In 2001, there was a breakthrough. Three groups independently and concurrently provided evidence that endocannabinoids act as retrograde messengers in the synaptic transmission in the brain (Ohno-Shosaku et al., 2001; Wilson and Nicoll, 2001; Kreitzer and Regehr, 2001a; for reviews, Montgomery and Madison, 2001; Christie and Vaughan, 2001; Wilson and Nicoll, 2002; Kreitzer and Regehr, 2002). Ohno-Shosaku et al. (2001) examined inhibitory postsynaptic currents (IPSCs) in neuron pairs in rat hippocampal cultures (Figure 5.2). They demonstrated that the application of WIN55212-2, a cannabinoid receptor agonist, induced a remarkable suppression of IPSCs in 16 of 26 pairs. The presynaptic locus for the action of WIN55212-2 was conrmed by the following criteria: (1) In a pair where IPSCs were greatly suppressed by WIN55212-2, inhibitory autaptic currents (IACs) were also depressed; (2) WIN55212-2 induced a clear increase in the paired-pulse ratio of IPSCs;
and (3) WIN55212-2 did not alter the postsynaptic sensitivity to iontophoretically applied GABA. They demonstrated that the depolarization of the postsynaptic neuron induced the elevation of [Ca2+]i as well as DSI. Importantly, in all neuron pairs where DSI was observed, WIN55212-2 suppressed IPSCs.
Moreover, DSI was totally eliminated by CB1 receptor antagonists, AM281 and SR141716A. These results strongly suggested that DSI is mediated through endocannabinoids. In support of this notion, DSI was not affected by various types of mGluR antagonists such as MCPG, (RS)-1-aminoindan-1,5-dicarboxylic acid (AIDA), 2-methyl-6-(phenylethynyl)pyridine (MPEP), and (RS)-"-cyclopropyl-4-phosphonophenylglycine (CPPG). A GABAB receptor antagonist (2S)-3-[[(1S)-1-(3,4-dichlorophenyl) ethyl]amino-2-hydroxypropyl](phenylmethyl)phosphinic acid (CGP55845A) also failed to affect DSI.
Wilson and Nicoll (2001) reported similar results using rat hippocampal slices. They demon-strated that DSI in a CA1 pyramidal neuron was eliminated by incubating slices with a CB1 receptor-specic antagonist AM251 or SR141716A. On the other hand, the effect of DSI was mimicked by WIN55212-2. WIN55212-2 did not affect the DSI-resistant component of the evoked IPSC. These results indicated that DSI is mediated by endocannabinoids. DSI does not appear to involve the vesicular release of endocannabinoids, because both botulinum toxin B and E, which prevent vesicular fusion, exerted no effect on DSI. They then examined the effect of 2-AG. They found that 2-AG had only a small effect compared with WIN55212-2. They assumed that this may be attributed to rapid removal from the extracellular space by an endogenous transporter. In fact, AM404, a structural analog of anandamide known as an inhibitor of the endocannabinoid trans-porter, depressed the baseline evoked IPSC amplitude without affecting the DSI-resistant component of the evoked IPSC. Regarding such an effect of AM404, however, it seems necessary to examine whether AM404 actually blocked the action of an endocannabinoid transporter, thereby exhibiting such an activity, because Glaser et al. (2003) recently reported that the uptake of endocannabinoids is not mediated by a specic transporter or carrier protein. Wilson and Nicoll (2001) also examined the effect of augmented cytoplasmic [Ca2+]i in the postsynapse on spontaneous IPSCs. They demonstrated that the effect of uncaging Ca2+ from a photolabile chelator is indistinguishable from the effect of a depolarizing step, indicating that the elevation of [Ca2+]i in the postsynapse is sufcient to trigger DSI. They also examined the effects of mGluR antagonists, (2S)-2-amino-2-[(1S,2S)-2-carboxycycloprop-1-yl]-3-(xanth-9-yl)propanoic acid (LY341495), and MCPG on DSI. Neither LY34195 nor MCPG was found to affect DSI, this being consistent with the notion that endocan-nabinoids but not excitatory amino acids are responsible for the induction of DSI. These results are in general agreement with the results of Ohno-Shosaku et al. (2001) on cultured hippocampal neurons mentioned before.
FIGURE 5.2 Blockade of depolarization-induced suppression of IPSCs by cannabinoid antagonists.
(A): Examples of IPSCs (left) and the summary (right) of the results showing that the depolarization-induced suppression can be elicited repeatedly without any run-down of its magnitude. Traces acquired before and 6 sec after the rst (control-1) or the second (control-2) depolarization in the normal external solution are shown. Averaged time courses of the changes in IPSC amplitudes induced by the rst (open circles) and the second (closed circles) depolarization (n = 10). (B) and (C): Examples of IPSCs (left) and the summary (right) of the results showing the blockade of the depolarization-induced suppression by 0.3 !M AM281 ([B]; n = 11) and 0.3 !M SR141716A ([C]; n = 3). IPSC traces and averaged time courses of the depolarization-induced changes in IPSC amplitudes are shown in the similar manners to (A). The asterisks attached to data points represent statistically signicant differences from the control (asterisk, p < 0.05; double asterisks, p < 0.01;
paired t test). (From Ohno-Shosaku, T., Maejima, T., and Kano, M. [2001] Endogenous cannabinoids mediate retrograde signals from depolarized postsynaptic neurons to presynaptic terminals, Neuron 29: 729–738. With permission.)
Possible roles of endocannabinoids as retrograde messengers in the cerebellum were reported by Kreitzer and Regehr (2001a). They examined the effects of postsynaptic depolarization of Purkinje cells on excitatory synaptic transmission between: (1) the parallel ber (cerebellar granule cells) and Purkinje cells and (2) the climbing ber (originating in the inferior olive) and Purkinje cells. They found that the depolarization of Purkinje cells induced a reduction of parallel ber excitatory postsynaptic currents (EPSCs). Similar inhibition was observed with climbing ber EPSCs. They referred to this phenomenon as DSE. The paired-pulse ratio was increased at both the parallel ber synapse and the climbing ber synapse, suggesting that the postsynaptic depolarization induced the reduced release of excitatory amino acids from both the parallel ber and the climbing ber. They next examined whether a rise in postsyn-aptic Ca2+ is required for DSE. They found that the inclusion of a Ca2+ chelator BAPTA in the postsyn-aptic recording pipette completely blocked DSE at both the parallel ber synapse and the climbing ber synapse. This clearly indicates that the elevation of postsynaptic Ca2+ is essential for the induction of DSE at both synapses. They also found that a reduction of Ca2+ inux in climbing ber takes place after the postsynaptic depolarization. The reduction of Ca2+ inux, followed the same time course as the inhibition of the EPSC. Noticeably, the inclusion of BAPTA in Purkinje cells resulted in the failure of the inhibition of the presynaptic Ca2+ inux. These results strongly suggested that the increase in postsynaptic Ca2+ induces the decrease in presynaptic Ca2+ inux, thereby reducing the release of excitatory amino acids, and that retrograde messengers must be involved in such a regulatory mechanism.
Excitatory amino acids, GABA, and adenosine appeared not to be involved in DSE, because a group II mGluR agonist, L-carboxycyclopropylglycine (L-CCG-1), a group II mGluR antagonist, LY34195, a GABAB receptor antagonist, CGP55845A, and an adenosine A1 receptor antagonist, 1,3-dipropyl-8-cyclopentyladenosine (DPCPX), did not affect DSE. On the other hand, the application of AM251, a CB1 receptor antagonist, greatly reduced DSE. In contrast, the application of WIN55212-2, a CB1 receptor agonist, inhibited EPSC at both the parallel ber synapse and the climbing ber synapse. These results strongly suggested that endocannabinoids act as retrograde messengers in these synapses and play crucial roles in the induction of DSE.
Details of endocannabinoid-mediated retrograde suppression of synaptic transmission (DSI and DSE) were further investigated by a number of investigators. Kreitzer and Regehr (2001b) examined DSI in the rat cerebellum and reported that it is mediated by endocannabinoids as in the case of DSI in the hippocampus and of DSE in the cerebellum. Endocannabinoid-mediated DSI in the cerebel-lum was also reported by Diana et al. (2002). Kreitzer et al. (2002) recently demonstrated that endocannabinoids released from cerebellar Purkinje cell dendrites suppress the spontaneous ring of nearby interneurons.
Wilson et al. (2001) investigated DSI in the mouse hippocampus. They showed that DSI is absent in the hippocampus of CB1 receptor-decient mice (CB1 receptor#/# mice), providing further concrete evidence that the CB1 receptor is crucially involved in DSI. The absence of DSI in the hippocampus of CB1 receptory#/# mice was also reported by Varma et al. (2001). DSI was impaired in the cerebellum of CB1 receptor#/# mice as well (Yoshida et al., 2002). Wilson et al. (2001) reported that CB1 receptor activation inhibits presynaptic Ca2+ channels possibly through direct action of the G') protein on voltage-dependent Ca2+ channels (VDCC) and that endocannabinoids exhibit a striking specicity in targeting a distinct class of interneurons, which are distinguished by their prole and exclusive use of the N-type Ca2+ channels for the neurotransmitter release. This is consistent with the
nding that DSI is initiated mainly by N-type Ca2+ channels (Lenz et al., 1998), although Lenz et al.
(1998) assumed at the time that postsynaptic rather than presynaptic N-type Ca2+ channel activation is necessary for DSI. Later, the same group reported that the block of presynaptic N-type Ca2+
channels is probably a major mechanism of DSI, yet they also described that the block of N-type Ca2+ channels does not fully explain the DSI in the hippocampus (Varma et al., 2002).
Ohno-Shosaku, Tsubokawa, et al. (2002) demonstrated that not only DSI but also DSE can be induced in the rat and mouse hippocampus. DSE in the hippocampus was mediated by the presynaptic CB1 receptor and endocannabinoids released from depolarized postsynaptic neurons.
They compared DSE and DSI in slices from the same animals and found that DSE was much less
prominent than DSI. For the induction of DSE, the necessary duration of depolarization was longer than for DSI, and the magnitude of DSE was much smaller than that of DSI. They showed that excitatory synapses were homogeneous and had moderate sensitivities to WIN55212-2, whereas inhibitory synapses were dichotomized into two distinct populations, one with a high sensitivity to WIN55212-2 and the other with no sensitivity. They concluded that the presynaptic cannabinoid sensitivity is a major factor that determines the extent of DSI and DSE.
Noticeably, the depolarization of the postsynaptic neurons is not necessarily required for the reduction of the neurotransmitter release from the presynaptic terminals. Maejima et al. (2001) reported that activation of the mGluR subtype 1 (mGluR1) expressed in cerebellar Purkinje cells with dihydroxyphenylglycol (DHPG), a group I mGluR agonist, reduced neurotransmitter release from climbing bers similar to the case of DSE. This phenomenon required the activation of G proteins but not Ca2+ elevation in the postsynaptic Purkinje cells. Thus, it is apparent that not only depolarization but also the activation of G proteins in Purkinje cells can induce the backward regulation of presynaptic neurotransmitter release. Ohno-Shosaku, Shosaku, et al. (2002) also inves-tigated this phenomenon using rat hippocampal neurons. They demonstrated that activation of group I mGluRs by DHPG suppressed IPSCs in about half of the neuron pairs. WIN55212-2 also suppressed IPSCs in all DHPG-sensitive pairs but not in most of the DHPG-insensitive pairs. Both DSI and DHPG- or WIN55212-2-induced suppression of IPSCs were abolished by treatment with AM281 or SR141716A, indicating that these responses were mediated through the CB1 receptor.
Interestingly, Varma et al. (2001) found that the bath application of DHPG or a group I and II mGluR agonist aminocyclopentane-1S, 3R-dicarboxylic acid (ACPD) induced marked acceleration of DSI. On the other hand, a mGluRs antagonist LY34195 reduced DSI. Ohno-Shosaku, Shosaku, et al. (2002) also reported that DSI was signicantly enhanced in the presence of DHPG. This en-hancement was much more prominent than expected from the simple summation of depolarization-induced and group I mGluR-depolarization-induced endocannabinoid release; it is probable that postsynaptic depolarization and the activation of group I mGluRs worked in concert to generate sufcient amounts of endocannabinoids in postsynaptic neurons.
Importantly, such cooperation in the generation of endocannabinoids is not restricted to the case of depolarization and activation of mGluRs. Recently, Kim et al. (2002) reported that activation of muscarinic acetylcholine receptors also enhances DSI in the rat hippocampus. They demonstrated that the application of carbacol enhanced DSI, and this effect of carbacol was blocked by atropine.
The addition of eserine, a cholinesterase inhibitor, also enhanced DSI, implying that muscarinergic acetylcholine receptor–mediated endocannabinoid release is a physiologically relevant event. Ohno-Shosaku et al. (2003) demonstrated that postsynaptic M1 and M3 receptors are responsible for such an enhancement in the hippocampus. These observations, together with previous ndings, suggested that the retrograde modulation by endocannabinoids is an important and widespread mechanism in the brain by which the activity of postsynaptic neurons can inuence the functions of various types of inhibitory and excitatory presynaptic neurons, although the neuroregulatory roles of endocannabinoids have already been postulated by a number of investigators as mentioned previ-ously (for reviews, see Pertwee, 1997; Di Marzo, 1998; Di Marzo et al., 1998; Mechoulam et al., 1998; Piomelli et al., 1998, 2000; Sugiura et al., 1998; Sugiura and Waku, 2000; Schlicker and Kathmann, 2001; Davies et al., 2002; Fride, 2002; Lichtman et al., 2002; Pertwee and Ross, 2002;
Schweitzer, 2002).
What then are the physiological implications of endocannabinoid-mediated DSI and DSE in vivo? This important issue has not yet been fully claried. There are several possibilities con-cerning the physiological meanings of DSI and DSE. In the hippocampus and neocortex, the CB1 receptor is mainly expressed on interneurons which form GABAergic synapses with fast kinetics (Wilson and Nicoll, 2001). Such interneurons are assumed to be responsible for regulating gamma oscillations which are synchronized over long distances in the brain and have been proposed to be involved in binding simultaneous perceptions. Endocannabinoids may be closely involved in the process of cognition by attenuating GABA release in these brain regions through DSI. In support
of this, Hajos et al. (2000) have reported that WIN55212-2 interferes with kainate-induced gamma oscillations in hippocampal slices. It is also plausible that DSE is one of the intrinsically fast inhibitory mechanisms of certain types of excitatory neurons. As mentioned previously, we have proposed that 2-AG, generated during increased synaptic transmission, plays an important role in attenuating subsequent neurotransmitter release by acting on the CB1 receptor expressed in the presynaptic terminals (Sugiura et al., 1998, 2000; Sugiura and Waku, 2000). Such a negative feedback mechanism would greatly contribute to the homeostasis of neurons and synaptic transmission.
Another possibility is that DSI and DSE are involved in the modication of synaptic plasticity such as LTP and long-term depression (LTD). For example, suppression of inhibitory neurotrans-mitter release by endocannabinoids (DSI) leads to hyperactivation of postsynaptic neurons, which appears to be favorable for stimulation or promotion of LTP. In fact, the modulation of LTP (potentiation or suppression) by endocannabinoids has already been demonstrated by several inves-tigators (Terranova et al., 1995; Collin et al., 1995; Stella et al., 1997; Carlson et al., 2002). In addition to LTP, endocannabinoids may also participate in LTD. Robbe et al. (2002) recently provided evidence that endocannabinoids, derived from the postsynapse, induce LTD in the nucleus accumbens. Several investigators have also demonstrated the involvement of endocannabinoids in LTD in the amygdala (Marsicano et al., 2002), the striatum (Gerdeman et al., 2002), and the hippocampus (Chevaleyre and Castillo, 2003). Thus, it seems conceivable that DSI and DSE are actually involved in several physiological phenomena in vivo, yet details still remain to be deter-mined. A thorough elucidation of the physiological signicance of DSI and DSE as well as other endocannabinoid-dependent neural cell functions awaits future investigations.