Abstract
Ventral subiculum (vSUB) is the major output region of ventral hippocampus (vHIPP) and sends major projections to nucleus accumbens medial shell (NAcMS). Hyperactivity of the vSUB-NAcMS circuit is associated with substance use disorders and the modulation of vSUB activity alters drug seeking and drug reinstatement behavior in rodents. However, to the best of our knowledge, the cell type-specific connectivity and synaptic transmission properties of the vSUB-NAcMS circuit have never been directly examined. Instead, previous functional studies have focused on total ventral hippocampal (vHIPP) output to NAcMS without distinguishing vSUB from other subregions of vHIPP, including ventral CA1 (vCA1). Using ex vivo electrophysiology, we systematically characterized the vSUB-NAcMS circuit with cell type- and synapse-specific resolution in male and female mice and found that vSUB output to dopamine receptor type-1 (D1R) and type-2 (D2R) expressing medium spiny neurons (MSNs) displays a functional connectivity bias for D2R MSNs. Furthermore, we found that vSUB-D1R and vSUB-D2R MSN synapses contain calcium-permeable AMPA receptors in drug-naive mice. Finally, we find that, distinct from other glutamatergic inputs, cocaine exposure selectively induces plasticity at vSUB-D2R synapses. Importantly, we directly compared vSUB and vCA1 output to NAcMS and found that vSUB synapses are functionally distinct and that vCA1 output recapitulated the synaptic properties previously ascribed to vHIPP. Our work highlights the need to consider the contributions of individual subregions of vHIPP to substance use disorders and represents an important first step toward understanding how the vSUB-NAcMS circuit contributes to the etiologies that underlie substance use disorders.
SIGNIFICANCE STATEMENT Inputs to nucleus accumbens (NAc) dopamine receptor type 1 (D1R) and D2R medium spiny neurons (MSNs) are critically involved in reward seeking behavior. Ventral subiculum (vSUB) provides robust synaptic input to nucleus accumbens medial shell (NAcMS) and activity of this circuit is linked to substance use disorders. Despite the importance of the vSUB to nucleus accumbens circuit, the functional connectivity and synaptic transmission properties have not been tested. Here, we systematically interrogated these properties and found that basal connectivity and drug-induced plasticity are biased for D2R medium spiny neurons. Overall, we demonstrate that this circuit is distinct from synaptic inputs from other brain regions, which helps to explain how vSUB dysfunction contributes to the etiologies that underlie substance use disorders.
Introduction
The nucleus accumbens medial shell (NAcMS) plays critical roles in reward and motivated behavior. NAc is primarily populated by medium spiny principal neurons (MSNs) that are defined by their expression of either dopamine receptor type 1 (D1R) or type 2 (D2R). The respective roles of D1R and D2R MSNs in drug seeking behavior has been contentious, the predominant model contends that D1R MSNs promote reward seeking whereas D2R MSNs inhibit it (Lobo et al., 2010; Lobo and Nestler, 2011; Bock et al., 2013). However, several studies point to a more complicated role for MSNs in drug seeking that appears to be dependent on the subregion of NAc studied (Soares-Cunha et al., 2016, 2022; Cole et al., 2018; Z. Liu et al., 2022). In dorsal NAcMS, activation of D2R MSNs promotes reward, a behavior presumably mediated via their unique connections to glutamatergic neurons in ventral pallidum (VP; Yao et al., 2021). Thus, it is critical to dissect MSN function within the context of discrete NAc subregions that are defined by their anatomic position and by their input and output circuitry, however, an understanding of the inputs that drive D2R MSN activity in this region is incomplete (Yang et al., 2018; Baimel et al., 2019; Castro and Bruchas, 2019; Yao et al., 2021).
The ventral hippocampus (vHIPP) provides the most robust source of glutamatergic input to NAcMS (Britt et al., 2012; Li et al., 2018). This excitatory circuit regulates drug-relevant locomotion and cue and context-induced reinstatement of drug seeking in rodents (Grace, 2010; Britt et al., 2012; Glangetas et al., 2015; Bossert et al., 2016; Marchant et al., 2016; Preston et al., 2019). vHIPP input displays biased synaptic strength onto D1R MSNs (MacAskill et al., 2014), and long-term potentiation induced by withdrawal from cocaine exposure has been observed at vHIPP-D1R MSN synapses, but not D2R MSNs (Pascoli et al., 2014). Importantly, vHIPP is a composite of afferents originating from functionally discrete subregions of vHIPP: ventral CA1 (vCA1) and ventral subiculum (vSUB). vSUB is the major output of vHIPP and sends robust projections to dorsal NAcMS (Lopes da Silva et al., 1984; Britt et al., 2012; Li et al., 2018; Yao et al., 2021). Further, vSUB is thought to be a key regulator of reward seeking behavior as the selective pharmacological manipulation of vSUB activity is sufficient to alter drug reinstatement in rodents (Bossert et al., 2016; Marchant et al., 2016). Distinct from vCA1, vSUB contains two classes of excitatory principal neurons identified electrophysiologically as regular-spiking or burst-spiking cells. Regular-spiking and burst-spiking neurons are equally represented in vSUB but exhibit distinct gene expression, connectivity, plasticity, and behavioral functions (Wozny et al., 2008; Graves et al., 2012; Böhm et al., 2015; Cembrowski et al., 2018; Boxer et al., 2021). Thus, regular-spiking and burst-spiking vSUB neurons may provide distinct input to dorsal NAcMS and possess unique synaptic properties compared with other subregions of vHIPP. While most studies have examined the synaptic properties of total vHIPP or vCA1 input to NAc, to our knowledge, the functional connectivity and synaptic properties of vSUB input to dorsal NAcMS are untested.
Here, we systematically dissected the connectivity and functional properties of the vSUB-NAcMS circuit and reveal unique qualities of vSUB input. Despite a disproportionate number of regular-spiking neurons projecting to total nucleus accumbens shell, we performed the first intersectional cell type-specific retrograde labeling of the vSUB-NAcMS circuit and found that regular-spiking and burst-spiking neurons project equally to D1R and D2R MSNs. We next functionally characterized vSUB input in dorsal NAcMS. First, vSUB displays a significant bias for D2R MSNs. Second, vSUB synapses onto both MSNs in drug naive animals unexpectedly contain calcium-permeable AMPA receptors. Third, cocaine administration selectively potentiates vSUB-D2R but not vSUB-D1R MSNs. Our findings provide critical insight into the unique synaptic properties of vSUB input to dorsal NAcMS D2R MSNs.
Materials and Methods
Animals
Male and female mice were bred at the University of Colorado Anschutz and were from a B6;129 or B6.Cg mixed genetic background. The B6.Cg-Tg(Drd1a-tdTomato)6Calak/J (“D1R-tdTomato,” Jax 016214) breeders, B6;129-Tg(Drd1-cre)120Mxu/Mmjax (“D1R-Cre,” #037156-JAX) breeders, and the B6.FVB(Cg)-Tg(Adora2a-cre)KG139Gsat/Mmucd (“A2a-Cre,” MMRRC 036158-UCD) breeders were generous gifts from Robert Malenka. Mice were genotyped in house. Mice were housed in a dedicated animal care facility maintained at 35% humidity, 21–23°C, on a 14/10 h light/dark cycle and housed in groups of two to five in ventilated cages with same-sex littermates with food and water ad libitum. Mice were stereotactically injected on postnatal day (P)21–P24, and electrophysiology experiments were performed at P56–P65 in visibly healthy animals. All experiments were conducted in cohorts, wherein at least two littermates were injected with either saline or cocaine. At P40–P45, mice received intraperitoneal injections of cocaine (20 mg/kg) or saline in the home cage for five consecutive days. All procedures were conducted in accordance with guidelines approved by Administrative Panel on Laboratory Animal Care at University of Colorado, Anschutz School of Medicine, accredited by Association for Assessment and Accreditation of Laboratory Animal Care International (AAALAC; 00235).
Viral constructs
AAV vectors were constructed from an empty AAV transfer plasmid where the expression cassette is as follows: left-ITR, human synapsin promoter, multiple cloning site, WPRE and right ITR. Cre-dependent plasmids contained: left-ITR, human synapsin promoter, 5′ LoxP site, multiple cloning site, 3′ LoxP site, WPRE and right ITR. Plasmids carrying mRuby and T2A cDNAs were generous gifts from Kevin Beier (Beier et al., 2017). To generate AAVs, HEK293T cells were transfected with a AAV transfer plasmid, pHelper and pRC-DJ or pRC-retro. AAVs were purified as previously described (Aoto et al., 2013). Briefly, 72 h after transfection, cells were harvested, lysed and virus was purified and harvested from the 40% iodixanol fraction after ultracentrifugation. Virus was concentrated in a 100K MWCO Amicon filter. The following AAVs were used: AAVDJ-hSYN-ChIEF-mRuby; AAV DJ-hSYN-mCherry-T2A-WGA:Cre; AAV DJ-hSYN-DIOLoxP-mRuby; AAV DJ-hSYN-DIOLoxP-WGA:Flp-T2A-GFP; AAV DJ-hSYN-DIOFrt-mRuby; AAV2-Retro-mRuby (rgAAV2-mRuby).
Stereotactic surgeries
Stereotactic injections were performed on P21–P24 mice. Animals were anesthetized with an intraperitoneal injection of 2,2,2-Tribromoethanol (250 mg/kg) then head fixed to a stereotactic frame (KOPF). Solutions containing AAVs (0.2–0.5 µl) were injected into ventral subiculum, ventral CA1, or nucleus accumbens medial shell at a rate of 10 µl/h using a syringe pump (World Precision Instruments). Coordinates (in mm) were vSUB: anterior-posterior: −3.4, mediolateral: ±3.2 (relative to bregma), and dorsoventral: −3.45 (relative to pia); vCA1: AP: −3.1, ML: ±3.2, DV: −3.45; NAcMS: AP: 1.5, ML: ±0.75, DV: −4.0.
Ex vivo whole-cell electrophysiology
At P56–P65, animals were deeply anesthetized with isoflurane and decapitated. Brains were rapidly dissected and 300-µm horizontal slices (for vSUB recordings) or coronal slices (for NAc recordings) were sectioned with a vibratome in cutting solution then moved to artificial CSF (ACSF), as previously described (Boxer et al., 2021), were superfused with 29.5°C oxygenated ACSF containing (in mm) 126 NaCl, 26.2 NaHCO3, 11 D-glucose, 2.5 KCl, 2.5 CaCl2, 1.3 MgSO4-7H2O, and 1 NaH2PO4. For NAc recordings, ACSF included 100 μm picrotoxin. To isolate AMPAR-mediated currents (AMPA I/O, strontium, and current–voltage experiments) 50 μm D-AP5 was also included in the ACSF, whereas NMDAR-mediated currents (NMDA I/O and current–voltage experiments) were isolated with 10 μm NBQX. For strontium experiments, CaCl2 was replaced with 2.5 mm SrCl2 in the recording ACSF; 20 μm 1-Naphthyl acetyl spermine (NASPM) was used for wash-in experiments.
MSNs in NAc were identified by their shape and size and determined as D1R+ or D1R− (putative D2R) by expression of tdTomato. Cells were visualized using an Olympus BX51W microscope and with a 40× dipping objective collected on a Hamamatsu ORCA-Flash 4.0 V3 digital camera using an IR bandpass filter.
Neurons were voltage-clamped at −70 mV in whole-cell configuration with a cesium-based internal solution containing (in mm) 117 cs-methanesulfonate, 15 CsCl, 10 TEA-Cl, 10 HEPES, 10 phosphocreatine, 8 NaCl, 4 Mg2-ATP, 1 MgCl2, 0.5 Na2-GTP, and 0.2 EGTA. For AMPA-to-NMDA ratio, cells were held at −70 or +40 mV from the experimentally determined reversal potential to record AMPAR-mediated currents and NMDAR-mediated currents, respectively; 10 μm spermine was added to the internal solution for AMPA current–voltage experiments; 5 mm QX-314 was added for AMPA-to-NMDA ratio and current–voltage experiments. To optogenetically stimulate ChIEF expressing vSUB fibers, slices were illuminated with 470-nm LED light (ThorLabs M470L2-C1) for 3 ms through the 40× dipping objective located directly over the recorded cell. With an illumination area of 33.18 mm2 the tissue was excited with an irradiance of 0.006–0.17 mW/mm2. To electrically stimulate MSN synapses, a homemade nichrome stimulating electrode was placed ∼200 µm from the patched cell and pulsed at 0.1 Hz at 100 µA (A-M Systems 2100 Isolated pulse stimulator).
For regular-spiking/burst-spiking ratio in vSUB, the ACSF contained no drugs, and mRuby+ neurons were patched in whole-cell configuration with a K-gluconate based internal containing (in mm): 95 K-gluconate, 50 KCl, 10 HEPES, 10 phosphocreatine, 4 Mg2-ATP, 0.5 Na2-GTP, and 0.2 EGTA. vSUB pyramidal neuron identity (regular spiking vs burst spiking) was determined by the suprathreshold action potential firing pattern, as previously described (Boxer et al., 2021). All recordings were acquired using Molecular Devices Multiclamp 700B amplifier and Digidata 1440 digitizer with Axon pClamp 9.0 Clampex software, lowpass filtered at 2 kHz and digitized at 10–20 kHz.
Analysis of electrophysiology recordings
Evoked EPSC peak amplitudes were determined using Axon pClamp10 Clampfit software. A total of 12–20 sweeps (0.1 Hz) were averaged to obtain peak amplitude at each light intensity (I/O) or voltage. Input/output slope was calculated using the SLOPE function in Microsoft Excel: (pA amplitude range/mW intensity range). Release probability was assessed by measurements of paired-pulse ratios at interstimulus intervals of 20–100 ms. Paired-pulse ratios were measured by dividing the average EPSC amplitude evoked by the second stimulus, by the average amplitude evoked by the first stimulus (R2/R1). AMPA-to-NMDA ratio was determined by dividing the peak EPSC amplitude at −70 mV (AMPA), by the EPSC amplitude measured at 50 ms after onset, at +40 mV (NMDA). Asynchronous EPSC (aEPSC) event amplitudes (strontium experiments) were analyzed using Clampfit event detection software. For NMDAR decay kinetics analysis, NMDAR EPSC sweeps were averaged per cell and decay time constant values and weights were calculated by fitting the decay slope to a standard, two-term exponential functions, fitted with Levenberg-Marquardt method in Clampfit. Weighted decay tau was calculated as
Experimental design and statistical analyses
Both female and male mice were used in this study. In relevant graphs, data from females are represented as open circles while those from males are represented as closed circles. Number of animals (N) and cells (n) used for each experiment is included in each figure or corresponding figure legend. All experiments were replicated in at least three animals. Experimenter was not blinded to animal sex, genotype, or cell identity, but was blinded to saline or cocaine assignments during data acquisition. No statistical method was used to predetermine sample size before the study. All data were tested for normality using D'Agostino and Pearson normality tests. If datasets exhibited normal distribution, Student's paired and unpaired t tests and two-way and three-way ANOVAs were used to determine statistical differences. Otherwise, Mann–Whitney or Kruskal–Wallis tests were used. Tests were corrected for multiple comparisons using Sidak's or Dunn's multiple comparisons tests. For strontium datasets, Kolmogorov–Smirnoff tests were used to compare the cumulative probability plots. Differences were considered statistically significant when p < 0.05. Unless otherwise stated, all bar graphs are presented as mean ± SEM. Statistical tests and graph making were performed using Prism 7 or Prism 9 (GraphPad), and figures were compiled in Adobe Illustrator.
Results
Cell type-specific circuit tracing
The cell type-specific connectivity of regular-spiking and burst-spiking neurons in vSUB with D1R versus D2R MSNs in NAcMS, has not been fully explored (Fig. 1A,B). We first aimed to confirm previous findings by assessing the identity of presynaptic vSUB neurons that generally project to the NAcMS. Using wild-type mice, we stereotactically injected mRuby expressing retrograde AAV2 (rgAAV2-mRuby) into NAcMS (Fig. 1C), then made ex vivo slices and performed whole cell patch clamp to electrophysiologically identify mRuby+ neurons in vSUB. Consistent with previous reports, mRuby+ NAcMS-projecting neurons were concentrated in proximal vSUB but also present in ventral entorhinal cortex and ventral CA1 (vCA1; Naber and Witter, 1998; Britt et al., 2012; Cembrowski et al., 2018; Lee et al., 2019; Williams et al., 2020). Although vSUB is composed of ∼50% regular-spiking and 50% burst-spiking neurons (Staff et al., 2000; Graves et al., 2012), we found that the majority of NAcMS-projecting neurons in vSUB were regular-spiking neurons (70 ± 4%; Fig. 1G), which is remarkably consistent with other studies using different retrograde labeling approaches such as retrobeads or cholera toxin B (Kim and Spruston, 2012; Lee et al., 2019). In light of our recent findings that the wiring of vSUB local circuitry is sexually dimorphic (Boxer et al., 2021), we reviewed whether the wiring of the vSUB-NAcMS circuit also exhibits sexual dimorphism. We separated our rgAAV2-mRuby experiment by sex but found that equivalent proportions of regular-spiking/burst-spiking neurons project to NAcMS in male and female mice (% regular-spiking: male: 75 ± 6, female: 66 ± 4; U = 2, p = 0.40, Mann–Whitney; Fig. 1G; males: closed circles, females: open circles).
vSUB regular and burst spiking neurons equally project to NAc medial shell D1R and D2R MSNs. A, Left, Schematic illustration of vHIPP in the horizontal plane. Ventral subiculum (vSUB) mainly consists of two distinct principal neuron types: regular (red) and burst (blue) spiking neurons. Right, Representative current-clamp traces of regular-spiking (above) and burst-spiking (below) principal neurons. Insets, Magnification of the first action potential in response to current injection. B, Simplified graphic of vSUB to NAcMS circuitry. The proportion of regular-spiking and burst-spiking principal neurons in vSUB projecting to D1R and D2R MSNs in NAcMS is unknown. C–E, Above, Schematic drawing of intersectional circuit-tracing approaches and mouse lines used for characterizing projection pattern of vSUB-NAcMS circuit. Below, Representative image of viral expression in NAcMS in coronal slices. Scale bar: 500 µm. C, AAV-Retro-mRuby was injected in NAcMS of WT mice, labeling all NAcMS projecting cells. D, WGA:Cre-recombinase is retrogradely transported from NAcMS to vSUB output neurons that are labeled by injection of Cre-dependent mRuby. This allows for the visualization of NAcMS-projecting vSUB cells. E, Cre-dependent WGA:Flp-recombinase is selectively expressed in D1R or D2R MSN populations in NAcMS and retrogradely transported to vSUB output neurons that express Flp-dependent mRuby. This permits the visualization of D1R or D2R MSN projecting vSUB neurons. F, Representative image of a horizontal vHIPP slice after intersectional tracing using AAV-DIO-WGA:Flp-2A-GFP with AAV-DIOFrt-mRuby shows labeling of NAcMS D1R MSN-projecting cells is largely restricted to vSUB. Scale bar: 300 µm. G, Summary graph of regular-spiking and burst-spiking principal neurons projecting to NAcMS using different viral approaches and transgenic mouse lines. Independent of postsynaptic cell type, regular-spiking neurons send more projections to overall NAcMS than burst-spiking neurons (left two bars). Similar proportions of regular-spiking and burst-spiking neurons project to D1R and D2R MSNs. There is no difference in proportion of regular-spiking and burst-spiking neurons projecting to NAcMS between males and females (males: closed circles, females: open circles). Number of cells recorded and animals used for each method is included in the graph. Error bars: ±SEM. RS, regular-spiking; BS, burst-spiking; MSN, medium spiny neuron; vHIPP, ventral hippocampus; vSUB, ventral subiculum; NAcMS, nucleus accumbens medial shell.
Given the strikingly disproportionate level of vSUB regular-spiking neuron input to NAcMS, we asked whether regular-spiking neurons also disproportionally innervate D1R and D2R MSNs. To begin to address this question, we employed an intersectional viral circuit tracing approach using wheat germ agglutinin (WGA), which functions as a transcellular retrograde tracer to label NAcMS-projecting vSUB neurons (Gradinaru et al., 2010). To test whether WGA efficiently retrogradely labels neurons in vSUB, we stereotactically injected AAVs carrying WGA fused to Flp recombinase (WGA-Flp) into NAcMS and Flp-dependent mRuby expressing AAVs into vSUB of wild-type animals (Fig. 1D). Indeed, consistent with our results using rgAAV2-mRuby, ∼70% of the mRuby+ vSUB neurons labeled by the WGA-mediated intersectional tracing approach were regular-spiking (Fig. 1G). Thus, WGA uptake and transport is efficient and permits the characterization of the cell type-specific connectivity of the vSUB-NAc circuit.
Next, to identify the subicular neurons that project to either D1R or D2R MSNs, we stereotactically co-injected D1R-Cre or A2a-Cre mice with AAVs harboring cre-dependent WGA-Flp (DIOLoxP-WGA-Flp) into the NAcMS and AAVs carrying Flp-dependent mRuby (DIOFrt-mRuby) into vSUB (Fig. 1E). Using this strategy, WGA-Flp expression is restricted to either Cre-expressing D1R or D2R MSNs, and is retrogradely transported into vSUB, where it selectively enables the expression of Flp-dependent mRuby in vSUB neurons that innervate D1R or D2R MSNs (Fig. 1F). mRuby+ neurons were then identified electrophysiologically in ex vivo slices. Using this intersectional approach, we found that vSUB regular-spiking and burst-spiking neurons surprisingly projected in roughly equal proportions to both D1R and D2R MSNs; regular-spiking neurons represented ∼45% of total vSUB input to each cell type, which is in contrast to retrograde labeling results of total NAcMS (Fig. 1G). This suggests that the regular-spiking neurons that represent the major input to total NAcMS, which includes MSNs as well as local GABAergic and cholinergic interneurons, may primarily target non-MSN cell populations. This is consistent with the finding that vHIPP excitatory synaptic strength is significantly greater to parvalbumin interneurons in nucleus accumbens shell than to MSNs (Yu et al., 2017; Scudder et al., 2018). Our findings also show that burst-spiking neurons are more highly represented in the vSUB to NAcMS MSN circuit than the vSUB to NAcMS interneuron circuit.
vSUB input to NAcMS is biased to D2R MSNs
The properties of vSUB synapses on D1R and D2R MSNs in NAcMS have not been directly assessed and may be distinct from other areas of the extended ventral hippocampal formation which also project to NAc, such as vCA1 and ventral entorhinal cortex (Okuyama et al., 2016). To examine the specific functional connectivity of vSUB-D1R and vSUB-D2R MSN synapses in NAcMS, we stereotactically injected vSUB of D1R-tdTomato mice with AAV-ChIEF-mRuby on P21 (Fig. 2A). We allowed ChIEF-mRuby to express before making ex vivo coronal slices of NAc. Immediately before the start of each experiment, we also generated serial horizontal brain sections, which permit precise examination of discrete hippocampal subfields, to confirm that ipsilateral vSUB injections had no, or very minimal, viral spread to CA1 or EC (Fig. 2B). Hemispheres with spillover to vCA1 or ventral entorhinal cortex were excluded from this study. We found that vSUB terminals were concentrated in the dorsal NAcMS (Fig. 2B), consistent with what has been observed following total vHIPP injections of ChR2 (Britt et al., 2012; MacAskill et al., 2014). We then performed dual whole cell recordings in ex vivo slices of nearby TdTomato-positive (D1R) and TdTomato-negative (putative D2R) MSNs in dorsal NAcMS and performed terminal stimulation to measure light-evoked excitatory postsynaptic currents (EPSCs). As vSUB fiber density varies along the rostral-caudal axis of NAcMS (Groenewegen et al., 1999), recording from neighboring pairs of D1R and D2R MSNs allows for the direct comparison of vSUB input to D1R versus D2R MSNs. We first sought to characterize the relative strengths of vSUB-D1R and vSUB-D2R MSN excitatory synapses by assessing the input-output relationship (I/O). As expected, given the robust input of vSUB to NAcMS, optogenetic stimulation of vSUB terminals produced reliable EPSCs in both classes of MSNs. Interestingly, we found that vSUB EPSC amplitudes at individual light intensities (D1R vs D2R at 1.8, 3.45, 5.78 mW, U = 78, 63, 47, p = 0.034, 0.008, 0.004, respectively, Mann–Whitney; Fig. 2C) and the corresponding slopes of the I/O curves were significantly stronger onto D2R MSNs compared with D1R MSNs (U = 49, Mann–Whitney: p = 0.0027; Fig. 2C).
vSUB input to NAc medial shell is biased to D2R MSNs. A, Experimental schematic to isolate vSUB input to D1R and D2R MSNs: AAV-ChIEF was injected in vSUB of ∼P21 D1R-TdTomato mouse. At P56–P63, coronal NAc slices were made and vSUB input onto D1R and D2R MSNs in NAcMS was measured followed by optical stimulation of vSUB terminals. B, Left, Representative image of ChIEF-mRuby injected vSUB in the horizontal plane. Scale bar: 300 µm. Right, Representative image of vSUB fibers infected with ChIEF-mRuby enriched in NAcMS of a coronal NAc slice. Scale bar: 500 µm. C, D2R MSNs receive stronger input from vSUB than D1R MSNs in NAcMS. Representative light-evoked EPSC traces (left), input/output curves (middle), and corresponding summary graph of the I/O slope (right) of light-evoked EPSC amplitudes of D1R and D2R MSNs. The slope represents the linear range (0.8–3.45 mW) of the I/O curve. Closed circles represent individual cells analyzed from males and open circles represent individual cells analyzed from females. D, Left, Representative image of ChIEF-mRuby expression in the entire vHIPP in horizontal plane. Scale bar: 300 µm. Right, Representative traces of light-evoked EPSCs recorded from simultaneous dual recordings of D1R and D2R MSNs. E, EPSC ratio of NAcMS D2R/D1R MSNs determined from optogenetic activation of vHIPP or vSUB terminals. Data are presented as geometric mean with 95% confidence intervals. Note logarithmic scale of y-axis. vHIPP n = 21 pairs from 4 mice; vSUB n = 30 pairs from 14 mice. Closed circles represent individual cells analyzed from males and open circles represent individual cells analyzed from females. Error bars: ±SEM. Number of cells and animals used for each experiment is included in the figure or corresponding figure legend. Exact p-values and statistical methods are presented in the results section. *p < 0.05, **p < 0.01.
The bias of vSUB-specific synaptic strength on D2R MSNs is unexpected because it diverges from previous findings from total vHIPP that revealed synaptic strength is biased for D1R MSNs (MacAskill et al., 2014). This could indicate that the functional organization of the vSUB projections in dorsal NAcMS is unique from the total vHIPP. Importantly, vHIPP output represents an amalgam of afferents originating from vCA1, vSUB, and possibly ventral entorhinal cortex. To directly compare the properties of synapses made by vHIPP and vSUB onto D1R and D2R MSNs, we infected entire vHIPP or vSUB with AAV-ChIEF (Fig. 2D) and stimulated terminals in dorsal NAcMS while performing dual recordings from pairs of neighboring D1R and D2R MSNs. We calculated the D2R/D1R EPSC ratios to compare the relative cell type-specific synaptic strengths of each input. Consistent with the previously observed vHIPP-D1R bias, the D2/D1 ratio in vHIPP injected animals was <1, whereas in vSUB injected animals, the D2/D1 ratio was >1, and the ratios were significantly different from one another [geometric means: vHIPP = 0.79 (95% CI = 0.57–1.1), vSUB = 1.394 (95% CI = 1.01–1.92) Mann–Whitney: p = 0.0224; Fig. 2E]. Our findings support the notion that the vSUB to NAcMS circuit possesses connectivity and synaptic properties unique from other hippocampal subfield inputs and argues that specific targeting of vSUB is necessary to reveal these subregion specific characteristics. This is further supported by a recent study that performed rabies-based circuit tracing experiments and found that more vSUB neurons project to D2R MSNs than to D1R MSNs in the shell of the nucleus accumbens, whereas vCA1 neurons exhibited no bias (Li et al., 2018).
Finally, the sexually dimorphic organization of vSUB local circuitry (Boxer et al., 2021) prompted us to ask whether the vSUB-specific bias for D2R MSNs was sex-specific. We separated our data by sex but found no sex differences (effect of sex: 0.913, effect of cell type: F(1,29) = 6.913, p = 0.014, effect of interaction: p = 0.614, ordinary two-way ANOVA; Fig. 2C; males: closed circles, females: open circles). Thus, although regular-spiking and burst-spiking neurons in vSUB are under local sex-specific inhibitory control, their output to NAc MSNs does not appear to be organized in a sex-dependent manner.
Light-evoked vSUB aEPSC frequency is greater at D2R versus D1R MSN synapses
To further characterize this unique circuit, we next aimed to identify the underlying synaptic properties that contribute to the biased synaptic strength of vSUB-D2R MSN synapses. This bias can be a result of greater presynaptic release, greater postsynaptic strength, and/or more numerous synaptic contacts. First, we quantified paired-pulse ratios of presynaptic vSUB boutons, an indirect measure of release probability, but found that the paired-pulse ratios were not different between D1R and D2R MSNs (20 ms Inter-stimulus Interval (ISI): U = 96.3, p = 0.755, Mann–Whitney; 50 ms ISI: t(27) = 0.828, p = 0.415, unpaired t test; 100 ms ISI: t(27) = 1.272, p = 0.214, unpaired t test; Fig. 3A). Next, we assessed the quantal properties of these synapses by monitoring strontium-evoked asynchronous EPSCs (aEPSCs). We substituted extracellular calcium for strontium which desynchronizes light-evoked vesicle release from vSUB terminals resulting in a reduced phasic EPSC followed by asynchronous release events that are quantal in nature (Bekkers and Clements, 1999). aEPSC amplitude is commonly thought to reflect postsynaptic strength, while aEPSC frequency may reflect the number of functional synaptic contacts (Sinnen et al., 2017). To assess whether we could effectively discern vSUB-mediated aEPSCs from spontaneous activity in dorsal NAcMS MSN synapses, we compared baseline activity to aEPSCs occurring shortly after vSUB terminal stimulation and observed a threefold increase in aEPSC frequency (baseline vs light-evoked: t(11) = 8, p < 0.0001, paired t test; Fig. 3B). Additionally, we observed higher baseline aEPSC frequency in D2R MSNs (D1R vs D2R baseline: t(4) = 7, p = 0.009, unpaired t test), which may suggest two nonmutually exclusive possibilities: that the majority of synapses made onto MSNs in NAcMS originate from vSUB and/or that there are other, yet to be characterized, input regions that also exhibit a synaptic bias onto D2R MSNs (Fig. 3B). While we did not observe differences in the amplitude of light-evoked vSUB aEPSCs, we did observe that aEPSC frequency was ∼twofold higher at vSUB-D2R MSN synapses compared with D1R MSN synapses (amplitude: t(24) = 0.8882, p = 0.3833; frequency: t(24) = 7.029, p < 0.0001, unpaired t tests; Cumulative frequency of amplitude and interevent interval: D = 0.045 and 0.103, p = 0.2713 and p < 0.0001, respectively, K-S test; Fig. 3C). Taken together, the higher aEPSC frequency at vSUB-D2R MSN synapses cannot be explained by presynaptic release (Fig. 3A) indicating that the biased evoked synaptic strength measured at vSUB-D2R MSN synapses (Fig. 2C) is driven by an imbalance of synaptic contacts made by vSUB onto D2R MSNs and D1R MSNs (Fig. 3C).
Light-evoked vSUB aEPSC frequency is greater at D2R versus D1R MSN synapses. A, Presynaptic release probability is similar at vSUB-D1R and vSUB-D2R MSN synapses. Representative traces (left) and quantification (right) of light-evoked paired-pulse ratios from D1R and D2R MSNs. R2/R1, second response/first response. B, Left, Representative traces of strontium-mediated asynchronous EPSCs in the absence of optical stimulation (Baseline; top) and after a single optical stimulation (Light-Evoked; bottom) recorded from the same cell. Right, Summary plot of aEPSC frequency at baseline and after optical stimulation of vSUB synapses on D1R and D2R MSNs. aEPSC frequency of D2R MSNs increases significantly during 500 ms following light stimulation. n = 9 pairs (6 pairs from male mice and 3 pairs from female mice). C, Frequency but not amplitude of vSUB aEPSCs is greater in D2R compared with D1R MSNs. Left, Representative traces of strontium-mediated aEPSCs from a D1R and D2R MSN dual recording. Analysis of aEPSC amplitude and frequency was done in 500-ms window poststimulation (gray dashed line). See inset for a magnification of exemplary single aEPSC. Middle and right, Cumulative frequency distribution of aEPSC amplitude and frequency, respectively, from D1R and D2R MSNs. Inset bar graphs show mean aEPSC amplitude and frequency. D1R and D2Rs from simultaneous dual recordings are connected. There is no difference in aEPSC amplitude and frequency from D1R and D2R MSNs between males and females (males: closed circles, females: open circles). Error bars: ±SEM. Number of cells and animals used for each experiment is included in the figure or corresponding figure legend. Exact p-values and statistical methods are presented in the results section. **p < 0.01, ****p < 0.0001.
vSUB-NAcMS synapses harbor GluA2-lacking AMPA receptors
The unexpected bias of excitatory synaptic transmission onto D2R MSNs prompted us to further interrogate the synaptic properties of these inputs. We next assessed the properties of the receptors that populate vSUB-D1R and D2R MSN synapses, starting with the AMPAR subunit composition. We performed AMPAR current–voltage experiments to examine the composition of AMPARs at these synapses. As a control, we first assayed total glutamatergic input onto MSNs by monitoring electrically evoked AMPAR EPSCs. Consistent with previous reports, we found that the current–voltage curves were linear, indicating the majority of MSN synapses possess GluA2-containing calcium-impermeable AMPARs (Fig. 4A; Britt et al., 2012; Pascoli et al., 2014; Terrier et al., 2016). Next, we assayed AMPAR composition at vSUB-MSN synapses by selectively activating vSUB terminals via optogenetic stimulation in drug-naive animals. Surprisingly, AMPARs at vSUB-D1R and vSUB-D2R MSN synapses exhibited inward rectification at depolarized potentials in the presence of intracellular spermine, which is a hallmark of GluA2-lacking calcium-permeable AMPARs (Fig. 4B). At glutamatergic synapses in NAc, calcium-permeable AMPARs are typically found following experience-dependent plasticity, and not in baseline conditions in drug-naive animals. To confirm that the observed rectification was because of spermine-blockade of calcium-permeable AMPARs, we stimulated vSUB terminals and recorded from MSNs using an internal solution without spermine, and as expected, found that the resulting current–voltage curve and corresponding rectification index were significantly more linear (H = 10.61, p = 0.005, Kruskal–Wallis test; Dunn's multiple comparisons: No spermine vs D1R: p = 0.0092, No spermine vs D2R: p = 0.0075, D1R vs D2R: p > 0.999; Fig. 4B). To further confirm that calcium-permeable AMPARs are expressed at these synapses, we washed-in the calcium-permeable AMPAR antagonist, NASPM, while optically stimulating vSUB terminals. We observed 34 ± 1% and 42 ± 3.5% depression of vSUB-AMPAR EPSC amplitudes in D1R and D2R MSNs, respectively, following 20 min of NASPM exposure (U = 7, p = 0.3524, Mann–Whitney; Fig. 4C). The series resistances monitored throughout each experiment were stable (<2% deviation from baseline; data not shown), indicating that the depression of light-evoked AMPAR-mediated currents was likely because of NASPM block of calcium-permeable AMPARs and not a result of cellular rundown or loss of voltage clamp. To our knowledge, the basal presence of calcium-permeable AMPARs at synapses made by vSUB neurons onto D1R and D2R MSNs in drug-naive mice is unique from other output subfields of vHIPP.
vSUB-NAc medial shell synapses harbor GluA2-lacking AMPA receptors. A, Nonspecific activation of glutamatergic synapses via local electrical stimulation reveals linear AMPAR current–voltage (I/V) relationships. From left, Schematic drawing of experiment set-up, and representative traces of electrically evoked EPSCs voltage-clamped at −70, 0, and +50 mV (overlaid) from D1R and D2R MSNs. Middle, AMPAR I/V curves. Right, Summary graphs of the rectification indices for electrically evoked EPSCs from D1R and D2R MSNs. B, AMPAR current–voltage relationships of vSUB-D1R and vSUB-D2R MSN synapses exhibit inward rectification. Left, Representative traces of AMPAR-mediated EPSCs at −70, 0, and 50 mV (overlaid) at vSUB to D1R and D2R MSN synapses. Middle and right, AMPAR I/V curves and rectification indices, respectively, of D1R MSNs (blue), D2R MSNs (orange), or MSNs recorded without spermine in internal solution (black). Exact p-values and statistical methods are presented in the results section. **p < 0.01. C, Extracellular application of NASPM depresses vSUB-MSN AMPAR EPSC amplitudes. Left, Representative scaled traces of D1R and D2R MSN EPSCs before and after application of NASPM. Middle, EPSC amplitudes normalized to 5 min baseline and recorded following 18–20 min of NASPM application. Right, Summary graph of AMPAR EPSC amplitudes averaged over the last 2 min of NASPM wash-in reveals a ∼40% depression. D, AMPAR rectification differs depending on presynaptic input. Left, Images of AAV-ChIEF-mRuby injections in vCA1 (left), vSUB (middle), and entire vHIPP (right). Scale bars: 300 µm. Right, Summary graph of rectification indices. No spermine control: n = 7/3, vCA1: n = 22/3, vSUB: n = 30/5, vHIPP: n = 42/6. H = 21.69, p < 0.0001, Kruskal–Wallis test; Dunn's multiple comparisons: vCA1 versus No spermine: p > 0.9999, vCA1 versus vSUB: ***p = 0.0005, vCA1 versus vHIPP: p = 0.0828, No spermine versus vSUB: **p = 0.0035, No spermine versus vHIPP: p = 0.0905, vSUB versus vHIPP: p = 0.3509. Error bars ± SEM. Males: closed circles, females: open circles. Number of cells and animals used for each experiment is included in the figure or corresponding figure legend.
Our finding that vSUB-NAcMS MSN synapses contain calcium-permeable AMPARs is particularly unexpected because optical stimulation of fibers from total vHIPP identified that these synapses contain calcium-impermeable AMPARs (Britt et al., 2012; Pascoli et al., 2014). However, vHIPP output is a composite of vCA1 and vSUB, thus the presence of calcium-impermeable AMPARs observed at vHIPP-MSN synapses might primarily reflect input originating from vCA1. To attempt to reconcile the differences in AMPAR composition at vSUB-MSN and vHIPP-MSN synapses, we selectively injected vCA1 with AAV-ChIEF (Fig. 4D) and assessed the AMPAR current–voltage relationship. We found that the AMPAR rectification index at vCA1-MSN synapses was significantly higher than that of the inwardly rectifying vSUB-MSNs, and was indistinguishable from the no-spermine control, suggesting that vCA1 synapses are populated by calcium-impermeable AMPARs (Fig. 4D). Furthermore, we performed total vHIPP injections with AAV-ChIEF, performed current–voltage experiments in MSNs. We validated that all subfields of vHIPP were infected with AAV-ChIEF and found that the vHIPP rectification index fell between the vCA1 and vSUB rectification indices (RI: vCA1 = 0.45 ± 0.02, vSUB = 0.34 ± 0.02, vHIPP = 0.40 ± 0.02) and had the largest spread of the three conditions (coefficient of variation: vCA1 = 20%, vSUB = 26%, vHIPP = 35%; Fig. 4D). Therefore, the AMPAR composition monitored at vHIPP-MSN synapses resembles a mixture of vCA1 and vSUB input. Together, the current–voltage relationship and NASPM experiments strongly suggest that in drug-naive animals, vSUB-MSN synapses contain calcium-permeable AMPARs, and that this property is unique from vCA1-MSN synapses (H = 21.69, p < 0.0001, Kruskal–Wallis; Dunn's multiple comparisons: vCA1 vs vSUB: p = 0.0005, No spermine vs vSUB: p = 0.0035; Fig. 4D). Additionally, the comparison of AMPAR rectification indices at vCA1, vHIPP with vSUB synapses suggests that differences in injection site (e.g., predominant infection of vCA1) likely results in the disproportionate representation of vCA1-MSN synaptic properties and provides an explanation why calcium-impermeable AMPARs have been observed at vHIPP-NAcMS MSN synapses. This unique feature of vSUB synapses prompted us to ask whether this phenotype is sexually dimorphic. We separated our current–voltage data by sex and found both males and females exhibit the same degree of inward rectification at vSUB-D1R and vSUB-D2R synapses (effect of sex: p = 0.9123, effect of cell type: p = 0.8232, effect of interaction: p = 0.4768, ordinary two-way ANOVA; Fig. 4B; males: closed circles, females: open circles).
NMDA receptor function is equal at vSUB-D1R and vSUB-D2R MSN synapses
Next, we interrogated the properties of NMDA receptors (NMDARs) at vSUB-NAcMS synapses. NMDARs are required for the induction of synaptic plasticity at vHIPP-NAcMS synapses following cocaine exposure or high frequency stimulation (MacAskill et al., 2014; Pascoli et al., 2014; LeGates et al., 2018). Interestingly, at vHIPP synapses made on unidentified MSNs, NMDARs are less sensitive to blockade by Mg2+ and exhibit strong inward current at negative holding potentials compared with inputs from basolateral amygdala or medial prefrontal cortex, indicating that NMDAR subunit composition is input-specific (Britt et al., 2012). Given that calcium-permeable AMPARs are unique to vSUB-MSN synapses, we questioned whether vSUB synapses in NAcMS also contain NMDARs with strong inward current and whether NMDAR composition exhibits synapse-specificity. To specifically address these questions, we pharmacologically isolated NMDAR-mediated EPSCs while optically stimulating vSUB terminals. First, we assessed the NMDAR I/O relationship at vSUB-MSN synapses and found that amplitudes and the corresponding I/O curves were similar between D1R and D2R MSNs (U = 89, p = 0.7168, Mann–Whitney; Fig. 5A). Additional analysis of NMDAR decay kinetics showed similar weighted decay values for D1R and D2R MSNs (t(16) = 0.127, p = 0.9005, unpaired t test; Fig. 5B). Next, we assayed NMDAR subunit composition by testing the current–voltage relationship. D1R and D2R MSNs NMDAR current–voltage curves overlapped and displayed very minimal inward current at negative potentials (fraction of +50 mV amplitude: 0.14 at −70 mV and 0.25 at −25 mV; D1R vs D2R: F(12,79) = 354.2, −70 mV: p = 0.9623; −50 mV: p = 0.9995; −25 mV: p ≥ 0.9999; 0 mV: p ≥ 0.9999; +25 mV: p = 0.1974; +50 mV: p ≥ 0.9999, ordinary one-way ANOVA; Fig. 5C). Thus, vSUB-D1R and vSUB-D2R MSN synapses contain NMDARs that are more sensitive to Mg2+ compared with the NMDARs previously observed at total vHIPP inputs (Britt et al., 2012), suggesting that the induction of plasticity at vSUB-NAcMS synapses could be distinct from vHIPP synapses. Together, our results indicate that both AMPAR-mediated and NMDAR-mediated synaptic transmission properties at vSUB inputs to D1R and D2R MSNs are distinct from vHIPP-MSN synapses.
NMDA receptor function is equal at vSUB-D1R and vSUB-D2R MSN synapses. A, Optically evoked NMDAR-mediated EPSC amplitudes are equal at vSUB to D1R and D2R MSN synapses. Left, MSNs are voltage-clamped at +50 mV, and EPSCs are recorded at different light intensities. Representative traces from D1R and D2R MSNs are overlaid. Middle, NMDAR-mediated I/O curve. Right, Summary graph of NMDAR I/O slopes. B, NMDAR decay kinetics are comparable between D1R and D2R MSNs. C, NMDAR current–voltage relationships. Left, Representative NMDAR-mediated EPSCs at −25, −50, and +50 mV. Right, NMDAR I/V curves at vSUB-D1R and vSUB-D2R MSNs. All values are normalized to +50 mV. Error bars ± SEM. Males: closed circles, females: open circles. Number of cells and animals used for each experiment is included in the figure. Exact p-values and statistical methods are presented in the results section.
Ventral subiculum to D2R MSN synapses is potentiated following cocaine administration
vHIPP-NAc glutamatergic inputs undergo plasticity following cocaine administration and withdrawal. In drug-naive mice, vHIPP exhibits a bias in synaptic strength for D1R MSNs and vHIPP-D1R synapses are selectively potentiated following withdrawal from contingent and noncontingent administration of cocaine. However, whether vSUB-MSN synapses undergo similar drug-induced plasticity is untested. Our results thus far indicate that vSUB possess distinct basal synaptic transmission properties relative to total vHIPP or vCA1 at MSN synapses. We therefore tested the impact of cocaine exposure on vSUB-MSN synapses. We injected D1R-TdTomato mice with AAV-ChIEF on ∼P21 then administered daily intraperitoneal injections of either saline or cocaine (20 mg/kg) for 5 d. After a 10–11 d withdrawal period, we interrogated the synaptic properties of vSUB-D1R and vSUB-D2R synapses in dorsal NAcMS ex vivo slices (Fig. 6A,B). We systematically tested for the manifestation of cocaine-induced presynaptic and postsynaptic plasticity by measuring paired-pulse ratios to assess changes in presynaptic release, AMPA-to-NMDA ratios to assess postsynaptically expressed potentiation, and AMPAR current–voltage curves and NASPM wash-ins to detect changes in the subunit stoichiometry of AMPARs. Presynaptic release was unaffected by cocaine withdrawal, as paired-pulse ratios were comparable between saline or cocaine injected animals at vSUB-D1R and vSUB-D2R synapses (D1R saline vs D1R cocaine: t(60) = 1.504, p = 0.150, D2R saline vs D2R cocaine: t(64) = 0.15, p = 0.8815, unpaired t tests; Fig. 6C). Next, we measured AMPA-to-NMDA ratios by quantifying the EPSC amplitudes at holding currents of −70 and +40 mV to isolate AMPAR-mediated and NMDAR-mediated EPSC amplitudes, respectively. Interestingly, while we did not observe cocaine-induced changes in the AMPA-to-NMDA ratio at vSUB-D1R synapses, we observed a significant increase in the AMPA-to-NMDA ratio at vSUB-D2R synapses in cocaine injected animals (D1R saline vs D1R cocaine: t(55) = 0.74, p = 0.462, D2R saline vs D2R cocaine: t(52) = 2.314, p = 0.025, unpaired t tests; Fig. 6D). The synaptic potentiation observed at vSUB-D2R synapses represents another unique property of these synapses and contrasts with the selective cocaine-induced potentiation at vHIPP-D1R synapses. We next tested whether the cocaine-induced plasticity changes the composition of AMPARs at vSUB-D2R synapses. We found that the current–voltage rectification indices at either D1R or D2R synapses was unchanged between cocaine compared with saline-injected groups (D1R saline vs D1R cocaine: t(30) = 0.297, p = 0.769, D2R saline vs D2R cocaine: t(32) = 0.559, p = 0.580, unpaired t tests; Fig. 6E–G). We confirmed this finding with application of extracellular NASPM, which produced a similar degree of AMPAR-mediated EPSC depression in saline and cocaine injected animals at both vSUB-D1R and vSUB-D2R synapses (D1R saline vs D1R cocaine: U = 10, p = 0.43, D2R saline vs D2R cocaine: U = 9, p = 0.329, Mann–Whitney; Fig. 6H–J). Together, these results indicate that while cocaine-induced plasticity at vSUB-D2R synapses manifests as an increase in synaptic AMPARs, the stoichiometry of CI-permeable and calcium-permeable AMPARs established basally is unexpectedly maintained after synaptic potentiation.
Cocaine potentiates vSUB-D2R MSN synapses. A, Simultaneous dual recording schematic to monitor vSUB-D1R and vSUB-D2R MSN synaptic properties. AAV-ChIEF was injected in vSUB of D1R-TdTomato mice, then vSUB input to D1R and D2R MSNs in NAcMS was measured by optically stimulating vSUB terminals. B, Experimental timeline: mice were injected with AAVs on P21 then three weeks later received daily noncontingent injections of saline or cocaine (20 mg/kg) for 5 d. Animals were used for ex vivo recordings following a 10- to 11-d withdrawal period. C, Exposure to cocaine does not change release probability in vSUB-NAcMS synapses. Left, Representative traces of light-evoked paired-pulse ratios from D1R and D2R MSNs in saline or cocaine-injected mice. Right, Quantification of paired-pulse ratios. R2/R1 (second response/first response). 50-ms interstimulus interval. D, Exposure to cocaine induces synaptic plasticity at vSUB-D2R MSN synapses. Left, Representative traces of light-evoked EPSCs at −70 and +40 mV from D1R and D2R MSNs in saline or cocaine-injected mice. Right, Quantification of AMPA-to-NMDA ratios. A direct sex-specific comparison of AMPA/NMDA ratios did not identify sex differences in cocaine-induced plasticity (females: closed circles, males: open circles). Sample size: female n = 3 mice, D1R saline n = 14, D1R cocaine n = 15, D2R saline n = 15, D2R cocaine n = 16; male N = 3 mice, D1R saline n = 15, D1R cocaine n = 13, D2R saline n = 12, D2R cocaine n = 11. E, F, Left, Representative traces of AMPAR EPSCs at −70 and +50 mV. Right, AMPAR I/V plots of saline and cocaine-injected mice from vSUB-D1R and vSUB-D2R MSN synapses, respectively. G, Exposure to cocaine does not change AMPAR inward rectification compared with saline controls. Rectification index of D1R and D2R MSNs in saline or cocaine-injected mice. H, Representative traces of AMPAR EPSCs before and after extracellular application of NASPM. I, Change in EPSC amplitude normalized to baseline over time in D1R (left) and D2R MSNs (right) from saline or cocaine-injected mice. J, Summary graph of EPSC amplitudes averaged over the last 2 min of NASPM wash-in from D1R and D2R MSNs in saline or cocaine-injected mice. Error bars ± SEM. Males: closed circles, females: open circles. Number of cells and animals used for each experiment is included in the figure or corresponding figure legend. Exact p-values and statistical methods are presented in the results section. *p < 0.05.
Finally, we tested whether the vSUB phenotype was sex specific by separating our AMPA-to-NMDA data by sex and performing a three-way ANOVA (Fig. 6D; males: closed circles, females: open circles). As expected, there were main effects of cell type and drug condition, but there was no main effect of sex, nor any interactions between the three variables (main effects: sex: p = 0.16, cell type: p = 0.0339, drug: p = 0.0209; interactions: sex-cell: p = 0.74, sex-drug: p = 0.14, cell-drug: p = 0.16, sex-cell-drug: p = 0.29, Three-way ANOVA). Thus, consistent with our findings that the vSUB-NAcMS circuit is unique from the total vHIPP-NAcMS circuit, we find that vSUB inputs to dorsal NAcMS display unique cocaine-induced plasticity distinct from total vHIPP input (which includes vSUB but also vCA1 and possibly ventral entorhinal cortex), which experiences potentiation at D1R synapses, whereas vSUB-specific input undergoes D2R MSN-specific alterations.
Discussion
vHIPP input to the NAcMS has received attention for its proposed role in substance use disorders. vCA1 and vSUB are surprisingly distinct subregions in vHIPP, they differ in principal neuron diversity, projection targets within NAc shell, and in their influence on behavior; however, a distinction between their functional contribution in the vHIPP-NAcMS circuit has not been explored. Here, we systematically interrogated the cell type-specific organization and synapse-specific properties of the vSUB-NAcMS circuit and report four fundamentally important findings. First, using an intersectional retrograde circuit tracing approach, we defined the connectivity of regular-spiking and burst-spiking neurons with D1R and D2R MSNs and found that equal proportions of regular-spiking and burst-spiking neurons innervate D1R and D2R MSNs. Second, we performed the first functional dissection of vSUB-NAcMS circuitry and found a striking bias of excitatory synaptic strength at D2R synapses, relative to D1R synapses, mediated by differences in synapse density. Third, we reveal that vSUB-D1R and vSUB-D2R synapses basally contain calcium-permeable AMPARs. Fourth, we report that vSUB-D2R synapses, but not D1R synapses, are selectively potentiated following cocaine administration. Importantly, the functional properties of vSUB-D1R and vSUB-D2R synapses in drug-naive and drug exposed mice reveal novel properties of these synapses not previously observed in vHIPP-NAc synapses.
Cell type-specific organization and synaptic properties of the vSUB-NAcMS circuit
Regular-spiking neurons represent about half of the total population of principal neurons in vSUB, however, regular-spiking cells provide ∼70% of vSUB output to total NAc. We used intersectional retrograde tracing to reveal that equal proportions of regular-spiking and burst-spiking neurons project to D1R and D2R MSNs. Our findings suggest that regular-spiking afferents may primarily innervate non-MSNs, such as local interneurons that provide strong inhibition within NAcMS (Scudder et al., 2018; Baimel et al., 2022). Although the functional relevance of regular-spiking and burst-spiking neurons is largely unknown, regular-spiking and burst-spiking neurons exhibit distinct forms of long-term potentiation and the activation of D1/D5 receptors on regular-spiking neurons reduces the threshold for LTP induction at CA1-SUB synapses (Wozny et al., 2008; Roggenhofer et al., 2013). Additionally, the intrinsic excitability properties of regular-spiking neurons undergo experience-dependent plasticity following exposure to novel contexts relevant for contextual fear learning (Dunn et al., 2018). Thus, regular-spiking neuron input to local interneurons in NAcMS may serve as an important feed-forward mechanism to sustain excitatory-inhibitory balance of MSNs following cocaine exposure. Future experiments are needed to elucidate the functional properties of regular-spiking input onto NAcMS local interneurons.
We also report that vSUB neurons exhibit a synaptic bias for D2R MSNs in dorsal NAcMS. Furthermore, in drug-naive animals, vSUB-D1R and vSUB-D2R synapses unexpectedly contain calcium-permeable AMPARs. By contrast, previous studies of vHIPP-MSN synapses revealed a bias for D1R MSNs and the presence of synaptic calcium-impermeable AMPARs (Britt et al., 2012; MacAskill et al., 2014; Pascoli et al., 2014). To reconcile the striking differences in AMPAR composition, we found that calcium-impermeable AMPARs populate vCA1-MSNs synapses suggesting that the previously reported synaptic properties of vHIPP-NAc synapses may primarily reflect the properties of vCA1 synapses. vCA1 and vSUB are molecularly distinct subregions of vHIPP, and while studies are still limited, vCA1 and vSUB could be functionally distinct as well (Bryant and Barker, 2020). vCA1 neurons encode reward location, cocaine CPP, and store social memory (Ciocchi et al., 2015; Okuyama et al., 2016; Zhou et al., 2019) while vSUB principal neurons encode the vigor of reward seeking (Lindenbach et al., 2022) and are required for context-induced drug relapse in rats (Bossert et al., 2016; Marchant et al., 2016). Perhaps it is not surprising that these regions exhibit differences in basal synaptic transmission properties and highlights the importance of studying each region separately.
Functional implications of the synaptic properties unique to vSUB
In NAc, calcium-permeable AMPARs typically manifest after drug exposure, thus, the use of calcium-permeable AMPARs at vSUB-MSN synapses of drug-naive mice is particularly intriguing and raises the question – what purpose does the basal expression of these calcium-permeable AMPARs serve? The basal expression of calcium-permeable AMPARs may maintain vSUB-MSN synapses in a metaplastic state to facilitate future plasticity (Park et al., 2021). Additionally, it has been proposed that synaptic plasticity can be triggered by Ca2+ influx via calcium-permeable AMPARs, independent of postsynaptic NMDARs (Lamsa et al., 2007; Mameli et al., 2011). This form of plasticity requires presynaptic excitatory transmission paired with postsynaptic hyperpolarization. Thus, feed-forward local inhibition in NAc, perhaps driven by vSUB regular-spiking neurons, may be required to induce synaptic plasticity at vSUB-MSN synapses. Additionally, Ca2+ influx through calcium-permeable AMPARs and NMDARs at the same synapse can additively induce LTP, which suggests that vSUB-MSN synapses may undergo exaggerated LTP in response to relevant stimuli (Jia et al., 1996).
Synapse-specific cocaine induced plasticity
We observed a selective cocaine-induced potentiation of vSUB-D2R synapses. By contrast, cocaine plasticity occurs selectively at vHIPP-D1R synapses (Britt et al., 2012; MacAskill et al., 2014; Pascoli et al., 2014). It is commonly thought that D1R and D2R MSNs participate in opposing but parallel pathways to regulate drug seeking behavior, D1R plasticity strictly promotes reward behavior whereas plasticity at D2R synapses encodes aversion. Thus, based on these putative roles, our findings seemingly contradict with the notion that hyperactivity of vSUB promotes drug seeking and drug reinstatement. However, our results add to a growing literature that NAc D2R MSNs play more complex roles in drug-induced plasticity and motivated behavior than previously appreciated and, in some cases, promote reward seeking behavior (Soares-Cunha et al., 2016; Terrier et al., 2016; Cole et al., 2018; Sjulson et al., 2018; Gong et al., 2021; Inbar et al., 2022). While NAc D2R MSNs mostly project to inhibitory neurons in VP to encode aversive behavior, D2R MSNs in dorsal NAcMS project to excitatory neurons in VP to promote reward (Yao et al., 2021). vSUB robustly projects to dorsal NAcMS, which is where we exclusively performed our experiments. However, given our incomplete understanding of vSUB-NAcMS circuity, we cannot exclude the possibility that vSUB-D2R synapses might promote aversive states and that reward behavior associated with vSUB hyperactivity is encoded by its afferents to other cell types in NAcMS. The selective potentiation of vSUB-D2R MSN synapses has not been described for vHIPP input and further underscores the need to study each output region of vHIPP individually.
In most subregions of striatum, dopamine signaling, acting via D1-receptors, is required for and enhances LTP on D1R-expressing MSNs, whereas activation of D2-receptors may inhibit LTP (Goto and Grace, 2005; Håkansson et al., 2006; Pawlak and Kerr, 2008; Schotanus and Chergui, 2008; LeGates et al., 2018; Iino et al., 2020). Synaptic D1-receptor and D2-receptor modulate excitability via G-protein signaling and can form heterodimers with the NMDAR subunit NR2B to either facilitate or inhibit NMDAR function (X.Y. Liu et al., 2006; Cahill et al., 2014; Pascoli et al., 2014). This raises the question: if dopamine signaling is required for LTP in NAc, how can we observe selective LTP at vSUB-D2R MSN synapses? However, at vHIPP-NAcMS synapses, D1-receptor and D2-receptor activation are neither required for nor do they enhance MSN LTP (LeGates et al., 2018). Dopamine signaling-independent LTP may be because of the fact that in NAcMS, dopamine terminals mostly contact MSN dendrites instead of spines, suggesting that dopamine signaling may be less able to directly influence glutamatergic input to spines (Zahm, 1992; Meredith et al., 2008). If LTP in dorsal NAcMS is independent of dopamine signaling, how do vSUB-D2R synapses selectively potentiate after cocaine exposure? It is possible that only vSUB neurons projecting to D2R MSNs experience increased activity following cocaine exposure, resulting in downstream potentiation. Additionally, microcircuit differences within NAc may drive synapse-specific changes. Biased feedforward inhibition onto D2R MSNs in dorsal NAcMS could provide the hyperpolarization necessary to enable potentiation at synapses that contain calcium-permeable AMPARs. Alternatively, vSUB-D2R synapses could be molecularly unique in other, yet untested ways. For example, cell adhesion molecules can specify synapse identity and govern synaptic plasticity (Südhof, 2017) and although largely unstudied in NAc, cell type-specific expression and/or localization of cell adhesion molecules may facilitate LTP at D2R-expressing MSN synapses (Rothwell et al., 2014; Fuccillo et al., 2015).
vSUB-D1R and vSUB-D2R synapses are not sexually dimorphic
Our recent work identified sex differences in PV-mediated inhibition of vSUB regular-spiking versus burst-spiking neurons and here we tested whether vSUB synapses in dorsal NAcMS also exhibit sexual dimorphism (Boxer et al., 2021) but found no differences in any of the circuit or synaptic properties measured. Importantly, these measurements were performed independent of real-time regulation by the vSUB local circuit. Sex specific differences are reported for substance use disorders in humans and reward seeking behaviors in rodents (Becker, 2016); however, a comprehensive understanding whether the sex-specific properties of vSUB local inhibition contribute to sexually dimorphic reward seeking behaviors is unknown. Future studies into the intact vSUB-NAcMS circuitry conducted in behaving animals will be critical to determine how vSUB local circuitry regulates excitatory output to NAcMS.
Footnotes
This work was supported by National Institutes of Health Grants R00MH103531 and R01MH116901 (to J.A.), T32NS099042 and 1F31MH125510 (to E.E.B.), and T32GM007635 (to BD) and by the Brain & Behavior Research Foundation Grant NARSAD24847 (to J.A.). We thank Nora Langer and Melanie Becher for their contributions to mice husbandry and AAV injections. We also thank the members of the Aoto lab for helpful discussions, Dr. Kevin Beier (University of California Irvine) for contributing plasmids and Dr. Robert Malenka (Stanford University) for contributing mouse lines, and the National Institute on Drug Abuse Drug Supply Program for supplying the Cocaine HCl used in this study.
The authors declare no competing financial interests.
- Correspondence should be addressed to Jason Aoto at jason.aoto{at}cuanschutz.edu
