MATERIALS AND METHODS

RESULTS

We recorded fEPSPs evoked in the NAc core by the stimulation of glutamatergic afferences. We found that LFS-4Hz (20-min stimulation at 4 Hz) induces a robust LTD (n = 8; Fig. 1, A and D; Table 1): when the recording was protracted for 180 min after the end of LFS-4Hz stimulation, the mean of fEPSP amplitudes recorded between 175 and 180 min was 74.87%, a value still significantly (P < 0.05) different from baseline (n = 3; data not shown).

We first tested whether this novel form of LTD required activation of the glutamate NMDAR, as previously demonstrated in other forms of LTD in the NAc (Grueter et al. 2010; Thomas et al. 2000). Thus the competitive NMDAR antagonist d-AP5 (50 μM) was added to the slice-perfusing medium 20 min before and during LFS-4Hz stimulation: this treatment did not affect LTD (n = 6; Fig. 1D; Table 1). To evaluate the involvement of postsynaptic ionotropic glutamate receptors, the slices were perfused with the non-subtype-selective ionotropic glutamate receptor antagonist kynureate (1 mM) 20 min before and during LFS-4Hz stimulation. As expected, kynureate completely blocked fEPSP but did not prevent LTD induction (n = 8; Fig. 1, B and D; Table 1). Control recordings indicate that the depressive effect of kynureate is fully reversed after washout when LFS-4Hz is not delivered (n = 6; Fig. 1B). Previous studies have shown that in addition to NMDAR, transient activation of mGluR2/3 (with either synaptically released glutamate or a specific exogenous agonist) triggers a presynaptic form of LTD at excitatory synapses in both the NAc (Robbe et al. 2002a) and the hippocampus (Tzounopoulos et al. 1998). Therefore, we next tested whether LFS-4Hz also would induce this form of LTD in our experimental conditions by applying the mGlu2/3 antagonist LY-341495 (0.2 μM) 20 min before and during LFS-4Hz stimulation. We found that antagonism at mGluR2/3 had no effect on LTD induction (n = 6; Fig. 1, C and D; Table 1).

In addition to NMDA- and mGlu2/3-dependent LTD, NAc synapses have been shown to display eCB-LTD, which depends on the production of eCBs acting at CB1 receptors (Grueter et al. 2010; Hoffman et al. 2003; Robbe et al. 2002b). Therefore, we tested the involvement of the eCB system in LFS-4Hz in the presence of the CB1 receptor antagonists AM251 (2 μM) or LY-320135 (10 μM) (Yu et al. 2013). As shown in Fig. 2, A and C, both antagonists completely prevented LTD induction (n = 6 and 6; Table 1). The two main eCBs produced upon neuronal stimulation in the brain are AEA and 2-arachidonoylglycerol (2-AG) (Piomelli 2003). We first attempted to evaluate the involvement of 2-AG in LFS-4Hz; slices were treated with THL (10 μM), an inhibitor of the 2-AG-producing enzyme DGL. As shown in Fig. 2, A and C, preincubation of the slices (>2 h) with THL completely prevented LTD induction (n = 9; Table 1). We next examined whether AEA was involved in LFS-4Hz-induced LTD. To test this possibility, we used a weaker stimulation protocol (10 min instead of 20 min at 4 Hz; LFS-weak), subthreshold to LTD induction in control conditions (n = 5; Fig. 2, B and C; Table 2). Endogenous AEA production was then increased by blocking the activity of the FAAH, the enzyme responsible for AEA degradation, by adding JNJ-1661010. As shown in Fig. 2, B and C, LFS-weak failed to induce LTD in the presence of JNJ-1661010 (1 μM) (n = 5; Table 2), suggesting that AEA is not involved in this form of LTD. To further test the involvement of 2-AG in LFS-4Hz-induced LTD, slices were incubated (>2 h) with JZL-184 (50 μM) (Puente et al. 2011), an inhibitor of MAGL, the enzyme responsible for 2-AG degradation, and LFS-weak was applied. As shown in Fig. 2, B and C, elevation of 2-AG levels rescued LTD, confirming the involvement of 2-AG in LFS-4Hz-induced LTD (n = 6; Table 2).

A recent study suggested that LTD induced in the NAc core by 10-Hz/5-min stimulation involves TRPV1 activation (Grueter et al. 2010); therefore, we performed a series of experiments in the presence of the TRPV1 antagonist capsazepine (10 μM) and found that capsazepine did not affect LTD induced by LFS-4Hz (n = 6; Fig. 2, A and C; Table 1). Moreover, we performed a series of experiments using the 10-Hz/5-min stimulation protocol and found that, in our experimental conditions, this protocol does not induce LTD (mean fEPSP was 106.52 ± 4.67% at 55–60 min after 10-Hz/5-min LFS; n = 6; data not shown).

Because it has been shown that eCB-LTD in the NAc requires activation of mGluR5 (Robbe et al. 2002b), we also tested its involvement in the LTD obtained in our experimental conditions: application of the mGluR5 antagonist MPEP (10 μM) during LFS-4Hz did not affect LTD induction (n = 4; Fig. 1, C and D; Table 1). We further investigated the involvement of group I mGluRs (which comprise both mGluR1 and mGluR5) by applying the mGluR1 antagonist LY-367385 (0.1 mM). Figure 1, C and D, shows that LFS-4Hz still elicited LTD (n = 4; Table 1).

To further clarify the mechanisms underlying LFS-4Hz-induced LTD, we measured the effect of the L-type VGCC blocker nifedipine (10 μM) on LFS-4Hz-LTD. As depicted in Fig. 3, A and E, nifedipine completely prevented LTD induction. According to the observation that the dense afferent DA innervation in the NAc regulates the activity of the medium spiny neurons (MSN), which comprise the great majority of NAc neurons, as well as their ability to pursue plastic changes (Kreitzer and Malenka 2005; Schotanus and Chergui 2008; Thomas et al. 2001), we next examined whether DA is also involved in LFS-4Hz-induced LTD. To this aim, the DA D1 and D2 receptor antagonists SCH-23390 (25 μM) and sulpiride (25 μM) were added to the slice-perfusing medium 20 min before and during LFS-4Hz stimulation. We found that blockade of both D1 and D2 DA receptors did not prevent LTD induction (n = 6; data not shown).

Because CB1 and 5-HT₂ receptors are both expressed in the NAc (Barnes and Sharp 1999; Pertwee 1997), and activation of 5-HT₂ receptors has been shown to cause production and release of eCBs leading to LTD in the inferior olive (Best and Regehr 2008), we investigated the involvement of 5-HT₂ receptors in LFS-4Hz-induced LTD. As shown in Fig. 3, A and E, the 5-HT₂ receptor blocker ketanserin (10 μM) completely blocked LTD. Data obtained in the presence of the 5-HT₁ receptor antagonist pindolol (10 μM) suggest that this receptor subtype does not contribute to LFS-4Hz-induced LTD (n = 4; Fig. 1, A and E; Table 1).

To confirm the involvement of 5-HT in this form of eCB-LTD, in the next set of experiments LFS-4Hz was replaced by bath application of 5-HT (5 μM) for 20 min. Consistent with previous observations at corticostriatal synapses (Mathur et al. 2011), 5-HT added to the slice-perfusing medium (5 μM, 20 min) produced an initial, large depression and a significant (P < 0.05) and stable LTD following washout (n = 4; Fig. 3, B and E; Table 3). To test if the effect of 5-HT-LTD was mediated by 5-HT₂ receptors, slices were perfused with the specific 5-HT₂ receptor agonist α-methyl 5-HT (20 μM). As shown in Fig. 3, B and E, the agonist mimicked the effect of 5-HT (n = 6) to induce LTD. A comparison of the amplitude of the LTD obtained in the presence of 5-HT vs. α-methyl 5-HT revealed that the magnitude of LTD was significantly increased by the latter treatment. To further test the involvement of 5-HT₂ receptors in 5-HT-LTD, slices were perfused with ketanserin (10 μM) for at least 20 min before and during application of 5-HT (5 μM). As depicted in Fig. 3, C and E, ketanserin completely prevented the effect of 5-HT on fEPSP amplitude (n = 6; Table 3). We then tested whether the effects of 5-HT on synaptic transmission was mediated by the CB1 receptors: coapplication of 5-HT and the CB1 receptors antagonist AM251 (2 μM) completely prevented 5-HT-LTD (n = 4; Fig. 3, C and E; Table 3). We next examined the involvement of the L-type VGCC in 5-HT-LTD by applying 5-HT (5 μM) along with the specific antagonist nifedipine (10 μM). As depicted in Fig. 3, C and E, nifedipine completely prevented 5-HT-LTD (n = 4; Table 3). To test whether endogenous 5-HT in the acute slice preparation would facilitate the induction of LTD, the subthreshold stimulation protocol LFS-weak was applied in the presence of the SSRI citalopram (4 μM). Figure 3, D and E, shows that this treatment enabled LFS-weak to induce LTD (n = 6; Table 2), indicating the involvement of endogenous 5-HT in this process.

Presynaptic CB1 receptors have been shown to inhibit glutamate release at the prelimbic cortex-NAc synapses (Robbe et al. 2001); therefore, if 5-HT receptor activation and LFS-4Hz share common mechanisms to inhibit excitatory synaptic transmission, LFS-4Hz-LTD and 5-HT-LTD may be mutually occlusive. We first verified whether repetition of LFS-4Hz caused saturation of LTD. As shown in Fig. 4A, a second LFS-4Hz applied during LTD consolidation failed to induce a further depression: fEPSP amplitude was 74.02 ± 5.44% (n = 5; P < 0.05 vs. baseline) between 25 and 30 min after the end of the first LFS-4Hz stimulation, and 73.74 ± 3.78% between 25 and 30 min after the end of the second LFS-4Hz stimulation, a value similar to the one recorded after the first LFS-4Hz [n = 5; no significance (NS) vs. first LFS-4Hz-LTD]. As shown in Fig. 4B, to verify whether 5-HT-LTD also occludes LFS-4Hz-LTD, we first perfused the slices with 5-HT (5 μM) and then applied LFS-4Hz: we found that fEPSP amplitude was 78.99 ± 2.96% (n = 5; P < 0.05 vs. baseline) between 25 and 30 min after the end of 5-HT application, and 76.20 ± 1.34% between 25 and 30 min after the end of LFS-4Hz stimulation (n = 5; NS vs. 5-HT-induced LTD). Conversely, 5-HT (5 μM) had no significant effect when applied after LFS-4Hz-LTD was already induced: fEPSP amplitude was 70.52 ± 4.46% of baseline (n = 7; P < 0.05 vs. baseline) between 25 and 30 min after LFS-4Hz, and 72.98 ± 4.41% between 25 and 30 min after 5-HT application (n = 7, NS vs. LFS-4Hz-LTD; Fig. 4C).

DISCUSSION

In recent years, LTD in the NAc has been investigated in several studies (Grueter et al. 2010; Hoffman et al. 2003; Ji and Martin 2012; Kahn et al. 2001; Kombian and Malenka 1994; Pennartz et al. 1993; Thomas et al. 2000, 2001; Robbe et al. 2002a, 2002b), but the present work appears to provide the first evidence of an eCB-dependent LTD requiring 5-HT₂ and CB1 receptor activation and increase in Ca²⁺ entry through L-type VGCC. Thus at least two distinct forms of eCB-dependent LTD can be induced in the NAc depending on the induction protocol used, because previous studies have shown that stimulations at higher frequencies induce LTDs requiring mGluR5 and CB1 receptor activation, a rise in postsynaptic Ca²⁺ mediated by intracellular stores (Grueter et al. 2010; Robbe et al. 2002b), and TRPV1 channel activation that triggers AMPA receptor endocytosis (Grueter et al. 2010).

eCBs are arachidonic acid-derived lipid signaling molecules with potent actions at CB receptors, namely CB1, widely expressed throughout the central nervous system (Herkenham et al. 1991), and CB2, mainly present in peripheral tissues and immune cells (Galiègue et al. 1995). CB1 receptors belong to the G protein-coupled receptor family and signal through the G_o/i family of G proteins; their activation can induce several effects, including adenylyl cyclase inhibition, mitogen-activated protein kinase stimulation, Ca²⁺ current reduction, and K⁺ current modulation. Previous studies in the mouse NAc have shown that CB1 activation inhibits glutamatergic synaptic transmission through the modulation of presynaptic K⁺ currents (Robbe et al. 2001) and underlies the LTD induced by 13-Hz/10-min (Robbe et al. 2002b), 5-Hz/10-min (Hoffman et al. 2003), and 10-Hz/5-min stimulation protocols (Grueter et al. 2010). These observations suggest the involvement of a presynaptic mechanism consisting of reduction of glutamate release induced by eCBs acting at CB1 receptors as retrograde messengers. This view is in accordance with previous evidence in the NAc (Hoffman and Lupica 2001) and at excitatory synapses in other brain regions, such as the striatum (Gerdeman and Lovinger 2001), neocortex (Sjöström et al. 2003), and dorsal principal olive (Best and Regehr 2008), as well as at inhibitory synapses in the hippocampus (Chevaleyre and Castillo 2003) and amygdala (Marsicano et al. 2002). CB1 receptors are activated by Δ⁹-tetrahydrocannabinol, the psychoactive compound of the hemp plant, as well as by endogenous ligands, such as AEA and 2-AG, the latter being prevalent in the brain. AEA has been shown to activate TRPV1 (De Petrocellis et al. 2001), intracellular membrane proteins belonging to the transient receptor potential family of store-operated Ca²⁺ channels. Given the colocalization of TRPV1 and presynaptic CB1 receptors in the NAc core (Micale et al. 2009), and according to the recent observation of their involvement in 10-Hz/5-min stimulation-induced eCB-LTD in the NAc core (Grueter et al. 2010), it was reasonable to hypothesize that LFS-4Hz-LTD resulted from the activation of TRPV1 by AEA. However, the results obtained by delivering LFS-4Hz in the presence of the TRPV1 blocker, or by elevating AEA levels in conjunction with LFS-weak, or in the experiments performed with 10-Hz/5-min stimulation all rule out this possibility and point toward the involvement of the selective CB1 receptor ligand 2-AG, which has no activity at TRPV1 (De Petrocellis et al. 2000). 2-AG is the most abundant eCB in the brain (Mechoulam et al. 1995; Sugiura et al. 1995); in particular, its concentration in the mesencephalon is 100 times higher than that of AEA (Bisogno et al. 1999). In addition, 2-AG acts as a potent and full agonist at CB1 receptors, whereas AEA is only a partial agonist (Savinainen et al. 2001; Stella et al. 1997; Sugiura et al. 1995, 1999). Biochemical studies indicate that the major enzymes involved in 2-AG biosynthesis are phospholipase C (PLC) and DGL (Piomelli 2003), although other pathways have also been suggested (Sugiura et al. 2006). The involvement of 2-AG in LFS-4Hz-LTD was confirmed by the experiments performed in the presence of the DGL inhibitor THL, which was proven to have no direct effect on CB1 receptors (Szabo et al. 2006; Uchigashima et al. 2007); moreover, it is further supported by the evidence that LTD can be induced by a subthreshold stimulation protocol when endogenous 2-AG levels are increased by an inhibitor of MAGL, the 2-AG-hydrolyzing enzyme. This finding is consistent with previous observations in the hippocampus (Ameri and Simmet 2000; Edwards et al. 2006; Hashimotodani et al. 2007; Stella et al. 1997), cerebellum (Maejima et al. 2005; Safo and Regehr 2005), and prefrontal cortex (Melis et al. 2004), supporting the involvement of 2-AG.

In addition to eCB, present results point toward the involvement of 5-HT, acting at 5-HT₂ receptors, in LFS-4Hz-LTD, as demonstrated by LFS-4Hz experiments performed in the presence of the 5-HT₂ receptor blocker and by the evidence that LTD can be induced by the subthreshold stimulation protocol when the endogenous 5-HT levels are elevated by the SSRI citalopram. This evidence is further supported by the observation that LTD can also be elicited by bath application of 5-HT or a specific 5-HT₂ receptor agonist.

It was recently shown that 5-HT increases the activity of GABAergic interneurons in the NAc through 5-HT_2C receptors, resulting in inhibitory action on MSN (Blomeley and Bracci 2009). One might argue that the 5-HT₂-dependent LTD observed in the current study could derive from increased GABAergic signaling; however, this possibility seems unlikely, because the experiments were carried out in the constant presence of the GABA_A receptor antagonist picrotoxin. Rather, present data support the view that 5-HT₂ receptor activation might trigger eCB production and release, in accordance with a previous observation in the dorsal principal olive (Best and Regehr 2008).

We also investigated the possible involvement of L-type VGCC in both LFS-4Hz- and 5-HT-induced LTD, because they are particularly important for down- to up-state transition of MSN (Carter and Sabatini 2004). Moreover, in vivo studies have demonstrated that these channels play a critical role in psychostimulant-induced behavioral and neuronal plasticity in various regions of the mesolimbic DA system, including the NAc (Gnegy 2000; Licata and Pierce 2003). We found that nifedipine prevented both LFS-4Hz- and 5-HT-induced LTD, indicating that these channels play a critical role in this process, as previously observed in eCB-induced LTD in dorsal striatum MSN (Kreitzer and Malenka 2005), as well as in other forms of synaptic plasticity in various brain areas, including the hippocampus (Bolshakov and Siegelbaum 1994; Grover and Teyler 1992; Morgan and Teyler 1999), lateral amygdala (Weisskopf et al. 1999), and perirhinal cortex (Seoane et al. 2009). It has been shown that large increases in intracellular Ca²⁺ (10–50 μM) are needed to activate phospholipase D and AEA biosynthesis, whereas lower intracellular Ca²⁺ elevations can activate PLC and DAG, yielding 2-AG production (Brenowitz and Regehr 2003; Hashimotodani et al. 2005). It seems reasonable to suppose that 5-HT₂ activation by 5 HT (either endogenously released in response to LFS-4Hz stimulation or exogenously applied) might stimulate 2-AG production by increasing both DGL activity and L-type VGCC permeability, resulting in intracellular Ca²⁺ level elevation compatible with 2-AG production and release.

Altogether, present results suggest that LFS-4Hz stimulation of glutamatergic afferences induces a novel form of eCB-LTD in the NAc core involving the release of 5-HT, which acts at 5-HT₂ postsynaptic receptors, increasing Ca²⁺ influx through L-type channels and 2-AG production and release; the eCB would travel retrogradely and bind to presynaptic CB1 receptors, causing a long-lasting decrease of glutamate release, resulting in LTD (Fig. 5).

Dysfunction in 5-HT neurotransmission plays a critical role in a variety of psychiatric pathologies, including drug addiction, major depression, food intake disorders, emesis, chronic pain, sleep disorders, autism, and schizophrenia (Malison et al. 1998; Mann et al. 1995; Pacher et al. 2006; Shahsvand-Ananloo et al. 2013). Present observations might be helpful to understand the mechanisms underlying these disorders, hopefully leading to the development of new pharmacological tools. Moreover, the present findings add a bit of knowledge to the growing body of evidence pointing toward a major role of the eCB system in short- and long-term synaptic plasticity in brain areas relevant to the mediation of reward.

GRANTS

This work was supported by Ricerca Fondamentale Orientata (RFO) grants from the University of Bologna and by a fundamental contribution from the “Fondazione del Monte di Bologna e Ravenna” (to G. Aicardi).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

C.B. and G.A. conception and design of research; C.B. and G.B. performed experiments; C.B., G.B., and F.T. analyzed data; C.B. and G.A. interpreted results of experiments; C.B. and G.B. prepared figures; C.B. drafted manuscript; C.B., G.B., F.T., and G.A. edited and revised manuscript; C.B., G.B., F.T., and G.A. approved final version of manuscript.