriguing form, but unfortunately the antibodies are not able to discriminate well among JNK isoforms. Immunoprecipitation assay suggests a functional overlapping between the two isoforms, JNK2 and JNK3. Further studies are needed to clarify this issue. These findings all together unravel a new and important role of JNK in the presynaptic site.

Conclusions
The JNK stress-signalling pathway is indicated as a key player in developmental and neurodegenerative pathologies. Its implication in excitotoxicity and synaptic dysfunction/dysmorphogenesis makes its importance clear. JNK represents a therapeutic target against neurodegenerative processes, but much information is missing to manipulate its action in order to reduce unwanted adverse effects. In this work we demonstrated for the first time that JNK leads to neurotransmitter release facilitation, due to a direct interaction with SNARE proteins. Thus an important new role of JNK in controlling synaptic plasticity emerges from this study, adding another part of the story to better clarify JNK function in synapse modulation.