Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2017 Jun 1.
Published in final edited form as: Neuron. 2016 May 19;90(5):1071–1085. doi: 10.1016/j.neuron.2016.05.001

Linking cholinergic interneurons, synaptic plasticity, and behavior during the extinction of a cocaine-context association

Junuk Lee 2,3,#, Joel Finkelstein 2,3,#, Jung Yoon Choi 2,3, Ilana B Witten 2,3,4
PMCID: PMC4893317  NIHMSID: NIHMS788935  PMID: 27210555

Abstract

Despite the fact that cholinergic interneurons are a key cell-type within the nucleus accumbens, a relationship between synaptic plasticity and the in vivo activity of cholinergic interneurons remains to be established. Here, we identify a three-way link between the activity of cholinergic interneurons, synaptic plasticity, and learning in mice undergoing the extinction of a cocaine-context association. We found that activity of cholinergic interneurons regulates extinction learning for a cocaine-context association and generates a sustained reduction in glutamatergic presynaptic strength onto medium spiny neurons. Interestingly, activation of cholinergic interneurons does not support reinforcement learning nor plasticity by itself, suggesting that these neurons have a modulatory rather than a reinforcing function.

Introduction

Cholinergic (ChAT) interneurons within the nucleus accumbens (NAc) and the dorsal striatum comprise a sparse population of interneurons that are well positioned to play an important role in the function of the circuit (Tepper and Bolam 2004; F.-M. Zhou, Wilson, and Dani 2002). They provide a major source of cholinergic innervation, they are activated by salient or reward-predicting events (Atallah et al. 2014a; Joshua et al. 2008a; Morris et al. 2004a), they locally regulate dopamine release (Cachope et al. 2012; Threlfell et al. 2012; F. M. Zhou, Liang, and Dani 2001), and their activity modulates the activity of the medium spiny neurons (MSNs), the output neurons in the striatum (English et al. 2011; Oldenburg and Ding 2011; Witten et al. 2010; Nelson et al. 2014; (Higley et al. 2011)). Indeed, ChAT interneurons in the NAc appear to play a role in certain forms of learning, including formation of cocaine-context associations (Witten et al. 2010), discriminative fear conditioning (Brown et al. 2012), and learning within an attentional set shift task (Aoki et al. 2015).

Despite the importance of ChAT interneurons, several key questions remain unaddressed. First, what, if any, synaptic plasticity in the NAc underlies ChAT-interneuron-mediated changes in behavior? Second, if ChAT interneurons do affect synaptic plasticity, under what conditions do they do so? One possibility is that ChAT interneuron activity is by itself sufficient to generate plasticity and to drive reinforcement. Alternatively, the activity of ChAT interneurons could instead function to modulate the extent of plasticity and learning for external reinforcers, without itself being reinforcing. Distinguishing between these possibilities is at the heart of characterizing how these neurons contribute to circuit function and behavior.

We address these questions in the context of mice undergoing extinction of a cocaine conditioned place preference (CPP). We focused on extinction learning because a critical feature of drug addiction is the persistence of maladaptive drug-seeking behavior, and thus a particular challenge in treating addiction is to find ways to reduce or extinguish drug-related associations. Therefore, mechanisms underlying extinction learning have the potential to be clinically relevant, and yet less is known about mechanisms of extinction in comparison to formation of a drug-context association. We examine extinction of a cocaine CPP rather than operant cocaine self-administration in order to specifically isolate mechanisms underlying contextual associations from other types of associations (e.g. operant associations, cue associations). This is a valuable approach given that contextual associations contribute to drug-seeking and depend on different neural mechanisms than operant or cue associations (Torregrossa, Gordon, and Taylor 2013; Fuchs et al. 2005).

Thus, our first objective was to determine if ChAT interneurons do in fact play a role in extinction learning for a cocaine CPP, a finding which would extend our previous observation that they play a role in formation of such associations (Witten et al. 2010). We then proceeded to combine in vivo manipulation of ChAT interneuron activity with ex vivo electrophysiology in order to connect ChAT interneuron activity with potential synaptic plasticity mechanisms.

Results

Control of activity in ChAT interneurons bidirectionally regulates extinction of a cocaine-context association

We first sought to determine if activity in ChAT interneurons contributes to extinction learning of a cocaine-associated context. Towards this end, an AAV2/5 virus expressing either Cre-dependent ChR2-YFP or YFP-only (control virus) was injected into the medial NAc of ChAT::IRES-Cre mice and optical fibers were implanted bilaterally above the structure (Figures 1A and 1B). This strategy resulted in selective ChR2 expression in ChAT interneurons of the medial NAc (Figures 1B and 1C). Before proceeding to in vivo experiments, the efficacy of optogenetic stimulation of ChAT interneurons was confirmed through whole cell recordings in brain slices (Figure 1D; 4-6 weeks after virus injection, 445 nm light, 5 mW/mm2, 15 Hz, 5 ms pulse duration, 2s activation period interleaved with 2s light off period).

Figure 1. Optogenetic activation of cholinergic interneurons in the NAc enhances the extinction of a cocaine CPP.

Figure 1

Open in a new tab

A. Cre-dependent ChR2-YFP (or YFP-only) virus was injected bilaterally into the NAc of ChAT::IRES-Cre mice and fibers were implanted above the injection sites. B. Top: ChR2 expression in ChAT interneurons. Scale bar: 300um. Bottom: summary of fiber tip locations. C. Co-localization between ChR2-YFP and ChAT immunohistochemistry. Scale bar: 20um. D. Whole cell recordings to confirm the functionality of ChR2. E. Cocaine conditioned place preference paradigm with optogenetic stimulation on “Test 1”. F. Heatmaps representing time spent in the saline and cocaine chamber on “Test 1” of an example mouse expressing ChR2-YFP virus (top) or YFP-only virus (bottom). Each pixel represents 0.2cm × 0.2cm. G. Cocaine chamber preference during the baseline and the three test days for mice undergoing the cocaine CPP described in panel 1E (F(1,17)=5.141 p=0.036 for group, repeated measures ANOVA. ChR2 group, (n=10) Pre-test: 1.9 ± 36.2, Test 1: 184.3 ± 38.7, Test 2: 63.2 ± 55.245, Test 3: 6.843 ± 41.46. YFP group, (n=9) Pre-test: −12.064 ± 50.183, Test 1: 301.835 ± 53.791, Test 2: 145.145 ± 44.072, Test 3: 82.624 ± 41.532). H. Average preference for stimulated chamber in mice undergoing a real time CPP test (p=0.53, two tailed t-test. ChR2 group, (n=8) 76.547 ± 78.112. YFP group, (n=7) 14.157 ± 64.274). In D-H, optogenetic stimulation of ChAT interneurons as follows: 447nm, 15 Hz, 5ms pulse duration, 2s light on interleaved with 2s light off. All error bars are SEM.

To probe the function of ChAT interneurons in cocaine-context extinction learning, mice were conditioned on a cocaine conditioned place preference (CPP; paradigm schematic in Figure 1E). During the first day of training (“Pre-test”), mice had access to both chambers of the apparatus. The following two days were conditioning days, during which they were trained to associate one environment with cocaine (15 mg/kg, i.p.) and the other environment with an equal volume of saline. Mice then underwent three extinction days to assess and extinguish their preference for the cocaine chamber; “Test 1” and “Test 2” took place on the two days subsequent to conditioning while “Test 3” took place a week later. Only during “Test 1”, mice received bilateral optogenetic activation of the ChAT interneurons as they explored the apparatus (activation was not contingent on the location of the mice in the chamber).

Mice that received optogenetic activation of ChAT interneurons during Test 1 spent less time in the cocaine-associated chamber relative to YFP-control littermates across the three post-conditioning test sessions (Figures 1F and 1G; F(2,34)=10.31 p=0.0003 for test day and F(1,17)=5.141 p=0.0367 for group, repeated measures ANOVA). Of note, the modulation of extinction behavior by ChAT interneurons generalizes beyond the specific mouse strain used in this experiment (strain C57BL/6J; Figure 1G), as we replicated this result in a separate cohort of mice with a different genetic background (strain 129S1; Figures S1A and S1B; F(2,26)=10.155 p=0.000551 for test day and F(1,13)=5.275 p=0.0389 for group, repeated measures ANOVA). The fact that the difference in preference persisted between the ChR2 and YFP groups during the 2nd and 3rd extinction test, even though ChAT activity was only manipulated on the first test day, suggests that ChAT interneurons effect drug-context extinction learning (rather than simply affecting the expression of the memory; p=0.021 for group, repeated measures ANOVA with group and day as factors, analyzing test day 2 and 3 from cohorts in Figure 1G and Figure S1B).

In addition, we examined the temporal evolution of chamber preference during the first extinction test, when ChAT interneurons were activated (Figure S1C). Over the course of the session, control mice do not show a significant decrease in chamber preference over time (p=0.38 for time as a predictor of preference in a linear regression), while mice that received ChAT interneuron activation do show a significant decrease (p=0.002 for time as a predictor of preference in a linear regression). Although this is suggestive that the time course of extinction is different even during the first test across the two groups, the interaction between group and time is not quite significant with the available data (p=0.001 for time, p<0.001 for group, p=0.081 for timeXgroup; linear regression to predict preference based on group, time, and groupXtime).

ChAT interneurons may contribute to cocaine-context extinction either by modulating ongoing learning, or by generating new learning. To determine if activity in ChAT interneurons is sufficient to generate reinforcement learning in naive mice, we activated ChAT interneurons in a real-time CPP test in which the neurons were activated whenever a mouse was in one chamber and not the other chamber (“Real time CPP”, Figure 1H). This manipulation had no effect on the time spent in the two chambers, indicating the activity in ChAT interneurons by itself does not drive reinforcement (Figure 1H; p=0.53, two-tailed t-test; additional stimulation frequencies in Figures S1D). We also tested if ChAT interneuron activation could support intracranial self-stimulation in an operant chamber, and again did not find that the manipulation was reinforcing (Figures S1E).

Given that activation of ChAT interneurons was sufficient to increase extinction learning (Figure 1G), we next investigated the possibility that inhibition of the same neurons would result in the reverse effect of reducing extinction learning. Towards this end, an AAV2/5 virus expressing either Cre-dependent NpHR-YFP or YFP-only (control virus) was injected into the medial NAc of ChAT::IRES-Cre mice and optical fibers were implanted bilaterally (Figures 2A-2C). Optogenetic inhibition of ChAT interneurons successfully suppressed action potentials, as confirmed with whole cell recordings (Figure 2D and Figures S2A and S2B). A cocaine CPP assay was implemented as before, but in this case constant illumination of 590 nm light was delivered to inhibit activity during the first post-conditioning test (“Test 1”, Figure 2E). Inhibition of ChAT interneurons caused a decrease in extinction learning relative to control littermates (Figure 2F and 2G; F(2,37)=15.23 p<0.0001 for test day and F(1,18)=5.63 p=0.029 for group, repeated measures ANOVA). Similar to the case of ChAT interneuron activation, the effect of inhibition on chamber preference persisted on the test days subsequent to the inhibition, meaning that these neurons modulate extinction learning rather than only having acute effects on expression of the memory (Figure 2G). Taken together with the results from activating ChAT interneurons (Figure 1G and Figure S1B), these experiments indicate that ChAT interneuron activity provides bi-directional regulation of extinction learning for a cocaine-context association.

Figure 2. Optogenetic inhibition of cholinergic interneurons suppresses the extinction of a cocaine CPP.

Figure 2

Open in a new tab

A. Cre-dependent NpHR-YFP (or YFP-only) virus was injected bilaterally into the NAc of ChAT::IRES-Cre mice and fibers were implanted above the injection sites. B. Top: NpHR expression in ChAT interneurons. Scale bar: 300um. Bottom: summary of fiber tip locations. C. Co-localization between NpHR-YFP and ChAT immunohistochemistry. Scale bar: 20um. D. Whole cell recordings to confirm the functionality of NphR (current injections of 120 pA at 1Hz). E. Cocaine conditioned place preference paradigm with optogenetic stimulation on “Test 1”. F. Heatmaps representing time spent in the saline and cocaine chamber on “Test 1” of an example mouse expressing either NpHR-YFP virus (top) or YFP-only virus (bottom). Each pixel represents 0.2cm × 0.2cm. G. Average cocaine chamber preference during the pre-test and the three test days for mice undergoing cocaine CPP as described in panel 2E. (F(1,17)=5.6 p=0.02 for group, repeated measures ANOVA. NpHR group, (n=11) Pre-test: −7.2 ± 36.0, Test 1: 361.8 ± 27.0, Test 2: 305.6 ± 35.0, Test 3: 182.4 ± 51.6. YFP group, (n=10) Pre-test: 18.5 ± 18.5, Test 1: 232.0 ± 44.3, Test 2: 198.1 ± 46.6, Test 3: 43.7 ± 46.1). H. Average preference for the stimulated chamber in mice undergoing a real time CPP test (p=0.62, two tailed t-test. NpHR group, (n=6) −22.7 ± 64.3. YFP group, (n=6) 29.9 ± 83.5). In D-H, constant illumination with 590nm light. All error bars are SEM.

In addition, similar to the case of activation, the inhibition of ChAT interneurons was not sufficient to drive a real-time CPP, again supporting the idea that ChAT interneurons modulate extinction learning without being able to drive new learning on their own (Figure 2H; p=0.62, two-tailed t-test). In addition, it is unlikely that these behavioral effects can be explained by nonspecific changes in anxiety-like behavior, as no effect was observed of activation nor inhibition of these neurons on anxiety assays (open field test and elevated plus maze; Figures S1F and S1G and S2D and S2E). Similarly, these manipulations did not generate a change in velocity, neither in the open field nor in the cocaine CPP extinction test (Figures S1H and S1I and Figures S2F and S2G).

ChAT interneurons mediate plasticity of excitatory synapses onto MSNs during extinction of a cocaine-context association

We next sought to test the hypothesis that ChAT interneurons could impact extinction learning by modulating synaptic plasticity in the NAc. Reward-related learning is associated with changes in the strength of excitatory synapses onto medium spiny neurons (MSNs), the output neurons of the NAc (Lüscher and Malenka 2011; Britt et al. 2012), and thus ChAT interneurons may play a role in modulating such plasticity. To address this possibility, we compared miniature excitatory postsynaptic currents (mEPSCs) in MSNs of mice in which ChAT interneurons had been optogenetically activated during the extinction test, relative to MSNs of control mice in which ChAT interneurons were not activated (Figures 3A-C). In either case, mice were sacrificed immediately after the first extinction test (behavioral schematic in Figure 3A). Consistent with our earlier results (Figure 1G and Figure S1B), activation of ChAT interneurons during the initial extinction test decreased chamber preference, which implies enhanced extinction learning (Figure 3B; F(1,15)=32.1 p<0.0001 for group and F(1, 15)= 8.396 p=0.011 for test, two-way ANOVA; p=0.0003 across groups for the test day, p=0.2843 for the pre-test day, Sidak’s post-hoc test). Comparison of the mEPSCs from the two groups revealed a substantial decrease in the frequency of mEPSCs as a consequence of ChAT interneuron activation and no significant change in amplitude (Figure 3D; p<0.001 for interevent interval and p=0.11 for amplitude, mixed-effect linear regression, see Methods for details; no difference in basic membrane properties or recording conditions across groups, Figure S3B).

Figure 3. ChAT interneurons mediate plasticity of excitatory synapses onto MSNs during cocaine-context extinction learning.

Figure 3

Open in a new tab

A. Mice were conditioned on a cocaine CPP and received optogenetic stimulation of ChAT interneurons during the post-conditioning test (447nm, 15Hz, 5ms pulse duration, 2s light on interleaved with 2s light off). Immediately after the test they were sacrificed for ex vivo recordings. B. Prior to ex vivo physiology, mice exhibited less preference for the cocaine-paired chamber if ChAT interneurons were activated during the test (relative to control mice which did not receive ChAT interneuron activation but otherwise underwent the same behavior) (F(1,15)=32.1 p<0.0001 for group and F(1, 15)=8.396 p=0.011 for test, two-way ANOVA with group and test as factors; p=0.2843 for Pre-test day, p=0.0003 for Test day, Sidak’s post-hoc test.ChAT activation group, (n=6) Pre-test: −46.880 ± 29.853, Test: 24.486 ± 70.486. Control group, (n=11) Pre-test: 49.730 ± 37.211, Test: 308.139 ± 40.474). C. Sample voltage clamp recordings of mEPSCs from an MSN from one mouse that received ChAT stimulation and from a control mouse. D. Left: Cumulative probability of interevent intervals for mEPSCs in mice that received ChAT activation during the test and for control mice. Decreased mEPSC frequency in mice that received stimulation. (p<0.001 for group in a linear, mixed effects regression. Left inset: Median frequency of mEPSCs, ChAT activation group, (n=15) 3.2 ± 0.5. Control group, (n=24) 7.0 ± 0.8). Right: Cumulative probability of the amplitude mEPSCs in mice that received ChAT activation and for control mice. No change in mEPSC amplitude in mice that received stimulation (p=0.11 for group in a linear, mixed effects regression. Right inset: Median amplitude of mEPSCs, ChAT activation group, 13.1 ± 0.3. Control group, 14.9 ± 0.6). E. Same as A, but this cohort was sacrificed for ex vivo recordings 24 hrs after the test. F. mEPSC amplitude and frequency 24 hrs after the test. Left: Cumulative probability of interevent intervals for mEPSCs in mice that received ChAT activation and for control mice. Decreased mEPSC frequency in mice that received stimulation (p=0.03 for group in a linear, mixed effects regression. Left inset: Median frequency of mEPSCs, ChAT activation group, (n=21) 4.6 ± 0.6. Control group, (n=16) 6.2 ± 0.6) Right: Cumulative probability of the amplitude of mEPSCs in mice that received ChAT activation and for control mice. No change in mEPSC amplitude in mice that received stimulation. (p=0.42 for group in a linear, mixed effects regression. Right inset: Median amplitude of mEPSCs, Chat activation group, 13.0 ± 0.2 Control group, 13.7 ± 0.5). All error bars are SEM.

The fact that ChAT interneurons mediate changes in chamber preference that persists during the days subsequent to their activation (Figures 1G and 2G) suggests that the concomitant plasticity in the NAc may also persist on the day following stimulation. Thus, we investigated whether or not ChAT interneuron mediated plasticity persisted on the day following stimulation by training a new cohort of mice on the cocaine CPP paradigm and activating ChAT interneurons during the first post-conditioning test day, as before (Figure 3E). In this case, mice were sacrificed and mEPSCs were measured on the day following the test day, rather than immediately after the test day as in the prior experiment (Figure 3E). As expected, ChAT interneuron activation on the test day was again associated with less cocaine chamber preference (Figure S3A; F(1, 16)=11.02 p=0.0043 for group and F(1,16)=5.87 p=0.027 for test, two-way ANOVA; p=0.035 across groups for the test day, p=0.29 for the pre-test day, Sidak’s post-hoc test; no difference in basic membrane properties or recording conditions across groups, Figure S3C). In addition, once again, ChAT interneuron stimulation was associated with a decrease in the frequency and no change in the amplitude of mEPSCs in MSNs (Figure 3F; p=0.03 for interevent interval and p=0.4 for amplitude, mixed-effect linear regression, see Methods for details). In addition, there was no statistically significant interaction between the time the mice were sacrificed (0hr vs 24hr) and the effect of ChAT stim in explaining mEPSC frequency (p=0.16 for Stim X Time interaction; mixed-effect linear regression, see Methods for details).

We observed a change in frequency and not amplitude of mEPSCs as a result of ChAT interneuron activation (Figures 3D and 3F), implying that the site of plasticity mediated by ChAT interneurons may be presynaptic and not postsynaptic. However, reductions of mEPSC frequency can occur with loss of postsynaptic sites as well. To provide additional information regarding the site of plasticity, in a new cohort of mice we measured the paired-pulse ratio (PPR) and the ratio of AMPA to NMDA receptor currents onto MSNs, the former being an assay of presynaptic plasticity and the latter being as assay of postsynaptic plasticity. To this end, mice underwent an identical cocaine CPP as described in Figure 3A, receiving activation of ChAT interneurons during the first extinction test followed immediately by ex vivo electrophysiological measurements (Figures 4A and 4B). We observed a large enhancement in PPR as a consequence of ChAT interneuron stimulation, consistent with a presynaptic plasticity mechanism (Figure 4C, F(1,16=15.62) p=0.0011 for group and F(2,32)=7.4 p=0.0023 for stimulation interval, two-way ANOVA; p=0.0043 at 20 ms, p=0.0019 at 50 ms and p=0.0343 at 100 ms, Sidak’s post-hoc test). In contrast, the change in AMPA/NMDA receptor current ratio was not statistically significant between the two groups (Figure 4D; p=0.1738, two-tailed t-test). Note that this negative result does not rule out the possibility that there is postsynaptic plasticity that we were not able to detect with our assay. For example, opposing postsynaptic effects may occur in parallel in different subpopulations of MSNs (e.g. D1R vs D2R expressing MSNs).

Figure 4. ChAT interneurons modulate paired-pulse ratio at excitatory synapses onto MSNs during cocaine-context extinction learning.

Figure 4

Open in a new tab

A. Mice were conditioned on a cocaine CPP and received optogenetic stimulation of ChAT interneurons during the first extinction test (447nm, 15Hz, 5ms pulse duration, 2s light on interleaved with 2s light off). Immediately after the test, they were sacrificed for ex vivo recordings. B. During the extinction test, mice exhibited less preference for the cocaine-paired chamber if ChAT interneurons were activated during the test (relative to control mice which did not receive ChAT interneuron activation but otherwise underwent the same behavior) (F(1,14)=4.76 p=0.0467 for group and F(1, 14)= 8.96 p=0.009 for test, two-way ANOVA with group and test as factors; p=0.9361 for Pre-test, p=0.0197 for test, Sidak’s post-hoc test. ChAT activation group, (n=8) Pre-test: −7.991 ± 68.717, Test: 59.530 ± 46.839. Control group, (n=8) Pre-test: 16.635 ± 38.180, Test: 268.885 ± 55.486). C. Left: Representative EPSCs in MSNs in response to interpulse intervals of 20, 50 and 100 ms. Top traces from an MSN of a mouse that received ChAT activation and the bottom traces from a control mouse. Right: Significant increase in paired-pulse ratio in MSNs of mice that received ChAT activation. (F(1,16)=15.62, p=0.0011 for group; F(2, 32)=7.4, p=0.0023 for stimulation interval; 2-way ANOVA with group and stimulation interval as factors; p=0.0043 for 20 ms interval (0.665 ± 0.064), p=0.0019 for 50 ms interval (0.940 ± 0.143), and p=0.0347 for 100 ms interval (1.043 ± 0.133), Sidak’s post-hoc test). D. Left: Representative traces of AMPA and NMDA receptor currents. Right: No significant difference in AMPA/NMDA receptor current ratio in MSNs between ChAT activation group versus control group. (p=0.1738, two-tailed t-test; ChAT activation group, n=12, 3.656 ± 0.4363; Control group, n=13, 4.642 ± 0.5414). All error bars are SEM.

So far, we have determined that manipulation of activity in ChAT interneurons during the extinction of a cocaine CPP generates changes in both extinction learning and in glutamatergic plasticity onto MSNs. The presence of these two consequences of ChAT stimulation suggests that changes in glutamatergic plasticity may represent a neural substrate for extinction learning for a cocaine CPP. To determine if it is indeed the case that the reduction in mEPSC frequency is a neural manifestation of extinction learning, we trained two cohorts of mice on a cocaine CPP. After conditioning, one cohort received four extinction tests while the control cohort did not undergo any extinction tests (Figures 5A and 5B). We found that multiple days of extinction training generates reductions in mEPSC frequency, and not amplitude (Figure 5C; p=0.0135 for interevent interval, p=0.697 for amplitude, mixed-effect linear regression, analysis details in Methods), changes that closely mimic the changes that occur as a consequence of ChAT interneuron activation during a single extinction test (Figure 3D). This suggests that ChAT stimulation during a single extinction test accelerates the plasticity that naturally occurs during the course of repeated extinction sessions on a cocaine CPP.

Figure 5. Repeated extinction of a cocaine-context association reduces the frequency of mEPSCs onto MSNs.

Figure 5

Open in a new tab

A. Cocaine CPP paradigm followed by 4 extinction sessions in the extinction group and no extinction in the control group. Both groups were sacrificed at the same time point, immediately after the final extinction test. B. Cocaine chamber preference during the baseline and four extinction tests for mice undergoing the CPP described in panel 5A. (Chamber preference for extinction group, (n=5) Pre-test: −3.457 ± 39.292, Test 1: 294.361 ± 740.072, Test 2: 250.524 ± 36.798, Test 3: −45.299 ± 66.656, Test 4: 33.560 ± 45.948. Chamber preference for control group, (n=6) Pre-test: 15.993 ± 51.390). C. Left: Cumulative probability of interevent intervals for mEPSCs in extinction group and control. Decreased mEPSC frequency in extinction group. (p=0.0135 for group in a linear, mixed effects regression. Left inset: Median frequency of mEPSCs, Extinction group, (n=12) 5.1 ± 0.6. Control group, (n=7) 8.1 ± 0.8). Right: Cumulative probability of the amplitude mEPSCs in extinction group and for control. No change in mEPSC amplitude in both groups (p=0.697 for group in a linear, mixed effects regression. Right inset: Median amplitude of mEPSCs, Extinction group, 12.7 ± 0.3 and Control group, 12.8 ± 0.3). All error bars are SEM.

ChAT interneuron mediated plasticity onto MSNs depends on context and experience

We next sought to determine if ChAT-interneuron-mediated plasticity depended on the mouse’s experience and context. These experiments were geared towards distinguishing between two hypotheses. The first hypothesis is that activity in ChAT interneurons can generate plasticity regardless of context and experience. Alternatively, it could be that ChAT interneurons regulate the extent of plasticity if and when cocaine-context extinction is occurring. As detailed below, we took several approaches to distinguish between these hypotheses.

Our first approach involved training mice on a cocaine CPP as before, and comparing the effect of ChAT interneuron activation during the first extinction test (as before, Figures 1 and 3), versus activation immediately before the extinction test in the home cage (experiment schematic in Figure 6A). On a behavioral level, we observed a reduction in preference when stimulation occurred in the extinction context, suggesting that ChAT interneuron activity only affects extinction learning if the neurons are activated during the extinction test (Figure 6B; F(1,16)=4.703 p=0.0455 for group and F(1,16)=14.24 p=0.0017 for test, two-way ANOVA; p=0.013 across groups for the test day, p=0.99 for the pre-test day, Sidak’s post-hoc test). We next sought to determine if there were differences in plasticity onto the MSNs between the groups which could underlie these differences in behavior. Indeed, we found a reduction in mEPSC frequency (and not amplitude) in MSNs when stimulation occurred during the extinction test in comparison to when it occurred beforehand in the home cage (mice were sacrificed immediately after the test; Figure 6C; p=0.02 for frequency, p=0.39 for amplitude; mixed-effect linear regression, see Methods for details; no difference in basic membrane properties or recording conditions across groups, Figure S6A). In addition, there was no significant difference in mEPSC frequency nor amplitude when comparing mice that received ChAT interneuron activation in the home cage relative to control mice that received no stimulation (control mice in Figure 3D; p=0.9 for frequency, p=0.9 for amplitude; mixed-effect linear regression, see Methods for details). These results support the hypothesis that ChAT interneurons regulate the extent of plasticity if and when it is occurring.

Figure 6. ChAT interneuron mediated plasticity depends on experience and context.

Figure 6

Open in a new tab

A. Mice were conditioned on a cocaine CPP and received optogenetic stimulation of ChAT interneurons either in the home cage or during the test (447nm, 15Hz, 5ms pulse duration, 2s light on interleaved with 2s light off). Immediately after the test they were sacrificed for ex vivo recordings. B. Mice in which ChAT interneurons were activated during the extinction test showed less preference for the cocaine-paired chamber before whole cell recordings (relative to mice which received the ChAT interneuron stimulation in the home cage but otherwise had the same behavior experience) (F(1,16)=4.703 p=0.0455 for group and F(1,16)=14.24 p=0.0017 for test, two-way ANOVA with group and test as factors; p=0.99 for Pre-test, p=0.013 for Test, Sidak’s post-hoc test) Stim during test group, (n=9) Pre-test: −57.394 ± 62.111, Test: 78.646 ± 95.978. Stim before test group, (n=9) Pre-test: −46.171 ± 72.113, Test, 371.631 ± 48.406). C. Left: Cumulative probability of interevent intervals for mEPSCs in mice that received ChAT activation during the test and mice that received ChAT activation in the home cage. Decreased mEPSC frequency in mice that received stimulation during the test. Right: Cumulative probability of the amplitude of mEPSCs in mice that received ChAT activation during either the test or the home cage (p=0.393). D. Mice were conditioned on a saline CPP and received optogenetics stimulation of ChAT interneurons on the test day. Immediately after the test they were sacrificed for ex vivo recordings. E. Saline-only conditioned mice in which ChAT interneurons were activated during the test exhibited no difference in chamber preference relative to control mice, nor any difference relative to pre-conditioning (F(1,9)=0.37 p=0.55 for group and F(1,9)=0.46 p=0.51 for test, two-way ANOVA with group and test as factors; p=0.23 for Pre-test, p=0.65 for Test day, Sidak’s post-hoc test). ChAT activation group, (n=6) Pre-test: 45.9 ± 58.75838, Test: −90.9 ± 37.9. Control group, (n=5) Pre-test: −80.1 ± 37.0, Test: −25.0 ± 77.4). F. Left: Cumulative probability of interevent intervals for mEPSCs in saline-only conditioned mice that received ChAT stimulation versus control mice. No significant difference in mEPSC frequency across groups (p=0.30, for group in a linear regression. Left inset: Median frequency of mEPSCs, ChAT activation group, (n=16) 4.6 ± 0.5. Control group, (n=12) 5.7 ± 0.6) Right: Cumulative probability of the amplitude of mEPSCs in mice that received stimulation versus the control condition. No significant difference in mEPSC amplitude across groups. (p=.37 for group in a linear regression. Right inset: Median amplitude of mEPSCs, ChAT activation group, (n=16) 13.4 ± 0.5. Control group, (n=12) 12.7 ± 0.5). All error bars are SEM.

Our second approach to identify the conditions under which ChAT interneurons mediate plasticity involved examining mEPSCs in MSNs of mice that underwent an identical conditioning procedure as before (Figure 3A), but instead of pairing cocaine injections to one chamber, they received saline injections in both chambers (“saline-only conditioning”; schematic in Figure 6D). In this case, there was no sign of learning (nor extinction) since there was no reason for them to form a spatial preference (Figures 6E). During the test session, one group of mice received ChAT interneuron activation, while the control group did not. As expected, neither group of mice exhibited a spatial preference and there was no effect on behavior of the optogenetic stimulation (Figure 6E; F(1,9)=0.37 p=0.55 for group and F(1,9)=0.46 p=0.51 for test, two-way ANOVA; p=0.65 across groups for the test day, p=0.23 for the pre-test day, Sidak’s post-hoc test). We sacrificed the mice immediately after this test, and compared mEPSCs ex vivo in MSNs of mice which received ChAT interneuron activation during the test, relative to MSNs of mice which received no ChAT interneuron activation. Comparison of mEPSCs in both groups revealed no effect of ChAT interneuron activation on the frequency nor amplitude of mEPSCs (Figure 6F; p=0.30 for frequency, p=0.37 for amplitude; mixed-effect linear regression, see Methods for details; no change in basic membrane properties or recording conditions across groups, Figure S6B). Thus, given that in this experiment, there was no spatial preference learning, these results further support the hypothesis that ChAT interneurons only regulate the extent of plasticity if and when it is occurring.

Given that ChAT interneuron stimulation had no observable effect on mEPSCs in MSNs when performed in the home cage rather than in the conditioning chamber (Figures 6A-6C), nor did it have an effect in cocaine-naive mice (Figures 6D-6F), we assumed that activation of ChAT interneurons is also not able to induced changes in glutamatergic plasticity when activated in slices from naive mice. Indeed, we found that the amplitude of electrically evoked EPSCs did not change when comparing before and after 15 minutes of ChAT interneuron activation, which again supports the hypothesis that ChAT interneurons modulate plasticity only if and when it is occurring (Figure S6C).

No significant effect of ChAT interneuron activation on extinction of a contextual association for food or fear

To determine if the effects of ChAT interneuron manipulations are specific to the extinction of a cocaine CPP or generalize to extinction of other contextual associations, we trained a new cohort of mice on a food CPP, in which one chamber contained food and the other chamber was empty. As we did for the cocaine CPP extinction test (Figure 1), we activated ChAT interneurons during the first (of three) extinction tests (Figure 7A). In contrast to the case of the cocaine CPP, we found that ChAT interneuron activation had no effect on extinction of a food CPP (Figure 7B; F(1,17)=0.713 p=0.41 for group and F(2,34)=7.863 p=0.00156 for test day, repeated measures ANOVA). Consistent with the lack of effect on behavior, we found that there was no ChAT-interneuron-mediated change in the frequency nor amplitude of mEPSCs that were measured immediately after the first extinction test in separate groups of mice that were sacrificed at that time point (Figure 7C; p=0.964 for frequency, p=0.565 for amplitude; mixed-effect linear regression, see Methods for details).

Figure 7. ChAT interneuron activation did not affect extinction of a food CPP.

Figure 7

Open in a new tab

A. Mice were conditioned on a food CPP and received optogenetic stimulation of ChAT interneurons during the first post-conditioning test (“Test 1”), but not the following 2 test days (stimulation parameters: 447nm, 15Hz, 5ms pulse duration, 2s light on interleaved with 2s light off). B. Mice in which ChAT interneurons were activated during the first extinction session of a food CPP exhibited no difference in chamber preference relative to control mice (F(2,34)=7.863, p=0.0156? for test and F(1,17)=0.713?, p=0.41 for group, repeated measures ANOVA. ChR2 group, (n=9) Pre-test: 12.198 ± 41.019, Test 1: 290.073 ± 67.514, Test 2: 121.4315 ± 51.2722, Test 3: 24.7315 ± 96.55. YFP group, (n=10) Pre-test: 10.654 ± 30.8001, Test 1: 250.2969 ± 95.04933, Test 2: 98.6788 ± 66.84659, Test 3: −47.5576 ± 87.39763). C. Left: Cumulative probability of interevent intervals for mEPSCs in food conditioned mice that received ChAT stimulation versus control mice. No significant difference in mEPSC frequency across groups (p=0.964 for group in a linear regression. Left inset: Median frequency of mEPSCs, ChAT activation group (n=13) 6.48 ± 0.56. Control group (n=13) 6.59 ± 0.69. Right: Cumulative probability of the amplitude of mEPSCs in mice that received stimulation versus the control condition. No significant difference in mEPSC amplitude across groups. (p=0.565 for group in a linear regression. Right inset: Median amplitude of mEPSCs, ChAT activation group, (n=13) 12.38 ± 0.38. Control group, (n=13) 12.7 ± 0.5. D. Mice were conditioned in a fear conditioning paradigm and received optogenetic stimulation of ChAT interneurons during the 1st post-conditioning extinction test, but not the 2nd. C. Mice in which ChAT interneurons were activated during the first extinction test of a fear conditioning paradigm exhibited no difference in freezing relative to control mice (F(1,14)=13.110, p=0.0278 for test and F(1,14)=0.62, p=0.444 for group, repeated measures ANOVA. ChR2 group, (n=8) Pre-test: 10.25 ± 2.373, Test 1: 49.9617 ± 5.112, Test 2: 29.958 ± 5.369. YFP group, (n=7) Pre-test:12.25 ± 2.7722, Test 1: 50.916 ± 4.704, Test 2: 37.083 ± 4.438). All error bars are SEM.

We also examined the effect of ChAT interneuron activation on the extinction of a contextual fear memory. In this case, mice were trained to associate a chamber with a foot shock, and then received ChAT interneuron activation during the first of two extinction tests (Figure 7D). We observed no significant effect of ChAT interneuron activation on freezing during the two extinction tests (Figure 7E; F(1,14)=13.110 p=0.00278 for test day, F(1,14)=0.62 p=0.444 for group, repeated measures ANOVA).

Discussion

ChAT interneurons modulate cocaine-context extinction learning and the associated synaptic plasticity

Our data support a model in which ChAT interneurons in the NAc modulate cocaine-context extinction learning, without being able to generate reinforcement on their own. Optogenetic activation or inhibition of ChAT interneurons bidirectionally modulates extinction learning for a cocaine-context association, while the same manipulations have no effect on a real-time CPP, in which one of two chambers is paired with optogenetic stimulation and the mice have access to both chambers.

Similarly, ChAT interneurons regulate glutamatergic synaptic plasticity onto MSNs during cocaine-context extinction learning, while they appear to not affect plasticity under conditions in which ChAT interneuron stimulation does not affect behavior. Specifically, ChAT interneuron activation during cocaine-context extinction learning affected the mEPSC interevent interval distribution and the paired-pulse ratio of MSNs, suggesting plasticity of the presynaptic inputs. This plasticity seems to depend on the activation occurring at the same time that the mice underwent extinction of the cocaine CPP. In fact, the same activation pattern in mice that had not been conditioned with cocaine, and therefore were not undergoing extinction learning, resulted in no ChAT-interneuron-mediated glutamatergic plasticity onto MSNs (Figure 6). Similarly, neither home cage stimulation before the extinction test of a cocaine CPP nor stimulation in brain slices resulted in detectable glutamatergic plasticity onto MSNs (Figures 6 and S6). The correspondence between conditions in which ChAT interneurons mediate changes in behavior and conditions in which ChAT interneurons mediate plasticity support a convincing relationship between the behavioral and plasticity effects mediated by ChAT interneurons.

Our interpretation of these results is that ChAT interneurons can hasten plasticity but cannot generate plasticity on their own. In support of this hypothesis, we found that multiple days of extinction training after a cocaine CPP causes reductions in mEPSC frequency onto MSNs that closely mimic the changes that occur as a result of ChAT stimulation during a single extinction session (Figure 5 for mEPSC changes after repeated extinction and Figure 3D for effect of ChAT stimulation of mEPSCs after a single extinction session). This suggests that ChAT stimulation during a single extinction test accelerates changes that naturally occur during the course of repeated extinction tests of a cocaine CPP.

This function of ChAT interneurons as modulating but not reinforcing appears to contrast with the reinforcing role of the midbrain dopamine neurons which innervate the NAc. Dopaminergic activity is sufficient to generate learning and plasticity on its own (Pascoli et al. 2015; Witten et al. 2011; Tsai et al. 2009; Ilango et al. 2014; Steinberg et al. 2013). Thus, dopamine neurons may contribute to determining when learning occurs, while ChAT interneurons may instead help determine how quickly or efficiently the learning progresses. The contrast between the function of ChAT interneurons and dopamine neurons is particularly surprising given that ChAT interneuron activation generates DA release in the NAc (through axonal interactions) (Cachope et al. 2012; Threlfell et al. 2012). Thus, it is likely that simultaneous effects of ChAT interneurons on other nodes of the circuit (e.g., the acute silencing of the MSNs) prevents ChAT-interneuron-mediated dopamine release from being reinforcing (Nelson et al. 2014; English et al. 2011; Witten et al. 2010; Faust et al. 2015).

Significance of glutamatergic plasticity in the NAc

Plasticity of glutamatergic inputs to the NAc is thought to be essential in supporting reward-seeking behaviors. A prominent hypothesis is that glutamatergic inputs are modulated by a reinforcement signal carried by dopamine neurons, such that the synaptic strength of inputs that predict reward is altered. Indeed, there is extensive experimental evidence of glutamatergic plasticity as a result of reward- and drug-related learning (Lüscher and Malenka 2011; Stuber et al. 2010; Martin et al. 2006; Britt et al. 2012). In addition, such plasticity has been linked causally to reward-seeking behaviors through optogenetic manipulations of specific glutamatergic inputs (Ma et al. 2014; Lee et al. 2013; Pascoli et al. 2014a).

Given the established importance of glutamatergic plasticity onto MSNs, our finding that ChAT interneurons are capable of dramatic modulation of such plasticity is significant. Note that our new results are distinct from previous optogenetic plasticity paradigms that involved directly inducing long term depression in specific glutamatergic inputs (Ma et al. 2014; Lee et al. 2013; Pascoli et al. 2014a). Those experiments involved optogenetic manipulations of glutamatergic inputs before the behavioral test. In direct contrast, here we show that ChAT interneuron activation only modulates glutamatergic plasticity if activated during ongoing learning (and not if activated beforehand). Another distinction between our findings and this previous work is that the ChAT interneuron mediated glutamatergic plasticity we report here appears to be largely presynaptic, whereas an extended withdrawal periods after operant self-administration generates robust postsynaptic plasticity (Conrad et al. 2008; Loweth et al. 2013; Pascoli et al. 2014a; Lee et al. 2013). However, note that in this study we did not distinguish between different classes of MSNs (e.g. D1R vs D2R expressing), and therefore there may be postsynaptic effects mediated by ChAT interneurons that were masked by averaging opposing effecst.

Relationship to endogenous dynamics in ChAT interneurons

In this study, we determined the behavioral and physiological consequence of manipulating ChAT interneurons by activating or inhibiting them optogenetically over the course of 15 minutes. We performed measurements in brain slices to determine the efficacy of optogenetic inhibition over that timescale (Figures S2A and S2B), and found that action potentials were effectively eliminated by NpHR, and that the action potential waveforms were similar before and after the inhibition period (Figure S2C). However, it is possible that there are changes in post-inhibition excitability that our assay does not capture (e.g. changes in the chloride reversal potential (Raimondo et al. 2012)). We believe that even if such effects exist, they are unlikely to impact our conclusions. This is because aberrant ChAT interneuron activity after the extinction test cannot affect the outcome of the prior test, and seem unlikely to affect behavior 24 hours later.

Tonically active neurons (TANs) in the striatum, which are likely to be ChAT interneurons, have been shown to have very interesting in vivo temporal dynamics, including brief pauses and also bursts of activity around the time of salient or reward-predicting events (Morris et al. 2004b; Atallah et al. 2014b; Joshua et al. 2008b). The prolonged optogenetic inhibition and repeated 2s burst of optogenetic activation that we employ in this study during the CPP extinction test do not replicate these endogenous dynamics within TANs. In general, contextual learning (e.g. CPP) provides a challenge for linking endogenous dynamics to behavior, given that the lack of specific behavioral events to align with stimulation. Despite this challenge, it is an important goal in systems neuroscience to discover neural mechanisms that underlie contextual associations. This is because contextual learning represents an essential type of learning which involves distinct brain regions and mechanisms compared to cue learning or operant learning (Torregrossa, Gordon, and Taylor 2013; Fuchs et al. 2005).

Despite not replicating the endogenous dynamics of TANs, there are reasons to believe our stimulation protocol was both reasonable and informative. First, the fact that our ChAT interneuron manipulation leads to bidirectional changes in behavior demonstrates that our manipulation has interpretable effects on behavior. Second, the synaptic changes onto MSNs mediated by ChAT interneuron activation during the extinction test resemble the synaptic changes that naturally occur over the course of multiple extinction tests, in the absence of a manipulation (Figure 5). The fact that ChAT interneuron activation causes changes in chamber preference and in synaptic function that resemble the changes that occur normally through extended extinction training suggests that our manipulation causes interpretable changes in synaptic function, despite not fully replicating the endogenous dynamics of ChAT interneurons.

In addition, the finding that our manipulation has significant effects on extinction of a cocaine memory, but not on a food or a fear memory, suggests that our manipulation may have clinical relevance, regardless of not replicating the endogenous dynamics of ChAT interneurons. The preferential effect on a cocaine memory may be related to the fact that cocaine is known to generate extensive and persistent changes in circuit function in the NAc (Wolf and Tseng 2012; Tukey et al. 2013; Pascoli et al. 2014b; Bock et al. 2013; Hopf et al. 2010). Note that a preferential effect of the cholinergic system in modulating cocaine-related behavior is consistent with a prior study that compared the effect of muscarinic blockade in the NAc during cocaine versus food self-administration (Mark et al. 2006). However, it seems likely that a manipulation that more closely mimicked the endogenous dynamics in ChAT interneurons might cause changes in learning for natural rewards.

Connecting acetylcholine, plasticity, and cocaine-context extinction learning

Our results provide the most definitive three-way connection to date between cholinergic interneurons, plasticity and learning. Cholinergic antagonists have been shown to modulate long term depression and potentiation in dorsal striatal brain slices (Threlfell et al. 2012; Cachope et al. 2012), but if and how such modulation relates to behavior has not been shown. In addition, there have been several studies demonstrating a relationship between striatal acetylcholine and certain forms of learning (Brown et al. 2012; Witten et al. 2010; Aoki et al. 2015), but those studies did not investigate synaptic plasticity. Our study builds on this work by demonstrating that in vivo activity of ChAT interneurons is sufficient to modulate both learning and the associated synaptic plasticity onto MSNs. It is interesting that the changes in synaptic function and in cocaine chamber preference that are mediated by ChAT interneuron activation during a single extinction test mimic the changes in synaptic function and behavior that normally occur over the course of multiple extinction sessions (Figures 3 and 5). This finding strengthens the link between the changes in synaptic function mediated by ChAT interneurons and the changes in behavior, and suggests that the changes in synapses that are mediated by ChAT interneurons are causally related to the increase in extinction in these mice.

There have been a number of prior studies demonstrating acute changes in glutamatergic and GABAergic inputs onto MSNs as a result of ChAT interneuron activity (English et al. 2011; Oldenburg and Ding 2011; Witten et al. 2010; Nelson et al. 2014; Higley et al. 2011). The synaptic plasticity reported in this study are distinct from such acute changes in synaptic currents, given that those effects do not outlast the stimulation. Here we show that ChAT interneurons can mediate plasticity onto MSNs, and that this plasticity is present not only immediately after the manipulation, but that it persists for at least 24 hours (Figure 3F).

Conclusions

We found that activity in cholinergic interneurons in the NAc regulates the extinction of a cocaine-context association. Mechanistically, the activation of cholinergic interneurons during extinction results in a decrease in the frequency, but not amplitude, of mEPSCs in the output neurons of the NAc (MSNs), a change which persists for at least 24 hours. This cholinergic modulation of plasticity only occurs in the context of ongoing learning. Together, the data provides a strong three-way link between the in vivo activity of cholinergic interneurons, synaptic plasticity onto MSNs, and cocaine-context extinction learning.

Experimental procedures

Animals and surgeries

ChAT::IRES-Cre mice (JAX stock 006410: B6;129S6-Chattm2(cre)Lowl/J [RRID:IMSR_JAX:006410]) were maintained on a C57/BL6J background. All experimental and surgical protocols were approved by Princeton University IACUC to meet guidelines of the National Institutes of Health guide for the Care and Use of Laboratory Animals. During surgery, the NAc was infused bilaterally with either AAV5-EFIa-DIO-ChR2-eYFP, AAV5-EFIa-DIO-eNpHR3.0-eYFP or AAV5-EFIa-DIO-ChR2-eYFP in the medial NAc (AP 1.4mm, ML0.7-.9mm, DV −4.8mm; and 1.4mm, ML +/−0.7-.9mm, DV −4.8mm at 1 μl per site). Optic fibers (300 μm core diameter) were implanted at a ten degree angle to target the following coordinates: AP1.4mm, ML.7mm, DV3.9mm.

Cocaine conditioned place preference paradigm

On the first day, each mouse was placed in the central portal of the CPP chamber while connected to patch cables that were not emitting light and allowed to freely explore the entire apparatus for 15 minutes (pre-test). Day 2 and 3 consisted of conditioning. Each mouse was confined to one of the side chambers for 20 minutes in the morning and then again, to the opposite chamber in the afternoon for the same period of time. Subjects received i.p. injections of cocaine (15 mg/kg) before placement in one chamber or i.p. injections of an equal volume of saline before being placed in the other chamber (0.1mL). Day 4 consisted of the 1st extinction trial. Mice received optogenetic stimulation during the 15 minutes in which they had access to the two chamber apparatus. On day 5 and 12, mice were placed in the center chamber and allowed to freely explore the entire apparatus for 15 min (post-test). CPP experiments that did not involve cocaine were performed identically, except that the i.p. injections of cocaine were omitted and replaced with saline.

Real-time conditioned place preference

Each mouse was placed in the central portal of a CPP chamber and then allowed to freely explore the entire apparatus for 20 minutes. Light stimulation was paired to one chamber of the apparatus and remained on until the animal moved to the other side.

Food CPP

On the first day (pre-test), each mouse was placed in the center of the chamber. They were then allowed to freely explore the entire apparatus for 15 minutes. Days 2-6 consisted of food conditioning. Each mouse was confined to one of the side chambers in the morning and to the opposite chamber in the afternoon (30min sessions). Kellogg’s Fruit Loops were spread evenly throughout one chamber while the other chamber was empty. The order that the mice experienced the two chambers alternated each day. Day 7 consisted of the 1st extinction test. Mice received blue light during the 15 minutes in which they explored the two chamber apparatus (extinction test 1). On day 8 (extinction test 2) and 15 (extinction test 3), mice were placed in the center chamber and allowed to freely explore the entire apparatus for 15 min.

Fear Conditioning

To induce fear conditioning, mice were placed in the conditioning chamber for 120s and a foot shock (0.5 mA, 2 s) was delivered and then repeated a second time 120s after the first shock. 30s after the delivery of the second shock, mice were returned to their home cage. Freezing was scored throughout the testing trial by an experienced experimenter blind to the treatment group. During the two extinction tests, mice were placed in the original conditioning cage, and freezing was tested for 5 min over the following two days, with optogenetic stimulation on the first of the two tests.

Optogenetic manipulations

The power was 9-10 mW at the tip of the implanted optical fiber for blue light (447 nm) and 2-3mW for yellow light (590 nm). Unless stated otherwise, for all ChR2 behavioral experiments, blue light was delivered in a burst pattern, with 5 ms long pulses at 15 Hz for 2s interleaved with 2s light off periods. For all NpHR behavioral experiments, yellow light was delivered continuously.

Immunohistochemistry

After perfusion, brains were fixed overnight in 4% PFA and then equilibrated in 30% sucrose in PBS for 1-2 days. 40 μm coronal slices were sectioned. For immunohistochemistry, individual sections were washed in PBS and then incubated for 30 min in 0.3% Triton-X and 3% normal donkey serum (NDS). ChAT primary antibody (1:200; Millipore, product# AB144P [RRID:AB_2079751]) incubations were performed overnight at 4°C in 3% NDS/PBS. Sections were washed and the left to incubate in secondary antibodies conjugated to AlexaFluor586 for 3 hrs at room temperature (1:1000; Life Technologies, Product# A11057 [RRID:AB_10564097]). Following a 20 min incubation with DAPI (1:50,000), sections were washed and mounted.

Ex vivo electrophysiology

Mice received a transcardial perfusion of ice-cold carbogenated NMDG ACSF (see Supplementary Materials). After extraction, the brain was immersed in ice-cold NMDG ACSF for 2 minutes. Afterwards coronal slices (300um) were sectioned using a vibratome (VT1200s, Leica, Germany) and then incubated in NMDG ACSF at 34°C for 15 minutes. Slices were then transferred into a holding solution of HEPES ACSF (see Supplementary Materials). During whole cell recordings, slices were perfused with a recording ACSF solution (see Supplementary Materials). Infrared differential interference contrast–enhanced visual guidance was used to select neurons that were 3–4 cell layers below the surface of the slices, which were held at room temperature while the recording solution was delivered to slices via superfusion driven by peristaltic pump. The pipette series resistance was monitored throughout the experiments; if the series resistance changed by >20 % during the recording, the data were discarded. Whole-cell currents were filtered at 1 kHz and digitized and stored at 20 KHz (Clampex 9; MDS Analytical Technologies). All experiments were completed within 4 hours after slicing the brain. Detail of PPR, AMPAR/NMDAR current, and mEPSC measurements are in Supplemental Materials.

Statistical Analyses

Comparisons of behavioral preference across groups for the CPP tests was performed with a two-way repeated measures ANOVA, with subject and session as the factors. To compare the mEPSC frequency and amplitudes across conditions, a mixed effect linear regression was used after first transforming the skewed distributions so that the data better approximated a normal distribution (see Supplemental Materials).

Supplementary Material

Acknowledgments

We thank the entire Witten lab with their help and support with this project as well as SSH Wang for providing comments on the manuscript. This work was funded by the Pew, McKnight, NARSAD and Sloan Foundations and an NIH DP2 New Innovator Award (IBW) as well as an NSF Graduate Research Fellowship (JF).

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Author contributions

J.F., J.L., and I.W. conceived the project and designed the experiments. J.F. and J.L collected the data. J.F., J.L, and J.C. analyzed the data. J.F., J.L., and I.W. wrote the paper.

References

  1. Aoki Sho, Liu Andrew W., Zucca Aya, Zucca Stefano, Wickens Jeffery R. Role of Striatal Cholinergic Interneurons in Set-Shifting in the Rat. The Journal of Neuroscience: The Official Journal of the Society for Neuroscience. 2015;35(25):9424–31. doi: 10.1523/JNEUROSCI.0490-15.2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Atallah Hisham E., McCool Andrew D., Howe Mark W., Graybiel Ann M. Neurons in the Ventral Striatum Exhibit Cell-Type-Specific Representations of Outcome during Learning. Neuron. 2014a;82(5):1145–56. doi: 10.1016/j.neuron.2014.04.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Bock Roland, Roland Bock, Hoon Shin J, Kaplan Alanna R., Alice Dobi, Eric Markey, Kramer Paul F., et al. Strengthening the Accumbal Indirect Pathway Promotes Resilience to Compulsive Cocaine Use. Nature Neuroscience. 2013;16(5):632–38. doi: 10.1038/nn.3369. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Britt Jonathan P., Benaliouad Faiza, McDevitt Ross A., Stuber Garret D., Wise Roy A., Bonci Antonello. Synaptic and Behavioral Profile of Multiple Glutamatergic Inputs to the Nucleus Accumbens. Neuron. 2012;76(4):790–803. doi: 10.1016/j.neuron.2012.09.040. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Brown Matthew T. C., Tan Kelly R., O’Connor Eoin C., Nikonenko Irina, Muller Dominique, Lüscher Christian. Ventral Tegmental Area GABA Projections Pause Accumbal Cholinergic Interneurons to Enhance Associative Learning. Nature. 2012;492(7429):452–56. doi: 10.1038/nature11657. [DOI] [PubMed] [Google Scholar]
  6. Cachope Roger, Mateo Yolanda, Mathur Brian N., Irving James, Wang Hui-Ling, Morales Marisela, Lovinger David M., Cheer Joseph F. Selective Activation of Cholinergic Interneurons Enhances Accumbal Phasic Dopamine Release: Setting the Tone for Reward Processing. Cell Reports. 2012;2(1):33–41. doi: 10.1016/j.celrep.2012.05.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Conrad Kelly L., Tseng Kuei Y., Uejima Jamie L., Reimers Jeremy M., Heng Li-Jun, Shaham Yavin, Marinelli Michela, Wolf Marina E. Formation of Accumbens GluR2-Lacking AMPA Receptors Mediates Incubation of Cocaine Craving. Nature. 2008;454(7200):118–21. doi: 10.1038/nature06995. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. English Daniel F., Osvaldo Ibanez-Sandoval, Eran Stark, Fatuel Tecuapetla, György Buzsáki, Karl Deisseroth, Tepper James M., Tibor Koos. GABAergic Circuits Mediate the Reinforcement-Related Signals of Striatal Cholinergic Interneurons. Nature Neuroscience. 2011;15(1):123–30. doi: 10.1038/nn.2984. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Faust Thomas W., Assous Maxime, Shah Fulva, Tepper James M., Koós Tibor. Novel Fast Adapting Interneurons Mediate Cholinergic-Induced Fast GABAA Inhibitory Postsynaptic Currents in Striatal Spiny Neurons. The European Journal of Neuroscience. 2015;42(2):1764–74. doi: 10.1111/ejn.12915. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Fuchs Rita A., Allison Evans K, Ledford Christopher C., Parker Macon P., Case Jordan M., Mehta Ritu H., See Ronald E. The Role of the Dorsomedial Prefrontal Cortex, Basolateral Amygdala, and Dorsal Hippocampus in Contextual Reinstatement of Cocaine Seeking in Rats. Neuropsychopharmacology: Official Publication of the American College of Neuropsychopharmacology. 2005;30(2):296–309. doi: 10.1038/sj.npp.1300579. [DOI] [PubMed] [Google Scholar]
  11. Higley Michael J., Gittis Aryn H., Oldenburg Ian A., Balthasar Nina, Seal Rebecca P., Edwards Robert H., Lowell Bradford B., Kreitzer Anatol C., Sabatini Bernardo L. Cholinergic Interneurons Mediate Fast VGluT3-Dependent Glutamatergic Transmission in the Striatum. PloS One. 2011;6(4):e19155. doi: 10.1371/journal.pone.0019155. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Hopf FW, Stuber GD, Chen BT, Bonci A. Encyclopedia of Behavioral Neuroscience. 2010. Cellular Plasticity in Cocaine and Alcohol Addiction; pp. 236–41. [Google Scholar]
  13. Ilango Anton, Kesner Andrew J., Keller Kristine L., Stuber Garret D., Bonci Antonello, Ikemoto Satoshi. Similar Roles of Substantia Nigra and Ventral Tegmental Dopamine Neurons in Reward and Aversion. The Journal of Neuroscience: The Official Journal of the Society for Neuroscience. 2014;34(3):817–22. doi: 10.1523/JNEUROSCI.1703-13.2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Joshua Mati, Adler Avital, Mitelman Rea, Vaadia Eilon, Bergman Hagai. Midbrain Dopaminergic Neurons and Striatal Cholinergic Interneurons Encode the Difference between Reward and Aversive Events at Different Epochs of Probabilistic Classical Conditioning Trials. The Journal of Neuroscience: The Official Journal of the Society for Neuroscience. 2008a;28(45):11673–84. doi: 10.1523/JNEUROSCI.3839-08.2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Lee Brian R., Yao-Ying Ma, Huang Yanhua H., Xiusong Wang, Mami Otaka, Masago Ishikawa, Neumann Peter A., et al. Maturation of Silent Synapses in Amygdala-Accumbens Projection Contributes to Incubation of Cocaine Craving. Nature Neuroscience. 2013;16(11):1644–51. doi: 10.1038/nn.3533. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Loweth Jessica A., Scheyer Andrew F., Mike Milovanovic, LaCrosse Amber L., Eden Flores-Barrera, Werner Craig T., Xuan Li, et al. Synaptic Depression via mGluR1 Positive Allosteric Modulation Suppresses Cue-Induced Cocaine Craving. Nature Neuroscience. 2013;17(1):73–80. doi: 10.1038/nn.3590. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Lüscher Christian, Malenka Robert C. Drug-Evoked Synaptic Plasticity in Addiction: From Molecular Changes to Circuit Remodeling. Neuron. 2011;69(4):650–63. doi: 10.1016/j.neuron.2011.01.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Mark Gregory P., Kinney Anthony E., Grubb Michele C., Xiaoman Zhu, Finn Deborah A., Mader Sarah L., Paul Berger S, Bechtholt Anita J. Injection of Oxotremorine in Nucleus Accumbens Shell Reduces Cocaine but Not Food Self-Administration in Rats. Brain Research. 2006;1123(1):51–59. doi: 10.1016/j.brainres.2006.09.029. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Martin Miquel, Chen Billy T., Woodward Hopf F, Scott Bowers M, Bonci Antonello. Cocaine Self-Administration Selectively Abolishes LTD in the Core of the Nucleus Accumbens. Nature Neuroscience. 2006;9(7):868–69. doi: 10.1038/nn1713. [DOI] [PubMed] [Google Scholar]
  20. Ma Yao-Ying, Lee Brian R., Wang Xiusong, Guo Changyong, Liu Lei, Cui Ranji, Lan Yan, et al. Bidirectional Modulation of Incubation of Cocaine Craving by Silent Synapse-Based Remodeling of Prefrontal Cortex to Accumbens Projections. Neuron. 2014;83(6):1453–67. doi: 10.1016/j.neuron.2014.08.023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Morris Genela, Arkadir David, Nevet Alon, Vaadia Eilon, Bergman Hagai. Coincident but Distinct Messages of Midbrain Dopamine and Striatal Tonically Active Neurons. Neuron. 2004a;43(1):133–43. doi: 10.1016/j.neuron.2004.06.012. [DOI] [PubMed] [Google Scholar]
  22. Nelson Alexandra B., Hammack Nora, Yang Cindy F., Shah Nirao M., Seal Rebecca P., Kreitzer Anatol C. Striatal Cholinergic Interneurons Drive GABA Release from Dopamine Terminals. Neuron. 2014;82(1):63–70. doi: 10.1016/j.neuron.2014.01.023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Oldenburg Ian Antón, Ding Jun B. Cholinergic Modulation of Synaptic Integration and Dendritic Excitability in the Striatum. Current Opinion in Neurobiology. 2011;21(3):425–32. doi: 10.1016/j.conb.2011.04.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Pascoli Vincent, Terrier Jean, Espallergues Julie, Valjent Emmanuel, O’Connor Eoin Cornelius, Lüscher Christian. Contrasting Forms of Cocaine-Evoked Plasticity Control Components of Relapse. Nature. 2014a;509(7501):459–64. doi: 10.1038/nature13257. [DOI] [PubMed] [Google Scholar]
  25. Pascoli Vincent, Terrier Jean, Hiver Agnès, Lüscher Christian. Sufficiency of Mesolimbic Dopamine Neuron Stimulation for the Progression to Addiction. Neuron. 2015;88(5):1054–66. doi: 10.1016/j.neuron.2015.10.017. [DOI] [PubMed] [Google Scholar]
  26. Raimondo Joseph V., Kay Louise, Ellender Tommas J., Akerman Colin J. Optogenetic Silencing Strategies Differ in Their Effects on Inhibitory Synaptic Transmission. Nature Neuroscience. 2012;15(8):1102–4. doi: 10.1038/nn.3143. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Steinberg Elizabeth E., Ronald Keiflin, Boivin Josiah R., Witten Ilana B., Karl Deisseroth, Janak Patricia H. A Causal Link between Prediction Errors, Dopamine Neurons and Learning. Nature Neuroscience. 2013;16(7):966–73. doi: 10.1038/nn.3413. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Stuber Garret D., Woodward Hopf F, Tye Kay M., Chen Billy T., Bonci Antonello. Neuroplastic Alterations in the Limbic System Following Cocaine or Alcohol Exposure. Current Topics in Behavioral Neurosciences. 2010;3:3–27. doi: 10.1007/7854_2009_23. [DOI] [PubMed] [Google Scholar]
  29. Tepper James M., Paul Bolam J. Functional Diversity and Specificity of Neostriatal Interneurons. Current Opinion in Neurobiology. 2004;14(6):685–92. doi: 10.1016/j.conb.2004.10.003. [DOI] [PubMed] [Google Scholar]
  30. Threlfell Sarah, Sarah Threlfell, Tatjana Lalic, Platt Nicola J., Jennings Katie A., Karl Deisseroth, Cragg Stephanie J. Striatal Dopamine Release Is Triggered by Synchronized Activity in Cholinergic Interneurons. Neuron. 2012;75(1):58–64. doi: 10.1016/j.neuron.2012.04.038. [DOI] [PubMed] [Google Scholar]
  31. Torregrossa Mary M., Gordon Jessica, Taylor Jane R. Double Dissociation between the Anterior Cingulate Cortex and Nucleus Accumbens Core in Encoding the Context versus the Content of Pavlovian Cocaine Cue Extinction. The Journal of Neuroscience: The Official Journal of the Society for Neuroscience. 2013;33(19):8370–77. doi: 10.1523/JNEUROSCI.0489-13.2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Tsai Hsing-Chen, Zhang Feng, Adamantidis Antoine, Stuber Garret D., Bonci Antonello, de Lecea Luis, Deisseroth Karl. Phasic Firing in Dopaminergic Neurons Is Sufficient for Behavioral Conditioning. Science. 2009;324(5930):1080–84. doi: 10.1126/science.1168878. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Tukey David S., Ferreira Jainne M., Antoine Shannon O., D’amour James A., Ninan Ipe, Cabeza Soledad, Vaca de, Incontro Salvatore, et al. Sucrose Ingestion Induces Rapid AMPA Receptor Trafficking. The Journal of Neuroscience: The Official Journal of the Society for Neuroscience. 2013;33(14):6123–32. doi: 10.1523/JNEUROSCI.4806-12.2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Witten Ilana B., Lin Shih-Chun, Brodsky Matthew, Prakash Rohit, Diester Ilka, Anikeeva Polina, Gradinaru Viviana, Ramakrishnan Charu, Deisseroth Karl. Cholinergic Interneurons Control Local Circuit Activity and Cocaine Conditioning. Science. 2010;330(6011):1677–81. doi: 10.1126/science.1193771. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Witten Ilana B., Steinberg Elizabeth E., Lee Soo Yeun, Davidson Thomas J., Zalocusky Kelly A., Brodsky Matthew, Yizhar Ofer, et al. Recombinase-Driver Rat Lines: Tools, Techniques, and Optogenetic Application to Dopamine-Mediated Reinforcement. Neuron. 2011;72(5):721–33. doi: 10.1016/j.neuron.2011.10.028. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Wolf Marina E., Tseng Kuei Y. Calcium-Permeable AMPA Receptors in the VTA and Nucleus Accumbens after Cocaine Exposure: When, How, and Why? Frontiers in Molecular Neuroscience. 2012 Jun;5:72. doi: 10.3389/fnmol.2012.00072. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Zhou FM, Liang Y, Dani JA. Endogenous Nicotinic Cholinergic Activity Regulates Dopamine Release in the Striatum. Nature Neuroscience. 2001;4(12):1224–29. doi: 10.1038/nn769. [DOI] [PubMed] [Google Scholar]
  38. Zhou Fu-Ming, Wilson Charles J., Dani John A. Cholinergic Interneuron Characteristics and Nicotinic Properties in the Striatum. Journal of Neurobiology. 2002;53(4):590–605. doi: 10.1002/neu.10150. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

NIHMS788935-supplement.pdf (1.4MB, pdf)

RESOURCES