Roles of phospholipase Cβ and NMDA receptor in activity-dependent endocannabinoid release

Abstract

Endocannabinoids are released from postsynaptic neurons, activate presynaptic cannabinoid receptors and cause various forms of short-term and long-term synaptic plasticity throughout the brain. Using hippocampal and cerebellar neurons, we have revealed that endocannabinoid release can be induced through two different pathways. One is independent of phospholipase Cβ (PLCβ) and driven by Ca²⁺ elevation alone (Ca²⁺-driven endocannabinoid release, CaER), and the other is PLCβ-dependent and driven by activation of G_q/11-coupled receptors (receptor-driven endocannabinoid release, RER). CaER is induced by activation of either voltage-gated Ca²⁺ channels or NMDA receptors. RER is functional even at resting Ca²⁺ levels (basal RER), but markedly enhanced by a small Ca²⁺ elevation (Ca²⁺-assisted RER). In Ca²⁺-assisted RER, PLCβ serves as a coincidence detector of receptor activation and Ca²⁺ elevation. We have also demonstrated that Ca²⁺-assisted RER is essential for the endocannabinoid release triggered by synaptic activity. Our anatomical data show that a set of receptors and enzymes required for RER are well organized so that the excitatory input can trigger RER effectively. Certain forms of spike-timing-dependent plasticity (STDP) are reported to depend on endocannabinoid signalling. The NMDA receptor and PLCβ might play key roles in the endocannabinoid-dependent forms of STDP as coincidence detectors with different timing dependences.

Our knowledge of the endocannabinoid system has been rapidly expanded in the last several years. In 2001, it was first demonstrated that endocannabinoids work as a retrograde messenger in the CNS and contribute to activity-dependent modulation of synaptic transmission (Kreitzer & Regehr, 2001; Maejima et al. 2001; Ohno-Shosaku et al. 2001; Wilson & Nicoll, 2001). Since then, the endocannabinoid-mediated retrograde modulation has been reported in various brain regions (Chevaleyre et al. 2006; Hashimotodani et al. 2007b). It is now known that the endocannabinoid signal contributes to many forms of short-term and long-term synaptic plasticity throughout the brain (Gerdeman & Lovinger, 2003; Chevaleyre et al. 2006; Hashimotodani et al. 2007b). In this report, we focus on molecular mechanisms underlying endocannabinoid release induced by synaptic activities.

Induction of endocannabinoid release

Endocannabinoids are produced on demand and released from postsynaptic neurons. The released endocannabinoids travel backward across the synapse, activate presynaptic CB1 cannabinoid receptors (CB1Rs) and cause suppression of neurotransmitter release. Endocannabinoid release can be induced through two different pathways. One is driven by Ca²⁺ elevation alone (Ca²⁺-driven endocannabinoid release, CaER). The other is driven by activation of G_q/11-coupled receptors (receptor-driven endocannabinoid release, RER).

CaER contributes to depolarization-induced transient suppression of synaptic transmission, which is known as DSI for inhibitory synapse (Llano et al. 1991; Pitler & Alger, 1992) and DSE for excitatory synapse (Kreitzer & Regehr, 2001). DSI/DSE is induced postsynaptically by strong depolarization (e.g. to 0 mV for 5 s), and expressed presynaptically as a transient reduction of transmitter release. Therefore, involvement of some retrograde signal was proposed. In 2001, it was revealed that endocannabinoids serve as a retrograde messenger in DSI/DSE (Kreitzer & Regehr, 2001; Ohno-Shosaku et al. 2001; Wilson & Nicoll, 2001). Induction of ‘pure’ DSI/DSE, which is induced by depolarization alone without any simultaneous activation of G_q/11-coupled receptors, requires a large increase in intracellular Ca²⁺ concentration (to a micromolar range) (Brenowitz & Regehr, 2003; Maejima et al. 2005), which is primarily caused by Ca²⁺ influx through voltage-gated Ca²⁺ channels. Later studies demonstrated that the ‘pure’ DSI/DSE is phospholipase Cβ (PLCβ) independent (Hashimotodani et al. 2005; Maejima et al. 2005). In addition to depolarization, activation of NMDA-type glutamate receptors also induces a transient suppression of synaptic transmission in a Ca²⁺- and CB1R-dependent manner (Ohno-Shosaku et al. 2007). We showed that this suppression is caused by Ca²⁺ inflow into postsynaptic neurons through NMDA receptors and not through voltage-gated Ca²⁺ channels activated secondarily by NMDA receptor-induced local depolarization (Ohno-Shosaku et al. 2007). Therefore, NMDA receptors can take the place of voltage-gated Ca²⁺ channels for CaER. Molecular mechanisms of CaER still remain to be elucidated.

RER was first reported in a study showing that postsynaptic activation of type 1 metabotropic glutamate receptor (mGluR1) induced CB1R-dependent retrograde suppression of synaptic transmission in the cerebellum (Maejima et al. 2001). Since this discovery, RER was found to be induced in various brain regions by activation of G_q/11-coupled receptors, such as group I mGluRs (mGluR1/5) (Varma et al. 2001; Ohno-Shosaku et al. 2002; Galante & Diana, 2004; Kushmerick et al. 2004; Narushima et al. 2006), M₁ and/or M₃ muscarinic acetylcholine receptors (Kim et al. 2002; Fukudome et al. 2004; Narushima et al. 2007; Newman et al. 2007), orexin receptor (Haj-Dahmane & Shen, 2005) and oxytocin receptor (Oliet et al. 2007). Molecular mechanisms of RER have been elucidated by using pharmacological and genetic tools. Figure 1 shows a current model for RER. Activation of G_q/11-coupled receptors stimulates PLCβ and yields diacylglycerol (DAG). DAG is then converted by DAG lipase (DAGL) to 2-arachidonoylglycerol (2-AG), one of the two major endocannabinoids. This model is supported by studies showing prevention of RER by pharmacological inhibition of PLC or DAGL (Melis et al. 2004; Haj-Dahmane & Shen, 2005; Maejima et al. 2005; Safo & Regehr, 2005; Hashimotodani et al. 2007c; Newman et al. 2007; Straiker & Mackie, 2007; Uchigashima et al. 2007) and absence of RER in PLCβ1- or PLCβ4-deficient mice (Hashimotodani et al. 2005; Maejima et al. 2005). RER was originally thought to be Ca²⁺ independent, because RER was functional even under the conditions that prevent Ca²⁺ elevation (‘basal RER’, Fig. 1A) (Maejima et al. 2001). However, our later studies have revealed that RER is highly sensitive to the Ca²⁺ level of the postsynaptic neuron and is markedly enhanced by a small Ca²⁺ elevation to a physiological (submicromolar) range (‘Ca²⁺-assisted RER’, Fig. 1B) (Hashimotodani et al. 2005; Maejima et al. 2005). We have also demonstrated that PLCβ plays a key role as a Ca²⁺ sensor for ‘Ca²⁺-assisted RER’ (Hashimotodani et al. 2005; Maejima et al. 2005), which is briefly described in the following section.

Figure 1. Models of basal receptor-driven endocannabinoid release (RER) and Ca²⁺-assisted RER.

A, at basal Ca²⁺ levels, strong activation of G_q/11-coupled receptors (e.g. mGluR1/5, M₁/M₃) stimulates PLCβ through Gα_q/11. PLCβ hydrolyses phosphatidylinositol 4,5-bisphosphate into diacylglycerol and inositol 1,4,5-trisphosphate. Diacylglycerol is then hydrolysed to 2-arachidonoylglycerol (2-AG) by diacylglycerol lipase (DAGL). 2-AG is released and activates presynaptic CB1 cannabinoid receptors, leading to the suppression of neurotransmitter release. B, when activation of G_q/11-coupled receptors coincides with Ca²⁺ increase through Ca²⁺-permeable channels (e.g. voltage-gated Ca²⁺ channel, NMDA receptor), PLCβ activation is enhanced. In this condition, 2-AG production can be induced even by weak activation of G_q/11-coupled receptors, which is subthreshold for basal RER.

Mechanisms of Ca²⁺-assisted RER

To determine the role of PLCβ in endocannabinoid release, we used cultured hippocampal neurons and monitored the endocannabinoid release by measuring cannabinoid-sensitive synaptic currents (Hashimotodani et al. 2005). We found that RER mediated by muscarinic receptors or group I mGluRs (I-mGluRs) was absent in mutant mice lacking PLCβ1, which is the predominant isoform in the forebrain including hippocampus (Watanabe et al. 1998) among the four isoforms of the PLCβ family (PLCβ1–4). This PLCβ1-dependent RER was augmented by increasing the postsynaptic Ca²⁺ concentration. We measured PLCβ1 activity in intact neurons by using the exogenous TRPC6 channel, a member of the canonical transient receptor potential family. Because the TRPC6 channel is activated by intracellular DAG, we used it as a biosensor for the PLC product DAG. Large inward currents were induced by activation of the G_q/11-coupled receptors in hippocampal neurons expressing TRPC6 channels, but not in control neurons. These currents were also negligible in TRPC6-expressing neurons from the PLCβ1-deficient mice, indicating that the receptor-driven TRPC6 current reflects PLCβ1 activity. By measuring the TRPC6 current at different Ca²⁺ levels, we found that the receptor-driven PLCβ1 activation showed a similar Ca²⁺ dependence to that of the receptor-driven endocannabinoid release (Fig. 2). These results indicate that PLCβ1 is most effectively activated when Ca²⁺ elevation and G_q/11-coupled receptor activation coincide, and contributes to ‘Ca²⁺-assisted RER’.

Figure 2. Intracellular Ca²⁺ dependence of RER and PLCβ1-mediated TRPC6 activation driven by muscarinic activation.

A, simultaneous whole-cell voltage clamp recordings were made from neuron pairs in rat hippocampal culture. The presynaptic neurons were stimulated and cannabinoid-sensitive IPSCs were recorded from the postsynaptic neurons dialysed with the solutions containing 30 mm BAPTA with the indicated pCa levels. Examples representing the effects of 2-AG (30 nm) and oxotremorine-M (oxo-M, 0.3 μm) at the four different pCa levels. IPSC traces acquired at the indicated time points are shown on the right. B, averaged data for oxo-M-induced suppression of IPSC at four different pCa levels buffered with 10–30 mm BAPTA or 10 mm EGTA. C and D, rat cultured hippocampal neurons expressing exogenous TRPC6 were dialysed with 10 mm BAPTA-containing solutions with the indicated pCa levels. Sample traces (C) and averaged amplitudes (D) of oxo-M (3 μm)-induced currents at the four different pCa levels. In the summary bar graphs (D), the amplitudes were normalized to the values for pCa 6. (Modified from Hashimotodani et al. 2005, with permission.)

We drew a similar conclusion from a study on mouse cerebellar slices (Maejima et al. 2005). We made whole-cell recordings from Purkinje cells (PCs) and examined their excitatory synapses arising from climbing fibres (CFs) and parallel fibres (PFs). We sampled PCs from the rostral half of the cerebellum where PLCβ4 is the predominant isoform in PCs (Watanabe et al. 1998). We found that the mGluR1-driven endocannabinoid release depended on postsynaptic Ca²⁺ levels within a physiological range, and that the enhancement of endocannabinoid release by the combination of depolarization (Ca²⁺ elevation) and mGluR1 activation was abolished in PLCβ4-deficient mice. These results indicate that PLCβ4 plays a key role in Ca²⁺-assisted RER in the cerebellum.

It has been reported that endocannabinoid release is markedly facilitated when depolarization and G_q/11-coupled receptor activation is combined in various brain regions, including the hippocampus (Varma et al. 2001; Kim et al. 2002; Ohno-Shosaku et al. 2002, 2003), neocortex (Fortin et al. 2004), cerebellum (Maejima et al. 2005) and striatum (Narushima et al. 2007). This phenomenon has often been described as ‘enhancement of DSI/DSE by receptor activation’, which gives a false impression that the ‘pure’ DSI/DSE mechanism (CaER) is enhanced by receptor activation. However, there is no experimental evidence for the enhancement of CaER by receptor activation. Our results indicate rather that the enhanced DSI/DSE includes two distinct components, namely ‘pure’ DSI/DSE caused by CaER and the component caused by Ca²⁺-assisted RER (Maejima et al. 2005). The ‘enhancement of DSI/DSE by receptor activation’ can be explained largely, if not totally, by the addition of Ca²⁺-assisted RER to CaER.

Physiological significance of Ca²⁺-assisted RER

Using mouse cerebellar slices, we also examined the role of Ca²⁺-assisted RER in the endocannabinoid release triggered by synaptic activity. Repetitive stimulation of PFs (10 pulses at 100 Hz) induced a transient suppression of transmission at PF–PC synapses. This suppression was blocked by a selective mGluR1 antagonist, CB1R antagonists, and postsynaptic injection of 30 mm BAPTA. The suppression was also abolished in the slices prepared from PLCβ4-deficient mice. These results indicate that Ca²⁺-assisted RER involving the mGluR1–PLCβ4 cascade is essential for effective endocannabinoid release by PF activity. Furthermore, we demonstrated by biochemical analysis that combined weak mGluR1 activation and mild depolarization in PCs effectively produces 2-AG, whereas either stimulus alone does not produce detectable 2-AG. Several other studies also reported the data suggesting a major contribution of Ca²⁺-assisted RER to short-term retrograde suppression triggered by synaptic activities (Brown et al. 2003; Melis et al. 2004; Brenowitz & Regehr, 2005).

Ca²⁺-assisted RER might also play a role in integrating different synaptic inputs. We have reported recently that DSI is constantly enhanced by tonic activation of the M₁ muscarinic receptor in the striatum (Narushima et al. 2007). In the striatal medium spiny (MS) neurons, the M₁ receptor is tonically activated by ambient acetylcholine that is released from tonically active cholinergic interneurons. The tonic activation of M₁ receptor–PLCβ cascade, which is insufficient to induce endocannabinoid release, makes the neurons ready to respond to upcoming depolarization, which will be caused by glutamatergic input under physiological conditions, and to release endocannabinoids. Therefore, it is expected that MS neurons can integrate glutamatergic and cholinergic inputs and regulate the function of cannabinoid-sensitive presynaptic terminals through Ca²⁺-assisted RER.

Arrangement of endocannabinoid signalling molecules

Anatomical studies have elucidated the subcellular distributions of endocannabinoid signalling molecules including mGluR1α, mGluR5, M₁ muscarinic receptor, Gα_q/11, PLCβs, DAGLα, monoacylglycerol lipase (MAGL) and CB1R. In general, the elements involved in endocannabinoid production are co-localized on postsynaptic sites (Lujan et al. 1996; Tanaka et al. 2000; Yoshida et al. 2006; Narushima et al. 2007; Uchigashima et al. 2007), whereas CB1R, and MAGL which degrades 2-AG and terminates the retrograde endocannabinoid signal (Dinh et al. 2002; Hashimotodani et al. 2007c), are located on presynaptic terminals (Dinh et al. 2002; Gulyas et al. 2004). The fine arrangement of these elements is, however, unique to each brain region.

In hippocampal pyramidal neurons, mGluR5, Gα_q/11 and DAGLα are distributed in postsynaptic spines (Lujan et al. 1996; Tanaka et al. 2000; Yoshida et al. 2006). CB1R is expressed at low levels on excitatory presynaptic terminals that face the DGLα-expressing spine heads (Katona et al. 2006; Yoshida et al. 2006), and expressed at high levels on cholecystokinin-positive inhibitory terminals that face the somatodendritic surface lacking DAGLα. In cerebellar PCs, mGluR1α, Gα_q/11 and PLCβ4 are all concentrated at the perisynaptic region of spine heads (Baude et al. 1993; Tanaka et al. 2000; Nakamura et al. 2004). However, DAGLα is not found on spine heads and necks, and is highly concentrated at the base of spine neck and somatodendritic surface (Yoshida et al. 2006). Therefore, DAG should travel from spine heads to the bases of spine necks to be converted to 2-AG. This distance between PLCβ4 and DAGLα might be physiologically significant, considering that DAG also functions as an activator for PKCγ, which plays a crucial role in motor coordination and synapse elimination in PCs (Chen et al. 1995; Kano et al. 1995). CB1R is accumulated at the perisynaptic region of PF terminals and inhibitory presynaptic terminals (Kawamura et al. 2006; Yoshida et al. 2006). In striatal MS neurons, mGluR5 and DAGLα are widely expressed on the extrasynaptic somatodendritic surface, and their densities are highest in spines (Uchigashima et al. 2007). In contrast to mGluR5, the M₁ receptor is sparse in spines and shows widespread somatodendritic distribution (Narushima et al. 2007). CB1R is expressed at high levels on inhibitory presynaptic terminals of parvalbumin-positive interneurons and MS neurons and at low levels on excitatory corticostriatal afferents (Uchigashima et al. 2007).

These anatomical data are largely consistent with our electrophysiological data. As a whole, a set of receptors and enzymes required for RER are closely localized so that the glutamatergic (and/or cholinergic) activity can trigger 2-AG production effectively (Fig. 3). 2-AG is then released and travels some distance to presynaptic terminals expressing CB1R. The density of CB1R is high at inhibitory and low at excitatory terminals, as if the spatial disadvantage of inhibitory terminals, which are located far from the 2-AG production site, is compensated by increasing the sensitivity to 2-AG (Fig. 3).

Figure 3. Schematic drawings showing the molecular, anatomical and physiological organization for endocannabinoid signalling of the three cell types in the brain.

A, in the Purkinje cell, diacylglycerol lipase-α (DAGLα) is densely concentrated at the base of spine neck. DAG produced at spine head diffuses to the base of spine neck and is converted to 2-arachidonoylglycerol (2-AG) by DAGLα. Released 2-AG activates CB1 cannabinoid receptors (CB1R) located on perisynaptic region of the parallel fibre (PF) terminal or nearby inhibitory terminal (In). B, in the hippocampal CA1 pyramidal cell, DAGLα is distributed in the spine head and neck. At this site, 2-AG is produced and travels to activate CB1R located on both excitatory (Ex) and cholecystokinin-positive inhibitory (CCK-In) terminals. The density of CB1R is low at the excitatory terminal, and high at the CCK-positive inhibitory terminal. PCD, Purkinje cell dendrite; PyD, pyramidal cell dendrite; PV-In, parvalbumin-positive inhibitory terminal. C, in the striatal medium spiny neuron, coincidental depolarization and mGluR5 activation are essential for PLCβ1/DAGLα-mediated production of 2-AG to induce retrograde suppression of corticostriatal synapse, because of low CB1 levels in excitatory corticostriatal afferents. This mGluR5-driven Ca²⁺-assisted RER is further facilitated with the aid of M₁, while M₁ activation alone fails to trigger 2-AG production because of its sparse distribution in the spines. This modulation will lead to the suppression of the hyperactivity of the medium spiny neuron (MSN). At MSN–MSN and parvalbumin (PV) interneuron–MSN synapses, which show high density of CB1R, both mGluR5 and M₁ can induce RER because of their widespread somatodendritic distributions. This retrograde suppression will lead to the increase of the excitability and striatal output of the MSN. (Modified from Yoshida et al. 2006 and Uchigashima et al. 2007, with permission.)

Roles of two coincidence detectors, PLCβ and NMDA receptor

Based on our experimental data, we can speculate on the mechanisms of endocannabinoid-mediated synaptic plasticity. In Ca²⁺-assisted RER, PLCβ functions as a coincidence detector of G_q/11-coupled receptor activation and Ca²⁺ elevation (Hashimotodani et al. 2007a). If firing of presynaptic neurons activates postsynaptic I-mGluR and postsynaptic spikes induce Ca²⁺ elevation through voltage-gated Ca²⁺ channels, PLCβ can detect a certain timing of presynaptic and postsynaptic spikes. Indeed, involvement of the endocannabinoid signal in spike-timing-dependent plasticity (STDP) has been suggested at several types of synapses in the sensory cortex (Sjostrom et al. 2003; Bender et al. 2006; Nevian & Sakmann, 2006). At these synapses, long-term potentiation (LTP) is induced when presynaptic firing repeatedly precedes postsynaptic firing by 0–20 ms (pre-post timing), whereas long-term depression (LTD) is induced by their repeated firings in an inverse order (post-pre timing). This post-pre timing-dependent LTD (LTD_post-pre) was reported to depend on CB1R, I-mGluR and voltage-gated Ca²⁺ channels but not on postsynaptic NMDA receptors (Bender et al. 2006; Nevian & Sakmann, 2006). These data suggest the possibility that PLCβ might function as a post-pre timing detector in these endocannabinoid-mediated forms of LTD_post-pre.

Recently, we demonstrated that NMDA receptors can take the place of voltage-gated Ca²⁺ channels in both CaER and Ca²⁺-assisted RER by using cultured hippocampal neurons (Ohno-Shosaku et al. 2007). The NMDA receptor is the best known coincidence detector, and is generally thought to be able to detect a pre-post timing. Interestingly, it was recently reported in interneurons of dorsal cochlear nucleus that LTD was induced by a pre-post timing protocol (Tzounopoulos et al. 2007), which usually induces LTP in the sensory cortex. This pre-post timing-dependent LTD (LTD_pre-post) was shown to be expressed presynaptically, and depend on postsynaptic Ca²⁺ elevation, postsynaptic NMDA receptors and presynaptic CB1R, but not on I-mGluRs (Tzounopoulos et al. 2007). These data suggest that the NMDA receptor might contribute to LTD_pre-post by functioning as a pre-post timing detector and inducing CaER.

These results suggest that PLCβ and the NMDA receptor may function differently in different central synapses as a coincidence detector for activity-dependent endocannabinoid release.

Conclusions

The endocannabinoid signal plays an essential role in many forms of synaptic plasticity. Generation of the endocannabinoid signal is finely regulated by neural activities through multiple mechanisms, which involve voltage-gated Ca²⁺ channels, NMDA receptors, I-mGluRs, M₁/M₃ muscarinic receptors, PLCβs and DAGL. These molecules are well organized so that neural activities can trigger the endocannabinoid release effectively. Among these signalling molecules, NMDA receptors and PLCβ can function as coincidence detectors of presynaptic and postsynaptic activities. Recent studies suggest that these molecules might play a key role as timing detectors in endocannabinoid-dependent STDP. Under physiological conditions, the endocannabinoid release might be induced by more complicated patterns of neural activities. The physiological significance of these two coincidence detectors in various forms of synaptic plasticity remains to be elucidated.