Pathway- and Cell-Specific Kappa-Opioid Receptor Modulation of Excitation-Inhibition Balance Differentially Gates D1 and D2 Accumbens Neuron Activity

Transcription

1 Article Pathway- and Cell-Specific Kappa-Opioid Receptor Modulation of Excitation-Inhibition Balance ifferentially Gates 1 and Accumbens Neuron Activity Highlights d NAcc KORs presynaptically inhibit glutamate synapses in a pathway-specific manner d d KORs inhibit 1 MSN collaterals more robustly than they inhibit MSN collaterals KORs preferentially inhibit glutamate and onto 1 and MSNs, respectively Authors Hugo A. Tejeda, Jocelyn Wu, Alana R. Kornspun,..., Brad B. Lowell, William A. Carlezon, Jr., Antonello Bonci Correspondence antonello.bonci@nih.gov d KORs inhibit excitatory drive of 1 MSNs and disinhibit excitatory drive of MSNs In Brief Tejeda et al. provide a circuit-based framework wherein nucleus accumbens kappa-opioid receptor modulation of excitatory and inhibitory synapses dynamically alters the activity of nucleus accumbens 1 and MSNs. Tejeda et al., 17, Neuron 93, January 4, 17 Published by Elsevier Inc.

2 Neuron Article Pathway- and Cell-Specific Kappa-Opioid Receptor Modulation of Excitation-Inhibition Balance ifferentially Gates 1 and Accumbens Neuron Activity Hugo A. Tejeda, 1 Jocelyn Wu, 1 Alana R. Kornspun, 1 Marco Pignatelli, 1 Vadim Kashtelyan, Michael J. Krashes, 1,3 Brad B. Lowell, 4 William A. Carlezon, Jr., 5 and Antonello Bonci 1,6,7, * 1 Synaptic Plasticity Section Neuronal Networks Section National Institute on rug Abuse Intramural Research Program, Baltimore, M 14, USA 3 iabetes, Endocrinology and Obesity Branch, National Institute of iabetes and igestive and Kidney iseases, National Institutes of Health, Bethesda, M 89, USA 4 ivision of Endocrinology, iabetes and Metabolism, epartment of Medicine, Beth Israel eaconess Medical Center, Harvard Medical School, Boston, MA 15, USA 5 Behavioral Genetics Laboratory, epartment of Psychiatry, Harvard Medical School, McLean Hospital, Belmont, MA 578, USA 6 Solomon H. Snyder epartment of Neuroscience, Johns Hopkins University, Baltimore, M 15, USA 7 Lead Contact *Correspondence: antonello.bonci@nih.gov SUMMARY Endogenous dynorphin signaling via the kappaopioid receptor (KOR) in the nucleus accumbens (NAcc) powerfully mediates negative affective states and stress reactivity. Excitatory inputs from the hippocampus and amygdala play a fundamental role in shaping the activity of both NAcc 1 and MSNs, which encode positive and negative motivational valences, respectively. However, a circuitbased mechanism by which KOR modulation of excitation-inhibition balance modifies 1 and MSN activity is lacking. Here, we provide a comprehensive synaptic framework wherein presynaptic KOR inhibition decreases the excitatory drive of 1 MSN activity by the amygdala, but not the hippocampus. Conversely, presynaptic inhibition by KORs of inhibitory synapses on MSNs enhances integration of excitatory drive by the amygdala and hippocampus. In conclusion, we describe a circuit-based mechanism showing differential gating of afferent control of 1 and MSN activity by KORs in a pathwayspecific manner. INTROUCTION ynorphin (yn), an endogenous opioid receptor peptide, and its receptor, the kappa-opioid receptor (KOR), are powerful effectors of stress-induced alterations in reward processing and dysphoric states (Chavkin et al., 198; Van t Veer and Carlezon, 13). yn and KORs are highly expressed in the nucleus accumbens (NAcc) (Gerfen et al., 199; Meng et al., 1993; Meshul and McGinty, ; Svingos et al., 1999). Activation of NAcc KORs, either pharmacologically or by endogenous yn, produces aversion and depressive-like states (Bruchas et al., 1; Muschamp et al., 11; Tejeda et al., 1; Wee and Koob, 1). Increased yn expression in ventral and dorsal striatum of addicted and suicidal individuals (Hurd and Herkenham, 1993; Hurd et al., 1997), and in animal models of addiction and depression (Bruchas et al., 1; Carlezon et al., 1998; Pliakas et al., 1; Schlosburg et al., 13; Shirayama et al., 4), has led to the hypothesis that the NAcc yn-kor system underlies negative affective states and heightened stress reactivity in various psychiatric disorders. However, the mechanism by which KORs shape NAcc activity to modify affective and stress-related behaviors has yet to be identified. The NAcc is a hub of a distributed motivational circuit of the brain. NAcc medium-sized spiny neurons (MSNs) integrate excitatory inputs, including basolateral amygdala () and ventral hippocampus (VH) afferents, with local inhibition to dictate spiking activity (Britt et al., 1; O onnell and Grace, 1995). NAcc MSNs are segregated into dopamine (A) 1 receptor-containing MSNs (1 MSNs) that express yn and receptor-containing MSNs ( MSNs) (Gerfen et al., 199; Kreitzer, 9). 1 MSN activity is a critical mediator of reward learning, whereas optogenetic activation of MSNs attenuates reward and is aversive (Calipari et al., 16; Ferguson et al., 11; Kravitz et al., 1; Lobo et al., 1). As MSNs do not intrinsically generate action potentials, MSN activity is driven by excitatory synapses (O onnell and Grace, 1995). Optogenetic activation of VH and afferents to the NAcc is rewarding, and these pathways mediate natural and drug-reward-related behaviors (Britt et al., 1; Pascoli et al., 14; Stuber et al., 11). VH and pathways to the NAcc have also been implicated in mediating negative behavioral outcomes after chronic stress (Bagot et al., 15). In neocortical regions, excitation-inhibition balance is critical for fine-tuned network activity (Isaacson and Scanziani, 11). How excitatory synapses interact with inhibition to gate NAcc 1 and MSN output is not well understood. Neuron 93, , January 4, 17 Published by Elsevier Inc. 147

3 KOR activation inhibits release of both glutamate and (Hjelmstad and Fields 3). However, an understanding of how KORs shape synaptic transmission from specific glutamatergic afferents and local-circuit ergic synapses is lacking. Moreover, it is not known whether KORs differentially regulate 1 and MSNs. Elucidating the role of KORs in regulating limbic glutamatergic afferents and ergic microcircuitry within the NAcc will provide a comprehensive physiological framework wherein the NAcc KOR system can control mood and emotion. To this end, we utilized a combination of wholecell slice electrophysiology, optogenetics, anatomy, and transgenic mice to delineate the cellular basis for KOR regulation of NAcc afferents and microcircuits and ultimately NAcc 1 and MSN output. RESULTS rgic Pathway-Specific Regulation by KORs In the NAcc, KOR activation inhibits glutamatergic release (Hjelmstad and Fields, 3; Mu et al., 11). In agreement with previous reports, bath application of the KOR agonist (U69) produces a long-term depression (LT) of electrically evoked EPSCs (eepscs) in NAcc MSNs in the medial shell (Figure S1). However, Oprk (KOR) mrna is expressed in various cortical and limbic excitatory regions innervating the NAcc (Meng et al., 1993). It is unclear whether KOR modulates VH and afferents in a pathway-specific manner. To determine whether KORs negatively modulate and VH inputs, we injected AAV-CaMKII-ChR-eYFP into the or the VH of wildtype (WT) mice and conducted whole-cell slice electrophysiology recordings from MSNs in the NAcc medial shell (Figure 1A). In agreement with previous findings, VH-evoked optical EPSCs (VH oepscs) were larger than oepscs and were mediated by glutamatergic transmission (Figure S; Britt et al., 1). U69 inhibited oepscs, but not VH oepscs (Figure 1B), suggesting that KORs confer pathway-specific effects. To determine whether KOR inhibition occurred via a presynaptic site of action, we examined the effects of U69 on paired-pulse ratios (PPRs) of and VH oepscs (Figure S3A). However, it is unclear whether KOR regulation of synapses occurs presynaptically, as there was no change in the PPR of VH or oepscs. To confirm that the lack of a U69 effect on VH oepscs was not due to a general lack of presynaptic inhibition in this pathway, we bath applied the -B receptor agonist baclofen and observed a reduction of VH oepscs (Figure 1C). Baclofen inhibition of VH oepscs was associated with an increase in the PPR (Figure S3B), suggesting a presynaptic site of -B receptor action on VH synapses. Inhibition of oepscs was prevented by pretreatment with the KOR antagonist nor-bni (Figure 1); however, inhibition was not reversed by nor-bni application after U69 (Figure 1E). These results suggest that, once KORs are activated, downstream signaling maintains KOR-dependent inhibition of synapses. To determine whether differential KOR modulation of VH and afferents was due to differential KOR expression, we injected the retrograde tracer fluorogold (FG) into the NAcc and performed KOR mrna in situ hybrization in the and VH (Figures 1F and 1G). The number of cells projecting to the NAcc from the VH exceeded the number of cells from the (Figure S; Britt et al., 1). FG-positive cells in the also contained KOR mrna, whereas in the VH, FG-positive cells did not express KOR mrna (Figures 1G and 1H). Collectively, these results suggest that KORs inhibit glutamatergic synapses onto NAcc MSNs in a pathway-specific manner and that inhibition is dictated by differential KOR expression (Figure 8A). KOR Inhibition of Afferents Is Mediated by Presynaptic KORs We next determined whether KORs on terminals mediate inhibition of oepscs. We injected a cocktail of AAV-Syn-Cre- GFP and AAV-EF1a-IO-ChR-eYFP (Mcevitt et al., 14) into the of WT and mice (Van t Veer et al., 13; Figure A). Based on this strategy, ChR-expressing cells should lack KOR expression in mice, while KOR expression should remain intact in ChR-expressing cells in WT mice. We confirmed that KOR expression was reduced in ChR-expressing cells by utilizing in situ hybridization of KOR and eyfp mrna. The percentage of eyfp-positive cells that were also KOR mrna positive was significantly decreased in mice relative to WTs (Figure B). If KOR inhibition is mediated by KORs on terminals, then oepscs in NAcc MSNs from mice should be insensitive to U69. Bath application of U69 inhibited -oepscs in NAcc MSNs of WT mice, an effect that was absent in mice (Figure C). This provides unequivocal evidence that KOR inhibition of synapses onto MSNs is mediated by KOR on terminals (Figure 8A), providing functional evidence of our knockdown approach and confirming specificity of U69 at the KOR. To test the possibility that genetic ablation of KOR in neurons renders synapses insensitive to presynaptic inhibition of glutamate release, we determined whether -B-mediated inhibition was still intact in synapses in mice. Baclofen produced a similar depression of oepscs in MSNs from WT control mice and mice (Figure ), suggesting that presynaptic inhibition is still present after KOR knockdown. We next examined whether KORs on terminals regulate basal synaptic transmission in the -to-nacc pathway of mice expressing AAV-Syn-Cre-GFP and AAV- EF1a-IO-ChR-eYFP in the. We first determined the input-output relationship between optical stimulation intensity and -evoked oepsc amplitude in MSNs of and WT mice. mice displayed increased oepsc amplitudes relative to WT controls at higher stimulation intensities (Figure E), suggesting that KORs inhibit the recruitment of synapses onto MSNs. PPRs of -oepscs across multiple pulse intervals were not significantly different between WT controls and mice (Figure F). Thus, KOR inhibition of synaptic efficacy is not mediated by a change in probability of release. The increased input-output relationship in mice could be a consequence of increased postsynaptic strength. Optically evoked AMPAR:NMAR ratios, which index postsynaptic strength, were similar in WT and mice (Figure G), suggesting that the increased input-output curve in mice is mediated by enhanced recruitment of afferents. Synapse-driven homeostatic 148 Neuron 93, , January 4, 17

4 A 1 pa F 1 ms Viral Injection AAV-CaMKII-ChR-eYFP NAcc 5 μm 5 μm VH Viral Injection AAV-CaMKII-ChR-eYFP L VH V M NAcc 5 μm 5 μm oepsc Amplitude (% ) 14 L V M L L V V M M nor-bni G NAcc MSN VH NAcc MSN VH E oepsc Amplitude (% ) pa ms B C VH oepsc Amplitude (% ) oepsc Amplitude (% ) 14 8 pa 1 ms 6 pa 1 ms 6 pa 1 ms Baclofen VH VH nor-bni Figure 1. Pathway-Specific KOR Modulation of rgic Afferents to NAcc (A) AAV-CaMKII-ChR-eYFP expression in the and VH and expression of ChR-eYFP-containing terminals in the NAcc. (B) Time course of the effect of KOR activation with U69 (1 mm) on the amplitude (expressed as a percentage of baseline) of optically evoked EPSCs (oepscs) in medial-shell NAcc MSNs from animals expressing ChR-eYFP in the (n = 8) or VH (n = 7, two-way ANOVA afferent 3 time interaction, F (19, 47) = 3.34, p <.1). Representative oepscs recorded in NAcc MSNs from (orange traces) and VH (green traces) during baseline (dark traces) and after the KOR agonist U69 (light traces) are shown. (C) Time course and representative traces of baclofenmediated inhibition of VH EPSCs (n = 4). () Time course and representative traces of nor-bni pretreatment of U69 on oepscs (n = 8). (E) Time course and representative traces of nor-bni treatment after U69 on oepscs (n = 5). (F) Low magnification of the fluorogold (FG) injection site in the NAcc. aca denotes anterior commissure; gcc denotes genu of the corpus callosum. (G) Top shows low magnification of FG-immunoreactive (IR) cells in the and VH. Squares delimit areas shown for high magnification images in middle and bottom. Middle shows FG-IR neurons in the and VH (yellow and red outline) and FG-labeled cells expressing KOR mrna (red outline). Bottom shows dark field illumination of KOR mrna in and VH. opt denotes optic tract; ec denotes external capsule. (H) Percentage of cells expressing KOR mrna and FG in the and VH (t (4) = 5.854, p =.4, n = 3). ata are expressed as mean ± SEM. FG-IR ec opt gcc aca FG - IR & KOR mrna μm μm FG - IR & KOR mrna 5 μm H % KOR mrna / FG-IR co-labeled ** VH KOR mrna 5 μm 5 μm KOR mrna 5 μm 5 μm Neuron 93, , January 4,

5 A C oepsc Amplitude (% ) E WT Controls or mice Intra- AAV-Syn-Cre-eGFP AAV-EF1-IO-ChR-eYFP WT pa NAcc MSN 1 ms eyfp KOR loxp WT WT Controls 1 pa oepsc Amplitude (pa) 1 ms 3 1 WT KOR mrna WT Controls WT Controls eyfp 1 ms Optical Stimulation Intensity (mw) pa oepsc Amplitude (% ) pa WT 1 ms KOR loxp WT Baclofen B pa Paired Pulse Ratio (P/P1) ms WT Controls WT Controls F KOR mrna μm μm 5 pa ms pa ms KOR mrna-positive / eyfp-positive (%) * 5 1 Interstimulus Interval (ms) Figure. KORs on Terminals Mediate KOR Inhibition of Synapses and Modulate Basal Synaptic Efficacy (A) Representative image of ChR-YFP and eyfp and KOR mrna (puncta) in the of a WT control and a mouse after injection of AAV-EF1a-IO- ChR-eYFP and AAV-Syn-Cre-eGFP into. (B) The percentage of eyfp-positive cells with KOR mrna was decreased in mice (n = 3) relative to WT controls (n = ) (t (3) = 4.881; p =.16), as was the proportion of cells expressing KOR mrna in eyfp cells in WT (49/15, 38.5%) and (5/173, 14.9%) mice (Fisher s exact test, p =.164). (C) Representative traces and time course of U69 (1 mm) effects on oepscs in WT controls (n = 7) and in mice (n = 6, ANOVA genotype 3 time interaction, F (19, 19) = 4.76, p <.1). () Representative traces and time course showing B receptor inhibition of oepscs with baclofen (5 mm) in WT (n = 5) and mice (n = 3, ANOVA genotype 3 time interaction, F (19, 114) =.7, p =.81). (E) Representative traces and mean oepsc amplitude in response to graded optical stimulation intensities in (n = 5) and WT mice (n = 6, ANOVA genotype 3 time interaction, F (6, 54) = 3.6, p =.4). (F) Mean PPR and representative traces of oepscs in MSNs of (n = 7) and WT mice (n = 9, ANOVA genotype 3 time interaction, F (4, 56) =.1, p =.98). (G) KOR genetic ablation does not alter to NAcc MSN synaptic strength (t (7) =.3, p =.76). Representative traces of AMPAR and NMAR EPSCs at +4 mv from WT control (n = 4) and mice (n = 5). (H) Intrinsic excitability of NAcc MSNs from WT (n = 1) and mice (n = 7) does not differ (ANOVA genotype 3 time interaction, F (1, 4) = 1.13, p =.34). ata are expressed as mean ± SEM. G WT 5 pa 5 ms AMPAR/NMAR H WT 5 mv 5 ms 5 pa Firing Rate (Hz) WT Controls 1 3 Somatic Current Input (pa) 15 Neuron 93, , January 4, 17

6 plasticity of intrinsic excitability, wherein long-term changes in synaptic activity modify intrinsic excitability, has been reported in the NAcc (Ishikawa et al., 9). We failed to detect a significant change in intrinsic excitability in MSNs from WT controls and mice, which suggested that lack of KOR inhibition does not change NAcc MSN intrinsic excitability (Figure H). Collectively, these results suggest that KORs negatively modulate synaptic efficacy of synapses in the NAcc via a presynaptic site of action (Figure 8A). KOR Regulation of MSN-MSN Collateral ergic Transmission KORs inhibit ergic synapses onto MSNs via a presynaptic site of action (Hjelmstad and Fields, 3). One fundamental question is whether KORs inhibit lateral inhibition from 1 and MSNs to other MSNs. In the dorsal and ventral striatum, yn mrna expression is restricted to 1 mrna-expressing MSNs (Gerfen et al., 199), while adenosinea (AA) receptor expression is restricted to MSNs (Kreitzer 9). We utilized Prodynorphin-IRES-Cre (Pyn-iCre) and AA-Cre mice to examine the role of KORs in regulating local 1 and MSN output, respectively, within the NAcc. We first validated that Pyn-iCre mice (Al-Hasani et al., 15; Krashes et al., 14) provide selective genetic access to 1 MSNs in the NAcc by injecting AAV-IO-eYFP into the NAcc and immunostaining for substance P, a peptide expressed in 1-MSNs (Figure 3A). In Pyn-iCre mice injected with intra-nacc AAV-IO-eYFP, 86.3% of eyfp-positive cells were also substance-p positive (Figure 3B; 776 eyfp and substance P double positive out of 899 total eyfp-positive cells). Moreover, with intra-nacc AAV- IO-ChR-eYFP injection, ChR-positive cells in AA-Cre mice, which express Cre in MSNs, had enhanced intrinsic excitability relative to ChR-positive cells from Pyn-iCre mice (Figure 3B), consistent with reports demonstrating decreased excitability in 1 MSNs relative to MSNs (Grueter et al., 1; Kreitzer and Malenka, 7). Thus, the Pyn-iCre mouse is a useful tool to gain genetic access to 1 MSNs in the NAcc (Al-Hasani et al., 15). We determined whether 1-MSN collaterals were inhibited by KORs by injecting AAV-IO-ChR-eYFP into the NAcc of PyniCre mice and recording in regions with dense ChR-eYFP expression. We recorded from ChR-negative neurons with optically evoked yn-msn -A oipscs (1-MSN oipscs) (Figure 3C). Under these conditions, recorded neurons are likely MSNs or the remaining fraction of uninfected 1 MSNs. As collaterals from MSNs to other MSNs elicit small inhibitory currents (Tepper et al., 4; Wilson, 7), we utilized a KCl-based internal solution to detect -A currents as large inward currents at a 7 mv holding potential. Consistent with a low probability of release and connection between MSNs (Planert et al., 1; Tunstall et al., ), yn-msn optical stimulation elicits -A currents with occasional failures (Figure 3C). U69 inhibited optically evoked 1-MSN oipscs (Figure 3) and increased the number of 1-MSN oipsc failures (Figure 3E), which is indicative of a reduction in presynaptic probability of release and/or number of release sites. 1-MSN oipsc inhibition and increase in failure by U69 were blocked by nor-bni pretreatment. Interestingly, basal 1-MSN oipsc failures in nor-bnitreated cells were significantly fewer than in untreated cells (Figure 3E), suggesting that KORs modulate basal 1 MSN lateral inhibition. To determine whether KORs modulate MSN collaterals, we injected AAV-IO-ChR-eYFP into the NAcc of AA-Cre mice and recorded from ChR-negative neurons while evoking -MSN oipscs (Figure 3F). Under these conditions, recordings are biased toward 1 MSNs and noninfected MSNs. MSN oipscs were larger in amplitude (Figure S4A) and exhibited fewer failures than 1 MSN collaterals (Figure S4B), consistent with a higher probability of synaptic connections between MSNs and neighboring 1 MSNs (Planert et al., 1). U69 inhibited AA-MSN oipscs; however, this effect was not as pronounced as that seen in 1-MSN oipscs (Figure 3F). Thus, KORs have a significantly stronger control over 1 MSN lateral inhibition than over MSN lateral inhibition. A possibility for greater KOR inhibition of 1 MSN collaterals than MSN collaterals is differential expression of KORs between 1 and MSNs. To this end, using triple-labeling in situ hybridization to detect 1,, and KOR mrna, we determined whether KOR is differentially expressed in 1 and MSNs (Figure 3G). The proportion of 1 MSNs expressing KOR mrna was significantly higher than that of MSNs (Figure 3H). Furthermore, the relative expression level of KOR mrna within KOR-positive MSNs cells was 85% of the level of KOR-positive 1 MSNs (Figure 3I). NAcc ergic interneurons are also a source of for MSNs, which may be regulated by KORs. To determine whether KORs were expressed in ergic interneurons, we utilized in situ hybridization to determine if somatostatin-mrna-positive and parvalbuminmrna-positive interneurons express KOR mrna (Figure S5). KOR mrna expression in somatostatin- and parvalbumin-positive interneurons was sparse and weak relative to MSNs (Figures S5A and S5B), suggesting that KORs are strategically poised to regulate MSN collaterals. Thus, KORs modulate MSN collaterals, and modulation of 1 MSN output to other MSNs is stronger than MSN output, presumably due to enhanced KOR mrna expression in 1 MSNs (Figure 8B). ifferential KOR Modulation of rgic Synapses onto 1 and MSNs It is unclear if KORs differentially inhibit glutamatergic synapses onto 1 and MSNs. As our previous experiments were agnostic to MSN identity, we utilized 1-tdTomato mice expressing tdtomato under control of the 1 receptor promoter to determine whether presynaptic KOR regulation differed between 1 and MSNs. 1 MSNs were identified by tdtomato fluorescence, while MSNs were identified by lack of tdtomato fluorescence and electrophysiological properties indicative of MSNs. 1 and MSNs can be distinguished with this strategy, as MSNs (tdtomato negative) had increased excitability relative to 1 MSNs (tdtomato positive; Figure S6), consistent with other reports (Grueter et al., 1; Kreitzer and Malenka, 7). We determined whether KOR modulation of eepscs would differ between 1 and MSNs (Figure 4A). U69 inhibited eepscs in 1 MSNs, while inhibition was not consistently observed in -MSNs (Figure 4B). We determined whether presynaptic KOR modulation Neuron 93, , January 4,

7 A AAV-EF1α-IO-eYFP eyfp Substance P Substance P eyfp B Pdyn-iCre ChR+ AA-Cre ChR+ Prodynorphin-IRES-Cre 5 μm 5 μm 5 μm Firing Rate(Hz) Somatic Current Input (pa) C F ChR-Negative NAcc MSN 1 MSN Post- G 1 ms 1 pa 1 mrna 1-MSN oipsc Amplitude (% ) acsf nor-bni %Failures Pretreatment acsf nor-bni-pretreated 6 4 KOR mrna mrna KOR mrna KORmRNA E 1 MSN MSN μm μm μm oipsc Amplitude (% ) ChR-Negative NAcc MSN ChR-Negative NAcc MSN yn-icre AA-Cre API 1 mrna mrna KOR mrna μm H 1 KOR-mRNA-Positive (%) MSNs MSNs *** I KOR-mRNA Expression Levels (Normalized to 1 MSNs) * Figure 3. KOR Modulation of ergic Transmission between MSNs (A) Experimental schematic depicting AAV-EF1a-IO-eYFP injection into the NAcc of Pyn-IRES-Cre (Pyn-iCre) mice. Colocalization of eyfp-positive and substance-p-positive NAcc MSNs in Pyn-iCre mice (n = 4). (B) Mean firing rate evoked by depolarizing current pulses in ChR-positive MSNs from Pyn-iCre (n = 7) and AA-Cre (n = 3) mice injected with AAV-EF1a-IO- ChR-eYFP into the NAcc. (legend continued on next page) 15 Neuron 93, , January 4, 17

8 of glutamatergic synapses differed between 1 and MSNs by examining the effects of U69 on miniature EPSCs (mepscs; Figure 4C). U69 decreased mepsc frequency, but not amplitude, in 1 MSNs, an effect not observed in vehicle-treated MSNs or U69-treated MSNs (Figure 4). Thus, KORs preferentially inhibit glutamate release onto 1 MSNs. We subsequently determined whether KOR modulation of afferents is cell type specific by injecting 1- tdtomato mice with intra- AAV-CaMKII-ChR-eYFP and recording oepscs in 1 and MSNs (Figure 4E). KOR activation inhibited oepscs in 1 MSNs, but not in MSNs (Figure 4F). As 1- tdtomato mice have been reported to display abnormal electrophysiological properties (Bagetta et al., 11), we injected AAV- CaMKII-ChR-eYFP into the of AA-Cre mice crossed with tdtomato reporter mice. Here, MSNs were identified by the presence of tdtomato, and 1 MSNs were lacking tdtomato. KOR activation inhibited oepscs in 1 MSNs, but not in MSNs, in AA-Cre-tdTomato mice (Figure 4G). Thus, differential regulation of excitatory afferents onto 1 and MSNs is not due to altered synaptic plasticity due to transgene expression in 1-tdTomato and AA-Cre-tdTomato mice. Collectively, these results suggest that the KORs are poised to regulate excitatory drive of 1-MSNs more reliably than excitatory synapses in -MSNs (Figure 8A). NAcc 1 MSNs project to the ventral tegmental area (VTA) and substantia nigra (SNc) in the midbrain and the lateral hypothalamus (LH) (Bocklisch et al., 13; O Connor et al., 15), but it is unclear whether presynaptic KOR regulation of glutamatergic synapses differs between 1 MSN efferents. We determined whether presynaptic KOR regulation of glutamate release onto 1 MSNs projecting to the VTA/SNc (herein referred to as midbrain) or the LH would differ. By injecting red retrobeads into the midbrain and green retrobeads into the LH (Figure 5A), we first identified whether midbrain- and LH-projecting NAcc MSNs were sending projections to only one target and/or collateralizing to both LH and midbrain. Midbrain- and LH-projecting NAcc MSNs were spatially segregated and largely nonoverlapping, with the majority of midbrain-projecting MSNs localized to lateral NAcc shell and core and with LH-projecting MSNs localized to medial and lateral shell and core (Figure 5A). To conservatively ascertain levels of overlap, we analyzed regions where midbrain- and LH-projecting cells were both present. A total of.4% (/91) of red-retrobead-positive cells (midbrain-projecting) also contained green retrobeads (LHprojecting). Similarly, 4.3% (/83) of green-retrobeadcontaining MSNs (LH-projecting) also contained red retrobeads (midbrain-projecting). Thus, midbrain- and LH-projecting MSNs are largely nonoverlapping populations, with subsets of cells projecting to both (Figure 5A). To confirm that midbrain and LH-projecting MSNs were primarily 1 MSNs, we injected green retrobeads into the LH or midbrain of 1-tdTomato mice; we observed that 95.6% (439/458) of midbrain-projecting and 94.1% (386/41) of LH-projecting MSNs were tdtomatopositive 1 MSNs (Figure 5B). Thus, NAcc outputs to the midbrain and LH primarily consist of largely nonoverlapping 1 MSNs. To examine whether KOR regulation of presynaptic glutamate release differed between LH- and midbrain-projecting MSNs, WT mice were injected with retrobeads into the midbrain or LH. Then, 6 to 7 days after injection, retrobead-positive MSNs in the NAcc core and shell were patched, and the effect of U69 application on mepsc frequency was determined (Figure 5C). Basal mepsc frequency and amplitude did not differ between LH- and midbrain-projecting cells (Figure 5). As in 1-tdTomato mice, U69 decreased mepsc frequency (Figure 5E), but not amplitude (Figure 5F), demonstrating that KORs presynaptically inhibit glutamate release onto 1 MSNs utilizing a nongenetic approach to identify 1 MSNs. Specifically, U69 inhibited mepsc frequency in a similar manner in LH- and midbrain-projecting cells independent of whether they were localized in the core or shell. However, U69 decreased mepsc frequency more robustly in shell 1 MSNs than in core 1 MSNs, regardless of whether they projected to the LH or the midbrain. Thus, KORs inhibit presynaptic glutamate release onto 1 MSNs projecting to both the LH and midbrain, and this effect is significantly more robust in the shell than in the core. We further studied subregional differences in KOR regulation of glutamate release onto 1 MSNs by determining the effects of U69 on mepscs in a large population of 1 MSNs across the NAcc in 1-tdTomato mice (Figure 5G). Basal mepsc frequency was significantly smaller in NAcc core versus shell subregions (Figure 5H), and there were no significant differences in mepsc amplitude (Figure 5H). KOR inhibition of mepsc frequency in 1 MSNs was observed throughout the (C) ChR-negative MSNs were recorded, and -A IPSCs from 1-MSNs (1-MSN oipscs) were evoked, in Pyn-iCre mice expressing ChR-eYFP in NAcc. Representative traces of 1-MSN oipscs in ChR-negative MSNs at baseline (top) and after U69 (bottom) recorded with a KCl-based internal solution. () U69 (1 mm) inhibition of 1 MSN oipscs recorded in ChR-negative MSNs from Pyn-iCre mice expressing AAV-EF1a-IO-ChR-eYFP in the NAcc (filled green circles; n = 6). Pretreatment with nor-bni (1 nm) blocked U69 inhibition of 1 MSN oipscs (open green circles; n = 5, two-way ANOVA treatment 3 time interaction, F (19, 171) = 5.84, p <.1). (E) Percentage of 1-MSN oipsc failures at baseline and after U69 (shaded region) in cells recorded in regular acsf (filled green circles) or in acsf-containing nor-bni (two-way ANOVA treatment 3 time interaction, F (1, 9) = 15.5, p =.36). (F) Time course of U69 (1 mm) inhibition of 1 MSN (filled green circles; n = 6) and MSN (open green circles; n = 5) oipscs recorded in ChR-negative MSNs from Pyn-iCre and AA-Cre mice, respectively (two-way ANOVA genotype 3 time interaction, F (19,9) = 1.7, p =.34). (G) Representative image of RNAscope in situ hybridization of API (blue), 1 mrna (green), mrna (red), and KOR mrna (white) expression. Top left shows 1 and KOR mrna expression; top middle shows and KOR mrna expression; top right shows KOR mrna expression; bottom left shows combined 1,, and KOR mrna expression. (H) Percentage of 1- and -mrna-positive MSNs that coexpress KOR mrna (n = 4, t (7) = 1.13, p <.1). (I) Mean relative expression levels of KOR mrna (mean integrated density/area) relative to 1 MSNs in KOR-mRNA-positive 1 and MSNs (t (7) = 5.84, p =.11). ata are expressed as mean ± SEM. Neuron 93, , January 4,

9 A 1 MSN MSN B C 1 pa 1 ms eepsc Amplitude (% ) MSN MSN MSN E pa 1 sec 1 MSN 8 pa 1 ms MSN MSN pa 1 ms F oepsc Amplitude (% ) mepsc Frequency (Hz) 6 4 ** mepsc Amplitude (pa) Vehicle MSNs 1 MSNs MSNs MSNs MSNs G oepsc Amplitude after (Mean % ) 1 MSNs MSNs 15 ** 1 5 AA-Cre-tdtomato 1-tdtomato Figure 4. KORs ifferentially Regulate rgic Synaptic Transmission in 1 and MSNs (A) 1 MSNs (tdtomato positive) and MSNs (tdtomato negative) were recorded from 1-tdTomato mice, and eepscs were evoked by electrical stimulation. Representative traces of eepscs in 1 and MSNs are shown at baseline and after U69. (B) U69 (1 mm) reliably inhibited eepscs in 1 MSNs (blue circles; n = 6), but not in MSNs (red circles; n = 1, two-way ANOVA cell type 3 time interaction, F (19,66) =.79, p =.1). (C) Representative traces of mepscs recorded from 1 (blue traces) and MSNs (red traces) before (top) and after application of U69 (bottom). () mepsc frequency (Hz) and amplitude (pa) during baseline (solid bars) and after U69 (patterned bars) in vehicle-treated MSNs (gray bars; n = 8) and U69-treated 1 MSNs (blue bars; n = 9) and MSNs (red bars; n = 8). U69 decreased mepsc frequency in 1 MSNs, but it had no significant effect in MSNs (two-way ANOVA cell type 3 time interaction, F (,) = 1.46, p =.6). U69 had no effect on mepsc amplitude (two-way ANOVA cell type 3 time interaction, F (,) = 1.95, p =.17). (E) 1 and MSNs were recorded in 1-tdTomato and AA-Cre/floxed-tdTomato mice, and U69 effects on oepscs were determined. (F) KOR inhibition of oepscs in 1 MSNs (n = 13), but not in MSNs (n = 1; two-way ANOVA cell type 3 time interaction, F (19,437) =.43, p =.7). (legend continued on next page) 154 Neuron 93, , January 4, 17

10 NAcc; however, this effect was significantly stronger in the mediodorsal NAcc shell than in the ventral shell and core (Figures 5G and 5I). Together, these results reveal that presynaptic KOR modulation of glutamate release onto 1 MSNs does not differ between LH- and midbrain-projecting 1 MSNs and that, though it is more robust in the dorsomedial NAcc shell, it is present throughout the NAcc. KORs ifferentially Regulate Excitation-Inhibition Balance in 1 and MSNs KORs inhibit ergic output from 1 and MSN collaterals; however, it is currently not known whether KORs differentially inhibit ergic inputs onto 1 and MSNs. We patched 1 and MSNs from 1-tdTomato mice and examined the effects of U69 on mipscs to determine KOR regulation of synapses (Figure 6A). mipsc frequency and amplitude did not differ between 1 and MSNs (Figure S7A). U69 significantly decreased mipsc frequency in MSNs without affecting amplitude, an effect not observed in vehicle-treated MSNs and U69-treated 1 MSNs (Figures 6B and 6C). U69 also decreased electrically evoked IPSC amplitude in both 1 and MSNs; however, this effect was larger in MSNs than it was in 1 MSNs (Figure S7B). Collectively, these results suggest that KORs preferentially inhibit ergic synapses onto MSNs relative to 1 MSNs via a presynaptic site of action. If KORs are exerting more control of glutamatergic synapses onto 1 MSNs than MSNs and more net inhibitory control over ergic synapses in MSNs than 1 MSNs, then KORs may serve as a gate to dynamically switch excitation-inhibition balance in 1 and MSNs. To determine whether KORs were differentially regulating excitation-inhibition balance in 1 and MSNs, we recorded biophysically isolated basal AMPAR eepscs at 7 or 55 mv and basal -A eipscs at mv (Figures 6 and 6E). We then applied U69 and subsequently recorded AMPAR eepscs at 7 or 55 mv and baseline -A eipscs at mv. eepscs and eipscs are blocked by NQX and picrotoxin (PTX), respectively (Figure 6F). U69 decreased the AMPAR eepsc amplitude relative to baseline in 1 MSNs, but not in MSNs (Figures 6E and 6G). Moreover, a significant decrease in -eipscs was only observed in MSNs (Figures 6E and 6H). Thus, U69 decreased the excitation-inhibition ratio in 1 MSNs and increased it in MSNs (Figure 6I). As the 1 and MSN excitation-inihibition ratio is similar in 1 and MSNs (Figure S7C), this would imply that KOR activation may shift the balance of excitation and inhibition in 1 and MSNs by decreasing the excitatory drive of 1 MSNs and favoring the disinhibition of MSNs. NAcc KOR Regulation of Pathway-Specific Excitation-Inhibition Balance ecreases Excitatory rive of 1 MSNs and isinhibits MSN Firing To directly delineate whether presynaptic KOR inhibition of glutamate and onto 1 and MSNs, respectively, would differentially impact synaptically driven spiking in 1 and MSNs, we examined the effects of U69 on electrical-stimulation-evoked spiking (Figure 7A). A ten-pulse, -Hz electrical train was utilized to evoke spikes in 1 and MSNs. Evoked spikes were mediated by glutamatergic transmission and inhibited by -A receptors, as PTX bath application enhanced MSN spiking while NQX/AP-5 blocked spiking (Figure S8A). In 1 MSNs, U69 decreased the number of evoked spikes per train relative to baseline (Figures 7B and 7C). In MSNs, an increase in synaptically driven spiking was observed after U69 application (Figures 7B and 7C). To determine whether KOR activation altered the pattern of evoked firing, we analyzed the interspike interval (ISI) and the latency to first spike. There was no difference in the ISI after vehicle or U69 in 1 or MSNs (Figures S8B and S8C). In MSNs, U69 decreased the latency to first spike, an effect absent in 1 MSNs and controls. Thus, KORs decrease excitatory drive of 1 MSNs without changing the pattern of activity, while KOR-mediated disinhibition in MSNs allows for spiking to occur earlier. We next determined whether KOR activation would inhibit spiking evoked by optogenetic activation of afferents in 1 and MSNs (Figure 7). Similar to that observed with electrical stimulation, -evoked spiking was enhanced by PTX, suggesting that feedforward inhibition was recruited to inhibit -evoked spiking; -evoked spiking was completely abolished by NQX/AP-5 (Figure S8). Interestingly, PTX increased MSN firing rate more robustly than it did with 1 MSNs, suggesting that -evoked spiking in MSNs is under stronger ergic control. U69 decreased -evoked spiking in 1 MSNs, while it increased spiking in MSNs (Figures 7E and 7F). In MSNs, U69 decreased both the latency to first spike and the ISI; this was not observed in 1 MSNs or vehicle MSNs (Figures S8E and S8F). Thus, KORs decrease excitatory drive of 1 MSNs and disinhibit MSNs by allowing them to spike earlier and at an increased firing rate (Figure 8C). As VH afferents are not directly modulated by KOR, KOR modulation of excitation-inhibition balance is predicted to result in a different pattern of 1 and MSN output than that observed for electrically evoked and -evoked spiking. To this end, we determined the effects of U69 on optogenetic VH-evoked spiking in 1 and MSNs (Figure 7G). PTX significantly increased VH-evoked spiking in a similar manner in both 1 and MSNs (Figure S8G). U69, however, did not modify VHevoked firing in 1 MSNs (Figures 7H and 7I), consistent with a lack of effect of U69 on VH-evoked oepscs. This result also suggests that KOR modulation of electrically evoked and evoked 1 MSN spiking is not due to a change in intrinsic excitability. KOR activation with U69 enhanced VH-evoked spiking in MSNs. Moreover, firing pattern, as assessed by the ISI and latency to first spike, was unchanged in vehicle-treated MSNs and U69-treated 1 MSNs (Figures S8H and S8I). In MSNs, U69 decreased the ISI, suggesting that KORs allow VH afferents to drive spiking at a higher rate (Figure S8H). Thus, the KOR (G) Mean percent of baseline oepsc in pooled 1 and MSNs from both genotypes (t (4) = 3., p =.37), which did not differ between 1 and MSNs identified in 1-tdTomato and AA-Cre/floxed-tdTomato mice (two-way ANOVA main effect of cell type 1 versus cell type, F (1, ) = 9.7, p =.51, no cell type 3 genotype interaction or main effect of genotype, p >.5). ata are expressed as mean ± SEM. Neuron 93, , January 4,

11 A Green Retrobeads B Red Retrobeads NAcc Midbrain (VTA/SNc) Red Retrobeads Injection L 1-tdtomato VTA/SNc LH Hypothalamus Green Retrobeads Injection Midbrain Projecting 1-tdtomato L M M V V 5 μm SNc VTA SNr LH 5 μm 1-tdtomato Hypothalamus-Projecting 1-tdtomato 5 μm 5 μm μm M Core L V Core Shell M μm C 5 μm 1 μm V E 1 sec mepsc Frequency (Hz) 8 mepsc Frequency (% ) pa 5 μm 5 μm L F **** ** 5 5 Core Shell Core Shell mepsc Amplitude (% ) Hypothalamus Projecting Midbrain Projecting L mepsc Amplitude (pa) Shell Core Shell Core Shell Midbrain-Projecting Hypothalamus-Projecting Hypothalamus % mepsc Amplitude (pa) 1% mepsc Frequency (Hz) mepsc Frequency (% ) 5 1 Midbrain I Core Ventral Shell orsal Shell mepsc Amplitude (% ) H 13% Hypothalamus mepsc Frequency (% ) G Midbrain Figure 5. Presynaptic KOR Regulation of rgic Synapses epends on NAcc Subregion, but Not Efferent (A) At top, WT mice were injected with unilateral fluorescent Green XI and Red XI retrobeads into the LH and ipsilateral midbrain, respectively. Bottom shows red and green retrobead labeling in the NAcc at low (left) and high (right) magnification. (B) Representative image of green fluorescent retrobead labeling in the NAcc of 1-tdTomato mice (n = 3) injected with green retrobeads in the LH or midbrain, demonstrating that 95.9% (439/458) of midbrain-projecting and 94.15% (386/41) of LH-projecting MSNs are 1 MSNs. (legend continued on next page) 156 Neuron 93, , January 4, 17

12 system acts as a pathway-specific filter to gate, but not VH, control of 1 MSNs while facilitating integration of the VH and excitatory drive of MSNs (Figure 8C). ISCUSSION Here we describe a novel synaptic framework wherein pathwayand cell-type-specific KOR modulation of excitatory and inhibitory balance shape information flow into and out of the NAcc (Figure 8C). Activation of NAcc KORs inhibits glutamatergic synaptic transmission from the, but not VH, via a presynaptic site of action. Accordingly, KOR mrna is expressed in NAcc-projecting neurons, but not those originating in the VH. Within local microcircuits, KORs inhibit ergic collaterals from 1 MSNs more robustly than MSN collaterals via enhanced KOR expression in 1 MSNs. rgic synapses in 1 MSNs are more sensitive than those on MSNs to KOR inhibition. This effect is independent on the output (hypothalamus versus midbrain) of 1 MSNs but is significantly more robust in NAcc shell than in core 1 MSNs. Conversely, KOR modulation of ergic synapses was significantly stronger onto MSNs than 1 MSNs, as KORs decreased excitation/inhibition balance in 1 MSNs while increasing it in MSNs. The net result of dichotomous KOR modulation of excitation and inhibition in 1 and MSNs is decreased, but not VH, synaptic drive of 1 MSNs and disinhibition of the and VH excitatory drives of MSNs. Thus, KORs play a role in NAcc information processing by inhibiting limbic excitatory drive of 1 MSNs in a pathwayspecific manner and by amplifying MSN output via disinhibition in a pathway-independent manner (Figure 8C). Here we demonstrate that the is a KOR-sensitive glutamatergic afferent. Moreover, we did not detect an effect of U69 on VH glutamatergic synapses. This is in agreement with our anatomical findings that KOR mrna is expressed in NAccprojecting neurons, but not in NAcc-projecting VH neurons. As terminal expression is patchy in the dorsomedial NAcc shell and KOR inhibition of glutamate release is strongest in this region, it is possible that there is at least one more KOR-sensitive input to the shell. afferents in the prefrontal cortex and bed nucleus of the stria terminalis are inhibited by KORs (Crowley et al., 16; Tejeda et al., 15), suggesting that presynaptic modulation of output is a shared principle among different efferents. Interestingly, baclofen inhibited both VH- and -evoked oepscs, suggesting that NAcc -B receptors may not confer the same pathway specificity as KORs. Thus, the NAcc may have signals that broadly decrease excitatory drive, presynaptically, across various pathways (i.e., -B) and modulatory signals that confer pathway-specific modulation (i.e., KORs). Indeed, mu-opioid receptors have been shown to produce pathway-specific effects in the dorsal striatum (Atwood et al., 14). This is consistent with our hypothesis that endogenous opioid receptor systems contribute to information processing by virtue of pathway-specific regulation of synaptic transmission. KOR inhibition of -evoked oepscs was observed in 1 MSNs; however, KOR activation failed to modify -evoked oepscs in MSNs. A possibility is that distinct neurons project to 1 and MSNs, and KORs are preferentially expressed in the former, not the latter. Another intriguing scenario is that KOR-expressing neurons target both 1 and MSNs, but functional KOR is preferentially targeted to synapses on 1 MSNs rather than MSNs. Thus, KORs may act as cell-type-specific filters on afferents by selectively depressing synapses onto 1 MSNs but largely leaving glutamatergic synaptic transmission onto MSNs intact (Figures 8A and 8C). 1 MSNs and MSNs are synaptically connected to both 1 and MSNs (Planert et al., 1). KORs inhibit 1 MSN collaterals more strongly than they inhibit MSN collaterals. In our experiments examining KOR regulation of 1 and MSN collaterals onto ChR-negative cells, it is likely that sampling was biased toward and 1 MSNs, respectively. It is tempting to speculate that 1 MSN collaterals to MSNs are more strongly inhibited than MSNs to 1 MSNs. This is consistent with more reliable effects of U69 on ergic transmission onto than 1 MSNs. KOR regulation of 1 MSN output may be restricted to NAcc collaterals, but not projections, as U69 does not modify synapses from NAcc to VTA A neurons (Matsui et al., 14). Connections between MSNs elicit weak IPSPs and IPSCs recorded at the soma, largely due to distribution of MSN collaterals on distal dendrites of MSNs (Koos et al., 4; Tunstall et al., ). In pyramidal neurons, synapses at distal dendrites inhibit synaptic integration (Lovett-Barron et al., 1; Palmer et al., 1). Likewise, MSN collaterals are hypothesized to play a role in sculpting synaptic integration by shunting incoming excitatory signals, while interneuronal inhibition regulates spike output (Tepper et al., 4). KOR mrna was sparsely and weakly expressed in somatostatin- and parvalbumin-expressing NAcc interneurons relative to 1 and MSNs, suggesting that KORs are preferentially expressed in NAcc MSNs and are poised to fine-tune dendritic inhibition by MSNs. By depressing MSN collateral dendritic inhibition, KORs may increase the window of synaptic integration of excitatory (C) Representative traces of mepscs during baseline (black) and after U69 (gray) from a midbrain-projecting MSN in the NAcc shell. () Basal mepsc frequency and amplitude in LH-projecting (n = 14) and midbrain-projecting (n = 1) 1 MSNs (frequency, t =.89 (4), p =.38; amplitude, t =.67 (4), p =.5). (E and F) Mean normalized mepsc (E) frequency (expressed as a percentage of baseline) and (F) amplitude after U69 in 1 MSNs in the core and shell projecting to the LH and midbrain (mepsc frequency, two-way ANOVA, subregion main effect, F (1,1) = 9.18, p =.6, **p <.1 pooled shell versus pooled core). (G) Map of presynaptic KOR inhibition of mepsc frequency (expressed as a percentage of baseline) in 1 MSNs from 1-tdTomato mice (n = 43). (H) Basal mepsc frequency and amplitude in 1 MSNs recorded in the NAcc core (n = 15), ventral shell (n = 16), and dorsomedial shell (n = 17, ANOVA, F (,45) = 3.14, p =.5, *p <.5 pooled shell versus core). (I) mepsc frequency and amplitude (expressed as a percentage of baseline) in 1 MSNs recorded in the NAcc core (n = 13), ventral shell (n = 16), and dorsomedial shell after U69 (n = 17, ANOVA, F (,43) = 3.749, p =.4, *p <.5). ata are expressed as mean ± SEM. Neuron 93, , January 4,

13 A 1 MSN E F eepsc eipsc eepsc eipsc 4 1 MSN pa MSN mipscs 1 sec PSC Amplitude (pa) PSC Amplitude (pa) MSN mipscs Time (Min) eepsc eipsc B C mipsc Frequency (% ) mipsc Amplitude (% ) 15 1 eepsc eipsc Time (Min) *** Vehicle MSNs 1 MSNs MSNs pa 1ms ** NQX Picrotoxin NQX Picrotoxin Figure 6. KOR Activation Alters Excitation-Inhibition Balance in a Cell-Type-Specific Manner (A) 1 MSNs or MSNs were patched from 1- tdtomato mice, and mipscs were recorded at mv. Representative traces are of mipscs during baseline and after U69 (1 mm) in 1 MSNs (blue traces) and in MSNs (red traces). (B) Mean mipsc frequency (expressed as a percentage of basal frequency) after vehicle (gray; n = 9) or U69 in 1 MSNs (blue; n = 7) and in MSNs (red; n = 8, one-way ANOVA, F (,3) = 15.3, p <.1, ***, **p <.5). (C) Mean mipsc amplitude after vehicle or U69 in 1 MSNs and in MSNs (one-way ANOVA, F (,3) = 1.17, p =.33). () 1 MSNs and MSNs were patched, and eepscs were recorded at the reversal potential of IPSCs, while eipscs were recorded at the reversal potential of glutamatergic EPSCs. (E) Representative postsynaptic currents (PSCs) of 1 (top) and (bottom) MSNs during baseline at 55 mv (eepscs) and mv (eipscs) and during U69 application (gray shaded region). (F) Traces from representative 1 and MSNs. eepscs recorded at 55 mv were abolished by application of NQX (1 mm), while eipscs were completely abolished by application of picrotoxin (1 mm). (G) Relative eepsc amplitudes (expressed as a percentage of baseline) after vehicle (n = 9) or U69 (1 mm) in 1 MSNs (n = 8) and MSNs (n = 8, one-way ANOVA, F (,4) = 7.39, p =.35, **, *p <.5). (H) Relative eipsc amplitudes (expressed as a percentage of baseline) after vehicle or U69 (one-way ANOVA, F (,4) = 5.9, p =.13, **p <.5). (I) Change in excitation-inhibition ratio after vehicle or U69 (eepsc/eipsc U69 eepsc/eipsc ) in 1 MSNs and MSNs (one-way ANOVA, F (,4) = 5.76, p =.97, **p <.5). ata are expressed as mean ± SEM. G H I eepsc Amplitude (% ) 1.5 ** * Vehicle MSNs 1 MSNs eipsc Amplitude (% ) * Change in Excitation/Inhibition Ratio ( EPSC/IPSC - EPSC/IPSC) ** MSNs 158 Neuron 93, , January 4, 17

14 A 1 MSN B C Vehicle MSNs 1 MSNs MSN 5 mv 5 ms Mean electrically-evoked spikes per train MSNs * * Change in electrically-evoked spikes (rug-) 4 *** MSN E F NAcc NAcc MSN 5 mv 5 ms Mean -evoked spikes per train Vehicle MSNs 1 MSNs MSNs * * Change in -evoked spikes (rug-) 4 *** - -4 G 1 MSN H I NAcc NAcc VH MSN VH 5 mv 5 ms Mean VH-evoked spikes per train Vehicle MSNs 1 MSNs MSNs * Change in VH-evoked spikes (rug-) * Figure 7. KOR Activation Preferentially ecreases Excitatory rive of 1 MSNs and isinhibits MSNs (A) 1 or MSNs were patched, and spiking was driven by extracellular stimulation using a bipolar stimulating electrode. Representative traces of synaptically driven spiking in 1 and MSN are shown during baseline and after U69. (B) Mean electrically evoked spikes during baseline (solid) and after vehicle or U69 (checkered) in MSNs treated with vehicle (gray; n = 1) or 1 (blue; n = 8) and (red; n = 7) MSNs treated with U69 (two-way ANOVA cell type 3 drug interaction, F (,) = 9.58, p =.1). (C) Mean change in evoked spiking in vehicle MSNs and U69-treated 1 and MSNs (One-way ANOVA, F (,4) = 9.585, p =.1, ***p <.5). (legend continued on next page) Neuron 93, , January 4,

15 Figure 8. Model of Pathway-Specific and Cell-Type-Specific KOR Modulation of 1 and MSN Activity (A) Pathway-specific KOR modulation of excitatory synapses in the NAcc. KORs inhibit, but not VH, afferents into NAcc MSNs. NAcc-projecting neurons express KOR mrna, whereas KOR mrna expression is absent in NAcc-projecting VH neurons. KORs only inhibit inputs to 1 MSNs. (B) KOR modulation of MSN collaterals. KORs inhibit 1 MSN collaterals more strongly than they inhibit MSN collaterals. KOR mrna expression is higher in 1 MSNs than in MSNs. (C) Afferent control of 1 and MSNs by the VH and during basal conditions (top) and after increased KOR signaling (bottom). The strength of excitation-spiking coupling between afferent and postsynaptic neurons is depicted at individual afferent synapses by the number of action potentials, the size of the presynaptic terminal, and the size of the glutamate or. KOR inhibition of synapses in 1 MSNs decreases driven spiking in 1 MSNs, whereas the VH afferent drive of 1 MSNs is unchanged. In MSNs, KOR disinhibits VH- and -driven spiking by inhibiting 1 MSN collaterals. ata are expressed as mean ± SEM. potentials and increase the probability of action-potential generation. Consistent with this notion, electrical and optogenetic excitation-driven spiking was facilitated by U69 in MSNs. Thus, KORs also indirectly control information flow from afferents by regulating local inhibitory connections between MSNs (Figures 8B and 8C). NAcc KORs may alter encoding properties of 1 and MSNs via presynaptic control of glutamate and. KORs may serve as a mechanism by which 1 MSN-mediated yn release, in response to strong glutamate and A activation of 1 MSNs (Atwood et al., 14; Gerfen et al., 199; Moratalla et al., 1996; Wang et al., 1994), can decrease excitationinhibition balance and, ultimately, excitation-spiking coupling in a pathway-specific manner in 1 MSNs. Conversely, presynaptic inhibition of release on MSNs increases excitation-inhibition balance, consequentially facilitating excitatory drive of MSNs. This effect is of interest, as disinhibition of spiking by delta-opioid systems has been recently reported in 1 MSNs in dorsal striatal striosomes (Banghart et al., 15). Since yn signaling is mobilized in an activity-dependent manner, yn-kor signaling could be a mechanism by which activity states of 1 MSNs influence ongoing activity of both 1 and () Spiking was evoked in 1 and MSNs by optogenetically activating glutamatergic terminals, which recruits feedforward and/or collateral inhibition. Representative traces of optogenetic -stimulation-evoked spiking in 1 and MSNs during baseline and after U69 in 1 and MSNs. (E) Mean -evoked spikes during baseline and after vehicle or U69 in MSNs treated with vehicle (n = 1) or 1 (n = 1) and (n = 7) MSNs treated with U69 (two-way ANOVA cell type 3 drug interaction, F (,4) = 9.5, p =.11). (F) The mean change in -evoked spiking in vehicle MSNs and in U69-treated 1 and MSNs (one-way ANOVA, F (,6) = 9.5, p =.11, ***p <.5). (G) Representative traces of optogenetic, VH-stimulation-evoked spiking in 1 and MSNs during baseline and after U69. (H) Mean VH-evoked spikes during baseline and after vehicle or U69 in MSNs treated with vehicle (n = 7) or 1 (n = 7) and (n = 7) MSNs treated with U69 (twoway ANOVA cell type 3 drug interaction, F (,18) = 5.14, p =.17). (I) Mean change in VH-evoked spiking in vehicle MSNs and in U69-treated 1 and MSNs (one-way ANOVA, F (,) = 5.14, p =.17, *p <.5). ata are expressed as mean ± SEM. 16 Neuron 93, , January 4, 17