Saturday, October 18, 2014

The Pin1 Enzyme and Synaptic Plasticity


This study looked specifically at inhibitory synapses - "The signal it transmits hinders activation of the postsynaptic neuron, making it less likely for it to become activated and emit its action potential,” according to Paola Zacchi, a SISSA [International School for Advanced Studies (Italian: Scuola Internazionale Superiore di Studi Avanzati, SISSA) researcher who coordinated the study.
“When Pin1 is absent from the synapse, signal transmission occurs “at full strength”, but also without control. Instead, when it is present, it regulates signal strength, making it weaker. We observed that Pin1 is able to modify the number of postsynaptic receptors”. The larger the number of receptors capable of binding to the neurotransmitter, the stronger the signal that reaches the postsynaptic membrane. “This also means that Pin1 plays a role in plasticity” explains Zacchi.
The whole article is open access, via Nature Communications, but it's pretty geeky, so we start with a summary of the press release from Neuroscience News, and then below that is most of the original article (minus the methods section which is not relevant to anyone outside the field of neuroscience and molecular biology).

An Enzyme and Synaptic Plasticity


Neuroscience News
October 10, 2014


Study reveals novel role for the Pin1 molecule.


A small, “empty” space teeming with activity: a synapse is a complex structure where the neural signal from the presynaptic neuron, as it travels towards its target turns into a chemical signal capable of crossing the synaptic space before becoming electrical again once on the other side. A synapse is a “dynamic” space not only because of the endless work that goes on there, but also for its ability to change its action over time (synaptic plasticity) as a result of either normal physiological processes (e.g., during learning) or because of disorders due to pathological conditions. A study, mainly carried out by SISSA researchers (which also involved the University of Zurich, LNCIB in Trieste, and EBRI in Rome), showed that a small enzyme (Pin1, peptidyl prolyl isomerase) that plays a mediating role in signal transmission has an effect on synaptic plasticity.

“The synapse we studied is of the inhibitory kind. The signal it transmits hinders activation of the postsynaptic neuron, making it less likely for it to become activated and emit its action potential”, explains Paola Zacchi, a SISSA researcher who coordinated the study. “When Pin1 is absent from the synapse, signal transmission occurs “at full strength”, but also without control. Instead, when it is present, it regulates signal strength, making it weaker. We observed that Pin1 is able to modify the number of postsynaptic receptors”. The larger the number of receptors capable of binding to the neurotransmitter, the stronger the signal that reaches the postsynaptic membrane. “This also means that Pin1 plays a role in plasticity” explains Zacchi.


This image depicts a neuron. Credit SISSA.

More in Detail

How does a synapse work? “A chemical synapse, the most common in vertebrates, is a small gap between nerve cells where the passage of a neural signal occurs”, explains Zacchi. In chemical synapses the two neurons are not in contact but they are separated by a distance of about 20 nanometres. For this reason, the electrical signal travelling along the presynaptic nerve ending is interrupted before resuming on the neuron on the other side of the gap. In between the two nerve cells the electrical signal is translated into a chemical signal (which then becomes electrical again).

“Arrival of the action potential on the presynaptic button causes release, into the interneural space, of molecules of neurotransmitter, which are picked up by receptors on the postsynaptic membrane”, says Zacchi. “If the synapse is excitatory, this leads to postsynaptic activation which, if sufficiently intense, triggers another action potential. If the synapse is inhibitory, as in our studies, the signal suppresses postsynaptic activation and inhibits firing of the electrical potential. In the process of neurotransmitter release and binding, other molecules come into play, such as scaffold proteins, which assemble receptors at the right place on the membrane in front of the neurotransmitter release sites, and neuroligins which act as bridges between the two ends of the synapse as well as interacting with the scaffold proteins. Pin1, the enzyme in the study, interacts with both neuroligins and scaffold proteins.

The Pin1 enzyme has long been known for its role in cancer and the development of neurodegenerative diseases such as Alzheimer’s and Parkinson’s (whereas neuroligins seem to be involved in autism). “Studies like this enhance our understanding of the biochemical mechanisms of synaptic plasticity, extending our knowledge of healthy mechanisms, but also helping those who are trying to understand what can be done in a wide range of pathological conditions”.

Notes about this neuroscience research
Contact: Federica Sgorbissa – SISSA
Source: SISSA press release
Image Source: The image is credited to SISSA and is adapted from the press release
Original Research: Full open access research for “Pin1-dependent signalling negatively affects GABAergic transmission by modulating ​neuroligin2/​gephyrin interaction” by Roberta Antonelli, Rocco Pizzarelli, Andrea Pedroni, Jean-Marc Fritschy, Giannino Del Sal, Enrico Cherubini and Paola Zacchi in Nature Communications; 5 (article #5066). Published online October 9 2014. doi:10.1038/ncomms6066

* * * * *

Pin1-dependent signalling negatively affects GABAergic transmission by modulating ​neuroligin2/​gephyrin interaction


Roberta Antonelli, Rocco Pizzarelli, Andrea Pedroni, Jean-Marc Fritschy, Giannino Del Sal, Enrico Cherubini & Paola Zacchi
Nature Communications 5, Article number (2014, Oct 9): 5066; doi:10.1038/ncomms6066
 

Abstract

The cell adhesion molecule ​Neuroligin2 (​NL2) is localized selectively at GABAergic synapses, where it interacts with the scaffolding protein ​gephyrin in the post-synaptic density. However, the role of this interaction for formation and plasticity of GABAergic synapses is unclear. Here, we demonstrate that endogenous ​NL2 undergoes proline-directed phosphorylation at its unique S714-P consensus site, leading to the recruitment of the peptidyl-prolyl cis–trans isomerase ​Pin1. This signalling cascade negatively regulates ​NL2’s ability to interact with ​gephyrin at GABAergic post-synaptic sites. As a consequence, enhanced accumulation of ​NL2, ​gephyrin and GABAA receptors was detected at GABAergic synapses in the hippocampus of ​Pin1-knockout mice (​Pin1−/−) associated with an increase in amplitude of spontaneous GABAA-mediated post-synaptic currents. Our results suggest that ​Pin1-dependent signalling represents a mechanism to modulate GABAergic transmission by regulating ​NL2/​gephyrin interaction.

Introduction


Structural and functional changes of post-synaptic density (PSD) components contribute to regulate synapse formation and plasticity. These remodelling events can affect trafficking, lateral mobility and turnover of several classes of structural and signalling molecules. They often involve interactions among specific proteins regulated by post-translational modifications, such as phosphorylation. At GABAergic synapses, the impact of phosphorylation on the gating properties, surface mobility and trafficking of the gamma-aminobutyric acid A receptors (GABAARs) has been extensively studied1, 2. Much less is known about the effects of phosphorylation of other post-synaptic proteins functionally linked to GABAARs.

An important class of molecules involved in synapse formation, maturation and stabilization comprizes the cell adhesion molecules of the neuroligin (NLs) family3. These post-synaptic proteins functionally coordinate pre and post-synaptic rearrangements by binding, via their extracellular domain, the presynaptically localized neurexins (NRXs) and via specific intracellular motifs, synapse-specific scaffolding molecules4, 5, 6. ​Neuroligin2 (​NL2) isoform is the only known adhesion molecule constitutively present at GABAergic PSDs7, where it drives the recruitment of inhibitory neurotransmitter receptors as well as the scaffolding molecule ​gephyrin6. ​Gephyrin, initially identified as a constituent of purified glycine receptor preparations (GlyR)8, 9, was soon recognized a key player in ​α2 and ​γ2 subunit-containing GABAARs clustering10, 11 and to be a central component of the GABAergic (and glycinergic) PSD8, 12. On the basis of its auto-oligomerization properties, ​gephyrin builds a bidimensional lattice underneath the synaptic membrane, which exposes a high number of binding sites to accumulate GlyR and GABAARs in front of the presynaptic releasing sites13, 14, 15, 16, 17.

NL2 interacts with ​gephyrin through a conserved stretch of amino acid residues highly conserved among all family members6. Site-directed mutagenesis within this binding module identified a specific tyrosine residue (Y770A) whose ​alanine substitution impairs ​NL2 ability to recruit recombinant and endogenous ​gephyrin to post-synaptic sites6. Notably, the corresponding ​tyrosine residue on ​NL1, the isoform enriched at excitatory synapses, was found to be phosphorylated in vivo, preventing ​NL1–​gephyrin interaction while favouring ​PSD95 recruitment at excitatory synapses18. Altogether, these findings point to the existence of intracellular signalling mechanisms able to modulate NL-scaffolding protein interactions by modifying specifically NL properties, leading to alteration in excitatory and inhibitory synaptic transmission.

In the present study, we have investigated whether post-phosphorylation prolyl-isomerization may affect GABAergic transmission in a similar manner. This signalling cascade targets ​serine and ​threonine residues preceding a ​proline residue to promote conformational changes on its substrate19. This effect is achieved by a unique enzyme, peptidyl-prolyl isomerase ​Pin1, whose catalytic activity facilitates the cis–trans isomerization of the peptide bond20, 21. Notably, ​Pin1 was found to interact with ​gephyrin and to alter its overall conformation, thus enhancing its ability to bind the GlyR22.
Here, we provide evidence that endogenous ​NL2 can be phosphorylated at its unique ​Pin1 consensus motif thus rendering it able to physically recruit the phospho-specific effector ​Pin1. We show that post-phosphorylation prolyl-isomerization can regulate ​NL2’s ability to complex with ​gephyrin. Specifically, ​Pin1-mediated propyl-isomerization of phosphorylated serine 714 negatively modulates ​NL2–​gephyrin complex formation, down-regulating GABAergic synaptic transmission.

Results

Endogenous ​NL2 undergoes proline-directed phosphorylation

The cytoplasmic domain (CD) of ​NL2 possesses a unique consensus motif for proline-directed phosphorylation, S714-P, located 15 amino acids apart from the transmembrane domain (Fig. 1a). To assess whether this site can undergo phosphorylation in vivo we used the mitotic phosphoprotein monoclonal 2 (MPM2) antibody that specifically recognizes phosphorylated S/T-P motifs (Davis et al.23). Endogenous ​NL2 was therefore immunoprecipitated from mouse brain homogenates using an affinity-purified polyclonal antibody raised against its CD or normal mouse IgG as negative control. Western blotting using the MPM2 antibody revealed a band at around 120 kDa that corresponds to the upper band of the doublet recognized by the ​NL2 antibody in parallel immunoprecipitation experiments (Fig. 1b), suggesting that at least a fraction of ​NL2 can be phosphorylated at its unique ​Pin1 consensus motif. To demonstrate that phosphorylation at serine 714 is the event responsible for ​NL2 detection by the MPM2 antibody, we generated the phospho-defective point mutant ​NL2HA-S714A. This mutation was introduced into a ​NL2HA hampered in ​gephyrin binding (​NL2HA-S714A-Δgephyrin-binding domain, GBD) (see Supplementary Fig. 1), to exclude the possibility that the MPM2 antibody would immune-react with phosphorylated ​Pin1 consensus motifs on endogenous ​gephyrin, which is, at the same time, a ​Pin1 target22 and an interacting partner of ​NL2 (ref. 6). Under these conditions, the MPM2 antibody efficiently immunoprecipitated only ​NL2HA-ΔGBD but not the corresponding point mutant, as indicated by the anti-HA immunoblot (Fig. 1c), thus demonstrating that S714 can be found phosphorylated on ​NL2.
Figure 1: ​NL2 is a proline-directed substrate. 
 
NL2 is a proline-directed substrate.
(a) Amino acid sequence of the ​NL2 CD. In bold is marked the unique ​Pin1 consensus motif (S714-P). The gephyrin-binding domain and the proline-rich region are highlighted in bold. (b) Representative immunoblotting of endogenous ​NL2 immunoprecipitated (IP) from mouse brain and probed with the anti-MPM2 that specifically recognizes phosphorylated S/T-P motifs and anti-​NL2. Rabbit IgGs were used as negative control (IgG) (n=4). (c) Representative immunoblotting of overexpressed ​NL2HA lacking the ​gephyrin binding domain (​NL2HA-ΔGBD) and the corresponding point mutant (​NL2HA-ΔGBDSer714Ala) immunoprecipitated by the phospho-specific MPM2 antibody. Western blot analysis was carried out with anti-HA monoclonal antibody. Mouse IgGs were used as negative control (n=5). (d) Co-immunoprecipitation (Co-IP) of endogenous ​NL2 and ​Pin1 from DSP cross-linked brain homogenates of ​Pin1+/+ or ​Pin1−/− mice. Western blots were performed with anti-​NL2 polyclonal and anti-​Pin1 monoclonal antibodies. Mouse IgGs were used as negative control. Asterisk indicate the IgG light chains (n=6). (e) FLAG epitopes from cross-linked samples of HEK293 cells co-expressing ​Pin1-​FLAG and ​NL2HA-ΔGBD or ​NL2HA-ΔGBDS714 were immunoprecipitated by anti-FLAG antibody. Western blot was performed with anti-HA and anti-FLAG monoclonal antibodies. Mouse IgGs were used as negative control (n=4). Full images of western blots are in Supplementary Fig. 5.
The essential feature of proline-directed phosphorylation as a signalling mechanism relies on the ability of phosphorylated S/T-P motifs to recruit the prolyl isomerase ​Pin1 (refs 19, 24). To test whether this unique phospho-epitope is able to recruit the effector molecule of the signalling cascade, we performed co-immunoprecipitation experiments from ​Pin1+/+ and ​Pin1−/− brain lysates. This approach unveiled that ​Pin1 can be detected in ​NL2, but not in control, immunoprecipitates or in the absence of ​Pin1 expression (Fig. 1d). To exclude the possibility that ​Pin1 co-precipitated by ​NL2 is bound to endogenous ​gephyrin, these assays were performed on co-expression of ​NL2HA-ΔGBD and ​Pin1-​FLAG in HEK293 cells. Cell lysates were immunoprecipitated with the anti-FLAG antibody and bound protein complexes analysed by western blotting using anti-HA and anti-FLAG antibodies for ​NL2 and ​Pin1 detection, respectively. As shown in Fig. 1e, while ​NL2HA-ΔGBD was still able to be immunoprecipitated from cells expressing ​Pin1-​FLAG, S714 to ​alanine mutagenesis completely abolished such interaction, indicating that S714 represents a newly identified ​Pin1 target.
 
Pin1 modulates ​gephyrin–​NL2 interaction

The observation that two fundamental components of the GABAergic PSD are both targets of proline-directed phosphorylation prompted us to investigate whether such signalling cascade would modulate their interaction. To this end, we initially co-expressed ​gephyrin-​FLAG and ​NL2HA in HEK293 cells and examined the amount of ​NL2HA that complex with ​gephyrin-​FLAG at 48 h after treating the cells with the selective and reversible inhibitor of ​Pin1 isomerase activity ​PiB (IC50 of approximately 1.5 μM) (ref. 25). As shown in Fig. 2a, even though the anti-FLAG antibody immunoprecipitated comparable amounts of ​gephyrin-​FLAG, a significant increase (64%) in the amount of co-precipitated ​NL2HA was observed on ​PiB treatment as compared with mock-treated cells (​dimethylsulfoxide, ​DMSO). Interestingly, a marked increase (140%) was detected on ​gephyrin-​FLAG co-precipitation by ​NL2HA-S714A as compared with ​NL2HA, indicating that ​Pin1 exerts a negative control on ​NL2–​gephyrin complex formation, at least in part, through ​NL2 prolyl-isomerization (Fig. 2b).
Figure 2: ​Pin1 negatively modulates ​NL2/​gephyrin interaction.
 
Pin1 negatively modulates NL2/gephyrin interaction.
(a) Representative IP of FLAG epitopes from samples of HEK293 cells co-expressing ​gephyrin-​FLAG and ​NL2HA and treated for 48 h with ​PiB 2.5 μM, ​DMSO (mock) or untreated. IP was also performed on ​NL2HA single transfected cells as a negative control. Nitrocellulose membranes were probed with anti-HA and anti-FLAG antibodies. The histogram on the right shows the relative amount of ​NL2 co-precipitated by ​gephyrin-​FLAG in control and ​PiB treated cells obtained from densitometric analysis (n=5, mean values±s.d., **P<0.001, Student’s t-test). (b) Lysates of HEK cells transfected with ​gephyrin-​FLAG in the presence of ​NL2HA or ​NL2HA-S714A or with ​gephyrin alone (as a negative control) were immunoprecipitated with anti-HA agarose. Immunoprecipitates were analysed by western blotting using anti-FLAG and anti-HA monoclonal antibodies. Arrowhead indicates the IgG heavy chains. The histogram on the right shows the relative amount of ​gephyrin-​FLAG in complex with either ​NL2HA or ​NL2HA-S714A co-precipitated by anti-HA agarose obtained from densitometric analysis (n=5, mean values±s.d., **P<0.001, Student’s t-test). (c) Co-IP of endogenous ​NL2/​gephyrin complexes from DSP cross-linked brain homogenates of ​Pin1+/+ or ​Pin1−/− mice. Western blots were performed with anti-​NL2 polyclonal and anti-​gephyrin monoclonal antibodies. Rabbit IgGs were used as negative control. An increased amount of ​gephyrin co-precipitates in complex with ​NL2 in the absence of ​Pin1 expression. Arrowhead indicates the IgG heavy chains. The histogram on the right shows the relative amount (obtained from densitometric analysis) of endogenous ​gephyrin co-precipitated by endogenous ​NL2 from both mouse genotypes (n=8, mean values±s.d., *P<0.01, Student’s t-test). (d) A similar experiment described in c was carries out on hippocampus isolated from of ​Pin1+/+ or ​Pin1−/− mice. The histogram on the right shows the relative amount (obtained from densitometric analysis) of endogenous ​gephyrin co-precipitated by endogenous ​NL2 from both mouse genotypes (n=4, mean values±s.d., **P<0.001, Student’s t-test). Full images of western blots are in Supplementary Fig. 5.
This issue was then investigated using a source of native ​NL2–​gephyrin complexes mouse brain homogenates from both genotypes. For these experiments, endogenous ​NL2 was immunoprecipitated using a rabbit polyclonal anti-​NL2 antibody and the co-precipitated ​gephyrin fraction was visualized by the monoclonal 3B11 antibody (Fig. 2c). In the absence of ​Pin1 expression, the amount of ​gephyrin co-precipitated by ​NL2 was increased by 40% as compared with ​Pin1 expressing neurons. This approach was also applied on hippocampal tissues isolated from both mouse genotypes. Here, the enrichment of ​gephyrin co-precipitated by ​NL2 in the absence of ​Pin1 expression was even more dramatic as compared with the amount detected from whole brain (130% increase; Fig. 2d), suggesting a strong impact of such signalling pathway on GABAergic synapses of the hippocampus.

Characterization of ​gephyrinPin1 sites S270-P and S319-P

The scaffolding molecule ​gephyrin possesses 10 putative ​Pin1 consensus motifs, the majority of them being concentrated in the central region (C-domain)26. To determine whether specific ​Pin1 sites may contribute to enhance ​NL2/​gephyrin complex formation, we decided to focus on those located close to, or within, the ​NL2 binding site on ​gephyrin. A previous yeast two-hybrid screening identified a large portion of ​gephyrin encompassing the E-domain and part of the C-domain as the region involved in ​NL2 interaction6. We re-examine this issue by generating eGFP-tagged ​gephyrin truncated version to be tested in GST-​NL2-CD pulldown assays. HEK293 cells transfected with different eGFP-​gephyrin variants were incubated with GST-​NL2-CD loaded beads or with GST alone as negative controls. As shown in Fig. 3a, while ​gephyrin 310–736 was recruited even better than the wild-type (WT) version, the mutants ​gephyrin 326–736 and ​gephyrin 1–310 (​gephyrin GC) displayed a reduced binding activity as compared to both ​gephyrin full-length (FL) and the truncated version 310–736 (Fig. 3a). Since the two E-domain ​gephyrin versions, showing such a striking difference in the binding affinity, differ only for a short stretch of amino acids, we generated the deletion mutant removing, from the FL protein, only the residues contained in this region but belonging to the E-domain itself (​gephyrinΔ319–329) and assayed it for ​NL2 binding. Interestingly, the lack of this short sequence almost completely abolished the interaction of ​gephyrin with ​NL2 (Fig. 3b), indicating that epitope(s) contained in the C-domain together with this minimal binding module are involved in ​gephyrin recruitment.
Figure 3: Impact of ​gephyrin S270A and S319A in ​NL2/​gephyrin interaction. 
 
Impact of gephyrin S270A and S319A in NL2/gephyrin interaction.
(a) GST-​NL2-CD pulldown from samples of HEK293 expressing EGFP-​gephyrin full-length (FL), EGFP-​gephyrin 310–736 (E-310), EGFP-​gephyrin 326–736 (E-326) and EGFP-​gephyrin GC. GST was used as negative control. Pulled down eGFP​gephyrin variants were detected using an anti-GFP monoclonal antibody. The bottom panels show the levels of GST and GST-​NL2-CD in the pulldown assays (Ponceau staining) (n=8). (b) EGFP-​gephyrin Δ319 to 329 was tested in similar pulldown assays. Western blots in a and b were performed using anti-GFP antibody. ​Gephyrin requires amino acid sequence 319–329 for its efficient recruitment by ​NL2 (n=6). (c) Representative IP of HA epitopes from samples of HEK293 cells co-expressing ​NL2HA and EGFP-​gephyrin WT, EGFP-​gephyrinS270A or EGFP-​gephyrinS319A. Nitrocellulose membranes were probed with anti-HA and anti-GFP antibodies. EGFP-​gephyrin single transfected cells incubated with HA agarose were used as negative controls. The histogram on the right shows the relative amount of eGFP-gephyrinWT and point mutants co-precipitated by ​NL2HA (n=4, mean values±s.d., P>0.05). (d) Representative images of hippocampal neurons transfected with EGFP-​gephyrin and EGFP-​gephyrinS270A point mutant immunolabeled for endogenous ​NL2 (magenta) and ​VGAT (blue) at DIV10. Enlarged boxed areas are shown aside to the corresponding full view image. Post-synaptic clustering is demonstrated by apposition of ​gephyrin/​NL2 clusters to ​VGAT positive terminals on the merge window. Scale bars, 20 μm in full view images and 5 μm in enlarged panels. (e) Distribution histograms of the % of ​gephyrin clusters colabeled with ​NL2 (79±5% in EGFP-​gephyrinWT versus 77±4% in EGFP-​gephyrinS270A), % of ​NL2 clusters colabeled with ​gephyrin (48±5% in EGFP-​gephyrinWT versus 71±4% in EGFP-​gephyrinS270A), % of ​NL2 synaptically localized (29±2% in EGFP-​gephyrinWT versus 43±6% in EGFP-​gephyrinS270A) and ​NL2 clusters intensity (119±15 a.u. in EGFP-​gephyrinWT versus 102 a.u.±6 in EGFP-​gephyrinS270A). The number of transfected hippocampal neurons investigated in each experiments (four independent experiments) were as follow: n=15 for eGFP-​gephyrinWT, n=10 for eGFP-gepyrinS270A (for each neurons at least 4 dendritic regions of interests were measured, mean values±s.d., *P<0.01, Student’s t-test).
On the basis of these results, two ​Pin1 consensus sites were further characterized, namely S319-P, located at the edge of the minimal binding module, and S270-P, positioned in its proximity, still contained, in the C-domain participating in ​NL2 binding. To this end, we introduced point mutations in eGFP-​gephyrin to create S319A and S270A mutants and tested them for their ability to interact with ​NL2HA. As judged by co-immunoprecipitation experiments, no significant differences were observed in binding capacity of the mutants as compared with ​gephyrin WT (Fig. 3c). These constructs were also overexpressed in cultured hippocampal neurons to analyse and quantify their impact on endogenous ​NL2 distribution using immunofluorescence staining and confocal microscopy. As previously reported, neurons expressing the S270A mutants had an increased number, unchanged in size, of ​gephyrin clusters compared with eGFP-​gephyrin WT27 (18.9±1.7 per 20 μm dendritic segment versus 6.5±0.6, P=0.00015). The expression of the S319A construct produced a dramatic decrease in cluster density associated with a diffuse cytoplasmic staining. This latter effect seems to correlate with the intrinsic instability of the mutant protein that undergoes a high rate of degradation on neuronal expression (data not shown), hampering its further characterization. Clusters formed by ​gephyrin S270A co-localized with ​NL2 at the same extent as the WT protein (around 78%; Fig. 3d,e). The fraction of ​NL2 clusters co-localizing with S270A mutant as well as their synaptic localization were increased as compared with ​gephyrin WT but their intensity values (calculated by normalizing cluster fluorescence intensity to cluster area and expressed in a.u.: 119 a.u.±15.2 versus 102±6.3) were unchanged (Fig. 3e). These data indicate that the increase in ​NL2/​gephyrin S270 interaction observed by immunoprecipitation is simply due to the augmented S270A cluster density and not to an enhance affinity of the mutant for ​NL2.

Pin1 selectively controls ​NL2 synaptic enrichment
 
Pin1 has emerged as a negative regulator of ​gephyrin–​NL2 interaction. Since these protein complexes are mainly localized at the plasma membrane, we tested whether ​Pin1 affects the amount of ​NL2 transported to, or maintained at, the neuronal plasma membrane. To this end, cultured hippocampal neurons derived from ​Pin1+/+ and ​Pin1−/− mice were subjected to surface biotinylation assay. Cell surface proteins were treated with the membrane-impermeant ​sulfo-NHS-biotin reagent, then isolated by binding to Streptavidin beads and probed with anti-​NL2 antibody. To check for unspecific protein binding during surface biotinylation experiments, hippocampal neurons not labelled with ​biotin were processed with biotinylated samples. Western blot detecting the intracellular glycophosphatidylinositol-anchored protein Flotilin1 was included to ensure that similar amount of associated membrane proteins, biotinylated or not, where incubated with Streptavidin beads. No major differences on the total content of membrane localized ​NL2 were observed between ​Pin1+/+ and ​Pin1−/− (Fig. 4a).
Figure 4: ​Pin1 enhances ​NL2 synaptic content not its surface abundance. 
 
Pin1 enhances NL2 synaptic content not its surface abundance.
(a) Surface ​NL2 derived from cultured hippocampal neurons of ​Pin1+/+ and ​Pin1−/− mice was isolated by biotinylation assay and detected by anti-​NL2 antibody. No biotinylated neuronal cells were processed in parallel to evaluate unspecific ​NL2 binding. Western blot detecting glycophosphatidylinositol-anchored Flotilin was used as loading control (n=4). Full images of western blots are in Supplementary Fig. 5. (b) Typical examples of hippocampal neurons from ​Pin1+/+ and ​Pin1−/− immunolabeled for endogenous ​gephyrin (magenta), ​NL2 (green) and ​VGAT (blue) at DIV10. Enlarged boxed areas are shown aside to the corresponding full view image. Post-synaptic clustering is demonstrated by apposition of ​gephyrin/​NL2 clusters to ​VGAT positive terminals on the merge window. Scale bars, 20 μm in full view images and 5 μm in enlarged panels. (c) Distribution histograms of ​NL2 cluster density (normalized to 100 μm2), the average cluster size and intensity in ​Pin1+/+ and ​Pin1−/− hippocampal neurons. (d) Distribution histograms of the percentage of ​NL2 co-localizing with gephryin and the percentage of double labelled ​NL2/​gephyrin puncta overlapping with the presynaptic marker ​VGAT. (e) Distribution histograms of ​gephyrin cluster density (normalized to 100 μm2), the average cluster size and intensity (calculated as described in c) in both mouse genotypes. The number of hippocampal neurons investigated in each experiments (three independent experiments) were as follows: n=10 for ​Pin1+/+, n=12 for ​Pin1−/−. For each neurons, at least five dendritic regions of interests were measured, mean values±s.d., **P<0.001, ***P<0.0001, Student’s t-test).
These results allow excluding the involvement of ​Pin1 in ​NL2 transport and/or turnover at the plasma membrane. Surface biotinylation represents an experimental approach that cannot provide an accurate analysis of protein distributions among different membrane domains. Since ​NL2 is enriched at GABAergic synapses, but is also distributed on extrasynaptic sites28, with this approach differences in ​NL2 partitioning between these two compartments might have been missed.

To this aim, immunocytochemical experiments were performed in dissociated ​Pin1+/+ and ​Pin1−/− hippocampal neurons co-labelled for ​NL2, ​gephyrin and ​VGAT, a specific marker of GABAergic innervations29 (Fig. 4b). In the absence of ​Pin1 expression a significant increase in ​NL2 cluster size (2.4 μm2±0.2 versus 1.7 μm2±0.2, P=0.00044) and intensity (92 a.u.±4.0 versus 58 a.u.±2, P<0.00048) was observed as compared with WT neurons, while no major changes in ​NL2 cluster density were detected (Fig. 4c). The fraction of ​NL2-positive clusters co-localized with endogenous ​gephyrin puncta was also enhanced in ​Pin1−/− cells (80±3.0% versus 60±5%, P=0.00013) and found enriched at post-synaptic sites, as demonstrated by the higher percentage of ​NL2/​gephyrin co-stained puncta overlapping with the presynaptic marker ​VGAT (48±4% versus 33±4%, P=0.0008; Fig. 4d). ​Gephyrin puncta appeared slightly, but significantly, increased in size while their density and intensity values were unchanged as compared with ​Pin1+/+ (Fig. 4e). These observations suggest that the absence of ​Pin1 promotes the formation and/or stabilization of ​NL2/​gephyrin complexes at GABAergic post-synaptic sites.

NL2/​gephyrin complex modulates synaptic abundance of GABAARs

The recruitment of GABAARs at synaptic sites is functionally coupled to NLs expression levels as well as to the ​gephyrin scaffold6. To assess whether the enhanced ​NL2/​gephyrin complex formation detected at GABAergic synapses similarly affects the distribution of synaptic ​γ2 subunit-containing GABAARs, we performed a quantitative evaluation of the ​γ2 subunit present in synaptosome suspensions isolated from the hippocampus of ​Pin1+/+ and ​Pin1−/− mice. Quantitative immunoblot analysis was also extended to ​NL2 and ​gephyrin to further verify their synaptic enrichment. As shown in Fig. 5a, the amount of all three markers investigated was significant increased in ​Pin1−/− mice as compared with ​Pin1+/+. The synaptic enrichment (synaptic fraction versus homogenate) was 35±5% for the ​NL2, 30±6% for the ​γ2 subunit and 20±4% for ​gephyrin.
Figure 5: Synaptic enrichment of GABAARs is achieved in ​Pin1−/−. 
 
Synaptic enrichment of GABAARs is achieved in Pin1−/−.
(a) Representative immunoblots of ​NL2, ​gephyrin and ​γ2 subunit of GABAA receptor extracted from the hippocampus of ​Pin1+/+ and ​Pin1−/− mice (littermates) in two different sets of experiments. Total proteins from the homogenates and synaptosome suspension fractions were analysed by western blotting. Below: quantification of the indicated antigens extracted from hippocampal tissues of ​Pin1+/+ and ​Pin1−/− mice. All markers analysed are enriched at inhibitory synapses. Western blot to actin was done as loading control. ​Pin1 immunoblot indicates hyppocampus from ​Pin1+/+ and ​Pin1−/− (n=6 littermate pairs, mean values±s.d, *P<0.05, Student’s t-test) Full images of western blots are in Supplementary Fig. 5. (b) Representative confocal micrographs of frontal brain sections showing segments of the SR and SO of the CA1 region of the hippocampus from adult ​Pin1+/+ and ​Pin1−/− mice immunolabeled for ​gephyrin (magenta) and ​VGAT (green). Scale bar, 5 μm. (c) Quantification of ​gephyrin punctum density (normalized to 100 μm2) and their percentage of colocalization with the presynaptic marker ​VGAT in both mouse genotypes. (d) Confocal micrographs as in a immunolabeled for ​GABAA receptor γ2 subunit (green) and ​VGAT (magenta). (e) Quantification of ​γ2 subunit punctum and their percentage of colocalization with ​VGAT in both mouse genotypes. The number of ​gephyrin, ​γ2, ​gephyrin and ​VGAT puncta was assessed in at least eight sections for each genotypes (​Pin1+/+ and ​Pin1−/−), by taking at least four images of SR and SO of the CA1 region of each hippocampus in each set of experiments (n=3). Mean values±s.d., *P<0.05, Student’s t-test. Scale bar, 5 μm.
We also examined the number of puncta labelled for ​gephyrin and ​γ2 subunit-specific antibodies, as well as their levels of colocalization with the presynaptic marker ​VGAT, in the CA1 region of the hippocampus of both genotypes. The staining pattern of ​gephyrin in ​Pin1−/− demonstrated a slight increase in the number of clusters both in the stratum oriens (SO) and stratum radiatum (SR) as compared with ​Pin1+/+ (SO 16±3 clusters per 100 μm2 and SR 28±3 clusters per 100 μm2 versus SO 10±2 clusters per 100 μm2 and SR 19±3 clusters per 100 μm2; P<0.05; Fig. 5b,c). This increase was paralleled by a small increase (around 6–8%) in ​gephyrin puncta co-localized with presynaptic ​VGAT (SO 30±2% and SR 39±1.4% versus SO 24±2% and SR 31±2%; P<0.05; Fig. 5b,c). The average cluster size and intensity were similar in both genotypes (3.6 μm2±0.2 versus 3.5 μm2±0.3 and 61±7 versus 65±4 a.u. for cluster size and intensity in ​Pin1−/− versus ​Pin1+/+, respectively).

The ​γ2 subunit staining pattern exhibited a similar cluster density in the two strata analysed in both genotypes (SO 8±2 and SR 18±2 versus SO 8±1 and SR 17±1.2; P>0.05; Fig. 5d,e). A small, although significant, increase in their intensity was evident (120±3 RFU versus 106±2 RFU in ​Pin1−/− versus ​Pin1+/+; P<0.05) but they were similar in size (4.3 μm2±0.5 versus 3.7 μm2±0.5). ​VGAT colocalization was increased by 10–15% in tissue from knockout animals (SO 38.9±2.7% and SR 52±3% versus SO 29±2% and SR 36±3%; P<0.05; Fig. 5d,e). The changes in ​gephyrin and ​γ2 subunit synaptic fraction are not due to an increase in synapses numbers, the density of inhibitory terminals being unaltered between the two genotypes, as assessed by quantification of ​VGAT immunolabeling (SO 14±2% and SR 22±3% versus SO 13±2% and SR 21±3%; P>0.05).
Altogether, these data indicate that the enhanced interaction between ​gephyrin and ​NL2 observed in the absence of ​Pin1 is associated with a concomitant increase in the synaptic recruitment of ​γ2 subunit-containing GABAARs.

Pin1 signalling affects the number of synaptic GABAARs

To functionally explore whether the enrichment of ​γ2 subunit-containing GABAARs in ​Pin1−/− mice affects GABAergic transmission, whole-cell recordings in voltage clamp configuration were performed from CA1 principal cells in hippocampal slices obtained from ​Pin1+/+ and ​Pin1−/− mice at postnatal (P) day P10–P13. These neurons presented similar resting membrane potential (Vrest) and input resistance (Rin) values (data not shown), thus indicating that ​Pin1 does not affect the passive membrane properties of principal cells. Spontaneous GABAA-mediated inhibitory post-synaptic currents (sIPSCs) were then recorded from both genotypes in the presence of ​6,7-dinitroquinoxaline-2,3-dione (​DNQX; 20 μM) to block AMPA-mediated excitatory post-synaptic currents (sEPSCs). As shown in Fig. 6a, recordings from ​Pin1−/− mice exhibited sIPSCs of higher amplitude values compared with control littermates (106±12 pA versus 62±8 pA; P<0.05), in the absence of any significant change in frequency (4.2±0.5 Hz versus 3.6±0.6 Hz; P>0.05; Fig. 6b). The amplitude distribution histogram of sIPSCs recorded in ​Pin1−/− unveiled a clear peak at ~200 pA (Fig. 6c). The observed effects were selective for sIPSCs since no significant differences in amplitude (22±2 pAin ​Pin1−/− mice and 27±4 pA in ​Pin1+/+; n=6 for both genotypes; P>0.05) or frequency (1.7±0.3 Hz in ​Pin1−/− mice and 1.3±0.4 Hz in ​Pin1+/+ mice; P>0.05) of sEPSCs (recorded in the presence of picrotoxin, PTX, 100 μM) were detected between the two genotypes (Supplementary Fig. 2a,b).
Figure 6: ​Pin1 affects the amplitude but not the frequency of sIPSCs.
 
Pin1 affects the amplitude but not the frequency of sIPSCs.
(a) Representative traces of sIPSCs recorded from CA1 principal cells at P11 in hippocampal slices from ​Pin1+/+ (black) and ​Pin1−/− mice (grey). Note higher amplitude events in ​Pin1−/− mice. (b) Each column represents the mean frequency and amplitude values of sIPSCs recorded from ​Pin1+/+ (black, n=9) and Pin−/− mice (grey, n=8). *P<0.05, Student’s t-test). (c) Amplitude distribution histograms of sIPSCs recorded in ​Pin1+/+ (1,030 events; black) and in ​Pin1−/− mice (1,412 events; grey). Note the appearance of a clear peak at ~200 pA in ​Pin1−/− mice.
Spontaneous inhibitory events from hippocampal neurons in culture overexpressing the ​NL2HA-S714A mutation exhibited, compared with ​NL2HA-transfected cells, a significant increase in amplitude (but not frequency), which in part mimicked the phenotype observe in ​Pin1−/− mice, suggesting that the interaction of ​Pin1 with ​NL2 is critical for this effect (Fig. 7a). As shown in the cumulative amplitude plot (Fig. 7b), the curve obtained from ​NL2HA-S714A transfected cells was shifted to the right as compared with cells expressing ​NL2HA (P<0.05).
Figure 7: Changes in amplitude of sIPSCs involve the interaction of ​Pin1 with ​NL2.
 
Changes in amplitude of sIPSCs involve the interaction of Pin1 with NL2.
(a) Samples traces of sIPSCs recorded from hippocampal neurons in culture expressing either the ​NL2HA or the ​NL2HA-S714A mutation. (b) Amplitude and inter-event interval (IEI) plots of sIPSCs recorded in cells transfected either with the ​NL2HA (black; n=7) or the ​NL2HA-S714A point mutant (grey; n=12). P<0.05; Kolmogorov–Smirnov test. Note the shift to the right of the cumulative amplitude distribution curve obtained from cells transfected with the mutant as compared to controls.
The selective increase in amplitude of sIPSCs detected in ​Pin1−/− mice suggest a post-synaptic site of action. This may involve an increase in the number of active GABAARs or changes in single-receptor channel conductance. To distinguish between these two possibilities, peak-scaled non-stationary fluctuations analysis of sIPSCs was performed only on stable recordings with no time-dependent changes in either peak amplitude, 10–90% rise time and decay time (Fig. 8a) (electrotonic filtering was excluded on the basis of no correlation between 10–90% rise time and decay time30). Plotting the mean current amplitude versus variance and fitting individual points with the parabolic equation (equation (2) in the methods; Fig. 8b), allowed estimating single-channel conductance and the number of channels open at the peak of spontaneous IPSCs. The single-channel conductance was calculated according to equation (3), assuming a reversal potential for chloride equal to 0. Interestingly, while the values of single-channel conductance were similar in both genotypes (Fig. 8c), the average number of active channels open at the peak of sIPSCs (Np) was significantly increased in ​Pin1−/− mice compared with controls (53±11 versus 26±5; P=0.03; Fig. 8c).
Figure 8: ​Pin1 controls the number of active receptor channels at GABAergic synapses. 
 
Pin1 controls the number of active receptor channels at GABAergic synapses.
(a) Individual sIPSCs from ​Pin1+/+ (black) and ​Pin1−/− mice (grey) are shown with the average currents (thick lines). (b) Current/variance relationships for sIPSCs shown in a (c) Summary plots of weighted mean channel conductance (43±3 pS and 43±3 pS, P=0.9, Student’s t-test) and number of GABAA receptor channels (Np) in wt (black; n=8) and in ​Pin1−/− mice (gray; n=5). *P=0.03, Student’s t-test.
To further evaluate the possibility that higher amplitude inhibitory events recorded in ​Pin1−/− mice may originate from GABAARs containing different subunits, we measured in both genotypes the decay time constants of small and large amplitude events. Spontaneous IPSCs were plotted against their decay half-widths and arbitrarily divided in two main classes whose amplitude was <or>150 pA (Fig. 9a, in green and blue, respectively). Notably, larger amplitude events (>150 pA) prevailed in ​Pin1−/− mice. No differences in decay of sIPSCs <or>150 pA were observed between ​Pin1+/+ and ​Pin1−/− mice, thus excluding the involvement of multiple receptor subtypes with different kinetics (the 90–10% decay (τ) of sIPSCs <150 pA was 9±1 ms in Pin−/− mice and 11±2 ms in ​Pin1+/+; P>0.05; τ of sIPSCs >150 pA was 11±2 ms in Pin−/− mice and 10±2 ms in ​Pin1+/+; P>0.05. The 90–10% decay time (τ90–10%) of all sIPSCs was 11±2 ms and 10±2 ms in ​Pin1+/+ and ​Pin1−/− mice, respectively; Fig. 9b,c, P>0.05). These data altogether suggest that the observed increase in amplitude of sIPSCs in ​Pin1−/− mice is exclusively due a genuine increase in number of GABAARs composed of the same subunits.
Figure 9: ​Pin1 does not affect the decay kinetics of spontaneous IPSCs. 
 
Pin1 does not affect the decay kinetics of spontaneous IPSCs.
(a) The peak amplitude of individual sIPSCs <150 pA (green) and >150 pA (blue) is plotted against their decay half-widths (τ50%) in ​Pin1+/+ and in ​Pin1−/− mice. (b) In the upper part, average traces of spontaneous IPSCs shown in a. In the lower part, average traces are normalized and superimposed. (c) Each column represents the mean 90–10% decay time constant of spontaneous IPSCs in ​Pin1+/+ and ​Pin1−/− mice, <150 pA (green), n=8 and 7, respectively and >150 pA (blue), n=6 and 7, respectively. For all comparisons, P>0.05, Student’s t-test.
GABA release and tonic inhibition are unaltered in ​Pin1−/−

In a previous study, we demonstrated that the functional knockdown of ​NL2 was accompanied by a reduction in the probability of ​GABA release31, thus underlying the role of NLs as retrograde regulators of presynaptic function. Therefore, we evaluated here whether ​Pin1-dependent modulation of ​NL2–​gephyrin interaction could also affect ​GABA release from presynaptic nerve terminals. To this end, we used ​1,2,5,6-tetrahydropyridin-4-yl methylphosphinic acid (​TPMPA), a low affinity competitive GABAAR antagonist32. This approach allowed to compare differences in presynaptic ​GABA transients between ​Pin1+/+ and ​Pin1−/− mice. Similar reduction of sIPSCs amplitude in both genotypes (51±6% versus 54±8%, P>0.05, Supplementary Fig. 3a,b) was detected on bath application of ​TPMPA (200 μM), thus excluding a transsynaptic action of ​Pin1 on ​GABA release.

Part of ​GABA released during synaptic activity may escape the cleft and invade the extracellular space to activate extrasynaptic high affinity GABAARs. This feature generates a persistent GABAA-mediated conductance33 that is involved in a number of physiological processes34. To determine whether ​Pin1 signalling affects extrasynaptic GABAARs, we analysed the tonic GABAA-mediated conductance in both ​Pin1+/+ and ​Pin1−/−mice. The tonic conductance was assessed by the shift of the holding current induced by application of the GABAAR channel blocker PTX (100 μM) (Supplementary Fig. 4a). This drug caused a similar shift in holding current in ​Pin1−/− and ​Pin1+/+ mice (Supplementary Fig. 4b,c), indicating that extrasynaptic GABAA receptors are not influenced by ​Pin1-mediated signalling.

Discussion


The present study shows that ​NL2 is a newly identified substrate of proline-directed phosphorylation. This post-translational modification, acting on its unique ​Pin1 consensus motif localized within the CD (S714-P), modulates the amount of ​NL2–​gephyrin complexes at synaptic sites. This modulation impacts on GABAergic transmission, by selectively affecting the total number of synaptic GABAARs. On the basis of these findings, post-phosphorylation prolyl-isomerization can play a crucial role in remodelling the GABAergic PSD to sustain plasticity processes.

Protein phosphorylation on ​serine and ​threonine residues preceding a ​proline, the so-called proline-directed phosphorylation, has emerged as a mechanism regulating signalling events through conformational changes that are catalysed by the phospho-dependent recruitment of the peptidyl-prolyl isomerase ​Pin1. While the different roles of ​Pin1 in dividing cells have long been established and characterized19, its function in post-mitotic neurons in general and at synapses in particular is still poorly understood. In a previous study, we identified ​gephyrin, the main scaffolding protein of inhibitory PSD, as a new target of post-phosphorylation prolyl-isomerization22.

Here, by inspecting the protein sequence of NL CDs, we identified S/T-P motifs that may provide ​Pin1 binding sites if phosphorylated in vivo. In particular, ​NL2 presents a unique ​Pin1 consensus site in its cytoplasmic region, S714-P, which is located 15 amino acids apart from the transmembrane domain. Even though this proximity to the plasma membrane raises doubts about its accessibility by a proline-directed kinase, several lines of evidence suggest that endogenous ​NL2 can undergo proline-directed phosphorylation. First, this isoform was recognized by the MPM2 antibody on ​NL2 immunoprecipitation from mouse brain homogenates. Second, MPM2-mediated ​NL2 immunoprecipitation was still maintained on removal of the ​NL2–gephyrin-binding domain, excluding the possibility of an indirect recognition mediated by endogenous ​gephyrin. Third, such detection was completely lost on ​NL2HA-S714A mutagenesis. This phosphorylation event is then able to directly recruit the effector molecule of the signalling cascade ​Pin1, as shown by co-immunoprecipitation experiments with endogenous neuronal proteins. Also in this case, ​Pin1 binding to ​NL2 was still maintained on the removal of the GBD, while it was completely abolished by mutating S714 to ​alanine, thus suggesting that the prolyl isomerase can be directly recruited by the unique ​NL2Pin1 consensus motif in a phosphorylation-dependent manner. These results altogether indicate that ​NL2 represents a newly identified substrate for proline-directed signalling cascade in vivo.

Our biochemical data demonstrate that ​NL2–​gephyrin interaction is negatively regulated by proline-directed phosphorylation. Co-immunoprecipitation experiments on recombinantly expressed ​gephyrin-​FLAG and ​NL2HA unveiled an enhanced complex formation on pharmacological inhibition of ​Pin1 catalytic activity. Similarly, endogenous ​NL2/​gephyrin complexes pulled down from whole brain or hippocampal tissues of ​Pin1−/− animals were significantly augmented as compared with the corresponding WT tissues. These biochemical findings were also validated by immunocytochemistry performed on cultured hippocampal neurons, where we could detect a high number of clusters co-labelled for ​NL2 and gephryin as well as their increased apposition to presynaptic GABAergic inputs in the absence of ​Pin1 expression. Interestingly, the ​NL2 point mutant unable to undergo prolyl-isomerization was capable to recruit ​gephyrin even more efficiently as compared with the WT form, whereas ​gephyrin mutagenesis at two putative ​Pin1 consensus motifs, S270A and S319A, located within, or close to, the minimal ​NL2 binding domain, was completely ineffective. The fact that this post-translational modification seems to control the strength of ​NL2 association with ​gephyrin by acting mainly on ​NL2, and not vice versa, further reinforces the emerging idea that cell adhesion molecules are key determinant in regulating synapse function. In a recent study by Giannone et al.18, it has been demonstrated that the level of ​NL1 phosphorylation at a specific ​tyrosine residue located within the GBD dictates the strength of ​NL1/​gephyrin interaction. In other words, ​NL1, the isoform enriched at excitatory synapses and therefore mostly associated with ​PSD95, can potentially recruit ​gephyrin as well as ​NL2, but its phosphorylation, promoted by neurexin–adhesion signalling, precludes such interaction while favouring ​PSD95 binding. Our experimental data indicate that proline-directed phosphorylation is acting similarly to ​tyrosine phosphorylation signalling. Since ​NL2 S714 is not positioned within the GBD, but is located just 50 amino acid upstream, it is reasonable to believe that ​Pin1-driven conformational changes, by affecting the overall folding of the CD, will induce ​gephyrin release (Fig. 10a). Alternatively, these conformational changes may promote ​NL2tyrosine phosphorylation, an event shown to impede NLs/​gephyrin interaction18 (Fig. 10b). Interestingly, ​tyrosine to ​alanine mutagenesis on ​NL2 was shown to completely abolish recombinant ​gephyrin recruitment by the mutant protein or to strongly reduce its interaction with endogenous ​gephyrin6. Whether ​NL2 phosphorylation occurs at tyrosine 770 and whether this event is able to hamper ​gephyrin binding is still unknown.
Figure 10: Model of the putative cross-talk between proline-directed phosphorylation and ​tyrosine phosphorylation.
 
Model of the putative cross-talk between proline-directed phosphorylation and tyrosine phosphorylation.
Phosphorylation of ​NL2 CD at S714 by a proline-directed kinase allows the recruitment of the proly isomerase ​Pin1. ​Pin1-driven conformational changes, by altering the folding of the ​NL2 CD, may represent the main cause responsible for ​gephyrin detachment (a). Alternatively, ​Pin1-mediated structural rearrangement may render the conserved ​tyrosine residue of the GBD (Y770) susceptible to phosphorylation, an event shown to prevent ​NL1/​gephyrin interaction (b).
The other partner of the complex is represented by ​gephyrin, a recognized target of ​Pin1 (ref. 22). ​Gephyrin contains 10 consensus motifs mostly concentrated in its C-domain, and all of them found to be phosphorylated in vivo35, 36. This region of the protein is positioned between the amino-terminal G- and carboxyl-terminal E-domains, which are directly involved in ​gephyrin multimerization. Conformational changes induced by phosphorylation, possibly followed by prolyl-isomerization, are expected to alter the conformation of the ​gephyrin C-domain and in turn, regulate specific functional properties of ​gephyrin, in particular its binding to interacting proteins, including possibly ​NL2. However, the complexity of the system under investigation makes it very difficult to determine whether and how a specific phosphorylation event can contribute, directly or indirectly, to enhance ​gephyrin association to ​NL2. Nevertheless, it should be emphasized that ​gephyrin is robustly phosphorylated at several residues in vivo, thus suggesting that a specific pattern of phosphorylation, rather than a single post-translational modification, is functionally determinant. In contrast, ​NL2 possesses a unique target for prolyl-isomerization suggesting that it could represent the master switch of the signalling cascade.

Our electrophysiological experiments clearly demonstrate that deletion of ​Pin1 specifically affects GABAergic transmission, causing a dramatic increase in amplitude, but not in frequency, of sIPSCs due to an increase in the number of GABAARs at post-synaptic sites. Notably, such enhancement was detected on neuronal overexpression of the ​NL2 mutant unable to undergo prolyl-isomerization, suggesting a functional link between the signalling cascade strengthening ​NL2/​gephyrin interaction and the increased synaptic recruitment of GABAARs. There is a large body of evidence underlying the key role played by ​NL2 in promoting clustering and/or stabilization of GABAARs at post-synaptic sites. By employing a heterologous expression system, it was shown that GABAARs are able to co-aggregate with ​NL2 and only the presence of this isoform can induce strong GABAergic presynaptic differentiation from co-cultured neurons and promote the establishment of fully functional hemi-synapses37. In ​NL2-deficient mice, the number of functional GABAARs detected in the retina was shown to be drastically reduced38. Furthermore, targeting of GABAARs and ​gephyrin scaffold appeared severely compromised in the pyramidal cell layer of the CA1 region of the hippocampus, a morphological phenotype accompanied by a strong deficit in synaptic inhibition6.

The increased recruitment of synaptic GABAA receptors in ​Pin1−/− mice may simply depend on the enhanced ​gephyrin targeting at synaptic sites. More scaffold deposition should offer a high number of binding sites available for the transient immobilization of GABAARs at inhibitory synapses. In addition, or alternatively, we cannot exclude the possibility that the extracellular domain of ​NL2 could also participate in GABAARs receptor recruitment. The unique S714-P consensus motif, located very close to the ​NL2 transmembrane domain, could influence the folding of the extracellular domain of ​NL2, rendering it incapable to interact in cis with GABAAR subunits. This type of mechanism has been shown to operate at excitatory synapses, where the abundance of NMDARs is controlled by the interaction occurring between the GluN1 subunit with ​NL1-specific sequences located in its extracellular domain39.

In conclusion, our findings unveil the existence of a new signalling pathway operating at GABAergic synapses to alter the efficacy of GABAergic transmission by modulating ​NL2/​gephyrin interaction. Although a comprehensive understanding of the molecular mechanisms underlying the action of ​Pin1 on ​NL2/​gephyrin interaction is still lacking, we believe that our study further emphasizes the key role played by ​NL2 in organizing and stabilizing GABAergic synapses.


How to cite this article: Antonelli, R. et al. ​Pin1-dependent signalling negatively affects GABAergic transmission by modulating ​neuroligin2/​gephyrin interaction. Nat. Commun. 5:5066 doi: 10.1038/ncomms6066 (2014).