Featured ArticleResearch Articles, Cellular/Molecular

μ-Opioid Receptor Modulation of the Glutamatergic/GABAergic Midbrain Inputs to the Mouse Dorsal Hippocampus

Haram R. Kim, Soumil Dey, Gabriella Sekerkova and Marco Martina
Journal of Neuroscience 23 October 2024, 44 (43) e0653242024; https://doi.org/10.1523/JNEUROSCI.0653-24.2024

Abstract

We used virus-mediated anterograde and retrograde tracing, optogenetic modulation, immunostaining, in situ hybridization, and patch-clamp recordings in acute brain slices to study the release mechanism and μ-opioid modulation of the dual glutamatergic/GABAergic inputs from the ventral tegmental area and supramammillary nucleus to the granule cells of the dorsal hippocampus of male and female mice. In keeping with previous reports showing that the two transmitters are released by separate active zones within the same terminals, we found that the short-term plasticity and pharmacological modulation of the glutamatergic and GABAergic currents are indistinguishable. We further found that glutamate and GABA release at these synapses are both virtually completely mediated by N- and P/Q-type calcium channels. We then investigated μ-opioid modulation of these synapses and found that activation of μ-opioid receptors (MORs) strongly inhibits the glutamate and GABA release, mostly through inhibition of presynaptic N-type channels. However, the modulation by MORs of these dual synapses is complex, as it likely includes also a disinhibition due to downmodulation of local GABAergic interneurons which make direct axo-axonic contacts with the dual glutamatergic/GABAergic terminals. We discuss how this opioid modulation may enhance LTP at the perforant path inputs, potentially contributing to reinforce memories of drug-associated contexts.

Significance Statement

Corelease of an excitatory (glutamate) and an inhibitory (GABA) neurotransmitter from the same synapse is a rare finding in the nervous system, and the detailed mechanisms of this transmission are still incompletely described. Here we show that in dual glutamatergic/GABAergic synapses from the midbrain to the dentate gyrus of the dorsal hippocampus, similar calcium microdomains control the release of both transmitters. Additionally, we show that activation of μ-opioid receptors (MORs) limits release by strong inhibition of presynaptic N-type calcium channels in the mixed glutamatergic/GABAergic terminals while likely potentiating release by inhibition of axo-axonic synapses from local inhibitory interneurons. Modulation of these synapses by MORs might help reinforce memories of drug-associated contexts.

Introduction

The dorsal hippocampus is a brain area principally associated with memory and spatial navigation (Moser and Moser, 1998; Pronier et al., 2023). The dentate gyrus (DG), the primary gateway structure of hippocampal formation, receives inputs from the cortical and subcortical areas and projects to the inner hippocampal circuit forming the well known trisynaptic circuit (Yeckel and Berger, 1990; Lisman, 1999; Basu and Siegelbaum, 2015). The DG plays crucial roles for various cognitive functions including spatial learning, pattern separation, and memory encoding/retrieval (Jonas and Lisman, 2014), but more recent work has shown that the DG is also involved in the regulation of affective state (Sun et al., 2023), including assigning emotional valence to memories (Han et al., 2020).

In addition to the lateral and medial entorhinal cortex as the main input sources, two anatomically adjacent subcortical regions, the supramammillary nucleus (SUM) and the ventral tegmental area (VTA), also provide glutamatergic inputs to granule cells, the principal neurons of the DG. Distinctively, both the SUM-DG and the VTA-DG projections originate from VGluT2 (vesicular glutamate transporter 2)- and VGaT (vesicular GABA transporter)-positive neurons and corelease glutamate and the inhibitory neurotransmitter GABA (Ntamati and Luscher, 2016; Hashimotodani et al., 2018). Previous studies found that the SUM/VTA-DG circuit is important for fear-induced context memory (Han et al., 2020), spatial memory retrieval (Li et al., 2020), and contextual novelty (Chen et al., 2020). In particular, Han et al. (2020) suggested that activation of the DG granule cells inputs originating from TH (tyrosine hydroxylase)-negative, VGluT2-/VGaT-positive VTA neurons is both necessary and sufficient to induce reinstatement of freezing after extinction. Interestingly, this paper also showed that the behavioral effects due to activation of these inputs are stronger in females than in males, possibly because of the higher density of these terminals in female mice. More in general, these inputs are believed to attach a valence to memory circuits of the dorsal hippocampus.

At the cellular level, not much is known concerning the release mechanisms at these synapses. Electron microscopy suggests that glutamate and GABA release on granule cells is mediated by active zones within the same presynaptic terminal (Root et al., 2018). However, little is known about possible differences in synaptic modulation between the GABA and glutamate release. For example, it is known that in the mossy fiber terminal, which contains a very large number of active zones (Rollenhagen et al., 2007), various types of voltage-gated calcium channels (VGCCs) within a presynaptic terminal can be coupled selectively to different release sites (Chamberland et al., 2017). These authors showed that activation of P/Q channels lead to calcium signals that are spatially more homogeneous than those generated by activation of N-type channels, which, on the contrary, generate spatially heterogeneous microdomains. A close coupling between VGCCs and release sensors also characterizes inhibitory central synapses such as that between basket and granule cells of the DG, where the fast calcium chelator BAPTA, but not the slower chelator EGTA, inhibits calcium release (Bucurenciu et al., 2008). Yet, it is unknown whether different types of VGCCs are specifically coupled with the glutamate or GABA release sites at this mixed synapse, which could result in different synaptic plasticities for glutamate and GABA release.

Interestingly, the release at the SUM/VTA input to DG granule cells is heavily modulated by μ-opioid receptors (MOR), which inhibit both the GABA and the glutamate release (Han et al., 2020). The cellular mechanisms of this inhibition, however, remain obscure. Electron microscopy studies investigating the expression of MORs in the DG show that the highest MOR expression in the DG is in parvalbumin-positive GABAergic interneurons (Milner and Drake, 2001; Drake and Milner, 2006), but an interneuron-dependent regulation at this synapse would be expected to result in MOR-dependent disinhibition, not in a strong inhibition, as described (Han et al., 2020).

In the present study, we examine how MOR activation modulates the glutamatergic/GABAergic release at SUM/VTA-DG synapses and what is the mechanism that mediates this modulation.

Materials and Methods

Animals

A total of 66 VGluT2-Cre mice (Slc17a6 tm2(Cre)Lowl, Jackson Laboratories) were used for the study. All VGluT2-Cre animals received virus injections. We used 49 VGluT2 mice (21 males and 28 females) for electrophysiological experiment and 17 (6 males, 11 females) for anatomical analysis. For testing the electrically evoked currents, we used four C57BL/6J wild-type mice (two males and two females) that did not receive any injection. Animals were housed with a 14:10 h light/dark cycle at 21 ± 2°C temperature and 30–50% relative humidity. Food and water were freely accessible by animals. All experiments were conducted in accordance with guidelines of the IACUC of Northwestern University.

Stereotaxic surgery

Virus injection

Mice were anesthetized with isoflurane (∼4%) in an induction chamber and once unresponsive were placed on an operating table and fitted with an anesthesia mask. Meloxicam (20 mg/kg) was injected intraperitoneally, and a single drop of 0.5% bupivacaine HCl solution (∼0.05 ml) was applied to the wound edge. Respiration was monitored throughout the procedure. The head was fixed in the stereotaxic device, and the injector was guided with a robotic stereotaxic system (Neurostar). The cre-dependent anterograde viral vector carrying a channelrhodopsin (ChR2) construct [AAV9-Ef1a-DIO-hChR2(E123T/T159C)-EYFP, Addgene#35509, Addgene] was infused uni- or bilaterally into the anterior-targeted (SUM, AP, −2.6 mm; ML, ±0.3 mm; DV, −4.8 mm; total of 18 mice, 7 males and 11 females) or posterior-targeted location (VTA, AP, −3.5 mm; ML, ±0.5 mm; DV, 4.3 mm; total of 31 mice, 14 males and 17 females) for patch-clamp recording. Because of the procedure used to obtain the slices for electrophysiological recordings, we were not able to verify post hoc the exact location of the injection sites and the extension of the viral infection in these animals, but, considering the proximity of the two structures, some degree of cross-contamination is likely. ChR2-fused EYFP was expressed abundantly in the axons of targeted neurons, but the synaptic terminals were generally not detectable (Fig. 1). Thus, this expression pattern is ideal electrophysiological recordings, as axon terminals are likely devoid of opsin and thus activate in natural way following the axonal depolarization, which prevents artifacts caused by ChR2 overexpression. On the other hand, this expression pattern is not conducive to immunostaining studies to investigate the synaptic terminals. Therefore, for immunohistochemistry and in situ hybridization, we used an anterograde AAV virus without ChR2 (AAV9-Ef1a-DIO-EYFP, Addgene#27056) that was injected into the anterior (five mice) or posterior (10 mice) target location or a retrograde virus (retroAAV-Ef1a-DIO-EYFP, Addgene#27056_rg; two mice) injected uni- or bilaterally into the dorsal DG (AP, −3.0 mm; ML, ±1.9 mm; DV, −2.0 to −2.2 mm). All injections were performed using a NanoFil microsyringe with a 33–34 gauge needle (World Precision Instruments). The viral vectors were infused at a flow rate of 0.1 µl/min; after the infusion, the needle was left in place of another 10 min. We tested different volumes (0.15–0.25 µl per site) and different concentrations [titer 1.0 × 1012–1.0 × 1013 GC/ml, 1–10 times dilution of the stock solution in phosphate-buffered saline (PBS)] of viral solution and found that the size of the recorded current responses was unaffected. Therefore, we assume that even with the smallest volume and titer, the cells in the injection targets were fully infected. For immunostaining, we used the least amount of virus particles (0.15 µl, 1.0 × 1012 GC/ml) to prioritize target selectivity overexpression amount. After completing the injection, the needle was slowly withdrawn from the brain, and the skin wound was sutured. Animals were returned to their home cage after 1 h of observed recovery. We waited for 4–7 weeks after surgery to ensure successful gene expression, 4–5 weeks for AAV9-Ef1a-DIO-EYFP, and 5–7 weeks for retroAAV and AAV9-Efla-DIO-hChR2-EYFP reaching maximized expression.

Figure 1.
Figure 1.

Differential staining of VGluT2-positive terminals with ChR2-fused EYFP versus EYFP only. A, Immunostaining for VGluT2 and ChR2-fused EYFP (mice infected with AAV9-Ef1a-DIO-hChR2(E123T/T159C)-EYFP; see Materials and Methods) in DG of the mouse dorsal hippocampus. ChR2-fused EYFP was mainly detected in axon fibers but generally failed to reach the terminals where VGluT2 puncta are located. B, In contrast to the findings in panel A, when the mice were injected with a construct only encoding the reporter (AAV9-Ef1a-DIO-EYFP), EYFP and VGluT2 staining were largely overlapping. All images obtained from 30-µm-thick sections and using confocal optical sections of ∼1 µm thickness. Scale bars, 5 µm.

Electrophysiological recordings

Slice preparation

Mice were deeply anesthetized using isoflurane and killed. The brain was quickly dissected in ice-cold solution, and 300-µm-thick transverse slices of the dorsal hippocampus were cut using a vibro-slicer (Leica VT-1200) in ice-cold artificial cerebrospinal fluid (ACSF) containing the following (in mM): 125 NaCl, 25 NaHCO3, 2.5 KCl, 1.25 NaH2PO4, 25 glucose, 2 CaCl2, and 1 MgCl2, saturated with 95% O2 and 5% CO2 to pH 7.4. Before being used for recordings, the slices were kept at 34°C for ∼15 min and then transferred to room temperature (22–24°C) and incubated in the same solution for ≥30 min. Although the SUM/VTA axon terminals were found throughout the dorsoventral level of DG (as confirmed by immunostaining; data not shown), observable responses were only detectable in slices from the ventral hippocampus. Thus, all electrophysiological recordings were obtained from dorsal hippocampal slices (up to 1.5 mm from the most dorsal transverse hippocampal slice; 2–4 slices per mouse harvested).

Whole-cell patch–clamp recording

Pipettes were pulled from the borosilicate glass (1.5 mm OD; Sutter Instruments) using a horizontal puller (Sutter Instruments P-97) and had resistances of 5–7 MΩ, when filled with KCl-based internal solution containing the following (in mM): 148 KCl, 6 NaCl, 2 MgATP, 0.2 Na3GTP, 0.1 EGTA, 10 HEPES, pH 7.3 (with 2 N KOH). Lidocaine N-ethyl bromide (10 mM final; N-5783, Sigma-Aldrich) was added to the internal solution to block most voltage-gated channels and prevent action potential firing. DG granule cells were visualized using an upright microscope (Scientifica) with a 60× water-immersion objective lens (Olympus) and connected to a digital camera (DVC). All electrophysiological measurements were performed at 30–32°C using an Axopatch 200B amplifier and pClamp9 software (Axon Instruments) and analyzed using Clampfit10 (Axon Instruments) running on a PC. Signals were filtered at 2 kHz and sampled at 10 kHz. Only cells with access resistance <30 MΩ and resting potential in the range of −70 to −85 mV (measured immediately after obtaining electrical access) were used for recordings. Cells with membrane resistance >500 MΩ were also discarded, as they were likely newly generated, immature granule cells (Schmidt-Hieber et al., 2004). Recordings were started 1–2 min after access, to allow the membrane potential to stabilize. The electrophysiological properties were monitored during the recordings using the membrane test function embedded in pClamp9 (+5 mV pulse; at −70 mV) to confirm that the seal condition remained stable before and between the stimulation protocols. For the recordings investigating the effect of voltage–calcium channels, we used a cesium chloride-based internal solution (in mM: 148 CsCl, 6 NaCl, 2 MgATP, 0.2 Na3GTP, 0.1 EGTA, 10 HEPES, titrated to pH 7.3 by 1 M CsOH) to improve clamp conditions, and when measuring GABAergic current, we also added 10 mM EGTA to prevent current down. Synaptic currents were elicited by electrical or optical stimulation. For electrical stimulation, a concentric bipolar electrode (FHC) connected to an isolator (A.M.P.I.) was placed in the outer molecular layer of the DG, 100–200 µm away from the recorded cell and used to deliver 50 µA pulses of 0.2 ms duration. For optical stimulation, brief pulses of blue LED light (2 ms) were delivered using a collimated 470 nm LED (M470L2, Thorlabs) mounted on an upright microscope that projects the beam through the objective lens. The light intensity was set to the maximum power using an LED driver (LED1DBm, Thorlabs). Both electrical and optical stimulators were connected to DIGIDATA 1322A, and the stimulation parameters (onset time, duration, and timing of the stimulus relative to the recording) were programmed using pClamp9 software (Axon Instruments). We repeated the same protocol at least five times at 0.03 Hz (one trial per 30 s), and the recorded traces were averaged to measure the peak amplitude of the current. The paired-pulse ratio (PPR) was measured using a 250 ms interval. Electrically evoked responses were recorded from four C57BL/6J wild-type mice. These recordings were obtained at −70 mV using a potassium gluconate internal solution containing the following in mM: 140 K-gluconate, 8 NaCl, 2 MgATP, 0.2 Na3GTP, 0.1 EGTA, and 10 HEPES, pH 7.3 (with 2 N KOH), containing 10 mM lidocaine N-ethyl bromide to block voltage-gated channels.

Dissection of the optically evoked current components

For these recordings, we used a KCl-based internal solution. Optically evoked currents were recorded at −70 mV holding potential in the absence and in the presence of blockers of the ionotropic GABA and glutamate channels (Fig. 2A,B). As expected in these experimental conditions where both current types are inward at −70 mV, the total response recorded in the absence of synaptic blockers consisted of the sum of the glutamatergic and GABAergic currents since the sequential application of kynurenic acid (a blocker of ionotropic glutamatergic current) and picrotoxin (GABAA current blocker) in the same recording first reduced and then completely abolished the response. The time latency of the responses (∼2 ms for both components) further showed that both glutamatergic and GABAergic currents are monosynaptic (Fig. 2C,D).

Figure 2.
Figure 2.

Activation of the SUM/VTA-DG synapses induces monosynaptic glutamatergic/GABAergic currents. A, The size of currents evoked by optical stimulation (460 nm, 2 ms) of SUM/VTA inputs to the DG. Inset, Traces recorded in control (without blockers, black trace), in the presence of glutamatergic current antagonist kynurenic acid (3 mM; blue trace), and in the presence of both glutamatergic and GABAergic current antagonists (kynurenic acid and 50 µM picrotoxin, flat black trace). Granule cells were voltage-clamped at −70 mV (KCl-based internal solution). B, The current size in each condition normalized to the control current. C, Current latency distribution for the glutamatergic (top) and the GABAergic current (bottom). Fitting the histograms (bin size = 0.1 ms) using a Gaussian curve yielded peak values of 1.9 ms (n = 110) for the glutamatergic and 2.0 ms (n = 39) for the GABAergic current. R2 of the fit was 0.82 and 0.93, respectively. D, The average latency of the two current types was similar; 1.94 ± 0.03 ms (mean ± SEM) for glutamatergic and 2.08 ± 0.05 for GABAergic currents (n = 110 and 39, respectively; p = 0.72; unpaired t test).

Drugs

Picrotoxin (Tocris Bioscience) was prepared as stock solution (100 mM in DMSO) and used at a final concentration of 50 µM. [d-Ala2, N-MePhe4, Gly-ol]-enkephalin (DAMGO; Tocris Bioscience or Abcam) stock solution (1 mM) was prepared in water. The working solution was 10 µM. Both stock solutions were stored at −20°C. On the day of the experiment, the stock was added to the ACSF and applied using the bath perfusion system (3–5 ml/min). The pharmacological effects were tested after at least 10 ml of drug containing ACSF (∼4 times of the volume of the recording chamber) was perfused. In a subset of neurons, DAMGO was used at 1 µM; because we did not detect any significant difference between the data obtained with the two different concentrations, these data were pooled with the 10 µM DAMGO group.

To isolate the GABAergic currents, the glutamatergic current was blocked using 3 mM kynurenic acid (Sigma-Aldrich; 500 mM stock solution made in 1 N NaOH kept at 4°C) or the combination of 20 µM CNQX disodium salt (Tocris Bioscience; 20 mM stock in water and kept at −20°C) and 50 µM D-APV (Hello Bio; 50 mM stock in water).

ω-Conotoxin GVIA (Sigma-Aldrich) and ω-agatoxin IVA (Peptide Institute) were stored at −20°C as 10× stock solution dissolved in HEPES-ACSF (in mM: 138 NaCl, 2.5 KCl, 10 HEPES, 25 glucose, 2 CaCl2, 1 MgCl2, titrated with 1 N NaOH), pH 7.4, and used at final concentrations of 1 µM for ω-conotoxin and 0.2 µM for ω-agatoxin. When using these drugs, the final concentration was dissolved in HEPES-ACSF containing 0.1% bovine serum albumin to saturate possible adhesion sites on tubing. These two drugs were delivered through a quartz glass capillary placed ∼300 µm above the cell. To determine the drug effect, we first waited until the size of light-induced currents was stable for three consecutive traces (1.5 min); after that, we recorded five traces that were averaged to obtain the control current, and then the drug of interest was applied. Then, we waited again until the current was stable and then recorded five traces that were averaged and used to measure the peak amplitude of the current. Finally, DAMGO was added to the bath using the perfusion line, and the same protocol was repeated.

Immunohistochemistry

Mice were anesthetized with intraperitoneal sodium pentobarbital at a dose of 60 mg/kg and then perfused using 4% freshly depolymerized paraformaldehyde (PFA) in phosphate buffer. Brains were dissected 1 h later, postfixed in the same solution overnight, and then moved to 30% sucrose in PBS. Coronal and sagittal, 30-µm-thick sections were cut on microtome equipped with a freezing stage. Coronal section containing the dorsal hippocampus or sagittal sections containing the SUM/VTA were incubated in blocking solution [3% normal goat serum, 1% bovine serum albumin with 0.2% Triton X-100 in Tris-buffered saline (TBS)] for 1 h, followed by incubation with the primary antibodies for 72 h at 4°C (rabbit anti-MOR, 1:1,000, Frontier Institute, catalog #MSFR104190; guinea pig anti-VGluT2, 1:5,000, Merck Millipore, catalog #AB2251-I; chicken anti-GFP, 1:5,000, Abcam, catalog #ab13970; mouse anti-TH, 1:50 Merck Millipore, catalog #MAB318; guinea pig anti-VGaT, 1:2,000, Synaptic Systems, catalog #213004) diluted in 1% normal goat serum, 1% bovine serum albumin with 0.2% Triton X-100 in TBS. After washing four times with TBS (20 min each), the sections were incubated with the secondary antibodies (Alexa Fluor 594 goat anti-mouse IgG, catalog #A-11005; Alexa Fluor 488 goat anti-chicken IgG, catalog #A-11039; Alexa Fluor 594 goat anti-rabbit IgG; catalog #A-11012; Alexa Fluor 647 goat anti-guinea pig IgG, catalog #A-21450; all from Thermo Fisher Scientific, 1:500 in the same solution as primary antibodies) for 1 h. The sections were washed again four times with TBS (20 min each), mounted on gelatin-coated slides and coverslipped with Mowiol mounting medium. The fluorescence images were captured with a Nikon AX R confocal microscope (Nikon Instruments) controlled using the NIS-Elements software. For triple color colocalization, high-resolution images (1,024 × 1,024) were captured using a CFI Plan Apochromat Lambda D 60× Oil (NA 1.42) objective. To minimize the channel spillover, different colored images were sequentially captured. Postacquisition, all the images were edited for adjustment of brightness and contrast in Adobe Photoshop (Adobe Systems). Note the EYFP is a genetic mutant of GFP and was visualized using the anti-GFP antibody, as previously reported in the literature (Morales et al., 2022; Yan et al., 2022).

Quantification of MOR colocalization with VGluT2, VGaT, and GFP

To quantify the colocalization of MOR, we used the Vglut2-cre mice anterogradely labeled using AAV virus injection. Every sixth hippocampal sections from three mice (two males one female) were used for quantification of MOR coexpression with GFP and VGluT2 or VGaT. The immunolabeled puncta were counted using the “cell counter” plugin of open-source software Fiji (Fiji/ImageJ http://imagej.nih.gov/ij/i) and expressed as the percentage of total MOR-positive puncta.

In situ hybridization assay

Brains from VGluT2-Cre mice injected with retrograde AAV-DIO-EYFP in the DG were prepared following the same procedure used for immunostaining, but the dissected brains were postfixed in 4% PFA for an additional 24 h and then transferred to 30% sucrose solution with 0.1% diethylpyrocarbonate. Coronal, 14-µm-thick sections were cut on a freezing stage microtome. Sections were mounted on glass slides and treated with 0.9% H2O2 and 10% methanol in PBS for 10 min, followed by a treatment with 9.86 mM sodium citrate buffer (pH 6.0) for 2 min at 91°C. Next, the sections were incubated with 1 µg/ml proteinase K (catalog #03115887001, Roche Diagnostics) for 15 min at 40°C. MOR probe (Mm-oprm1, Advanced Cell Diagnostics) was hybridized to the sections for 2 h at 40°C. Hybridization signals were amplified using single-probe assay kit reagents (RNAscope 2.5 High-Definition Brown Assay kit, Advanced Cell Diagnostics). Between each amplification step, sections were washed in 0.1× SSC buffer at 33°C, with the last three washes at room temperature. The amplified mRNA probes were detected using 3,3′-diaminobenzidine (brown color). The sections were next immunostained with anti-GFP antibody to detect the virus-infected neurons. We used ABC method (Vector Laboratories) and SG substrate (blue color; Vector Laboratories) to visualize the bound antibody. Sections were coverslipped with Vectamount (Vector Laboratories).

Data presentation, quantification, and statistical analyses

Unless noted otherwise, all data in the results section are presented as mean ± SEM. Whisker plots show median (lines inside boxes), 25th and 75th percentiles (box margins), and 10th–90th percentiles (whiskers). Individual points on the box plots represent data outside the 10th and 90th percentile range. Crosses within the boxes indicate the means. P values were calculated using the two-tailed Mann–Whitney U test when comparing two independent groups and paired t test when comparing conditions before and after drug application in the same cell. In case comparing three groups, one-way ANOVA and various post hoc multiple-comparison tests were adopted: Bonferroni’s test (three groups were independent), Holm–Sidak test (three conditions were paired), and mixed-effect model (three conditions were paired, but in some case the value was missing at one condition). All statistical calculations were performed using the Prism 8 software (GraphPad Software).

Results

The dual glutamatergic/GABAergic current elicited in DG granule cells by activation of SUM/VTA inputs is monosynaptic

To characterize subcortical glutamatergic projections from the SUM and the VTA to the hippocampus, we injected Cre-dependent retrograde virus encoding EYFP into the DG of VGluT2-Cre mouse. Optical activation of these synapses elicited both glutamatergic and GABAergic currents with monosynaptic latencies (∼2 ms for both currents; Fig. 2; Kelly and Martina, 2018). We found that a majority of retrogradely infected neurons were located in SUM, although numerous positive neurons were also present in the rostral part of the VTA (Fig. 3A). Based on these observations, we adopted two coordinate schemes for ChR2 virus injection targeting SUM and VTA for electrophysiological recordings (Fig. 3B). In line with previous literature (Chen et al., 2020; Han et al., 2020; Ajibola et al., 2021), ChR2-infected fibers were found in the DG and CA2 of the hippocampus (Fig. 3C, inset) and were densely packed in the DG granule cell layer (Fig. 3C). Optical stimulation using short flashes of blue (470 nm) LED light evoked in postsynaptic granule cells glutamatergic and GABAergic currents (Fig. 3D) that could be isolated pharmacologically. Picrotoxin (50 µM) was used to detect the glutamatergic and kynurenic acid (3 mM) to detect the GABAergic currents.

Figure 3.
Figure 3.

Inputs from anterior and posterior midbrain elicit dual glutamate–GABA currents in DG granule cells. A, Top, Sagittal brain atlas showing both the presynaptic areas (supramammillary nucleus, SUM; ventral tegmental area, VTA) and the postsynaptic target (hippocampal dentate gyrus, DG). Bottom, Retrogradely labeled VGluT2-positive SUM and VTA neurons (green) in a sagittal section. TH-positive dopaminergic neurons (red) allow the identification of the VTA. Although the majority of retrogradely labeled VGluT2 neurons were detected in SUM, the rostral part of VTA (marked by arrows) was also abundantly stained. fr, fimbria. B, Injection sites of channelrhodopsin-containing anterograde virus for electrophysiological recordings. Left, Anterior-target site (in mm: AP, −2.6; ML, ±0.3; DV −4.8). Right, posterior site (in mm: AP, −3.5; ML, ±0.5; DV, −4.3). Medial mammillary nucleus (MM) and interpeduncular nucleus (IPN) are shown to help identify the injection location. C, Abundant expression of VGluT2-positive presynaptic fibers in the DG following viral infection of SUM/VTA. The lower-magnification image (inset) shows the entire structure of the dorsal hippocampus. The infected synaptic terminals are mainly located in the granule cell layer. D, Amplitude of the synaptic currents evoked in DG granule cells by photoactivation of the SUM/VTA axons in transverse hippocampal slices. Responses were elicited using a 2 ms flash of blue light (blue line above the traces) and recorded in the presence of either 50 µM picrotoxin (glutamatergic current) or 3 mM kynurenic acid (GABAergic current) at −70 mV. Orange color represents glutamatergic and purple GABAergic responses. Note that the individual data points only represent data outside the 10–90th percentile range. Showing each individual data point would make the plot to crowded (the total sample size is 110 and 39 cells for the glutamate and GABA currents, respectively).

Because the currents elicited in slices from mice injected in the anterior or posterior midbrain sites were all dual glutamatergic/GABAergic and monosynaptic and did not differ in the decay time constant of the glutamatergic component [the time constant was 5.6 ± 0.62 ms, for the anterior injection targets (14 cells from five male and nine female mice) and 5.1 ± 0.91 ms for the posterior injection targets (14 cells from six male and eight female mice; data not shown)], from here on they are pooled for analysis and collectively referred to as the SUM/VTA currents.

Activation of MOR disinhibits the DG but inhibits SUM/VTA-DG inputs

Next, we tested the effect of MORs on the currents elicited in granule cells by activation of SUM/VTA inputs. It is well known that the MOR activation induces disinhibition of the DG (Neumaier et al., 1988; Mayer et al., 1994; Drake et al., 2007), which is attributable to the abundant expression MORs in parvalbumin-positive GABAergic interneurons (Drake and Milner, 2006) that are therefore strongly inhibited by MOR activation causing granule cell disinhibition. Accordingly, when we performed indiscriminate electrical stimulation of the DG, which primarily excites the perforant pathways, MOR activation (10 µM DAMGO in the bath) produced the expected potentiation of the DG responses, and this disinhibitory effect was eliminated by blocking GABAA channels (Fig. 4A). In contrast to the electrical stimulation, the current evoked by optical stimulation, which selectively excited SUM/VTA-DG synaptic terminals, was heavily inhibited by DAMGO (Fig. 4B,C), as previously reported (Han et al., 2020). MOR activation decreased the SUM/VTA-DG synaptic currents in 69 of 70 neurons, regardless of whether we measured the GABA or the glutamate-mediated responses (Fig. 4B,C). We did not observe a clear reduction in only 1 of the 70 recorded responses (this was a GABAergic current). On average, DAMGO increased the size of electrically evoked currents to ∼185% of the control value (n = 10), while it decreased the size of optically evoked currents to ∼24% of the control value (n = 70; Fig. 4D). Bath application of DAMGO decreased the amplitude of light-evoked glutamatergic currents from 158.7 ± 18.6 pA to 26.8 ± 4.0 pA (n = 42; p < 0.0001; paired t test) and of GABAergic currents from 288.1 ± 38.7 pA to 82.9 ± 10.7 pA (n = 28; p < 0.00001; Fig. 4B,C). Therefore, activation of MORs inhibits release from SUM/VTA-DG inputs, while, in line with previous literature (Drake et al., 2007), it causes a general disinhibition of DG indicating that the effect of MOR activation in DG is complex and synapse-specific.

Figure 4.
Figure 4.

Activation of MORs disinhibits the DG but inhibits both the glutamatergic and the GABAergic current at the SUM/VTA-DG synapses. A, Representative currents recorded in response to electrical stimulation in the absence (black trace) or in the presence (red trace) of the MOR agonist DAMGO (10 µM). The currents were elicited by stimulation (50 µA, 0.2 ms; black vertical line) in the DG outer molecular layer and measured using K-gluconate internal solution at −70 mV and thus represent mostly glutamatergic currents (top). When the GABAA signaling was blocked using picrotoxin (50 µM), the disinhibitory effect of DAMGO was completely abolished (bottom). B, Optogenetically activated glutamatergic and GABAergic responses elicited by activation of SUM/VTA terminals in the absence (black traces) and in the presence (colored traces) of DAMGO. Left, Glutamatergic currents (n = 42). Right, GABAergic currents (n = 28). Currents were elicited using 2 ms, blue light flashes (vertical blue lines) at a holding potential of −70 mV. Glutamatergic currents were isolated using picrotoxin (50 µM) and GABAergic currents using kynurenic acid (3 mM) in the bath. Scale bars for traces, 100 pA (vertical) and 50 ms (horizontal). C, Summary plot showing the average effect of DAMGO on the optogenetically evoked glutamatergic and GABAergic currents. D, Current responses elicited by electrical (10 cells collected from 4 mice; see Materials and Methods) and optical stimulations (70 cells from 20 mice) in DAMGO normalized to the responses in control condition. For the optical stimulation, we combined both glutamatergic (n = 42) and GABAergic (n = 28) current responses shown in Figure 4B, where the data are shown disaggregated (p < 0.0001 for DAMGO effect on optically vs electrically evoked responses; Mann–Whitney test).

MOR-mediated inhibition of SUM/VTA VGluT2-positive synapses is mainly presynaptic

To investigate whether the MOR activation acted at the pre- or postsynaptic level, we tested its effect on paired-pulse plasticity using two paired light pulses with 250 ms intervals (Fig. 5). In the absence of MOR activation, we observed ∼50% depression in the PPR (Fig. 5A, black trace). In control conditions (no DAMGO), the paired-pulse depression was slightly larger for glutamatergic compared with GABAergic currents (the current amplitude elicited by the second pulse was 0.44 ± 0.02 of that by the first pulse for glutamatergic and 0.48 ± 0.01 for GABAergic currents, n = 42, 28, respectively; p = 0.012 by Mann–Whitney test; Fig. 5B,C, “CTR”). Overall, DAMGO increased the PPR (Fig. 5A, red trace) of both glutamatergic (n = 42 and p < 0.0001; paired t test for control vs DAMGO) and GABAergic currents (n = 28 and p = 0.0006; Fig. 5B). Only in 16% (7 of total 42) of the neurons recorded for glutamatergic and in 20% (7 of 35) of neurons recorded for GABAergic current DAMGO did not increase the PPR (Fig. 5B, pie charts). There appeared not to be any correlation between these outliers and either extent of current amplitude reduction by DAMGO or site of viral infection. On average, the PPR in DAMGO was similar for the two currents (0.73 ± 0.04 and n = 42 for glutamatergic and 0.63 ± 0.04 and n = 28 for GABAergic current; p = 0.46 Mann–Whitney test for PPR in glutamate vs GABA currents; Fig. 5C).

Figure 5.
Figure 5.

Increase in PPR suggests a presynaptic origin for MOR modulation of SUM/VTA-DG synapses. A, Representative glutamatergic current evoked by paired-pulse stimulation (250 ms apart; blue bar on the traces) in control (black) and in the presence of μ-opioid agonist (10 µM DAMGO; red). B, PPR of synaptic currents in individual neurons before and during MOR activation. Inset, A pie chart showing the proportion of neurons that either decrease (color-filled) or increase (empty) their response in DAMGO compared with control condition. Numbers within the charts indicate the number of neurons. C, Group averages of PPR. For glutamatergic current, n = 42 and p < 0.0001 for control versus DAMGO by paired t test. For GABAergic current, n = 28 and p = 0.0006.

The change in PPR shows that presynaptic mechanisms are involved in the MOR-dependent modulation of SUM/VTA-DG corelease synapses (Regehr, 2012). This is in keeping with previous studies showing that MOR expression in granule cells is very low (Drake et al., 2007). Thus, the effects of DAMGO must be mediated by MORs expressed in the terminals of VGluT2-positive SUM and VTA neurons projecting to the hippocampus. To verify this hypothesis, we injected Cre-dependent retrograde viral constructs encoding EYFP into the DG and combined the retrograde staining (using anti-GFP antibody) with in situ hybridization for MOR. These experiments confirmed that MOR transcript is expressed in SUM and VTA and is often colocalized with the retrograde tracer. A qualitative analysis of these data suggests that most (if not all) of the retrogradely traced neurons did express MOR mRNA (Fig. 6A). Moreover, immunostaining of the DG confirmed that MOR1 protein is expressed in SUM/VTA terminals in the granule cell layer, where it is often colocalized with GFP and VGluT2 or VGaT (GABA vesicular transporter; Figs. 6B–D, 8C,D).

Figure 6.
Figure 6.

VGluT2-positive SUM and VTA neurons projecting to the hippocampus express MORs. A, In situ hybridization of MOR1 transcript in hippocampal-projecting VGluT2–positive neurons in SUM/VTA. Left, Low-magnification images showing SUM/VTA area (scale bar, 250 µm). Middle, Higher-magnification images of the SUM (left) and the VTA (right; scale bar, 10 µm). Brown dots represent MOR1 mRNA. SUM and VTA VGluT2-positive neurons projecting to the DG were identified using Cre-dependent retrograde virus injections in the DG and visualized using GFP immunostaining. Arrows indicate VGluT2-positive cells retrogradely stained and expressing MOR transcript. B, Immunostaining via anti-GFP antibody of VGluT2-positive synaptic terminals in DG. EYFP-containing virus was injected into SUM/VTA. Scale bar, 100 µm. C, Triple immunostaining images for VGluT2, GFP, and MOR1 in the DG. EYFP expression was driven in VGluT2-positive SUM/VTA terminals by virus injection, while MOR1 and VGluT2 staining identifies native protein expression. Scale bar, 5 µm. D, Quantification of MOR staining relative to VGlut2- and GFP-positive terminals. 36.1 ± 2.0% (3 animals) of MOR staining was colocalized with both VGluT2 and GFP, while 7.4 ± 0.5% of MOR-positive staining was located in VGluT-positive but GFP-negative structures.

MOR inhibition of corelease synapses is largely mediated by N-type VGCCs

Finally, we investigated the identity of the presynaptic calcium channels affected by MOR activation. In glutamatergic synapses, both N- and P/Q-type calcium channels mediate the calcium influx that drives synaptic transmission, although with different efficacies (Luebke et al., 1993; Li et al., 2007). In GABAergic synapses, however, this is not always the case, as either N- or P/Q-type channels were shown to mediate mostly synchronous or asynchronous release in different interneuron types (Hefft and Jonas, 2005). Our data show that at the SUM/VTA to DG synapse N- and P/Q-type channels each accounted for ∼50% of the release (Fig. 7A,B,D,E). Then we investigated whether the MOR activation selectively affects one or the other calcium channel types, similar to what described in the spinal cord (Heinke et al., 2011). We found that DAMGO completely blocked the residual glutamate current recorded in the presence of the P/Q-type antagonist ω-agatoxin IVA (0.2 µM), while the current recorded in the presence of the N-type antagonist ω-conotoxin GVIA (1 µM) was only partly affected by DAMGO (∼60% of the glutamatergic currents was DAMGO-resistant in the presence of conotoxin; Fig. 7A–C). Additionally, we also tested the effect of the combined application of ω-conotoxin and agatoxin, which resulted in the almost complete block of the glutamate current (95.9 ± 0.2% blocked; n = 3; Fig. 7E). The differential effect of DAMGO on N- and P/Q-mediated release was similar for the GABAergic current (Fig. 7F–G). These results show that neurotransmitter release at the SUM/VTA-DG synapses is almost entirely mediated by only N- and P/Q-type calcium currents and that MOR modulation of release is preferentially mediated by presynaptic N-type calcium channels.

Figure 7.
Figure 7.

MOR inhibition of corelease synapses is preferentially mediated by N-type calcium channels. A, Plots representing the effect of ω-conotoxin GVIA (1 µM) and ω-conotoxin GVIA + DAMGO on glutamatergic currents. Left, Data points from individual cells. Right, Averages for each condition. Insets show representative current traces. Blue traces represent the currents recorded in the presence of DAMGO. n = 10; p = 0.0002 for control versus conotoxin; p = 0.0002 for conotoxin versus conotoxin + DAMGO; p < 0.0001 for control versus conotoxin + DAMGO [repeated-measure one–way ANOVA followed by Holm–Sidak test (F = 76)]. B, Same as A, for 0.2 µM ω-agatoxin IVA. n = 6; p = 0.0005 for control versus agatoxin; p = 0.0005 for agatoxin versus agatoxin + DAMGO; p < 0.0001 for control versus agatoxin + DAMGO [repeated-measure one–way ANOVA followed by Holm–Sidak test (F = 235)]. C, Summary plot showing the differential effect of DAMGO on glutamate release mediated by P/Q- and N-type calcium channels. The effect of DAMGO in the presence of conotoxin is normalized to the size of the current recorded in conotoxin only, and the effect of DAMGO in the presence of agatoxin is normalized to the recordings in agatoxin only. The “DMG only” box shows the effect of DAMGO on the current obtained without any calcium blocker (these are the same cells showed in Fig. 4C, left). The sample size was 10, 6, and 42, respectively. p < 0.0001 for conotoxin + DAMGO versus agatoxin + DAMGO and also for conotoxin + DAMGO versus DAMGO only; p = 0.67 for agatoxin + DAMGO versus DAMGO only, calculated by one-way ANOVA followed by Bonferroni’s test (F = 39). D, Inhibition of the glutamatergic current by the two different VGCC blockers. n = 10 versus 6 for conotoxin versus agatoxin; p = 0.43 by Mann–Whitney test. E, Glutamatergic current responses in the presence of both conotoxin and agatoxin. Inset shows representative traces obtained in control, in conotoxin (black traces), and in conotoxin + agatoxin (red). The sample size is 3. p = 0.046 for control versus conotoxin; p = 0.046 for conotoxin versus conotoxin + agatoxin; p < 0.0001 for control versus conotoxin + agatoxin [repeated-measure one–way ANOVA followed by Holm–Sidak test (F = 101)]. F, Plots representing the effect of ω-conotoxin GVIA (1 µM) and ω-conotoxin GVIA + DAMGO on GABAergic currents. Insets show representative current traces (DAMGO traces are in blue). Left, Currents recorded in control condition (kynurenic acid only), in the presence of 1 µM ω-conotoxin GVIA, and in the presence of ω-conotoxin + DAMGO. Recordings from six cells; p = 0.01 for control versus conotoxin; p = 0.05 for conotoxin versus conotoxin + DAMGO; and p = 0.004 for control versus conotoxin + DAMGO (mixed-effect model analysis). Right panel, Same as the left panel but using 0.2 µM ω-agatoxin IVA; n = 5; p = 0.062 for control versus agatoxin; p = 0.039 for agatoxin versus agatoxin + DAMGO; p < 0.0001 for control versus agatoxin + DAMGO (6 cells; mixed-effect model analysis). G, Summary plot showing the differential effect of DAMGO on GABA release mediated by P/Q- and N-type calcium channels. The effect of DAMGO in the presence of conotoxin is normalized to the size of the current recorded in conotoxin only, and the effect of DAMGO in the presence of agatoxin is normalized to the recordings in agatoxin only. The “DMG only” box shows the effect of DAMGO on the current obtained without any calcium blocker (these are the same cells showed in Fig. 4C, right). The sample size was 5, 4, and 28, respectively. P = 0.01 for conotoxin/DAMGO versus agatoxin/DAMGO and p = 0.26 for conotoxin/DAMGO versus DAMGO only; p = 0.07 for agatoxin/DAMGO versus DAMGO only, calculated by one-way ANOVA followed by Bonferroni’s test (F = 5).

Discussion

We investigated the synaptic dynamics, calcium channel dependence, and MOR modulation of the glutamate and GABA corelease from SUM/VTA inputs to the dorsal hippocampal DG. Because no differences in any of these parameters were detectable, the data obtained with the SUM and VTA targeted injections were pooled.

These are peculiar synapses that corelease glutamate and GABA. There are several mechanisms that may mediate corelease of different transmitters (Tritsch et al., 2016). In the case of these synapses, it was previously shown using electron microscopy analysis that the corelease is mediated by spatially segregated active zones within the same terminals (Root et al., 2018). Yet, numerous details concerning the mechanism of the release were still missing. Because GABA- and glutamate-releasing synapses can differ with regard to the identity of the VGCCs mediating the release (Hefft and Jonas, 2005; Li et al., 2007; Wender et al., 2023), we used a pharmacological approach to dissect the identity of the calcium channels mediating the release at this mixed synapse, and we found that N- and P/Q-type channels have broadly equal contributions (Fig. 7D). Additionally, when agatoxin and conotoxin were coapplied, we observed an almost complete block (Fig. 7E), suggesting that the contribution from other channels is minimal, if any. We also studied the paired-pulse plasticity of these synapses using a 250 ms interval between stimulations. At this time interval, the synapses showed strong depression, which was the same in animals where the intracranial injections were preferentially targeting the more anterior (SUM) or more posterior structure (VTA). In regard to the study of paired-pulse plasticity, it is worth stressing that, as noted in the methods section, when we drove expression of EYFP-fused ChR2, the successive immunostaining showed that expression was abundant in axons but was not detectable in terminals. Thus, it might be argued that release at these terminals was mostly mediated by physiologically occurring opening of VGCCs caused by axonal depolarization and not by any direct ChR2-mediated depolarization at the terminal.

Modulation by MORs

In keeping with the published literature, bath application of the MOR agonist DAMGO produced disinhibition of granule cells (Neumaier et al., 1988; Mayer et al., 1994; Bramham and Sarvey, 1996; Akaishi et al., 2000) but strong inhibition of the SUM/VTA synapses (Han et al., 2020). Pairing MOR in situ hybridization with retrograde labeling from the dorsal DG showed abundant MOR RNA expression in the SUM and VTA neurons projecting to the hippocampus, suggesting that the inhibitory effect of DAMGO is presynaptic. This hypothesis is supported by the change in paired-pulse plasticity caused by DAMGO and by the localization of MORs in presynaptic terminals. Additionally, our data show that PPR, identity of calcium channels driving release, and the calcium channel specificity of the DAMGO effect was similar for the GABA and the glutamate currents. These findings are in line with electron microscopy data showing that glutamatergic and GABAergic active zones are located within the same terminals (Root et al., 2018), However, close inspection of the data revealed one interesting difference in the DAMGO effect between the two currents: the fraction of DAMGO-insensitive current was significantly larger for the GABA component compared with the glutamatergic component (Fig. 8A,B). We thought that this unexpected difference could be explained by the different experimental conditions in which the two currents are recorded: the glutamate current was recorded in the presence of the GABA blocker picrotoxin, while the GABA current in the presence of kynurenic acid (or APV and CNQX). Therefore, when recording the GABA current, any disinhibitory effect of DAMGO potentially mediated by spontaneously firing interneurons may still be present and, if local interneurons provide inhibitory control on the SUM/VTA presynaptic terminals, would lead to a DAMGO-mediated potentiation of synaptic release. This potentiation would thus explain the differential inhibition of the two current components. If that is indeed the case, we predicted that MORs are present in VGaT-positive terminals that do not colocalize with GFP staining (the marker of SUM/VTA terminals) in sections from VGluT2-Cre mice infected with EYFP-encoding viral constructs injected in the SUM/VTA. In line with these predictions, quantification of the costaining data revealed that a large fraction (∼30%) of the MOR-expressing structures in the DG did not colocalize with EYFP (which is an accurate reporter for the VTA/SUM terminals; Fig. 6D) but overlapped with VGaT (Fig. 8C,D), likely representing synaptic terminals originating from parvalbumin-positive interneurons, which are known to express MOR at a high level (Drake and Milner, 2006; Drake et al., 2007). Thus, our data strongly suggest that activation of MORs has a dual effect on SUM/VTA synapses onto granule cells: a direct inhibitory effect mediated by inhibition of presynaptic N-type calcium channels and a disinhibitory effect mediated by inhibition of GABAergic interneurons innervating the SUM/VTA terminals (summarized in Fig. 9). It might be suggested that the net functional effect can be further modulated if the μ-opioid modulation in the midbrain terminals and on local interneurons have different kinetics, resulting in a dynamic control of the synapse that may have important consequences for highly timing-dependent mechanisms such as LTP. Finally, it is worth noticing that ∼35% of MOR-positive structures do not colocalize with either GFP staining or VGaT. It is possible that these MORs are expressed in VGluT3- or VChat-positive terminals in the DG (Soares et al., 2017; Fazekas et al., 2022) or in astrocytes, as in previously reported in the CA1 area (Nam et al., 2019).

Figure 8.
Figure 8.

GABAergic interneurons form inhibitory axo-axonic contacts on SUM/VTA terminals in the granule cell layer. A, The size of glutamatergic and GABAergic currents in DAMGO normalized to the currents measured in control condition. The effect of DAMGO on granule cell glutamatergic currents evoked by optogenetic stimulation of the SUM/VTA inputs was significantly stronger than on GABAergic currents (the residual current in DAMGO was 18.3 ± 2.0% of the control for glutamatergic currents and 32.0 ± 4.1% for GABA currents. n = 42 and 28, respectively; p < 0.0001; Mann–Whitney test). B, Change in PPR in the presence of DAMGO. The change was stronger for the glutamatergic than the GABAergic currents (174.1 ± 12.4% vs 134.0 ± 8.1%, n = 42 and 28, respectively, p = 0.043; Mann–Whiney test). C, Triple immunostaining images using antibodies against VGaT, GFP, and MOR1 in the DG. GFP expression was driven by injection of anterograde viral tracers and identifies VGluT2-positive SUM/VTA terminals. MOR1 and VGaT represent naturally occurring epitopes. Scale bar, 5 µm. Thick arrows with asterisk indicate a VGaT-positive/GFP-negative MOR–expressing terminal. Thin arrows indicate MOR-expressing terminal positive for both VGaT and GFP. D, Quantification of MOR expression relative to VGaT- and GFP-positive terminals. 25.5 ± 4.1% of MOR was colocalized with both VGaT and GFP, while 31.0 ± 2.3% was located at VGaT-positive but GFP-negative structures (likely representing terminals from local interneurons; data from 3 animals).

Figure 9.
Figure 9.

Mechanisms regulating neurotransmitter release at MOR-containing SUM/VTA-DG synapses in hippocampal DG. Top, Schematic representing the mechanisms regulating neurotransmitter release at MOR-containing SUM/VTA-DG corelease synapses in control condition (no MOR activation). Bottom, Effects of MOR activation on transmitter release from these synapses. Notice the dual effect of DAMGO, which reduces release from the SUM/VTA inputs by inhibiting presynaptic N-type channels while disinhibiting release by reducing the inhibitory control from local interneurons.

Possible network effects of MOR activation in the dorsal hippocampus

The network effects of MOR activation in the hippocampus are very complex. Even when considering a single synapse, multiple mechanisms simultaneously affect the response. In the case of the SUM/VTA synapses, we found at least three mechanisms: an overall disinhibition of granule cells due to inhibition of interneurons, the direct disinhibition of the SUM/VTA terminals due to the direct interneuronal control of the terminals, and the direct inhibition of release at these terminals due to the location within the terminals of MORs that inhibit N-type calcium channels. The overall picture, however, is even more complex, as another mechanism was recently described in the CA1 area of the hippocampus, where activation of MORs expressed in astrocytes leads to astrocytic glutamate release (Woo et al., 2018) via activation of presynaptic mGluR1 (Nam et al., 2019). Whether a similar mechanism is also present in the DG is an open question to be addressed in future studies.

Hippocampal disinhibition and astrocyte-dependent potentiation of synaptic release are believed to account for the MOR-dependent induction of LTP at the mossy fiber to CA3 synapses (Derrick et al., 1991; Derrick and Martinez, 1994) as well as the perforant path→DG synapses (Xie and Lewis, 1991). Thus, upon MOR activation, the hippocampal circuitry is disinhibited, LTP is reinforced, and yet the VGluT2/VGaT-positive inputs are heavily inhibited. This suggests that MOR activation changes the weight of these synapses on the overall hippocampal function. While the direct net effect of the corelease on granule cells is prevalently inhibitory for both the VTA (Ntamati and Luscher, 2016) and SUM (Ajibola et al., 2021) inputs, the global network effect was shown to favor LTP formation (Ajibola et al., 2021). Thus, the acute effect of MOR activation may lead to a relative decreased impact of the SUM/VTA synapses, in the context of a generally increased LTP of other inputs. These other inputs likely include the glutamatergic projections from the entorhinal cortex to the DG, which are enhanced during context-induced reinstatement of heroin seeking (Ge et al., 2017) and believed to be potentiated by opioids (Wiesner et al., 1986). Thus, it may be suggested that input modulation due to acute activation of MORs enhances LTP at the entorhinal cortex inputs and contributes to the reinforcement of memories of drug-associated contexts.

Footnotes

  • This work was supported by National Institutes of Health Grants DA044121 and NS112292. We thank Dr. Yen-Hsin Cheng for the outstanding management of mouse colonies and Dr. Anis Contractor for the critical reading of the manuscript. Part of the imaging work was performed using instrumentation at the Northwestern University Center for Advanced Microscopy generously supported by National Cancer Institute CCSG P30 CA060553 awarded to the Robert H. Lurie Comprehensive Cancer Center.

  • ↵*H.R.K. and S.D. contributed equally to this work.

  • The authors declare no competing financial interests.

  • Correspondence should be addressed to Marco Martina at m-martina{at}northwestern.edu.

References

  1. ↵
    1. Ajibola MI,
    2. Wu JW,
    3. Abdulmajeed WI,
    4. Lien CC
    (2021) Hypothalamic glutamate/GABA cotransmission modulates hippocampal circuits and supports long-term potentiation. J Neurosci 41:8181–8196. https://doi.org/10.1523/JNEUROSCI.0410-21.2021 pmid:34380766
    OpenUrlAbstract/FREE Full Text
  2. ↵
    1. Akaishi T,
    2. Saito H,
    3. Ito Y,
    4. Ishige K,
    5. Ikegaya Y
    (2000) Morphine augments excitatory synaptic transmission in the dentate gyrus through GABAergic disinhibition. Neurosci Res 38:357–363. https://doi.org/10.1016/S0168-0102(00)00177-2
    OpenUrlCrossRefPubMed
  3. ↵
    1. Basu J,
    2. Siegelbaum SA
    (2015) The corticohippocampal circuit, synaptic plasticity, and memory. Cold Spring Harb Perspect Biol 7:a021733. https://doi.org/10.1101/cshperspect.a021733 pmid:26525152
    OpenUrlAbstract/FREE Full Text
  4. ↵
    1. Bramham CR,
    2. Sarvey JM
    (1996) Endogenous activation of mu and δ-1 opioid receptors is required for long-term potentiation induction in the lateral perforant path: dependence on GABAergic inhibition. J Neurosci 16:8123–8131. https://doi.org/10.1523/JNEUROSCI.16-24-08123.1996 pmid:8987837
    OpenUrlAbstract/FREE Full Text
  5. ↵
    1. Bucurenciu I,
    2. Kulik A,
    3. Schwaller B,
    4. Frotscher M,
    5. Jonas P
    (2008) Nanodomain coupling between Ca2+ channels and Ca2+ sensors promotes fast and efficient transmitter release at a cortical GABAergic synapse. Neuron 57:536–545. https://doi.org/10.1016/j.neuron.2007.12.026
    OpenUrlCrossRefPubMed
  6. ↵
    1. Chamberland S,
    2. Evstratova A,
    3. Toth K
    (2017) Short-term facilitation at a detonator synapse requires the distinct contribution of multiple types of voltage-gated calcium channels. J Neurosci 37:4913–4927. https://doi.org/10.1523/JNEUROSCI.0159-17.2017 pmid:28411270
    OpenUrlAbstract/FREE Full Text
  7. ↵
    1. Chen S, et al.
    (2020) A hypothalamic novelty signal modulates hippocampal memory. Nature 586:270–274. https://doi.org/10.1038/s41586-020-2771-1
    OpenUrlCrossRefPubMed
  8. ↵
    1. Derrick BE,
    2. Martinez JL
    (1994) Opioid receptor activation is one factor underlying the frequency-dependence of messy fiber Ltp induction. J Neurosci 14:4359–4367. https://doi.org/10.1523/JNEUROSCI.14-07-04359.1994 pmid:7913121
    OpenUrlAbstract/FREE Full Text
  9. ↵
    1. Derrick BE,
    2. Weinberger SB,
    3. Martinez JL Jr.
    (1991) Opioid receptors are involved in an NMDA receptor-independent mechanism of LTP induction at hippocampal mossy fiber-CA3 synapses. Brain Res Bull 27:219–223. https://doi.org/10.1016/0361-9230(91)90071-Q
    OpenUrlCrossRefPubMed
  10. ↵
    1. Drake CT,
    2. Chavkin C,
    3. Milner TA
    (2007) Opioid systems in the dentate gyrus. Prog Brain Res 163:245–263. https://doi.org/10.1016/S0079-6123(07)63015-5
    OpenUrlCrossRefPubMed
  11. ↵
    1. Drake CT,
    2. Milner TA
    (2006) Mu opioid receptors are extensively co-localized with parvalbumin, but not somatostatin, in the dentate gyrus. Neurosci Lett 403:176–180. https://doi.org/10.1016/j.neulet.2006.04.047
    OpenUrlCrossRefPubMed
  12. ↵
    1. Fazekas CL,
    2. Szabo A,
    3. Torok B,
    4. Banrevi K,
    5. Correia P,
    6. Chaves T,
    7. Daumas S,
    8. Zelena D
    (2022) A new player in the hippocampus: a review on VGLUT3+ neurons and their role in the regulation of hippocampal activity and behaviour. Int J Mol Sci 23:790. https://doi.org/10.3390/ijms23020790 pmid:35054976
    OpenUrlCrossRefPubMed
  13. ↵
    1. Ge F,
    2. Wang N,
    3. Cui C,
    4. Li Y,
    5. Liu Y,
    6. Ma Y,
    7. Liu S,
    8. Zhang H,
    9. Sun X
    (2017) Glutamatergic projections from the entorhinal cortex to dorsal dentate gyrus mediate context-induced reinstatement of heroin seeking. Neuropsychopharmacology 42:1860–1870. https://doi.org/10.1038/npp.2017.14 pmid:28106041
    OpenUrlCrossRefPubMed
  14. ↵
    1. Han Y, et al.
    (2020) Excitatory VTA to DH projections provide a valence signal to memory circuits. Nat Commun 11:1466. https://doi.org/10.1038/s41467-020-15035-z pmid:32193428
    OpenUrlCrossRefPubMed
  15. ↵
    1. Hashimotodani Y,
    2. Karube F,
    3. Yanagawa Y,
    4. Fujiyama F,
    5. Kano M
    (2018) Supramammillary nucleus afferents to the dentate gyrus co-release glutamate and GABA and potentiate granule cell output. Cell Rep 25:2704–2715 e2704. https://doi.org/10.1016/j.celrep.2018.11.016
    OpenUrlCrossRefPubMed
  16. ↵
    1. Hefft S,
    2. Jonas P
    (2005) Asynchronous GABA release generates long-lasting inhibition at a hippocampal interneuron-principal neuron synapse. Nat Neurosci 8:1319–1328. https://doi.org/10.1038/nn1542
    OpenUrlCrossRefPubMed
  17. ↵
    1. Heinke B,
    2. Gingl E,
    3. Sandkuhler J
    (2011) Multiple targets of μ-opioid receptor-mediated presynaptic inhibition at primary afferent Adelta- and C-fibers. J Neurosci 31:1313–1322. https://doi.org/10.1523/JNEUROSCI.4060-10.2011 pmid:21273416
    OpenUrlAbstract/FREE Full Text
  18. ↵
    1. Jonas P,
    2. Lisman J
    (2014) Structure, function, and plasticity of hippocampal dentate gyrus microcircuits. Front Neural Circuits 8:107. https://doi.org/10.3389/fncir.2014.00107 pmid:25309334
    OpenUrlCrossRefPubMed
  19. ↵
    1. Kelly CJ,
    2. Martina M
    (2018) Circuit-selective properties of glutamatergic inputs to the rat prelimbic cortex and their alterations in neuropathic pain. Brain Struct Funct 223:2627–2639. https://doi.org/10.1007/s00429-018-1648-7 pmid:29550939
    OpenUrlCrossRefPubMed
  20. ↵
    1. Li Y,
    2. Bao H,
    3. Luo Y,
    4. Yoan C,
    5. Sullivan HA,
    6. Quintanilla L,
    7. Wickersham I,
    8. Lazarus M,
    9. Shih YI,
    10. Song J
    (2020) Supramammillary nucleus synchronizes with dentate gyrus to regulate spatial memory retrieval through glutamate release. eLife 9:e53129. https://doi.org/10.7554/eLife.53129 pmid:32167473
    OpenUrlCrossRefPubMed
  21. ↵
    1. Li L,
    2. Bischofberger J,
    3. Jonas P
    (2007) Differential gating and recruitment of P/Q-, N-, and R-type Ca2+ channels in hippocampal mossy fiber boutons. J Neurosci 27:13420–13429. https://doi.org/10.1523/JNEUROSCI.1709-07.2007 pmid:18057200
    OpenUrlAbstract/FREE Full Text
  22. ↵
    1. Lisman JE
    (1999) Relating hippocampal circuitry to function: recall of memory sequences by reciprocal dentate-CA3 interactions. Neuron 22:233–242. https://doi.org/10.1016/S0896-6273(00)81085-5
    OpenUrlCrossRefPubMed
  23. ↵
    1. Luebke JI,
    2. Dunlap K,
    3. Turner TJ
    (1993) Multiple calcium channel types control glutamatergic synaptic transmission in the hippocampus. Neuron 11:895–902. https://doi.org/10.1016/0896-6273(93)90119-C
    OpenUrlCrossRefPubMed
  24. ↵
    1. Mayer JH,
    2. Steffensen SC,
    3. Henriksen SJ
    (1994) Electrophysiological effects of selective opioid agonists on spontaneous and evoked neuronal activity in the dentate gyrus of the hippocampus in vivo. Neuropharmacology 33:963–975. https://doi.org/10.1016/0028-3908(94)90154-6
    OpenUrlCrossRefPubMed
  25. ↵
    1. Milner TA,
    2. Drake CT
    (2001) Ultrastructural evidence for presynaptic mu opioid receptor modulation of synaptic plasticity in NMDA-receptor-containing dendrites in the dentate gyrus. Brain Res Bull 54:131–140. https://doi.org/10.1016/S0361-9230(00)00415-9
    OpenUrlCrossRefPubMed
  26. ↵
    1. Morales JE,
    2. De A,
    3. Miller AA,
    4. Chen Z,
    5. McCarty JH
    (2022) Mlc1-expressing perivascular astrocytes promote blood-brain barrier integrity. J Neurosci 42:1406–1416. https://doi.org/10.1523/JNEUROSCI.1918-21.2021 pmid:34965971
    OpenUrlAbstract/FREE Full Text
  27. ↵
    1. Moser MB,
    2. Moser EI
    (1998) Functional differentiation in the hippocampus. Hippocampus 8:608–619. https://doi.org/10.1002/(SICI)1098-1063(1998)8:6<608::AID-HIPO3>3.0.CO;2-7
    OpenUrlCrossRefPubMed
  28. ↵
    1. Nam MH, et al.
    (2019) Activation of astrocytic μ-opioid receptor causes conditioned place preference. Cell Rep 28:1154–1166.e4=5. https://doi.org/10.1016/j.celrep.2019.06.071
    OpenUrlCrossRefPubMed
  29. ↵
    1. Neumaier JF,
    2. Mailheau S,
    3. Chavkin C
    (1988) Opioid receptor-mediated responses in the dentate gyrus and CA1 region of the rat hippocampus. J Pharmacol Exp Ther 244:564–570. https://jpet.aspetjournals.org/content/244/2/564.long
    OpenUrlAbstract/FREE Full Text
  30. ↵
    1. Ntamati NR,
    2. Luscher C
    (2016) VTA projection neurons releasing GABA and glutamate in the dentate gyrus. eNeuro 3:e0137–16.2016. https://doi.org/10.1523/ENEURO.0137-16.2016 pmid:27648470
    OpenUrlPubMed
  31. ↵
    1. Pronier E,
    2. Morici JF,
    3. Girardeau G
    (2023) The role of the hippocampus in the consolidation of emotional memories during sleep. Trends Neurosci 46:912–925. https://doi.org/10.1016/j.tins.2023.08.003
    OpenUrlCrossRefPubMed
  32. ↵
    1. Regehr WG
    (2012) Short-term presynaptic plasticity. Cold Spring Harb Perspect Biol 4:a005702. https://doi.org/10.1101/cshperspect.a005702 pmid:22751149
    OpenUrlAbstract/FREE Full Text
  33. ↵
    1. Rollenhagen A,
    2. Sätzler K,
    3. Rodriguez EP,
    4. Jonas P,
    5. Frotscher M,
    6. Lübke JHR
    (2007) Structural determinants of transmission at large hippocampal mossy fiber synapses. J Neurosci 27:10434–10444. https://doi.org/10.1523/JNEUROSCI.1946-07.2007 pmid:17898215
    OpenUrlAbstract/FREE Full Text
  34. ↵
    1. Root DH,
    2. Zhang S,
    3. Barker DJ,
    4. Miranda-Barrientos J,
    5. Liu B,
    6. Wang HL,
    7. Morales M
    (2018) Selective brain distribution and distinctive synaptic architecture of dual glutamatergic-GABAergic neurons. Cell Rep 23:3465–3479. https://doi.org/10.1016/j.celrep.2018.05.063 pmid:29924991
    OpenUrlCrossRefPubMed
  35. ↵
    1. Schmidt-Hieber C,
    2. Jonas P,
    3. Bischofberger J
    (2004) Enhanced synaptic plasticity in newly generated granule cells of the adult hippocampus. Nature 429:184–187. https://doi.org/10.1038/nature02553
    OpenUrlCrossRefPubMed
  36. ↵
    1. Soares JI,
    2. Valente MC,
    3. Andrade PA,
    4. Maia GH,
    5. Lukoyanov NV
    (2017) Reorganization of the septohippocampal cholinergic fiber system in experimental epilepsy. J Comp Neurol 525:2690–2705. https://doi.org/10.1002/cne.24235
    OpenUrlCrossRefPubMed
  37. ↵
    1. Sun D,
    2. Mei L,
    3. Xiong WC
    (2023) Dorsal dentate gyrus, a key regulator for mood and psychiatric disorders. Biol Psychiatry 93:1071–1080. https://doi.org/10.1016/j.biopsych.2023.01.005
    OpenUrlCrossRefPubMed
  38. ↵
    1. Tritsch NX,
    2. Granger AJ,
    3. Sabatini BL
    (2016) Mechanisms and functions of GABA co-release. Nat Rev Neurosci 17:139–145. https://doi.org/10.1038/nrn.2015.21 pmid:26865019
    OpenUrlCrossRefPubMed
  39. ↵
    1. Wender M,
    2. Bornschein G,
    3. Brachtendorf S,
    4. Hallermann S,
    5. Eilers J,
    6. Schmidt H
    (2023) Cav2.2 channels sustain vesicle recruitment at a mature glutamatergic synapse. J Neurosci 43:4005–4018. https://doi.org/10.1523/JNEUROSCI.1279-22.2023 pmid:37185239
    OpenUrlAbstract/FREE Full Text
  40. ↵
    1. Wiesner JB,
    2. Henriksen SJ,
    3. Bloom FE
    (1986) Opioid enhancement of perforant path transmission: effect of an enkephalin analog on inhibition and facilitation in the dentate gyrus. Brain Res 399:404–408. https://doi.org/10.1016/0006-8993(86)91537-4
    OpenUrlCrossRefPubMed
  41. ↵
    1. Woo DH, et al.
    (2018) Activation of astrocytic μ-opioid receptor elicits fast glutamate release through TREK-1-containing K2P channel in hippocampal astrocytes. Front Cell Neurosci 12:319. https://doi.org/10.3389/fncel.2018.00319 pmid:30319359
    OpenUrlPubMed
  42. ↵
    1. Xie CW,
    2. Lewis DV
    (1991) Opioid-mediated facilitation of long-term potentiation at the lateral perforant path-dentate granule cell synapse. J Pharmacol Exp Ther 256:289–296. https://jpet.aspetjournals.org/content/256/1/289.long
    OpenUrlAbstract/FREE Full Text
  43. ↵
    1. Yan JJ, et al.
    (2022) A circuit from the ventral subiculum to anterior hypothalamic nucleus GABAergic neurons essential for anxiety-like behavioral avoidance. Nat Commun 13:7464. https://doi.org/10.1038/s41467-022-35211-7 pmid:36463200
    OpenUrlCrossRefPubMed
  44. ↵
    1. Yeckel MF,
    2. Berger TW
    (1990) Feedforward excitation of the hippocampus by afferents from the entorhinal cortex: redefinition of the role of the trisynaptic pathway. Proc Natl Acad Sci U S A 87:5832–5836. https://doi.org/10.1073/pnas.87.15.5832 pmid:2377621
    OpenUrlAbstract/FREE Full Text
Back to top

In this issue

Email
Print
View Full Page PDF
Citation Tools
Respond to this article
μ-Opioid Receptor Modulation of the Glutamatergic/GABAergic Midbrain Inputs to the Mouse Dorsal Hippocampus
Haram R. Kim, Soumil Dey, Gabriella Sekerkova, Marco Martina
Journal of Neuroscience 23 October 2024, 44 (43) e0653242024; DOI: 10.1523/JNEUROSCI.0653-24.2024
Twitter logo Facebook logo Mendeley logo

Jump to section

Keywords

Responses to this article

Respond to this article

Jump to comment:

No eLetters have been published for this article.