The neuropeptide pituitary adenylate cyclase-activating polypeptide 38 (PACAP38) has been implicated in the induction of synaptic plasticity at the excitatory glutamatergic synapse. In particular, recent studies have shown that it is involved in the regulation of N-methyl-D-aspartate (NMDA) and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor activation. Here we demonstrate the effect of PACAP38 on the modulation of dendritic spine morphology through a disintegrin and metalloproteinase 10 (ADAM10)–N-cadherin–AMPA receptor signaling pathway. Treatment of primary hippocampal neurons with PACAP38 induced an accumulation of ADAM10 at the postsynaptic membrane. This event led to a significant decrease of dendritic spine head width and to a concomitant reduction of GluR1 colocalization with postsynaptic markers. The PACAP38-induced effect on dendritic spine head width was prevented by either treatment with the ADAM10-specific inhibitor or transfection of a cleavage-defective N-cadherin construct mutated in the ADAM10 cleavage site. Overall, our findings reveal that PACAP38 is involved in the modulation of dendritic spine morphology in hippocampal neurons, and assign to the ADAM10–N-cadherin signaling pathway a crucial role in this modification of the excitatory glutamatergic synapse.
The neuropeptide pituitary adenylate cyclase-activating polypeptide 38 (PACAP38) is expressed in the different regions of the hippocampus starting during embryogenesis, diminishing only slowly towards adulthood (Jaworski and Proctor, 2000). It has been shown that PACAP38 modulates a variety of signaling pathways within the excitatory glutamatergic synapse (MacDonald et al., 2005; Yang et al., 2009; Yang et al., 2010), ranging from activation of different protein kinases, i.e. protein kinase A (PKA) and the mitogen-activated protein (MAP) kinase to the mobilization of calcium (Harmar et al., 1998; Vaudry et al., 2000; Kojro et al., 2006; Ohnishi et al., 2008; Yang et al., 2010). Interestingly, induction of endogenous PACAP38 responded to different signals, i.e. cAMP signals and KCl-dependent membrane depolarization (Fukuchi et al., 2004).
A recent study showed that treatment with PACAP38 at a very low concentration reduced α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor excitatory postsynaptic currents (EPSCs) through PAC1-receptor independent signaling (Costa et al. 2009). By contrast, another study showed that the application of PACAP38 enhanced AMPA-evoked currents (Michel et al., 2006). PACAP38 has been also shown to induce a form of long-term depression (LTD) in hippocampal neurons that requires interaction of AMPA receptor subunits with scaffolding proteins (Kondo et al. 1997; Roberto et al. 2001; Ster et al., 2009), thus validating its role in learning and memory. This has been further confirmed by the use of transgenic animals (Yang et al., 2010). Nonetheless, the mechanism by which PACAP38-dependent signaling can modify AMPA receptors remains almost unknown and no information is available on the effect of PACAP38 on AMPA receptor subunit localization at the synapse.
Interestingly, PACAP38-dependent activation of PAC1 receptor can also activate key enzymes, i.e. a disintegrin and metalloproteinase 10 (ADAM10), localized within the excitatory postsynaptic density (Kojro et al., 2006; Marcello et al., 2007; Marcello et al., 2010; Malinverno et al., 2010). In particular, Kojro and coworkers showed that PACAP38 leads to a strong increase of ADAM10 activity through the intervention of three signaling cascades (PKC, MAP kinase and PI 3-kinase) (Kojro et al., 2006). Notably, we have recently described that inhibition of ADAM10-dependent cleavage of the cell-adhesion molecule N-cadherin induces a modification of the number and the current of synaptic AMPA receptors in hippocampal neurons, and a significant increase in the size of dendritic spines (Malinverno et al., 2010). Accordingly, several reports demonstrated the existence of a close coordination between AMPA receptor content at synapses, N-cadherin activation and spine size (Kasai et al., 2003; Kopec et al., 2007; Xie et al., 2008).
Here we show that in hippocampal neurons PACAP38 induces dendritic spine shrinkage correlated with a decrease of synaptic AMPA receptors through modulation of ADAM10-dependent cleavage of N-cadherin.
Results and Discussion
Treatment with PACAP38 leads to a modification of ADAM10 and GluR1 subunit levels at synapses in primary hippocampal neurons
A previous report described the ability of PACAP38 to stimulate ADAM10 activity in neuroblastoma cells (Kojro et al., 2006). We performed immunocytochemical labeling to verify whether treatment of hippocampal neurons with PACAP38 also induces a modification of ADAM10 localization at synaptic sites, where most of the ADAM10 neuronal substrates are located (Uemura et al., 2006; Malinverno et al., 2010). As shown in Fig. 1A,B, treatment with PACAP38 (300 nM, 30 minutes) led to an increase of ADAM10 colocalization with the postsynaptic marker Shank (PACAP38 300 nM vs control, 61.8±6.7%, **P<0.001). Treatment with PACAP38 at lower concentrations (1–10 nM) (MacDonald et al., 2005) did not induce any effect on ADAM10 localization in the postsynaptic compartment (PACAP38 1 nM vs control, −1.9±10.9%, P=0.902; PACAP38 10 nM vs control, 10.2±4.4%, P=0.344).
PACAP38 can modulate different intracellular pathways within the excitatory synapse (Macdonald et al., 2005; Yang et al., 2009, 2010). In particular, PACAP38 induces PKA, and several cellular effects of PACAP38 are mediated by MAP kinase activation (Kojro et al., 2006). Treatment of hippocampal neurons with PACAP38 in the presence of a p38-specific inhibitor (SB203580, 5 μM) did not abolish the effect of PACAP38 on ADAM10 localization (PACAP38 + SB203580 vs control, 64.8±4.4%, P<0.0001; PACAP38 + SB203580 vs PACAP38, 1.8±2.8%, P=0.788). Conversely, co-treatment with a PKA inhibitor (H89, 5 μM) blocked the PACAP38-induced increased in the colocalization of ADAM10 with Shank (Fig. 1A,B; PACAP38 + H89 vs control, 6.2±7.2%, P=0.431; PACAP38 +H89 vs PACAP 38, P<0.001).
PACAP38 has a high affinity for its specific receptor (PAC1), but it also has an affinity for the VIP-specific receptors (VPACRs) VPAC1 and VPAC2, similar to VIP (Yang et al., 2010). As shown in Fig. 1A,B co-treatment of hippocampal neurons with PACAP38 and VIP antagonist [VIP(6-28), 1 μM] partially blocked the increased colocalization of ADAM10 with Shank induced by PACAP38 [PACAP38 + VIP(6-28) vs control, 41.4±3.6%, P<0.005; PACAP38 + VIP(6-28) vs PACAP38, P<0.005].
Modulation of ADAM10 levels at synaptic sites is a crucial way to regulate AMPA receptors at hippocampal excitatory synapses, leading to a modification of AMPA receptor currents and subunits composition (Malinverno et al., 2010). Confocal analysis revealed that treatment with PACAP38 (300 nM, 30 minutes) induces a reduction of GluR1 localization in Shank-positive postsynaptic clusters (Fig. 1C,D; PACAP38 vs control, −37.3±3.5%, **P<0.001), thus indicating a decrease in AMPA receptor availability at synaptic sites following activation of PACAP38-dependent signaling.
Treatment of hippocampal neurons with PACAP38 in the presence of a PKA inhibitor (H89, 5 μM) partially blocked the reduction of GluR1 colocalization with Shank induced by PACAP38 (Fig. 1C,D; PACAP38 + H89 vs control, −14.5± 2.5%, P=0.016; PACAP38 + H89 vs PACAP38, P<0.001), suggesting a role for PKA in the complex mechanism by which PACAP38-dependent signaling can modify AMPA receptors (Yang et al., 2010). Interestingly, co-treatment with a VIP antagonist [VIP(6-28), 1 μM] also partially blocked the reduction of GluR1 localization at synaptic sites induced by PACAP38 [Fig. 1C,D; PACAP38 + VIP(6-28) vs control, −21.7±3.5%, P<0.005; PACAP38 + VIP(6-28) vs PACAP38, P=0.013].
Treatment with PACAP38 leads to a significant decrease of spine head width in primary hippocampal neurons
ADAM10 plays a crucial role in the complex sequence of events that regulates dendritic spine maturation and/or stabilization and in the modulation of the structural organization of the glutamatergic synapse (Malinverno et al., 2010).
Accordingly, we investigated the onset of a possible effect of PACAP38 on dendritic spine morphology by time-lapse confocal imaging. In control neurons, no significant modification of average spine head width was observed over 30 minutes (data not shown). A decrease of spine head width (Fig. 2A,B), that became statistically significant 21 minutes after treatment, was observed in PACAP38-treated neurons compared with control ones (Fig. 2A,B; P<0.0001 PACAP38 vs control). Furthermore, a significant increase in the percentage of shrinking spines (control 24.7±1.2%, PACAP38 46.3±7.0%, P=0.038) and a corresponding decrease in the percentage of growing ones (control 28.9±0.2%, PACAP38 7.5±3.4%, P=0.049) was detected 30 minutes after treatment with PACAP38 (Fig. 2C).
For a more detailed morphological analysis, dendritic spines were categorized according to their shape (mushroom, thin and stubby) using a highly validated classification method (see Materials and Methods section). As shown in Fig. 2D, treatment with PACAP38 induced a significant reduction in the proportion of mushroom spines and a concomitant significant increase in the proportion of thin and stubby ones (mushroom: P=0.00009, GFP + PACAP38 vs GFP; stubby: P=0.0002, GFP + PACAP38 vs GFP; thin: P=0.002, GFP + PACAP38 vs GFP). No significant change in the mean spine density was found (GFP 4.108±0.329 spines/10 μm; GFP + PACAP38 4.141±0.498 spines/10 μm; P=0.89).
Inhibition of ADAM10 activity by treatment with TIMP-1 blocks the effects of PACAP38 on the ADAM10–GluR1 pathway and dendritic spine morphology
The results described above suggest that PACAP38 induces a profound modification of dendritic spine morphology in hippocampal neurons correlated with an alteration of the synaptic localization of ADAM10 and GluR1 subunits of the AMPA receptor. Interestingly, we have recently demonstrated that treatment of hippocampal neurons with the ADAM10-specific inhibitor TIMP-1, which was previously shown to block the ADAM10-dependent pathway downstream of PAC1 receptor activation (Kojro et al., 2006), induced a substantial increase in GluR1 localization in Shank-positive clusters (Malinverno et al., 2010).
Based on these considerations, we treated hippocampal neurons with PACAP38 in the presence or absence of TIMP-1. As expected, incubation with TIMP-1 did not affect ADAM10 localization at the postsynaptic compartment in PACAP38-treated hippocampal neurons (TIMP-1 + PACAP38 vs PACAP38, 6.3±11.9%, P=0.68), indicating that TIMP-1 inhibits ADAM10 activity without affecting its subcellular localization. However, co-treatment of hippocampal neurons with PACAP38 + TIMP-1 led to a significant rescue of GluR1 colocalization with Shank-positive postsynaptic clusters (Fig. 3A,B; TIMP-1 + PACAP38 vs PACAP38, 30.7±10.3%, P=0.02; TIMP-1 + PACAP38 vs control, P=0.896).
We then performed morphological analysis of primary hippocampal neurons after treatment with PACAP38 in the presence or absence of TIMP-1. Statistical analysis revealed a significant increase of spine head width in neurons treated with PACAP38 + TIMP-1 compared with PACAP38-treated ones (Fig. 3C,D; P=0.0002, PACAP38 vs control; P=0.03, TIMP-1 + PACAP38 vs PACAP38; P=0.059, TIMP-1 + PACAP38 vs control).
Cumulative frequency plots of spine head width confirmed a significant shift towards smaller spine size in the presence of PACAP38 that was rescued in the presence of TIMP-1 (Fig. 3E). No significant changes in the mean spine density were found (control 3.087±0.299 spines/10 μm, PACAP38 2.820±0.118 spines/10 μm, PACAP38 + TIMP-1 3.0792±0.140 spines/10 μm; P=0.366, PACAP38 vs control; P=0.132, PACAP38 + TIMP-1 vs PACAP38; P=0.97, PACAP38 + TIMP-1 vs control).
Transfection of the ADAM10 cleavage-defective N-cadherin construct blocks any effect of PACAP38 on spine morphology
Interfering with ADAM10 activity at synapses is sufficient to induce a significant decrease in ADAM10-mediated N-cadherin cleavage, leading to accumulation of N-cadherin full length (FL) and to a significant enlargement of dendritic spine head width (Malinverno et al., 2010). Accordingly, we checked whether the morphological effects induced by PACAP38 could be correlated to an alteration of ADAM10-dependent N-cadherin cleavage. By using an antibody raised against N-cadherin C-terminal intracellular domain (C-32), we detected two main immunoreactive bands corresponding to the N-cadherin FL and to its C-terminal fragment (CTF), which derives from ADAM10-mediated cleavage (Uemura et al., 2006; Malinverno et al., 2010). As shown in Fig. 4A, treatment with PACAP38 induced a significant decrease in the N-cadherin FL/CTF ratio (PACAP38 control, −34.7±9.7%, P=0.02) that was completely rescued by co-treatment with TIMP-1 (TIMP-1 + PACAP38 vs PACAP38, P=0.001; TIMP-1 + PACAP38 vs control 4.5±10.6%, P=0.68). These results suggest an involvement of ADAM10-dependent N-cadherin cleavage in the morphological outcome of the treatment with PACAP38. Accordingly, treatment with PACAP38 and subsequent morphological analysis were repeated in neurons transfected with GFP, GFP–N-cadherin wild-type (wt) or with the cleavage-defective GFP–N-cadherin GD construct, mutated in the ADAM10 cleavage site (2006; Malinverno et al., 2010). Both N-cadherin constructs had the same expression pattern at the postsynaptic site in hippocampal neurons as demonstrated by colocalization analysis with Shank-positive clusters (data not shown). Statistical analysis revealed a significant increase of spine head width in neurons treated with GFP–N-cadherin GD (Fig. 4B,C; P=0.002, N-cadherin GD control vs GFP control) but not in those treated with GFP–N-cadherin wt, thus confirming that blocking ADAM10-dependent cleavage of N-cadherin leads per se to an increase of spine size (see also Malinverno et al., 2010). Treatment with PACAP38 induces a significant decrease in spine head width in neurons transfected with GFP (Fig. 4B,C; P=0.02 GFP + PACAP38 vs GFP control) or with N-cadherin wt (Fig. 4B,C; P=0.0003 N-cadherin wt + PACAP38 vs N-cadherin wt control) but not in those transfected with GFP N-cadherin GD (Fig. 4B,C; P>0.05 N-cadherin GD + PACAP38 vs N-cadherin GD control), thus indicating that blocking ADAM10-dependent cleavage of N-cadherin prevents any effect of PACAP38 on dendritic spine size.
Morphological regulation of dendritic spines at excitatory synapses represents one of the main relevant and efficient cellular mechanisms involved in the induction of different forms of plasticity within the central nervous system (Bourne and Harris, 2011). Accordingly, formation, stabilization and elimination of dendritic spines are highly organized and complex mechanisms based on a variety of cellular and molecular regulatory steps. Here we demonstrate that treatment of hippocampal neurons with PACAP38 leads to a significant modulation of dendritic spine morphology dependent on activation of the ADAM10–N-cadherin signaling pathway. In fact, co-treatment with the ADAM10-specific inhibitor TIMP-1 or transfection of N-cadherin GD construct, mutated in the ADAM10 cleavage site, are both sufficient to block any morphological effect observed after incubation with PACAP38. Notably, PACAP38-dependent shrinkage of dendritic spines is strictly paralleled by a reduction of AMPA receptor GluR1 subunit localization at synaptic sites.
PACAP38 induces a form of LTD in hippocampal neurons that depends on Rap-1 and p38-MAPK activation and on synaptic levels of AMPA receptors, and that was mutually occluded by the Rap guanine nucleotide exchange factor (Epac) LTD (Ster et al., 2009). However, we did not observe any effect of the Epac activator (8pCPT-2Me-cAMP, 200 μM) on ADAM10 colocalization with the postsynaptic marker Shank compared with controls (−1.91±6.78%, P=0.8464), further indicating that the Epac–p38 pathway is not involved in PACAP38-dependent modulation of ADAM10 function.
In conclusion, our findings demonstrate that the neuropeptide PACAP38 plays a key role in the modulation of dendritic spine morphology through activation of the ADAM10–N-cadherin–AMPA receptor signaling pathway in hippocampal neurons.
Materials and Methods
Antibodies and reagents
The following monoclonal antibodies were used: anti-GFP and anti-pan Shank, purchased from NeuroMab (Davis, CA). Anti-ADAM10 N-terminal polyclonal antibody was purchased from Abcam (Cambridge, MA) and anti-GluR1 polyclonal antibody from Millipore (Billerica, MA); Alexa Fluor secondary antibodies were purchased from Invitrogen (Carlsbad, CA). TIMP-1 and PACAP38 were purchased from Calbiochem (Darmstadt, Germany). p38 mitogen-activated protein kinase inhibitor (SB203580), VIP antagonist [VIP(6-28)], EPAC activator (8CPT-2Me-cAMP) and protein kinase A inhibitor (H89) were purchased from Tocris (Bristol, UK).
The plasmid encoding for GFP-tagged N-cadherin was kindly provided by Maria Passafaro (Milano, Italy); the cleavage-defective (N-cadherin GD) construct was created by site-direct mutagenesis of R714G and I715D. Transfection assay in COS7 cells was performed to verify that the N-cadherin GD mutant form leads to a dramatic decrease in the formation of the C-terminal fragment (CTF) products, thus indicating a significant decrease in N-cadherin cleavage (data not shown).
Neuronal cultures preparation and transfection
Hippocampal neuronal primary cultures were prepared from embryonic day 18–19 (E18–E19) rat hippocampi as previously described (Piccoli et al., 2007). Neurons were transfected between DIV7 using the calcium-phosphate method.
For morphological studies, transfected neurons were fixed in 4% paraformaldehyde with 4% sucrose at room temperature and immunostained for GFP; primary and secondary antibodies were applied in GDB buffer (Sala et al., 2001) (30 mM phosphate buffer, pH 7.4, containing 0.2% gelatin, 0.5% Triton X-100 and 0.8 M NaCl). Cells were chosen randomly for quantification. Fluorescence images were acquired using a Zeiss confocal LSM510 system with a 63× objective and a sequential acquisition setting at 1024×1024 pixels resolution; for each image two to four 0.5 μm sections were acquired and a z-projection was obtained.
For colocalization studies, hippocampal neurons were fixed in methanol at −20°C and immunostained for GluR1, ADAM10 and Pan-Shank; primary and secondary antibodies were applied in GDB buffer. Cells were chosen randomly for quantification.
Time-lapse images were obtained in an environmentally controlled chamber with 5% CO2 at 37°C using a Zeiss Confocal LSM510 (a gift from Fondazione Monzino, Milano, Italy) system with 63× objective and zoom function set on 4; for each image, three to four 1 μm sections were acquired and a z-projection was obtained. Images of non-treated cells used as control were acquired every 3 minutes for 30 minutes; after injection of PACAP38 images were acquired every 3 minutes for 30 minutes. Morphological analysis was conducted at each time point.
Quantification and statistical analysis
Quantification of western blotting analysis was performed with ImageJ software and values were expressed as means ± s.e.m. Colocalization analysis of confocal experiments was performed using Zeiss AIM 4.2 software. Analysis of dendritic spine morphology was performed with ImageJ software; for each dendritic spine length, the head and neck width were measured, which was used to classify dendritic spines into three categories (thin, stubby and mushroom) (see also Harris et al., 1992), as shown in Fig. 2D. In particular, the length and the ratio between the width of head and the width of neck (Wh/Wn) were used as parameters for the classification as follows: protrusions having a length of more than 3 μm were considered as filopodia, the others as spines; spines with a Wh/Wn ratio bigger than 1.7 were considered mushrooms; spines with a Wh/Wn ratio smaller than 1.7 were divided in stubby, if shorter than 1 μm, and thin if longer than 1 μm. Statistical evaluation of all confocal experiments was performed by using one-way ANOVA, followed by Bonferroni's post-hoc test. An operator who was ‘blind’ to the experimental conditions performed both image acquisition and quantification.
This work was supported by Fondazione Cariplo [grant numbers 2319-2008 to M.D.L. and 264-2009 to C.S.].