Direct and indirect interactions between cannabinoid CB1 receptor and group II metabotropic glutamate receptor signalling in layer V pyramidal neurons from the rat prefrontal cortex

Preparation of brain slices

Coronal slices (300 µm) were made from 15 to 20 days Sprague–Dawley rats as in Auclair et al. (2000) with slight modifications. After inhalation anaesthesia with halothane, rats were decapitated, and the brain was rapidly removed and placed in ice-free (0–1 °C) normal saline solution. Cutting was performed in 0–2 °C solution composed of (in mm): 120 choline Cl, 2 KCl, 8 MgSO₄, 1 KH₂PO₄, 26 NaHCO₃, 0.1 CaCl₂, osmolality adjusted to 330 mOsm/kg with choline Cl, pH 7.4 when bubbled with 95% O₂ : 5% CO₂. Slices were kept until use at room temperature in normal saline composed of (in mm): 126 NaCl, 2 KCl, 2 CaCl₂, 2 MgSO₄, 1.15 KH₂PO₄, 26 NaHCO₃, osmolality adjusted with glucose to 300 mOsm/kg, pH 7.4 when bubbled with 95% O₂ : 5% CO₂.

Recordings

Pyramidal neurons in layer V were visually identified with infrared Nomarski optics using a ×40 water immersion objective. Synaptic currents were recorded in whole-cell voltage clamp at a holding potential of −70 mV with an Axopatch 200A (Axon Instruments, Union City, CA, USA). Normal saline was added with bicuculine methochloride (1 µm) to isolate excitatory postsynaptic currents (EPSCs) (Auclair et al., 2000). No series resistance compensation was used. Data were usually filtered at 2 kHz. Internal solution was composed of (in mm): 100 potassium gluconate, 40 KCl, 10 Hepes, 8 NaCl, 2 Mg-ATP, 0.5 ethylene glycol bis-(beta-aminoethyl ether)-N,N,N′,N′-tetraacetic acid, pH 7.3, osmolality 300 mOsm/kg. Excitatory synaptic currents (EPSCs) were evoked with a glass monopolar electrode placed in layer V, 20–40 µm from the soma and 10–20 µm laterally from the apical dendrite. EPSCs were evoked during a −10 mV hyperpolarizing voltage step to monitor cell capacitance and input resistance. When these parameters changed by more than 20% in a recording session, the cell was discarded from analysis. Agonists and HFS consisting of four 1 s, 100 Hz trains separated by 10 s were applied 5–8 min after break-in to avoid cell wash that interfered with the experiments. HFS were delivered in current-clamp mode at the cell resting potential (−70 to −65 mV). Recordings were performed at room temperature (24–27 °C). Paired pulses ratios were calculated as the ratio between the amplitude of the second EPSC against that of the first for individual current traces and monosynaptic EPSCs.

Immunoblot analysis

Detection of activated extracellular signal-regulated kinases (ERK) was performed as in Otani et al. (1999). Frozen brain slices were homogenized at 4 °C in a lysis buffer containing 1% Nonidet P-40, 1% deoxycholic acid, 0.1% SDS, 158 mm NaCl, 1 mm phenylmethylsulphonyl fluoride, 1 mm Na₃VO₄, 10 mm Tris, pH 7.8. The protein concentration was determined with Bradford microassays. Equal amounts of protein from lysates (50 µg) were separated by electrophoresis on SDS-10% polyacrylamide gel and transferred onto a nitrocellulose membrane. Membranes were incubated for 2 h at room temperature with 0.25% gelatin in Tris-buffered saline containing 0.05% Tween to block nonspecific binding, and then for 2 h with the polyclonal antiactive ERK antibody at 1 : 20 000 (Promega, Madison, WI, USA), which recognizes dually phosphorylated type 2 ERK/p42. After washings with Tris-buffered saline added with Tween, membranes were incubated with horseradish peroxidase-conjugated goat antirabbit IgG (Dako, Glostrup, Denmark) at a dilution of 1 : 5000, then washed with blocking buffer. Proteins were visualized after chemiluminescence staining (ECL, Amersham, Arlington Heights, IL, USA). Immunoreactive bands were analyzed by densitometry using NIH image software.

Reprobing with specific antibodies was performed after incubation for 2 h at 65 °C in 200 mm glycine, pH 2.5, 1% SDS, followed with two washes with 1 m Tris-HCl, pH 8 and one with Tris-buffered saline added with Tween. The membrane was then incubated with polyclonal anti type 2 ERK C-14 antibody (1 : 1000, Santa Cruz Biotechnology, Santa Cruz, CA, USA), which recognizes type 2 ERK/p42 and weakly type 1 ERK/p44.

Immunocytochemistry of activated ERK in brain sections

Twenty minutes following intraperitoneal treatment with saline, THC, SR 141716A or MK801, mice were deeply anaesthetized by intraperitoneal injection of pentobarbital (Sanofi, Montpellier, France) and brains were fixed by intracardiac perfusion of 4% paraformaldehyde in 0.1 m Na₂HPO₄/NaH₂PO₄ buffer, pH 7.5, delivered with a peristaltic pump at 20 mL/min during 5 min. Brains were postfixed overnight in the same fixative solution. Sections (30 µm) were cut with a vibratome (Leica, Paris, France) and then kept in a solution containing 30% ethylene glycol, 30% glycerol, 0.1 m phosphate buffer, and 0.1% diethyl pyrocarbonate (Sigma, Deisenhofen, Germany) at −20 °C until they were processed for immunocytochemistry.

The immunocytochemical procedure was previously described (Valjent et al., 2001). For detection of phosphorylated proteins, 0.1 mm NaF was included in all buffers and incubation solutions. Day 1: free-floating sections were rinsed in Tris-buffered saline (0.25 m Tris and 0.5 m NaCl, pH 7.5), incubated for 5 min in Tris-buffered saline containing 3% H₂O₂ and 10% methanol, and then rinsed three times (10 min each) in Tris-buffered saline. After a 15-min incubation in 0.2% Triton X-100 in Tris-buffered saline, the sections were rinsed three times in Tris-buffered saline, and incubated with the primary antibody (antiphospho Thr202–Tyr1204 type 1 ERK, cat-9101S; Cell Signal Technology, Beverly, MA, USA) (dilution of 1 : 400) overnight at 4 °C. Day 2: after three rinses in Tris-buffered saline, sections were incubated for 2 h at room temperature with the secondary biotinylated antiboby (anti-IgG), using a dilution twice that of the first antibody in Tris-buffered saline, and then incubated for 90 min in avidin-biotin-peroxidase complex (ABC) solution (final dilution, 1 : 50; Vector Laboratories, Peterborough, UK). The sections were then washed in Tris-buffered saline and twice in Tris-buffered saline (0.25 m Tris, pH 7.5) for 10 min each, placed in a solution of Tris-buffered saline containing 0.1% 3,3′Δdiaminobenzidine (50 mg/100 mL), and developed by H₂O₂ addition (0.02%). After processing, tissue sections were mounted onto gelatin-coated slides and dehydrated through alcohol to xylene for light microscopic examination.

ERK-positive neurons were plotted at ×10 magnification using a computerized image analyzer (Biocom, France). Cell counts were performed, for each animal, in the whole prefrontal cortex divided into layer II/III and layer V. In each layer, the total amount of ERK-positive neurons, evaluated on the basis of a cytoplasmic and nuclear staining, was counted. The mean number of ERK-positive neurons is given per brain from four to five different animals.

Drugs

SR 141716A was a generous gift from Sanofi Recherche (Montpellier, France), WIN 55,212–2 (RS)-alpha-methylserine-O-phosphate monophenyl ester (MSOPPE), the MAP kinase inhibitor PD98059 and (2S,2′R,3′R)-2-(2′,3′-dicarboxycyclopropyl) glycine (DCG IV) were purchased from Tocris (Illkirch, France). Stock solutions were made in water except for SR 141716A and WIN 55,212-2, which were dissolved at 10 mm in dimethyl sulfoxide. Dimethyl sulfoxide (0.1%) had no detectable effects on intrinsic properties and synaptic currents of pyramidal neurons. For intraperitoneal injections, Δ⁹-tetrahydrocannabinol (THC) (Sigma, Saint Quentin-Fallavier, France) was dissolved in a 5% ethanol, 5% cremophor and 90% distilled water solution, MK801 (Sigma, Saint Quentin-Fallavier, France) was dissolved in 0.9% NaCl and SR 141716A was dissolved in a solution of 10% ethanol, 10% cremophor and 80% distilled water. SR 141716A (3 mg/kg) and MK801 (1 mg/kg) were administrated 15 min before THC intraperitoneal injections (1 mg/kg).

Group II mGlu induce a long-lasting depression of synaptic transmission at proximal synapses of layer V pyramidal neurons

We previously reported that the group II mGlu agonist (2S,2′R,3′R)-2-(2′,3′-dicarboxycyclopropyl) glycine (DCG IV) induces a long-lasting depression of synaptic transmission at distal synapses from layer V pyramidal neurons of the neocortex (Otani et al., 1999, 2002). The induction involves postsynaptic factors including postsynaptic group II mGlu linked to phospholipase C and intracellular calcium increases (Otani et al., 2002). In the present study, we examined the interactions between group II mGlu and CB1 receptors at proximal synapses from the same neurons. These synapses were chosen to allow comparisons with our previous study on CB1 receptors at these synapses (Auclair et al., 2000). DCG IV (100 nm) was bath applied for 10 min after a 5–8 min control period. In six out of six cells, DCG IV reduced the initial EPSC slope to 38.2 ± 12.4% (Fig. 1A and B). EPSC amplitudes were 90.2 ± 8.7 pA in control and 43.5 ± 5.3 pA in the presence of DCG IV. No change in EPSC kinetics was observed (Fig. 1A). The depression persisted during the wash-out of DCG IV for at least 30 min and up to 50 min. Postsynaptic calcium was needed in the DCG IV-induced long-term depression. In agreement, the inclusion of the calcium chelator, bis-(O-aminophenoxy)-N,N,N′,N′-tetraacetic acid (BAPTA) 30 mm in the internal pipette solution prevented the long-lasting component of the depression (Fig. 1C, n = 5 cells). Using BAPTA, the amplitude of the transient DCG IV-induced depression measured 5 min after the application of DCG IV was not statistically different from the amplitude of the control depression (Mann–Whitney U-test P > 0.1). However, the long-lasting component of the DCG IV-induced depression was inhibited with BAPTA (Mann–Whitney U-test P < 0.015, Fig. 1C). We conclude that the transient activation of group II mGlu induces a long-lasting depression of proximal excitatory synapses that depends on postsynaptic calcium.

Group II mGlu have been shown to mediate presynaptic changes in synaptic transmission and paired pulse ratios (Macek et al., 1996; Lin et al., 2000). In contrast, the induction of the DCG IV-induced depression in the neocortex needed postsynaptic calcium. However, the locus of expression of this depression remained uncertain. We thus tested for a presynaptic expression of DCG IV-induced long-lasting depression using paired-pulse experiments. Paired-pulse ratios were calculated from the peak current amplitude of monosynaptic EPSCs (Fig. 1D–G). In control conditions, mean paired-pulse ratio was 87.7 ± 14.4% (n = 5 cells, Fig. 1D). Upon application of DCG IV, paired-pulse ratio was increased to 200.3 ± 20.1%, while EPSC amplitude was reduced (n = 5 cells, Mann–Whitney U test P < 0.01, Fig. 1D–G). Furthermore, changes in paired-pulse ratios and EPSC amplitudes were correlated upon wash-out of DCG IV in an untypical cell showing a transient effect of DCG IV (Fig. 1E). These data suggest that the DCG IV-induced synaptic depression is expressed presynaptically.

Presynaptic interactions during activation of CB1 receptors and group II mGlu

The apparent discrepancy between a postsynaptic induction and a presynaptic expression of DCG IV-induced depression suggested that a retrograde signalling might be involved. Several studies suggest that mGlu activation releases endocannabinoids that inhibit synaptic transmission (Varma et al., 2001; Maejima et al., 2001; Ohno-Shosaku et al., 2002; Robbe et al., 2002a). We thus tested if the CB1 antagonist SR 141716A blocked the depression induced by DCG IV. Slices were pretreated with SR 141716A (2 µm) for 30 min to fully block presynaptic CB1 receptors (Auclair et al., 2000). In control experiments, SR 141716A applied after a test period did not change the passive properties of cells but increased the amplitude of EPSCs by 45.2 ± 4.8% (n = 4 cells) in agreement with Auclair et al. (2000). In the presence of SR 141716A (2 µm), mean EPSC amplitude was 186.5 ± 53.4 pA (n = 4 cells) and DCG IV induced a long-lasting depression to 23.0 ± 2.7% (n = 4 cells; Fig. 2A). Thus, we conclude that endocannabinoids were not involved in the long-lasting synaptic depression induced by DCG IV.

The CB1 receptor agonist WIN 55,212-2 (1–5 µm) also mediates a long-lasting depression of synaptic transmission at proximal synapses of layer V pyramidal neurons (Auclair et al., 2000). Application of WIN 55,212–2 (1–5 µm) induced a long-lasting reduction in EPSC amplitude of −30.0 ± 8.0% (n = 6 cells, data not illustrated). This depression was completely blocked by SR 141716A indicating that it is mediated by CB1 receptors. However, a novel non-CB1 receptor as that found in the hippocampus may be involved (Hajos et al., 2001). This long-lasting depression recalls that induced by DCG IV, as both are expressed presynaptically. The negative coupling to adenylyl cyclase of the CB1 receptor and group II mGlu supports the recent finding that both receptors compete for G-proteins (Vasquez & Lewis, 1999). We thus tested for an occlusion between the depressions induced by WIN 55,212–2 and DCG IV. The continuous presence of WIN 55,212–2 (2 µm) partially inhibited the initial phase of the depression induced by DCG IV (Mann–Whitney U-test, P < 0.05; Fig. 2B). Although reduced in amplitude, the long-term depression induced by DCG IV was not inhibited by WIN 55,212–2. EPSC amplitudes were 71.5 ± 10.8 pA and 49.8 ± 5.7 pA in the presence of WIN 55,212–2 and WIN 55,212–2 + DCG IV, respectively. WIN 55,212–2 (1–2 µm) did not induce detectable inward current, nor changed the bioelectrical properties of recorded cells. Furthermore, the interaction between WIN 55,212–2 and the DCG IV-induced depression could not be due to an indirect release of glutamate as WIN 55,212–2 had a depressant effect on synaptic transmission in the presence of the group II mGlu antagonist MSOPPE (200 µm). In six out of six cells, WIN 55,212–2 applied in the presence of MSOPPE depressed the EPSC initial slope to 54.2 ± 3.5% (n = 6 cells, data not shown). This was not significantly different from the effect of WIN 55,212-2 alone (Mann–Whitney U test, P > 0.5). Taken together, these data strongly support that WIN 55,212-2 has a direct depressant effect on synaptic transmission that interacts with the transient depression induced by DCG IV leading to a reduced depression in both the initial and late phases.

Conversely, when slices were preincubated with DCG IV (100 nm) for 30 min, WIN 55,212-2 (2 µm) did not reduce EPSC amplitudes (Mann–Whitney U-test, P > 0.3, n = 8 cells, Fig. 2C). DCG IV was shown previously to be effective in reducing evoked EPSCs in 100% of cells. Furthermore, the mean EPSC amplitude was 96.1 ± 9.0 pA in control conditions (n = 25) and 50.8 ± 5.4 pA (n = 17 cells) in the presence of DCG IV, with similar stimulus conditions (Mann–Whitney test, P < 0.005). Thus the effect of DCG IV was present and stationary when WIN 55,212-2 was applied. Altogether, these data further suggest an interaction between the depressions induced by activation of CB1 receptors and group II mGlu.

Postsynaptic interactions during activation of group II mGlu and CB1 receptors

Previous data suggested a presynaptic interaction between CB1 receptor and group II mGlu. In the next series of experiments, we wished to examine a potential postsynaptic interaction between these receptors. Previous data from our laboratory demonstrated that DCG IV activated mitogen activated protein kinase of the extracellular signal-regulated kinase (ERK) subfamily via postsynaptic group II mGlu (Otani et al., 1999, 2002). Thus, we sought to determine if CB1 receptors could interact with these receptors. ERK activities were assayed using antiactive ERK antibodies on slice lysates. Slices were previously incubated in oxygenated normal saline at 28 °C in the presence of WIN 55,212-2 (1 µm) or DCG IV (100 nm) or both for 2 min. When WIN 55,212-2 was applied alone, type 2 ERK activity increased to 182.0 ± 31.0% compared to control levels (n = 4, Mann–Whitney U-test, P < 0.05, Fig. 3A). Thus, activation of CB1 receptors clearly mediated type 2 ERK phosphorylation. In the next series of experiments, slices incubated with DCG IV gave similar results and ERK phosphorylation was increased to 176.8 ± 20.9% (n = 4, Mann–Whitney U-test, P < 0.05, Fig. 3A). ERK phosphorylation was not saturated when increases in the range of 200% were observed. When slices were incubated with WIN 55,212-2 or DCG IV for 5 min, instead of 2 min, ERK activities were further enhanced to values above 340%. Thus, type 2 ERK immunoreactivities were not saturated with 2 min incubations. However, when slices were incubated with WIN 55,212-2 + DCG IV for 2 min, type 2 ERK activity was only increased to 165.0 ± 23.3% (n = 4, Fig. 3A). This value was not significantly different from those obtained with WIN 55,212-2 or DCG IV alone (Mann–Whitney U-test, P > 0.3). Similar results were obtained with type 1 ERK activities (Fig. 3A). Type 1 ERK activity was increased to 225.0 ± 37.7% (n = 5), 193.2 ± 48.45% (n = 5) and 221.8 ± 39.9% (n = 5) in the presence of WIN 55,212-2, DCG IV or both, respectively. Again, there was no statistical difference between type 1 ERK phosphorylation when either WIN 55,212-2, DCG IV or both were used (Mann–Whitney U-test, P > 0.3). Taken together, these data suggest an occlusion between WIN 55,212-2 and DCG IV for ERK activation in the prefrontal cortex.

Immunocytochemical localization of activated ERK

In order to confirm that activated ERK were localized in pyramidal neurons, immunocytochemistry was performed using an antibody that recognizes the active form of type 2 ERK (Valjent et al., 2001). Animals were treated with intraperitoneal injections of Δ⁹-tetrahydrocannabinol (THC) 1 mg/kg, or vehicle. After intraperitoneal injection of pentobarbital, brains were fixed with intracardiac perfusion. In vehicle treated animals, a postsynaptic labelling was observed in some cells (Fig. 3B). Such a basal ERK activity is usually observed in various brain structures (Valjent et al., 2001). In particular, such a basal activity is frequently seen in layer II and probably relies on a basal neurogenic activity. At basal level, the number of activated ERK-positive neurons was 25 ± 9 cells (n = 4 animals) and 29 ± 6 cells (n = 4 animals) in layers II/II and V, respectively. In THC-treated animals (1 mg/kg for 20 min), the number of positive neurons was increased to 560.3 ± 228.2% (n = 4) and 302.4 ± 86.4% (n = 4) in layers II/II and V, respectively (Fig. 3B). A comparative analysis of ERK-positive neurons was performed at various times after THC administration (10, 20, 30 and 60 min). The number of positive neurons increased between 10 and 20 min with a maximum number of 70 ± 7 cells (n = 4) and 72 ± 7 cells (n = 4) in layers V and II/III, respectively. These values were significantly different from those from vehicle treated animals (Mann–Whitney U-test, P < 0.01). The number of positive neurons fell to 26 ± 8 cells (n = 4) and 59 ± 5 cells (n = 4) 30 min after THC treatment. Thus, following experiments were performed 20 min after THC treatment.

To control that positive neurons resulted from activation of CB1 receptors, animals were pretreated with an intraperitoneal injection of SR 141716A (3 mg/kg) 15 min before THC administration (n = 5). The number of positive neurons was reduced to basal levels and were not different from those observed in vehicle-treated animals (Mann–Whitney U-test, P > 0.3). These data clearly show that activation of CB1 receptors in the prefrontal cortex leads to the phosphorylation of ERK in pyramidal neurons. Thus, our results obtained from quantitative immunoblot analysis likely reflect phosphorylation of ERK in the cytosol of pyramidal neurons.

As postsynaptic CB1 receptors were not detected in neocortical pyramidal neurons (Egertova et al., 1998), the postsynaptic interaction between CB1 receptors and group II mGlu rather seems indirect. The postsynaptic phosphorylation of ERK by THC could involve a release of glutamate following CB1 receptor activation (Palazzo et al., 2001). To test for this hypothesis, the N-methyl-d-aspartate (NMDA) receptor antagonist MK801 was injected in some animals (1 mg/kg), 15 min prior to the injection of THC. In these experiments, MK801 reduced the number of positive layer V neurons to 50.3 ± 12.9%, compared to THC treated animals (Mann–Whitney U-test, P < 0.01, n = 4 animals). The numbers of positive neurons were 97 ± 9 (n = 4) and 45 ± 9 (n = 4) for THC and THC + MK801 treated animals, respectively. This effect was restricted to layer V, as MK801 had no effect on the number of positive neurons in layer II/III (81 ± 10, n = 4 and 87 ± 10, n = 4, respectively, Mann–Whitney U-test, P > 0.3). The effect of MK801 could not be attributed to an independent effect on the basal number of ERK positive neurons as injection of MK801 alone was without effect (Valjent and Caboche, unpublished data). We conclude that postsynaptic ERK are activated in neocortical pyramidal neurons following CB1 receptor activation. This activation required NMDA receptors, in agreement with a release of glutamate in the local neuronal network. Consequently, CB1 receptors might interact indirectly with group II mGlu in the phosphorylation of ERK by releasing glutamate on layer V pyramidal neurons.

Are group II mGlu involved in long-term synaptic plasticity?

In the view of the preceding results and the role of ERK in synaptic plasticity (Sweatt, 2001), we asked whether CB1 receptors and group II mGlu could interact during long-term changes in synaptic efficacy following HFS. We previously showed that CB1 receptors interfered with LTP and LTD at layer V excitatory synapses (Auclair et al., 2000). However, the involvement of group II mGlu and their possible interactions with CB1 receptors were never investigated.

Group II mGlu have been implicated in the induction of various forms of long-term synaptic plasticity notably in the hippocampus (Huang et al., 1997; Manahan-Vaughan, 1997). Thus, we sought to determine whether group II mGlu could modulate long-term plasticity induced by HFS in our model. Cells were continuously bathed in the presence of the group II mGlu antagonist MSOPPE (200 µm) and voltage-clamped at −70 mV. Control EPSCs were recorded during a 5–10-min period to minimize wash-out. HFS were delivered in current-clamp at the cells resting potential (≈ −65 to −70 mV). In six out of 15 cells, a strong LTD was observed and the mean data is illustrated in Fig. 4A. One cell showed LTP and the eight remaining cells showed no plasticity (Fig. 4B). These results contrast with the data obtained in control conditions, i.e. in the absence of MSOPPE, in which LTP and LTD were observed with approximately the same frequencies (Fig. 4B and Auclair et al., 2000). In particular, the percentage of cells showing LTP was significantly different compared to control conditions (χ², P < 0.05, Fig. 4B). Thus, these data show that MSOPPE reduced the number of cells expressing LTP by the blockade of group II mGlu.

To further support these results, the effect of DCG IV on synaptic plasticity was studied. Slices were preincubated with DCG IV (100 nm) for at least 15 min. In the presence of DCG IV, a robust 50–150% post-tetanic potentiation was always observed immediately following HFS (Fig. 4C). In these conditions, five cells out of 15 expressed LTP, three cells LTD and the seven remaining cells showed no plasticity. These percentages were not statistically different from control data (χ², P > 0.2).

Overall, the data suggest a role for group II mGlu activated by HFS in synaptic plasticity. However, the pharmacological activation of group II mGlu by DCG IV had no obvious effects on LTD or LTP. As will be discussed later, this may also be due to activation of both pre- and postsynaptic group II mGlu with opposite effects.

Do CB1 receptors interact with group II mGlu in HFS-induced synaptic plasticity?

We next searched for interactions between group II mGlu and CB1 receptors. In a first series of experiments, HFS were delivered in the presence of both WIN 55,212-2 and MSOPPE. In these conditions, seven out of 10 cells showed no plasticity, three cells showed LTD, while no cell showed LTP (Fig. 5). The proportions of cells showing LTD or no plasticity were greater than those obtained in the presence of MSOPPE or WIN 55,212-2 alone (χ² test, P < 0.025, Fig. 5B). These data fit with the generalized Hebbian scheme of plasticity according to Sejnowski (1977), Bienenstock et al. (1982) and Cho et al. (2001) (Fig. 5C). This model explains why nonplastic cells coexist with cells expressing LTP or LTD in control conditions. According to this model, decreasing postsynaptic calcium during HFS results in an increase in LTD expressing cells and a decrease in LTP expressing cells. Furthermore, a larger decrease in postsynaptic calcium results in a marked increase in nonplastic cells at the expense of plastic cells (Fig. 5B). Thus, the large proportion of nonplastic cells when MSOPPE and WIN 55,212-2 were applied together suggests that both drugs have additive effects on postsynaptic calcium.

Finally, synaptic plasticity was tested in the presence of MSOPPE and SR 141716A. In agreement with both drugs showing opposite effects on synaptic plasticity, the number of cells expressing LTD or LTP in the presence of MSOPPE and SR 141716A were intermediate between values obtained with MSOPPE or SR 141716A alone (Fig. 5B). In particular, the proportions of cells expressing LTD or LTP were significantly different when SR 141716A was applied alone or in the presence of MSOPPE (χ² test, P < 0.05, Fig. 5B). These data further support that the actions of group II mGlu and CB1 receptors are additive on layer V long-term changes in synaptic efficacy.

Interaction between group II mGlu and CB1 receptors in the long-lasting depression of proximal synapses

DCG IV was shown to induce a long-lasting depression of proximal synapses of layer V pyramidal neurons. The expression was presynaptic, in agreement with group II mGlu-dependent depression found in other central nervous system structures including amygdala, nucleus accumbens and hippocampus (Macek et al., 1996; Lin et al., 2000; Robbe et al., 2002b). However the long-lasting component of the depression required postsynaptic calcium. When BAPTA was used, a transient presynaptic depression was observed, suggesting the involvement of presynaptic group II mGlu in this effect. In agreement, this transient depression was associated with changes in paired-pulse ratio.

The long-lasting depression induced by WIN 55,212-2 was shown previously to be presynaptic and no CB1 receptors were detected on the somata of pyramidal neurons (Auclair et al., 2000; Egertova et al., 1998). The action of WIN 55,212-2 was not due to an indirect activation of group II mGlu by a local release of glutamate. Furthermore, the activation of ERKs by WIN 55,212-2 was not necessary for the induction of the long-lasting depression. Therefore, the long-lasting depression induced by WIN 55,212-2 most likely involved presynaptic CB1 receptors.

WIN 55,212-2 and DCG IV were shown to be mutually occlusive in the induction of the long-lasting depression. As the effects of WIN 55,212-2 and DCG IV involve presynaptic CB1 receptors and presynaptic group II mGlu, respectively, the occlusion between CB1 receptors and group II mGlu is likely expressed presynaptically on the same fibres. The occlusion might involve a common signalling pathway between presynaptic CB1 receptors and group II mGlu. The involvement of ERKs was ruled-out by showing that both WIN 55,212-2 and DCG IV-induced depressions were not inhibited by the MAP kinase inhibitor PD98059 (20 µm) (depressions to 27.9 ± 2.6%, n = 3 and 42.6 ± 8.1%, n = 3, respectively, data not illustrated). Both receptors being negatively coupled to cAMP formation through G-proteins (Kemp et al., 1996; Jung et al., 1997), interactions between G-proteins is a possible candidate (Vasquez & Lewis, 1999), but other mechanisms may be involved as well.

Alternatively, postsynaptic group II mGlu and a retrograde signal might occlude presynaptic CB1 receptors. However, our data demonstrate that such a mechanism did not require endocannabinoids as a retrograde messenger, as SR 141716A was without effect on the DCG IV-induced depression. Furthermore, the transient presynaptic depression induced by DCG IV was independent of postsynaptic calcium and was also occluded by WIN 55,212-2. Thus an indirect interaction between postsynaptic group II mGlu and presynaptic CB1 receptors seems unlikely. It is thus possible that an interaction takes place presynaptically in the signalling cascades of these receptors. Such an interaction could be physiologically relevant to central nervous system actions of endocannabinoids. Recently, cannabinoid-induced antinociception was shown to be modulated by mGlu antagonists (Palazzo et al., 2001) and a direct presynaptic interaction between CB1 receptors and mGlu may explain these results (Maejima et al., 2001; Varma et al., 2001; Ohno-Shosaku et al., 2002; Robbe et al., 2002a).

Interactions between group II mGlu and CB1 receptors for the postsynaptic activation of ERK

Our data demonstrate that both group II mGlu and CB1 receptors can activate postsynaptic ERK in an occlusive manner. The activation of ERKS was shown to take place in the cytosol of pyramidal neurons. Activation by DCG IV may involve postsynaptic group II mGlu, in agreement with Otani et al. (2002), but may involve presynaptic group II mGlu as well. The lack of CB1 receptors in neocortical pyramidal neurons (Egertova et al., 1998) suggest that CB1 receptors located on glial cells, interneurons or layer II/III pyramidal neurons may rather be involved. In agreement, NMDA receptors were shown to be required in the activation of ERK in the cytosol of layer V pyramidal neurons following activation of CB1 receptors. Thus, activation of CB1 receptors in the neocortex may release glutamate, which in terms activates NMDA receptors, postsynaptic and/or presynaptic group II mGlu in layer V pyramidal neurons. How might activation of CB1 receptors release glutamate? These receptors are expressed in GABAergic interneurons in the neocortex (Marsicano & Lutz, 1999), and their activation was shown to reduce GABAergic transmission in the hippocampus (Hajos et al., 2000; Carlson et al., 2002), amygdala (Katona et al., 2001) and nucleus accumbens (Hoffman & Lupica, 2001). While, no such modulation was reported in the neocortex, activation of CB1 receptors in the neocortex might decrease the inhibition/excitation ratio and increase basal glutamate levels. However, the site of interaction in the occlusion of ERK activation by WIN 55,212-2 and DCG IV is unknown. The interaction may involve NMDA receptors, pre- and/or postsynaptic group II mGlu activated indirectly by WIN 55,212-2 and group II mGlu activated by DCG IV. In any case, there is no direct interaction between CB1 receptors and group II mGlu in this occlusion. The recruitment of NMDA receptors by CB1 receptors could account for the involvement of NMDA receptors in the cannabinoid-induced antinociception (Palazzo et al., 2001).

Involvement of group II mGlu in synaptic plasticity

Our results showed that group II mGlu were involved in HFS-dependent synaptic plasticity. While DCG IV had only a moderate effect, MSOPPE was shown to favour LTD. These results cannot be solely related to the depressant effect by group II mGlu activation. Indeed, presynaptic depression would rather favour LTD, as shown for CB1 receptors (Misner & Sullivan, 1999; Auclair et al., 2000). Conversely, antagonists for such receptors would favour LTP as observed with SR 141716A (Auclair et al., 2000; Robbe et al., 2002b). Thus the effects of MSOPPE and DCG IV were not consistent with the involvement of a DCG IV-induced presynaptic depression and pointed out to a postsynaptic effect. In agreement, postsynaptic group II mGlu are present in the neocortex (Petralia et al., 1996), where they mediate calcium increases in layer V pyramidal somata (Otani et al., 2002). Therefore, the finding that MSOPPE favours LTD is in agreement with a postsynaptic rise in calcium mediated by group II mGlu activated during HFS. Conversely, the small effects of DCG IV would be best explained by an action on both pre- and postsynaptic group II mGlu with opposite effects on plasticity.

Lack of interactions between group II mGlu and CB1 receptors in HFS-dependent synaptic plasticity

No interaction between group II mGlu and CB1 receptors was found in the modulation of long-term plasticity changes induced by HFS. These results were based on the additive effects of WIN 55,212-2 and MSOPPE, and MSOPPE and SR 141716A. First, coapplications of WIN 55,212-2 and MSOPPE resulted in an increase in the number of nonplastic cells. Thus, both drugs were shown to favour LTD. Second, coapplications of MSOPPE and SR 141716A gave intermediate numbers of cells expressing either LTP or LTD. We suggest that in contrast to our previous data, the effects of WIN 55,212-2, SR 141716A and MSOPPE are rather additive on HFS-dependent forms of synaptic plasticity. Thus, the direct or indirect interactions between CB1 and glutamate receptors do not seem involved in these processes. The data suggest that CB1 receptors and group II mGlu might modulate HFS-dependent forms of synaptic plasticity by independent mechanisms. This points to an independent group II mGlu-dependent signalling pathway which might include phospholipase C and inositol triphosphate (see References in Otani et al., 2002; Otani et al., 2002).

Overall, the simplest interpretation of our data suggests a presynaptic interaction in the signalling cascades involved in the long-lasting depressions induced by activation of presynaptic CB1 receptor and group II mGlu. CB1 receptors located on other cells mediate a release of glutamate that activates glutamate receptors and induces ERK phosphorylation in layer V pyramidal neurons. However, these mechanisms do not seem to play a role in the long-lasting depressions induced by activation of CB1 receptor or group II mGlu nor in HFS-dependent synaptic plasticity. Other experiments will be needed to clarify the complex actions of endocannabinoids on glutamatergic transmission in the neocortex.