The current study demonstrates that the mechanism of MOR-mediated LTD is similar to the mechanism of induction of glutamatergic CB-LTD in that it involves cAMP-PKA-protein translation signaling and both occur at AIC-DLS synapses (Lonart et al., 2003; Yin et al., 2006; Mato et al., 2008; Yasuda et al., 2008; Atwood et al., 2014b; Younts et al., 2016). However, we found notable differences such as a lack of dependence on mTOR signaling (Fig. 8) and a novel mechanism involving HCN1 signaling for MOR-LTD (Fig. 10 and 11). Figure 13 summarizes the findings of the paper and illustrates the differential mechanisms involved in the expression of MOR-mediated LTD. Furthermore, the MOR-LTD mechanism is different than that of MOR-STD (Fig. 13), suggesting that MOR-mediated plasticity does not utilize a common pathway at all glutamatergic synapses.

Our previous report demonstrates that MORs expressed in AIC are the sole source of MOR-mediated LTD at glutamate inputs to the DLS(Munoz et al., 2018). However in the DMS, MORs on medial prefrontal cortex, anterior cingulate cortex, and basolateral amygdala inputs are capable of inducing LTD(Muñoz et al., 2020). MOR-LTD also occurs locally at cholinergic interneuron synapses that co-release glutamate on to MSNs(Munoz et al., 2018; Muñoz et al., 2020). However, alcohol only disrupts MOR-LTD at AIC inputs to DLS and at medial prefrontal and anterior cingulate cortex inputs to DMS leaving MOR plasticity at thalamostriatal, amygdalostriatal and cholinergic interneurons unaffected(Munoz et al., 2018; Muñoz et al., 2020). We also found that the opioid oxycodone disrupts MOR-LTD in DLS(Atwood et al., 2014a), and given that this is mediated by AIC-DLS MORs(Munoz et al., 2018), it follows that alcohol and opioids have similar actions on MOR plasticity at this synapse. Given that alcohol (and presumably oxycodone and other opioids) do not affect MOR plasticity at these other non-corticostriatal synapses, differential mechanisms that underlie MOR plasticity at each type of synapse may confer susceptibility to the deleterious effects of these drugs and could account for their mechanisms of action in the development of habitual or compulsive use of the drugs.

Previous work has shown that the inhibition of PKA disrupts CB-LTD in the nucleus accumbens(Mato et al., 2008) and that presynaptic PKA is required to induce cannabinoid LTD in cerebellum(Lonart et al., 2003) and cannabinoid synaptic plasticity in hippocampus(Yasuda et al., 2008). Moreover, previous reports had shown that the activation of AC by forskolin enhances corticostriatal EPSC amplitudes(Cho et al., 2008) and blocked delta opioid receptor-mediated LTD of inhibitory postsynaptic currents(Patton et al., 2016). cAMP/PKA signaling has also been implicated in presynaptic type 2/3 metabotropic glutamate receptor- and type 1b serotonin receptor-mediated LTD at various synapses in the brain(Atwood et al., 2014a). As previously reported(Duffy & Nguyen, 2003), we confirmed that the activation of postsynaptic PKA is necessary for the induction of LTP in CA1 pyramidal neurons from the hippocampus, but not for MOR-LTD in DLS (Fig. 5). Until now, the role of cAMP/PKA signaling pathway in MOR-mediated plasticity had not been addressed. Our results demonstrate that presynaptic PKA inhibition blocked AIC-DLS MOR-LTD (Fig. 4 and 13), but not thalamostriatal MOR-STD (Fig. 9D-F and 13). Interestingly, the magnitudes and time courses of eEPSCs and oEPSCs did not differ between measures (Fig. 4A, D, H, K, O and R), suggesting that PKA is not required for an acute MOR-mediated inhibition of glutamate release, but it is required for the maintenance of MOR-LTD. On the other hand, we found that AC activation disrupted both acute MOR-mediated inhibition (in fact turning it into a transient potentiation) and MOR-LTD (Fig. 6 and 13), but had no effect on MOR-STD (Fig. 9G-I and M and 13). Altogether these data suggest that cAMP/PKA pathway may be a general mechanism whereby GPCRs that signal through Gαi may induce LTD, but a receptor’s ability to couple to Gαi does not necessarily mean that it will induce cAMP/PKA-dependent LTD upon activation.

Protein translation is necessary for striatal CB-LTD(Yin et al., 2006) and is involved in GPCR-mediated LTD at some synapses, but not others(Atwood et al., 2014a; Younts et al., 2016) suggesting that protein translation can be an important component of glutamatergic LTD, depending on the synapse. Indeed, local protein translation can take place in the presynaptic region, where the activation of CB1 receptors enhances protein translation, being critical for the induction and not the maintenance of inhibitory LTD(Younts et al., 2016). Here we demonstrated that protein translation is necessary to produce MOR-LTD (Fig. 7 and 13), but not MOR-STD (Figure 9J-M and 13). Our data is consistent with the previous reports on protein translation-dependent CB-LTD, and suggest that it is presynaptic protein translation that is involved in MOR-LTD expression. However, cycloheximide has been reported to also activate p38/MAPK signaling (Lockhead et al., 2020) and several studies have shown a critical role of p38 kinases in the induction of LTD and LTP (Asih et al., 2020). Therefore, more data is needed to determine this is definitively protein translation-mediated in the DLS.

Although we have previously reported that MOR- and CB-LTD are mutually occlusive in the DLS(Atwood et al., 2014a), and here we found that there were similarities between MOR-LTD and CB-LTD mechanisms, we found some differences which make both types of LTD distinct. In the DLS, MOR-LTD of glutamate input is restricted to the AIC(Munoz et al., 2018), which differs from the more broadly expressed (and overlapping at AIC-DLS synapses, Fig. 3) CB-LTD(Wu et al., 2015; Gremel et al., 2016), suggesting that these two forms of LTD may occur in distinct synaptic environments that could utilize differing mechanisms. It is likely that even in synapses where both receptors are expressed (e.g. in AIC terminals) that each receptor shares partially overlapping, but not identical signaling pathways.

A clear example of distinct signaling mechanisms identified here is in the role of the mTOR pathway in CB-LTD and MOR-LTD. The CB1 receptor can signal via mTOR in the hippocampus, where the increase of protein translation induced by CB1 receptor activation is mTOR-dependent producing inhibitory CB-LTD(Younts et al., 2016). We predicted that we would see the same effect in the DLS for both CB-LTD and for MOR-LTD. Partially correct in our prediction, we found that CB-LTD is blocked after the inhibition of mTOR in the DLS (Fig. 8A-D), but the mechanism of MOR-LTD did not utilize mTOR (Fig. 8E-H). These novel data regarding CB1-mTOR signaling in DLS provide rationale for future experiments addressing the role of CB1-mTOR signaling in dorsal striatal-related behaviors.

Presynaptic cations such as Na⁺ and K⁺ are important for the propagation of action potentials and subsequent glutamate release(Chen & Lui, 2021). HCNs channels are permeable to Na⁺ and K⁺ and modulated by cAMP(Sartiani et al., 2017), and they can decrease neurotransmission by restricting presynaptic Ca²⁺ influx(Huang et al., 2011). We did identify a critical role for the HCN1 channel in MOR-LTD expression (Fig. 10, 11 and 13). It has been reported that HCN1 channels can be found at presynaptic sites in the globus pallidus, where the inhibition by ZD7288 produced an increase of miniature inhibitory postsynaptic current frequency, with no changes in amplitude(Boyes et al., 2007). The block or deletion of HCN1 causes an increase in the frequency of miniature EPSCs in entorhinal cortical pyramidal neurons(Huang et al., 2011). Here we showed that broad blockade of HCN channels alters glutamatergic synaptic transmission release (Fig. 10E-I). However, when we deleted just HCN1 from AIC inputs to DLS (Fig. 11C), we did not observe any changes in basal glutamate transmission (Fig. 11I-M). The differences in effects of broad blockade and synapse-specific deletion may be attributed to the overall contribution of HCN1s at AIC inputs relative to the population of glutamate synapses. Genetic deletion did not completely ablate MOR-LTD expression (Fig. 11G). This could be due to an incomplete AAV vector transduction (most recordings had a complete loss of LTD) or perhaps that HCN1 contributes to LTD differently at AIC inputs to different subtypes of MSNs. Therefore, in the future, it will also be important to investigate the role of HCN1 in MOR-LTD at AIC inputs to D1- or D2-expressing MSNs (direct or indirect pathway respectively). It has recently been reported that HCN1 contributes to alcohol preference(Salling & Harrison, 2020). Given that alcohol disrupts MOR-LTD in DLS, this mechanism of MOR-HCN1 signaling may be a target of alcohol’s effects and contribute to alcohol-related, dorsal striatal-dependent behaviors such as habitual and compulsive alcohol drinking. As the opioid oxycodone also disrupts MOR-LTD it will be intriguing to determine the impact of both oxycodone and alcohol on AIC-DLS HCN1 function. HCN1 may therefore have therapeutic potential in alcohol and opioid use disorders.

In conclusion, these findings indicate that glutamatergic synapses in the DLS have distinct mechanisms of plasticity induction mediated by the same types of receptors. The specific signaling processes that result in these different forms of plasticity might be the answer to why different types of plasticity are differentially sensitive to drugs of abuse and alcohol (Fig. 13) (Atwood et al., 2014b; Munoz et al., 2018; Muñoz et al., 2020). However, further work is required to resolve what components of these mechanisms render some synapses susceptible to the deleterious effects of these drugs while others are resistant. It will also be important to decipher the behavioral relevance of AIC-DLS synapse HCN1 channels and CB1-mTOR signaling in the DLS.