Nicotine Modifies Corticostriatal Plasticity and Amphetamine Rewarding Behaviors in Mice

Abstract Corticostriatal signaling participates in sensitized responses to drugs of abuse, where short-term increases in dopamine availability provoke persistent, yet reversible, changes in glutamate release. Prior studies in mice show that amphetamine withdrawal promotes a chronic presynaptic depression in glutamate release, whereas an amphetamine challenge reverses this depression by potentiating corticostriatal activity in direct pathway medium spiny neurons. This synaptic plasticity promotes corticostriatal activity and locomotor sensitization through upstream changes in the activity of tonically active cholinergic interneurons (ChIs). We used a model of operant drug-taking behaviors, in which mice self-administered amphetamine through an in-dwelling catheter. Mice acquired amphetamine self-administration under fixed and increasing schedules of reinforcement. Following a period of abstinence, we determined whether nicotinic acetylcholine receptors modified drug-seeking behavior and associated alterations in ChI firing and corticostriatal activity. Mice responding to conditioned reinforcement showed reduced ChI and corticostriatal activity ex vivo, which paradoxically increased following an amphetamine challenge. Nicotine, in a concentration that increases Ca2+ influx and desensitizes α4β2*-type nicotinic receptors, reduced amphetamine-seeking behaviors following abstinence and amphetamine-induced locomotor sensitization. Nicotine blocked the depression of ChI firing and corticostriatal activity and the potentiating response to an amphetamine challenge. Together, these results demonstrate that nicotine reduces reward-associated behaviors following repeated amphetamine and modifies the changes in ChIs firing and corticostriatal activity. By returning glutamatergic activity in amphetamine self-administering mice to a more stable and normalized state, nicotine limits the depression of striatal activity in withdrawal and the increase in activity following abstinence and a subsequent drug challenge.

Chronic psychostimulant use alters corticostriatal signaling ). Withdrawal induces a further adaptation by promoting a chronic presynaptic depression (CPD) in glutamate release from cortical projections within the dorsal striatum that disrupts normal DA filtering of corticostriatal activity (Bamford et al., 2008;Wang et al., 2013a). A drug challenge reverses CPD by increasing glutamate availability through a process called paradoxical presynaptic potentiation (PPP), when DA exerts an excitatory, rather than an inhibitory, effect on the corticostriatal pathway. By normalizing corticostriatal activity in withdrawal, PPP may provide a mechanism by which drug re-administration promotes physiological and behavioral stability, a feature supported in models of addiction (Ahmed and Koob, 2005).
The dorsal striatum is a key mediator of many of the reward-related behaviors. This structure is associated with the formation of compulsive drug-seeking habits (Yin et al., 2004;Everitt andRobbins, 2005, 2015), locomotor sensitization (Bamford et al., 2008), cue-dependent and reversal learning, cognitive flexibility (Quintana et al., 2012;Bradfield et al., 2013;Wang et al., 2013a), and sensorimotor conditioning . The changes in synaptic glutamate that occur during amphetamine withdrawal are analogous to measurements of extrasynaptic glutamate in the nucleus accumbens (NAc) that follow repeated cocaine (Pierce et al., 1996;McFarland et al., 2003). However, withdrawal from cocaine, but not amphetamine, modifies glutamater-gic signaling in the ventral and dorsal striatum by upregulating postsynaptic AMPA and NMDA receptor surface expression (Nelson et al., 2009). This suggests that contributions from upstream circuitry differentially regulate corticostriatal activity following cocaine or amphetamine withdrawal.
Within the dorsal striatum, ChIs play a critical role in these behaviors and contribute toward CPD and PPP (Bamford et al., 2008;Wang et al., 2013a). Through an extensive arborization network, ChIs exert a strong influence on striatal information processing (Wilson et al., 1990). Although morphologically similar across striatal regions in the rodent brain (Gonzales and Smith, 2015), ChIs show distinct region-specific responses to cocaine (Benhamou et al., 2014). Such heterogeneity may be a result of restricted DA and glutamate cotransmission (Chuhma et al., 2004). For example, in the dorsolateral striatum DA induces a pause in ChI firing followed by a rebound in activity. DA increases ChI activity in the NAc shell, but reduces ChI firing in the NAc core. A single exposure to amphetamine increases this heterogeneity across regions , suggesting that local DA release induces region-specific striatal responses.
ACh acts at nicotinic and muscarinic receptors in the striatum to regulate DA (Zhang and Sulzer, 2004), glutamate (McGehee et al., 1995), and ACh availability in a region-specific manner (Zhou et al., 2001;Threlfell and Cragg, 2011). Nicotinic receptors, through their profile of activation and desensitization, modulate the probability of DA and glutamate release to affect psychostimulantinduced reward behaviors (Hansen and Mark, 2007;Bamford et al., 2008;Wang et al., 2013a;Li et al., 2014). We propose that the ChIs in the dorsolateral striatum, and their repertoire of nicotinic receptors, are a key mediator of the striatal response to amphetamine-induced DA and glutamatergic plasticity.
Using a model of contingent amphetamine exposure in mice, we have examined striatal adaptations induced by amphetamine. This model pairs single or multiple lever press with a drug infusion through an indwelling jugular catheter to generate a behavioral profile of amphetamine use. We then examined the electrophysiological profile of the effects of chronic amphetamine on the ChIs in the dorsolateral striatum, an important hub of habitual drugseeking behaviors. Finally, we probed the involvement of the ‫-ء2␤4␣‬ and ␣7‫-ء‬type nicotinic acetylcholine receptor subtypes in blocking the presynaptic corticostriatal plasticity induced by amphetamine.

Animals
All animal procedures were performed in accordance with the authors' university animal care committee's regulations. Experiments started when male and female C57BL/6J mice (Jackson Laboratories) were 2-3 months old and completed by 6 months. Mice were housed together in a modified specific pathogen free vivarium with a 12 h light/dark cycle with ad libitum access to food and water. For some mice, a catheter (0.2 mm inner diameter, 0.4 mm outer diameter; Camcaths) was implanted in the right jugular vein; the mice were housed singly thereafter. The catheters were maintained by flushing daily with heparinized 0.9% saline (Thomsen and Caine, 2005;Cui et al., 2014) with Timentin (120 mg/ml) added for the first 7 d. We anesthetized mice with Beuthanasia (320 mg/kg, i.p.) or with ketamine (650 mg/kg, i.p.) and xylazine (44 mg/kg, i.p.) prior to euthanasia.

Operant intravenous amphetamine selfadministration
Stage 1: sucrose pretraining. After 2 d of a graded food restriction protocol with unrestricted access to water, all mice obtained and maintained a 10% reduction in body weight. A further 2 d of food restriction followed but with a 20% sucrose solution provided for 8 h each day. The mice habituated to the testing room containing the operant chambers during on these days (Med Associates). Over the following 5 d, they were trained to enter the magazine (2 d) and then press either of two operant levers (3 d) to obtain a 20% sucrose reward (delivered to the magazine) to a maximum of 30 reinforcers per session.
Stage 2: amphetamine self-administration. Food and water were provided ad libitum after the last sucrose training session and for the remainder of the experiment. Intravenous jugular catheters were implanted the day after the last sucrose session.
A: acquisition. Following 3 d of recovery, the mice were returned to the operant boxes for intravenous amphetamine (0.05 mg/kg) self-administration. The preferred lever during sucrose pretraining was designated as the active lever for amphetamine self-administration. A response on the designated active lever resulted in an amphetamine reinforcer (0.67 l/g body weight) and a cue (house light extinguished for 30 s), during which no further drug could be obtained. A maximum of 50 reinforcers was allowed during the 2 h session and the session concluded when either parameter was obtained. Mice underwent a minimum of 10 d of acquisition training at a fixed ratio of 1 (FR1), during which one lever press resulted in one reinforcer. Mice with Յ20% variation and a minimum of 20 earned reinforcers over the preceding 3 d were advanced to the next stage, FR2, where two lever presses resulted in one reinforcer. Mice trained at this level for a minimum of 3 d until they obtained stable rates of reinforcers. Thereafter, they proceeded to FR5 for a minimum of 3 d, where five presses resulted in one reinforcer. Mice with Յ20% variation in earned reinforcers then progressed to the next phase. We tested catheter patency by infusing 1% propofol (20 l) after FR1 and again after FR3 and FR5. We removed mice from the study if they failed to become transiently limp during the patency test.
B: amphetamine challenge following incubation. As the measured presynaptic adaptations are stable for 140 d following amphetamine injections (Bamford et al., 2008) and maintaining catheter patency in mice is technically challenging, a within-subject design was used to assess the effect of nicotine on amphetamine reward-seeking behaviors following 7 or 14 d of abstinence. Mice were randomly assigned to either an amphetamine group or an amphetamine with nicotine group, received a noncontingent amphetamine injection alone (1 mg/kg, i.p.) or amphetamine (1 mg/kg, i.p.) with nicotine (0.25 mg/kg, i.p.), and were then placed immediately in the same operant chamber as used during acquisition. Under an FR1 schedule of reinforcement, the number of active lever presses, inactive lever presses, and reinforcers (or cue presentations) were recorded for 30 min. After 7 d, these same reward-seeking behaviors were recorded in mice receiving the alternative treatment. Electrophysiology experiments were conducted within 30 d of the final FR5 session.

Locomotor sensitization
Locomotor responses were measured using animal activity monitor cages (San Diego Instruments). Computer monitoring of four infrared beams, separated by 8.8 cm that cross the width of each chamber, recorded the number of times mice broke each beam. We measured locomotor activity in ambulations (2 consecutive beam interruptions) summated over 5 min intervals. On each test day, animals acclimated to individual activity chambers for 90 min to allow the animal to become accustomed to its behavioral cage before subsequent injections of either amphetamine (2 mg/kg, i.p.) or saline (10 l/g, i.p.). Following each injection, the mice were placed back into their respective activity chamber and their ambulations were recorded for 90 min. To separate the effects of novelty from the pharmacological effects of the drug, animals were acclimated to the locomotor chambers and injected with saline on experiment days 1 and 2.
Cell-attached recordings from ChIs and whole-cell recordings from MSNs in the dorsal striatum were obtained in voltage-clamp mode. MSNs were clamped at Ϫ70 mV.
Cells were visualized in slices with the aid of infrared videomicroscopy coupled with differential interference contrast optics. ChIs were identified by size (ϳ18 -25 m) and repetitive firing in gap-free mode. Cell identification was confirmed by measuring passive and active membrane properties in whole-cell configuration (Wang et al., 2013a) and by labelling with 1% biocytin, according to published protocols (Joshi et al., 2009). The pipette internal solution contained the following (in mM): 119 KMeSO 4 , 1 MgCl 2 , 0.1 CaCl 2 , 10 HEPES, 1 EGTA, 12 phosphocreatine, 2 Na 2 ATP, and 0.7 Na 2 GTP, pH 7.2, 280 -300 mOsm/L (Bennett and Wilson, 1999;Maurice et al., 2004). Currents were Bessel filtered at 2 kHz and were allowed to stabilize for 5 min after achieving a seal resistance Ͼ1 G⍀. Cells were removed from analysis if the seal resistance fell Ͻ1 G⍀ or if the firing rate was Ͻ0.3 Hz or changed by Ͼ20% during the baseline (Bennett and Wilson, 1999).
EPSCs were evoked by electrical stimulation of the deep cortical layers of the motor cortex using a twisted tungsten bipolar electrode at stimulation strengths adjusted to 1.5ϫ threshold (0.6Ϯ0.2 mA). Paired current pulses of 200 s in duration were delivered at 20 Hz every 30 s. eEPSC currents were Bessel filtered at 1 kHz. The paired-pulse ratio (PPR) was calculated by dividing the amplitude of the second pulse by that of the first pulse and then multiplying by 100. Cells demonstrating eEPSCs with variable latencies or prolonged durations suggesting polysynaptic responses were rejected. mEPSCs were Bessel filtered at 4 kHz and recorded in gap free mode. The Na ϩ channel antagonist tetrodotoxin (1 M) was added to block spontaneous cortically-derived action potentials and isolate presynaptic terminal activity. Currents were digitized at 50 s using Digidata 1440A data acquisition and pClamp10.2 software (Molecular Devices). Cell activity was analyzed with Clampfit (Molecular Devices) and Mini Analysis (Synaptosoft). Chemicals and ligands were obtained from Sigma-Aldrich.

Spike analysis
The power spectra for recorded spike trains were calculated as the magnitude-squared of the Fourier transform (St-Pierre et al., 2014) using custom software designed in MATLAB (MathWorks). The nonuniform sampling rate of spike time data required a method for resampling at a uniform sampling rate. The time of each spike was rounded to the nearest thousandth to guarantee an equivalent time in a resampled train at 1000 Hz (⌬tϭ0.001 s) in which corresponding time instances were set to 1 in the new train and all others to 0. A sampling rate of 1000 Hz provided a high Nyquist frequency at 500 Hz, while limiting aliasing contribution for frequency bins Ͻ200 Hz (Bair et al., 1994).
Frequency spectra were computed by a Fast Fourier transform (FFT) according to Welch's method using a division of 15 windows with 50% overlap (Table 1; Welch, 1967). A Hanning window was used to improve frequency accuracy and to reduce spectral leakage (Wickramarachi, 2003). Window overlap preserved spikes located near the tapered ends of the Hanning windows. The spectrum of each window was squared then normalized to its average power to preserve the relative peaks and dips at the lower frequencies of interest when averaging (Bair et al., 1994). We applied these processing operations at the Welch window level to improve spectral accuracy for each cell. Averaging a large number of windows decreased spectral variance at the expense of frequency resolution (Jokinen et al., 2000), which was compensated by greater window length of overlapping windows than that of non-overlapping windows. To reduce the direct current spike, we subtracted windows by their means to create a zeromean spike train. Peak frequencies were the center of the bin with the maximum integral (Table 2). Bins started at each calculated frequency and extended out for a total of 10 points in each window. The peak frequencies of each cell were normalized to their corresponding power, which roughly approximated the peaks seen in the frequency spectrum. Frequency distributions were determined by creating a probability distribution of the 1/interspike interval (ISI) frequencies (Table 3). We averaged the distributions for each spike train in an experimental group to create an average distribution of that group and then calculated the peak frequencies using the same method used for the frequency spectra.
We used the Robust Gaussian Surprise (RGS; Ko et al., 2012) methods to analyze the burst and pause firing patterns in ChIs. The RGS method can accurately detect small bursts, small pauses, and strings of pause activity in individual ChIs (Tables 4 and 5). Unlike the Poisson Surprise method, which is limited by its assumption of a Poisson process (Legéndy and Salcman, 1985), and the Rank Surprise method, which does not accurately detect small bursts, the RGS method exhibits a robust adaptability to varying firing rates (Ko et al., 2012). It also provides many facets for statistical significance including both burst and pause information. A p value of 0.05 was used in the calculation of the central location of all ISI distributions and in the Bonferroni correction of the results. This method's robustness is due to its normalization algorithm, which takes the base 10 logarithm of all the ISIs and locates the central location through the median and median absolute deviation of the distribution of a standard window length around each logarithmic ISI. Then, each central location was subtracted from its corresponding logarithmic ISI to form a normalized log 10 ISI train. The median and median average deviation of the distribution of normalized ISIs were used to generate a cumulative Gaussian probability distribution, which was then used to determine surprise values for burst and pause seeds. Burst seeds were set as normalized log 10 ISIs less than Ϫ2.58 times the median average deviation, whereas pause seeds were set as normalized log 10 ISIs Ͼϩ2.58 times the median average deviation. Normalized log 10 ISIs (Ͼ0.05 s) in front or behind of the seeds were added if they decreased the likelihood, assuming a Gaussian distribution with mean and SD taken from the normalized log 10 ISI distribution, of the occurrence of the burst or pause string according to the cumulative probability distribution. This method's precise selection of spikes in each burst or pause string event by only including spikes that decrease the probability of the occurrence of the event, as well as the normalization that prevents the stringing of multiple small bursts in regions contribute to  (1):dt:Spikes(end)ϩdt); %New uniformly sampled times nTime(2,:) ϭ 0; %Initialize spike train %Find equivalent times and set value at that time to 1 nTime(2,ismember(round(nTime(1,:).‫/1(ء‬dt)).‫ء‬dt, round(Spikes.‫/1(ء‬dt)).‫ء‬dt)) . . . ϭ 1; %Spike train equals binary train (1,0) spikeTrain ϭ nTime(2,:); Nϭlength(spikeTrain); %Length of spike train fs ϭ 1/dt; %Sampling rate of new spike train L ϭ floor(N/(numBins-pOverlap/100ϫnumBinsϩpOverlap/100)); %Number of points in %length of window Overlap_N ϭ floor(LϫpOverlap/100); %Number of points in each overlapping section %Set endpoints of the windows windowEndpoints ϭ [linspace(1,1ϩ(L-Overlap_N)ϫ(numBins-1),numBins); . . . L:L-Overlap_N:length(spikeTrain)ϩ(L-Overlap_N)/2]; %Fix error from floor due to percent input spikeTrain ϭ [spikeTrain zeros(1,windowEndpoints(2,end)-length(spikeTrain))]; %Initalize windows windows ϭ zeros(numBins, 1); %Compute fft for all windows for i ϭ 1:numBins %Take current window from the spike train currTrain ϭ [spikeTrain(windowEndpoints(1,i):windowEndpoints(2,i)) . . . zeros(1,Padding)]; %Can change window here by replacing line 71 with %options below: windowed ϭ hanning(length(currTrain))'.‫(ء‬currTrain); %windowed ϭ blackman(length(currTrain))'.‫(ء‬currTrain); %windowed ϭ flattopwin(length(currTrain))'.‫(ء‬currTrain); %windowed ϭ hamming(length(currTrain))'.‫(ء‬currTrain); %Take zero-mean, fft, and magnitude-squared windows(i,1:length(windowed)) ϭ abs(fft(windowed-mean(windowed))).^2; %Normalize to the mean windows(i,:) ϭ windows(i,:)./mean(windows(i,:)); end %Frequencies up to the nyquist f ϭ linspace(0,fs/2,floor(length(windows)/2))'; %Average results of windows of corresponding spectra values spec ϭ mean(windows(:,1:floor(length(windows)/2)))'; end ; % %% PEAKDETECT %The number of points to extend forward or backward %Step size of the distribution dx ϭ A(2)-A(1); %Initialize integral array integral ϭ zeros(length(B)-binsizeϩ1,1); %Calculate the integrals for i ϭ 1:length(B)-binsize ϩ 1 %Take the sum of the x values multiplied by y values in the bin integral(i) ϭ sum(dxϫB(i:iϩbinsize-1)); end %Find the index of the center of the bin with the max integral j ϭ find(integralϭϭmax(integral))ϩround(binsize/2); %Find the value of the max integral peak ϭ A(j); %Return the average dist around the max distribution value dist ϭ mean(B(j-round(binsize/2):jϩround(binsize/2))); end its accurate and robust performance. Patterns in bursts and pauses were determined using variable thresholds to connect these events (Table 6). All algorithms were implemented in MATLAB.
[hits,ia,ib] ϭ intersect(BSPi(:,2),SPBi(:,1)); %Concatenate the common BSP and SPB patterns into a 3 event pattern.   Normal distribution Paired t test 0.7 Fig. 4F. Intra-burst frequency in self-administering mice (vehicle vs amphetamine) Normal distribution Paired t test 0.1 Fig. 4F. Intra-burst frequency in self-administering mice (amphetamine vs amphetamine with nicotine) Normal distribution Paired t test 0.07 Fig. 4F. Intra-burst frequency in self-administering mice (vehicle vs amphetamine with nicotine) Normal distribution Paired t test 0.1 Fig. 4F allowed, after 10 d at FR1, during which one lever press resulted in one reinforcer. We designated this group as amphetamine self-administering mice. Of the 10 remaining mice, five were removed due to occluded catheters. Five additional mice that had not obtained more than an average of 20 reinforcers during the last three sessions of FR1 were designated as nonresponding mice and excluded from the rest of the behavioral study. The 10 amphetamine self-administering mice progressed from FR1 to FR2 and FR5 schedules of reinforcement. A further 10 saline-treated mice underwent the identical procedure, but received saline instead of amphetamine. Salinetreated mice were included in all parts of the acquisition phase, and 5 of 10 underwent abstinence and a saline challenge. Figure 1B-E demonstrates the different measures recorded during the acquisition phase. The number of active lever presses by the amphetamine self-administering mice increased from 45 to 76 during the first 3 d of FR1 and then plateaued at 60 -70 presses for the remaining 7 d of FR1 (Fig. 1B). Thereafter, the number of active lever presses increased to ϳ120 at FR2 and to ϳ240 at FR5. This increase in the number of active presses required to receive the same number of infusions demonstrates the motivation of this group to receive amphetamine.
The active and inactive lever pressing of saline-treated and nonresponding mice was minimal (Fig. 1C). Likewise, amphetamine self-administering mice did not often press the inactive lever, but their active lever presses increased with the fixed ratio of reinforcers (saline-treated vs amphetamine self-administering active lever presses over time; F (15,120) ϭ26.55, pϽ0.0001 a , two-way ANOVA; Fig.  1B). The percentage ratio of active to total (active ϩ inactive) lever presses assessed the ability to discriminate the effect of pressing the active from inactive lever. The amphetamine self-administering mice achieved an average of 90Ϯ3, 95Ϯ1, and 95Ϯ1% active/total lever presses during FR1, FR2, and FR5, respectively (Fig. 1D). This was greater than the active/total lever ratio in salinetreated mice (F (1,7) ϭ149, pϽ0.0001 b , two-way ANOVA) who achieved 55Ϯ3, 65Ϯ6, and 75Ϯ3% during FR1, FR2, and FR5. The amphetamine self-administering mice obtained a greater number of drug reinforcers than salinetreated mice (F (15,120) ϭ3.837, pϽ0.0001 c , two-way ANOVA) and consistently reached 50, the maximum allowed, during the 2 h session of each schedule (Fig. 1E).
We assessed the effect of abstinence on drug-seeking behavior. After 7 d of forced abstinence, we injected the amphetamine self-administering mice with either amphet-amine alone (1 mg/kg, i.p.) or with amphetamine (1 mg/kg, i.p.) and nicotine (0.25 mg/kg, i.p.). We then placed the mice in the same operant chamber as used during the acquisition phase for 30 min. This amphetamine challenge resulted in a slight incubation effect when compared with the first 30 min period of the final FR5 session (76Ϯ4 vs 63Ϯ4 active lever presses during the incubation vs FR5 session, respectively; pϭ0.01 d , tϭ3.7, dfϭ6, paired t test). Nicotine coadministration with amphetamine reduced both the number of active lever presses (Ϫ44Ϯ14%; pϭ0.02 e , tϭ3.04, dfϭ5, paired t test) and the number of reinforcing cues obtained (Ϫ37Ϯ9%; pϭ0.02 f , tϭ3.5, dfϭ5, paired t test; Fig. 1F-I), resulting in a similar profile of drug-seeking responses as seen in salinetreated mice challenged with a saline injection. Nicotine treatment did not alter lever discrimination, as neither the number of inactive lever presses (pϭ0.07 g , tϭ2.2, dfϭ6, paired t test) nor the percentage ratio of active/total lever presses changed (pϭ0.15 h , tϭ1.7, dfϭ5, paired t test; Fig.  1G,H).
Through their widespread dendritic and axonal fields, ChIs modulate corticostriatal excitability through changes in the frequency of their tonic firing (Wilson et al., 1990). We developed computer algorithms to examine the power and relative distribution of individual firing frequencies in ChIs from saline-treated and amphetamine selfadministering mice (Tables 1-6). The Welch's power spectral density estimate of ChI activity from salinetreated mice revealed a unimodal distribution, with a single local maximum of 3.68Ϯ0.4 Hz centered on the average frequency of 3.58Ϯ0.58 Hz (Fig. 2B). Most (ϳ90%) individual ChIs from amphetamine self-ad-  Figure 1. Mice self-administer amphetamine. A, Acquisition of amphetamine self-administration: Following sucrose pretraining, mice were trained to self-administer amphetamine under increasing schedules of reinforcement. After 10 d on FR1, mice that had not achieved stable responding (Ͻ20 reinforcers during the last 3 d of FR1) were designated as nonresponding mice and were excluded from the remainder of the behavioral study. Thereafter the saline-treated mice and amphetamine self-administering mice underwent a further 6 d on FR2 and FR5 schedules of reinforcement. Following this acquisition phase, the amphetamine self-administering mice underwent 7 d of abstinence. This was followed by an amphetamine challenge in the same operant boxes and protocol as used during intravenous self-administration to assess the incubation of drug seeking behaviors. The effect of nicotine on this amphetamine challenge was also assessed. B, In comparison with saline-treated mice, amphetamine self-administering mice increased the number of active lever presses. C, The number of inactive lever presses was similar across all groups. D, Compared with saline-treated mice, the amphetamine self-administering mice showed drug-lever association, as assessed by the ratio of active to total (inactive ϩ active) lever presses (%). E, During FR1, FR2, and FR5 schedules of reinforcement, amphetamine self-administering mice achieved a greater number of reinforcers than mice treated with saline, often reaching 50, the maximum allowed during the 2 h session. F, Following forced abstinence, nicotine reduced the total number of active lever presses observed after an amphetamine challenge (G) without ministering mice also demonstrated a unimodal frequency distribution, but their averaged contributions appeared bimodal, with the uppermost peak frequency centered on the average frequency of ChIs from saline-treated mice, whereas the second maxima was centered on lower frequencies ϳ1 Hz (Fig. 2B-D). The peak frequencies identified by the power spectral density estimate approximated the frequency distributions derived from their ISIs and showed that ChI spiking in amphetamine selfadministering mice was lower than saline-treated mice due to an increased distribution of low-frequency activity between 0.5 and 1.5 Hz (Fig. 2C,D).

An amphetamine challenge increases tonic ChI firing in amphetamine self-administering mice
To determine whether an amphetamine challenge modified ChIs activity ex vivo, amphetamine ( Fig. 2E-G). Spectral analysis showed that amphetamine reduced the average peak frequency in ChIs from saline-treated mice by augmenting low-frequency activity between 0.5 Hz and 1.5 Hz (Fig.  2H). In ChIs from amphetamine self-administering mice, an amphetamine challenge increased activity by diminishing the contribution of those same frequencies (Fig. 2I). Treatment with amphetamine did not reorganize the bimodal frequency distribution of ChI activity in amphetamine self-administering mice, but modified the skewed, fat-tailed firing frequency to better approximate the spectra of ChIs from saline-treated mice (Fig. 2E, F, and J).

Nicotine opposes the change in tonic ChI firing caused by an amphetamine challenge
Nicotine attenuated amphetamine-induced drug-seeking behaviors (Fig. 1), and we determined whether nicotine (100 nM) ex vivo, in a concentration that approximates serum levels in smokers (Dani and Harris, 2005), modifies the effect of an amphetamine challenge on tonic ChI firing. When administered with amphetamine in vitro, nicotine blocked amphetamine inhibition of ChI activity in saline-treated mice (2.64Ϯ0.53 Hz in vehicle vs 2.61Ϯ0.31 Hz with amphetamine and nicotine; pϭ0.3 p , paired t test) by enhancing frequencies between 2 and 3 Hz (Fig. 2E-H). Conversely, in ChIs from amphetamine self-administering mice, nicotine suppressed amphetamine-mediated excitation (1.82Ϯ0.27 Hz in vehicle vs 2.16Ϯ0.51 Hz with amphetamine and nicotine; pϭ0.3 q , paired t test) by diminishing these same frequencies ( Fig.  2E-G,I). Nicotine modified higher frequencies and blocked the change in tonic firing caused by an amphetamine challenge.

Amphetamine self-administration reduces burst firing in ChIs
Reward-reporting stimuli increase the activity of dopaminergic neurons and correlate with a change in ChI activity throughout the striatum . These conditioned responses modify the tonic activity of ChIs by producing burst, pause, and rebound burst-firing patterns of various lengths that modify downstream network activity (Graybiel et al., 1994;Fig. 3A). Thus, minute, as well as volume-transmitted changes in ACh availability at nicotinic receptors may encourage gaiting at corticostriatal synapses. Electrophysiological studies have shown that although tonic activity may remain stable, changes in the burst and pause activity of DA neurons may play a role in degenerative disease (Lobb, 2014;Lobb and Jaeger, 2015) and could be altered by psychostimulant exposure. Indeed, ChIs show distinct region-specific responses to cocaine (Benhamou et al., 2014), but alterations in ChI burst or pause activity that accompany amphetamine self-administration are unclear. We used the RGS method to detect and quantify how amphetamine self-administration modifies unique burst and pause patterns in individual ChIs. The RGS method has been successful in identifying changes in bursting patterns of DA neurons (Branch et al., 2013;Lobb and Jaeger, 2015), as it exhibits a robust adaptability to varying firing rates through two steps: normalization of local ISI distributions and iterative addition of normalized ISIs, based on the p value obtained from global Gaussian statistics of the whole spike train, to burst and pause strings. (Ko et al., 2012). RGS thresholds for burst and pause activity were determined in ChIs from salinetreated (nϭ13 cells) and amphetamine self-administering mice (nϭ12 cells) using Gaussian probability distribution of the ISIs for each cell (see Materials and Methods; continued altering the inactive lever presses. H, There was no effect of nicotine on lever discrimination, as shown by the ratio of active/total lever presses. I, The administration of nicotine with amphetamine reduced the number of reward-associated cues. The effect of a saline injection on saline-treated mice is shown for comparison with amphetamine self-administering mice. For all panels, $ pϽ0.05, paired t test.  Fig. 3B). To determine whether amphetamine selfadministration modified the cycle and extent of burst activity, we expanded our burst-firing analysis to measure any change in the burstiness of the cell, defined by its intraburst frequency, burst length, and the percentage of time spent bursting. We also measured discrete pauses (single pauses in activity without intervening spikes), pause strings (contiguous discrete pauses with intermit-tent low-frequency spikes), intrapause frequency, pause length, and the percentage of time spent pausing. Compared to saline-treated mice, ChIs from amphetamine self-administering mice had a lower frequency of discrete bursts (1.65Ϯ0.19 bursts/min vs 1.12Ϯ0.15 bursts/min for amphetamine self-administering mice; pϭ0.03 r , Student's t test) and a reduction in their burstiness; the average burst length (612Ϯ63 ms vs 447Ϯ42 ms Figure 2. Amphetamine self-administration modifies ChI firing. A, Representative traces of cell-attached recordings in ChIs from saline-treated and amphetamine self-administering mice (left). The average baseline firing frequency (right) was lower in ChIs from amphetamine self-administering mice; This is shown in box-and-whisker plots where the median is shown as a solid line, the mean value is dotted red, the ends of the box indicate the 25 th and 75th percentiles, and the bars indicate the 10 th and 90 th percentiles. For all panels, $ pϽ0.05, Student's t test; ‫ء‬pϽ0.05 and ‫ءء‬pϽ0.01, paired t test. B, The normalized power distribution (left) and the average peak frequency (right) show prominent low-frequency activity in ChIs from amphetamine self-administering mice. C, The ISI distribution for ChIs from saline-treated and amphetamine self-administering mice (arrow). D, Frequency distributions (left) and their average peaks (right) show the prominent low-frequency distribution (1/ISI) of ChI firing from amphetamine self-administering mice (arrow). E, MeanϮSE frequencies of ChIs from saline-treated and (F) amphetamine self-administering mice in response to amphetamine (Amph) or amphetamine with nicotine. G, Normalized firing frequencies over time for the experiments shown in E and F. H, The frequency distribution (left) and the average peak frequency (right) of ChIs from saline-treated mice. Amphetamine produced a small increase in activity at 0.5 Hz (arrow). Nicotine blocked the inhibition caused by amphetamine and increased activity at 2.5 Hz (double arrow). I, Frequency distribution (left) and the average peak frequency (right) of ChIs from amphetamine self-administering mice show that amphetamine reduced 1-2 Hz activity (arrow). The addition of nicotine moderated the potentiating effect of amphetamine by reducing activity between 2 and 3 Hz, while increasing 3-4 Hz activity (double arrows). J, When exposed to amphetamine, the bimodal frequency distribution (left) and average peak frequency (right) of ChIs from amphetamine self-administering mice converged with the unimodal frequency distribution of ChIs from saline-treated mice. Saline-treated mice Self-administering mice G I H Figure 3. Amphetamine self-administration alters burst firing. A, The representative traces from a cell-attached recording of ChIs from saline-treated and amphetamine self-administering mice show burst, pause, and pause-string activity. The pause response of ChIs begins with an initial depolarizing phase followed by a pause in spike firing and ensuing rebound excitation. In this example, the ChI from a self-administering mouse lacks rebound excitation. B, The representative normalized log 10 ISI distribution demonstrates the RGS method for determining burst and pause thresholds for each ChI. The log 10 ISI values of burst (Ϫ0.324) and pause thresholds (0.311) for this cell correspond to the top 0.5 percentile and bottom 0.5 percentile. C, The average frequency (bursts/min) of discrete bursts was lower in ChIs from amphetamine self-administering mice. For all panels, $pϽ0.05, $$pϽ0.01, Student's t test. D, The histogram (left) compares the length of each burst with the frequency of spikes contained within that burst. The average intraburst frequency (spikes/s), average burst length (s), and the average percentage of time spent bursting determines the cell's burstiness. Compared to saline-treated mice, ChIs from amphetamine self-administering mice had a similar intraburst frequency, because the bursts contained a lower number of spikes and were of shorter duration. E, The average frequency (pauses/min) of discrete pause activity. F, The histogram (left) compares the length of each pause string with the frequency of spikes contained within that pause string. The average intrapause string frequency (spikes/s), average pause string length (s), and the average percentage of time used by pause strings was similar in ChIs from saline-treated and amphetamine self-administering mice. G, The minimum time thresholds for amphetamine self-administering mice; pϭ0.02 s , Student's t test) and the percentage of time spent bursting (1.57Ϯ0.26% vs 0.83Ϯ0.13% for amphetamine selfadministering mice; pϭ0.01 t , Student's t test) decreased, whereas the intraburst frequency remained constant (8.99Ϯ0.1 spikes/s vs 10.1Ϯ0.13 spikes/s for amphetamine self-administering mice; Fig. 3C,D). There was no change in discrete pauses or pause strings (Fig. 3E,F). Thus, amphetamine self-administration reduced both tonic firing and bursting and uncoupled burst and pause activity in ChIs (Fig. 3G-I).

Amphetamine and nicotine modify burst and pause activity in ChIs
Nicotine opposes the change in tonic ChI firing caused by an amphetamine challenge. We therefore determined if a nicotine challenge would modify the burst and pause activity in ChIs from saline-treated and amphetamine selfadministering mice. In ChIs from saline-treated mice (nϭ7 cells), amphetamine in vitro reduced the frequency of discrete bursts (1.55Ϯ0.31 vs 0.94Ϯ0.17 bursts/min without or with amphetamine; pϭ0.02 u , paired t test), but did not modify pausing (Fig. 4A-D). When combined with amphetamine, nicotine blocked the reduction in discrete bursts (1.55Ϯ0.31 bursts/min vs 1.26Ϯ0.28 bursts/min with amphetamine and nicotine) and had no effect on pausing.
In ChIs from amphetamine self-administering mice, an amphetamine challenge in vitro had no effect on burst activity. Instead, amphetamine increased the frequency of discrete pauses (2.53Ϯ0.99 pauses/min vs 5.62Ϯ1.02 pauses/min with amphetamine; pϭ0.04 v , paired t test) and enhanced the pause strings by increasing the intrapause frequency (1.19Ϯ0.42 spikes/s vs 1.96Ϯ0.27 spikes/s with amphetamine; pϭ0.03 w , paired t test), as well as the percentage of time spent pausing (3.79Ϯ1.14% vs 7.54Ϯ1.02% with amphetamine; pϭ0.04 x , paired t test; Fig. 4G,H). When added to amphetamine, nicotine increased the burst length and blocked the rise in discrete pausing and pause strings. Therefore, an amphetamine challenge reduced tonic firing and bursting in ChIs from saline-treated mice and reversed the depression in the ChI activity of amphetamine self-administering mice by increasing tonic firing. In summary, nicotine attenuated drug-seeking behaviors and opposed the changes in ChI activity that occurred following an amphetamine challenge.

Amphetamine self-administration promotes a CPD in corticostriatal activity
ChIs modulate corticostriatal activity through pre-and postsynaptic muscarinic receptors (Calabresi et al., 2000;Witten et al., 2010), but their control over corticostriatal activity derived through presynaptic nicotinic receptors remains unclear. We used whole-cell recordings in acute striatal slices to measure mEPSCs in MSNs from salinetreated, amphetamine self-administering, and nonresponding mice. The baseline frequency of mEPSCs continued that connect burst and pause events are compared with rate of burst-pause, (H) pause-burst, and (I) burst-pause-burst occurrence.

Figure 4.
Nicotine and amphetamine modify burst and pause activity. A, Nicotine blocked the reduction of discrete burst activity by amphetamine in ChIs from saline-treated mice. For all panels, ‫ء‬pϽ0.05, ‫ءء‬pϽ0.01, paired t test. B, The burstiness of ChIs from saline-treated mice was unaffected by amphetamine or the coadministration of nicotine. C, Amphetamine with nicotine or without did not change discrete pausing or (D) pause strings in ChIs from saline-treated mice. E, In ChIs from amphetamine self-administering mice, amphetamine or coadministered nicotine did not modify discrete bursting (F) but nicotine increased the length of bursting. G, Nicotine blocked the increase in discrete pausing by amphetamine in ChIs from amphetamine self-administering mice. H, In ChIs from amphetamine self-administering mice, an amphetamine challenge produced a nicotine-dependent enhancement of pause strings by in MSNs from amphetamine self-administering mice (3.45Ϯ0.42 Hz) was lower than mEPSCs in MSNs from saline-treated mice (4.92Ϯ0.59 Hz; pϭ0.04 ai , Student's t test; Fig. 6A), consistent with a depression in corticostriatal activity. The baseline frequency of mEPSCs in MSNs from nonresponding mice (13.59Ϯ3.82 Hz) was much higher than saline-treated (pϭ0.003 aj , Student's t test) or amphetamine self-administering mice (pϭ0.0003 ak , Student's t test). Differences in mEPSCs between these groups of mice mainly affected low-amplitude currents and there were no differences in the mEPSC amplitude distribution (Fig. 6B), suggesting that corticostriatal plasticity was presynaptic (Van der Kloot, 1991). The depression of mEPSC in MSNs from amphetamine selfadministering mice was secondary to an increase in lowfrequency activity distributed between 0.5 and 2 Hz, whereas the prominence of mEPSCs in MSNs from nonresponding mice was due to the persistence of frequencies Ͼ7 Hz (Fig. 6C).

An amphetamine challenge promotes PPP in self-administering mice
Having established that amphetamine self-administration depresses corticostriatal activity, we then determined how an amphetamine (10 M) challenge in vitro would modify this plasticity. In saline-treated mice, amphetamine reduced mEPSC frequency by 20Ϯ8% (4.42Ϯ1.1 Hz in vehicle vs 3.58Ϯ1.03 Hz with amphetamine; nϭ8 cells; pϭ0.04 al , paired t test; Fig. 6D), consistent with D2R-mediated depression of corticostriatal activity (Flores-Hernández et al., 1997;Bamford et al., 2004b). In the presence of amphetamine, nicotine (100 nM) further reduced the frequency of mEPSCs (Ϫ8Ϯ2%; 3.28Ϯ0.96 Hz in amphetamine and nicotine; pϭ0.04 am , paired t test compared with amphetamine alone; pϭ0.01 an , compared with vehicle, paired t test). Both amphetamine and amphetamine with nicotine in vitro reduced low-amplitude currents, but not the mEPSC amplitude distribution (Fig.  6E). The frequency spectra showed that both amphetamine and amphetamine with nicotine reduced mEPSC frequencies by increasing low-frequency activity between 0.5 and 1.5 Hz (Fig. 6F).

Nicotine modifies behaviors and corticostriatal activity in noncontingent amphetamine-treated mice
Behavioral sensitization characterizes the progressive and enduring enhancement found in many amphetamineinduced behaviors (Robinson and Camp, 1987). More specifically, locomotor sensitization depends on psychostimulant-induced plasticity of ACh-releasing ChIs in the dorsal striatum that promote long-lasting changes in corticostriatal activity (Bamford et al., 2008;Witten et al., 2010;Wang et al., 2013a). In contrast to noncontingent amphetamine-treated mice, mice self-administering amphetamine (contingent amphetamine-administration) had a 57% larger reduction in baseline ChI activity in withdrawal (24% vs 37% for amphetamine self-administering mice) but a 34% smaller increase in activity following an amphetamine challenge in vitro (88Ϯ27% vs 57Ϯ34% for amphetamine self-administering mice; data not shown in figures). Similarly, the baseline frequency of mEPSC in MSNs was depressed following amphetamine self-administration, but was unchanged following noncontingent treatment. Further, an amphetamine challenge in vitro produced a much smaller increase in mEPSC frequency in amphetamine selfadministering mice (107Ϯ21% vs 21Ϯ11% for amphetamine self-administering mice; data not shown in figures). Therefore, the contingent administration of amphetamine produced a greater depression in ChI firing and corticostria-tal activity, but a lower rebound following an amphetamine challenge.
To test whether nicotine modified locomotor ambulations in withdrawal, mice received either low-(0.1 mg/kg, i.p.) or high-dose (1 mg/kg, i.p.) nicotine on withdrawal day 5. Neither dose of nicotine had an effect on locomotion in saline-treated mice (pϭ0.9 ba , two-way, rm-ANOVA; Fig. 8D). In amphetamine-treated mice challenged with amphetamine, low-dose nicotine had no effect on locomotor ambulations (F (17,102) ϭ0.1, pϭ1 bb , two-way rm-ANOVA), whereas the higher dose of nicotine reduced locomotor ambulations (F (17,102) ϭ3.19, pϭ0.01 bc , twoway rm-ANOVA; Fig. 8E). Ambulations following highdose nicotine were similar to those following the fourth dose of amphetamine, indicating that nicotine suppressed the expression of locomotor sensitization. Locomotor testing following an amphetamine challenge in nonsensitized, saline-exposed mice showed that nicotine (1 mg/kg) did not block the acute stimulating effect of amphetamine on locomotion (pϭ0.9 bd , two-way rm-ANOVA; Fig. 8F). These same saline and nicotine (1 mg/kg) challenges in amphetamine-treated mice revealed similar ambulations compared with saline-treated mice challenged with saline (pϭ0.9 be , two-way rm-ANOVA; Fig. 8G). This indicates that amphetamine-experienced mice lack conditioned locomotor activity elicited by placement into the amphetamine-associated chamber and show that nico-continued self-administering mice. For all panels, $ pϽ0.05, Student's t test and ‫ء‬pϽ0.05, ‫,100.0‪pϽ‬ءءء‬ paired t test. B, Normalized-firing frequencies over time for the experiments shown in A. C, In ChIs from saline-treated mice, the spectra (left) show that nicotine (double arrows) increased low frequencies, whereas reducing higher frequencies and the average peak frequency (right). Nicotine shifted the unipolar distribution of ChIs from saline-treated mice toward the bimodal distribution of average frequencies in amphetamine self-administering mice. D, In ChIs from amphetamine self-administering mice, nicotine increased low frequencies, reduced high frequencies, and suppressed the average peak frequency. E, Nicotine shifted the bimodal distribution of average frequencies in amphetamine self-administering mice toward the unipolar distribution of ChIs from saline-treated mice. F, In ChIs from saline-treated mice, nicotine suppressed discrete bursting, (G) the percentage of time spent bursting, and (H) discrete pausing, but (I) did not alter pause strings. J, In ChIs from amphetamine self-administering mice, nicotine suppressed discrete bursting, (K) the percentage of time spent bursting, and (L) discrete pausing. M, Amphetamine self-administration modified pause strings; the intra-pause spike frequency decreased whereas the length of the pause string increased. tine did not modify locomotor activity in the absence of amphetamine.
It was previously shown that withdrawal from repeated injections of amphetamine produced CPD and PPP following reinstatement (Wang et al., 2013a). In another group of similarly treated mice, we performed electrophysiological recordings to determine whether nicotine in vitro modifies corticostriatal activity in withdrawal. eEP-SCs in MSNs were obtained in response to 50 ms pairedpulses delivered to the motor cortex. In MSNs from saline-treated mice, puff-applied nicotine decreased the amplitude of the first current of the pair by 19Ϯ5% (Ϫ99Ϯ11 pA vs Ϫ78Ϯ10 pA with nicotine; nϭ12 cells; pϭ0.02 bf , paired t test) and increased the PPR by 34Ϯ12% (1.18Ϯ0.06 vs 1.56Ϯ0.15 with nicotine; pϭ0.009 bg , paired t test; Fig. 8H), indicating a reduction in corticostriatal excitation. In MSNs from amphetaminetreated mice, nicotine increased the amplitude of the first current by 30Ϯ6% (Ϫ66Ϯ6 pA vs Ϫ83Ϯ5 pA with nicotine; nϭ10 cells; pϽ0.001 bh , paired t test) and decreased the PPR by 23Ϯ3% (1.25Ϯ0.11 vs 0.95Ϯ0.07 with nicotine; pϭ0.013 bi , paired t test; Fig. 8I). Compared with MSNs from saline-treated mice (nϭ12 cells), the amplitude of the first pulse of the pair was lower in MSNs from amphetamine-treated mice in withdrawal (nϭ10 cells; Ϫ99Ϯ11 pA for saline-treated vs Ϫ66Ϯ6 pA in withdrawal; pϭ0.02 bj , Student's t test; Fig. 8, compare H, I). By increasing the amplitude of the first eEPSC, nicotine blocked the synaptic depression in withdrawal (Ϫ99Ϯ11 pA for saline-treated in vehicle vs Ϫ83Ϯ5 pA for withdrawal with nicotine; pϭ0.2 bk , Student's t test). Therefore, alterations in ACh availability extend to models of noncontingent amphetamine-treated mice because nicotine suppressed the expression of locomotor sensitization and although nicotine in vitro reduced eEPSCs in saline-treated mice, it blocked synaptic depression in withdrawal by enhancing corticostriatal activity.

Discussion
The transition from psychostimulant use to dependence involves changes in corticostriatal activity in consequence of intermittent DA release Sulzer, 2011). Repeated exposure to amphetamine, which releases DA (Bamford et al., 2004b), promotes long-lasting and parallel changes in drug-primed rewarding, locomotor sensitization, ChI firing, and corticostriatal activity. This suggests that ChIs in the striatum participate in addictive behaviors and contribute to normal motor and cognitive function by promoting allostasis and sensitized responses (Ahmed and Koob, 2005). Our data show that the rewarding properties of amphetamine and the potentiation of locomotor sensitization are dependent on ACh because nicotine attenuated both responding to an amphetamine challenge after abstinence and sensitized locomotor responses to an amphetamine challenge in withdrawal.
Preclinical and clinical evidence suggests that cigarette smoking and nicotinic receptors participate in psychostimulant addiction (Rahman, 2013;Mello et al., 2014;Verrico et al., 2014). Contingent models of drug-seeking behavior in rodents have shown an interaction between nicotine and methamphetamine incubation (Shepard et al., 2004;Hiranita et al., 2006;Neugebauer et al., 2010;Pipkin et al., 2014). However, this interaction is complex and altered by diverse factors such as age, dose, drug pairing, and frequency of use. No studies have investigated the effect of acute nicotine on amphetamine incubation in mice. Using a standard self-administration protocol based on fixed schedules of reinforcement, we demonstrate that mice acquire amphetamine selfadministration behaviors and develop the motivation to ob-  (right) show that the frequency of mEPSC in MSNs from saline-treated mice was greater than amphetamine self-administering mice but far less than those in nonresponding mice. $ pϽ0.05, $$ pϽ0.01, $$$ pϽ0.001, Student's t test. B, Compared with saline-treated mice, amphetamine self-administration reduced the frequency of low-amplitude mEPSCs, while the frequency of mEPSCs was broadly higher in nonresponding mice. $ pϽ0.05, saline-treated versus amphetamine self-administering mice; ## pϽ0.01, saline-treated versus nonresponding mice; Bonferroni t test. Inset, The cumulative mEPSC amplitude distribution was unchanged across groups. C, Frequency distributions (left) and the average peak frequency (right) show a prominent low-frequency distribution of mEPSCs from amphetamine self-administering mice (small arrow) and more broadly distributed firing frequencies in MSNs from nonresponding mice (arrowhead; inset). $$ pϽ0.01, $$$ pϽ0.001, Student's t test. D, Responses of individual MSNs from saline-treated mice show that amphetamine, and the addition of nicotine, reduced the frequency of mEPSCs. ‫ء‬pϽ0.05, paired t test. E, Amphetamine with nicotine or without decreased the frequency of low-amplitude mEPSCs. ‫ء‬pϽ0.05, Bonferroni t test, amphetamine compared with either vehicle or amphetamine with nicotine. Inset, The mEPSC amplitude distribution was unchanged. F, Frequency distributions (left) and the average peak frequency (right) of mEPSCs from saline-treated mice shows that both amphetamine (arrow) and amphetamine with nicotine (double arrow) increased low-frequency events. G, In MSNs from amphetamine self-administering mice, nicotine blocked the increase in mEPSCs that followed amphetamine. ‫ء‬pϽ0.05, paired t test. H, Amphetamine augmented the frequency of low-amplitude mEPSCs, but had no effect on their amplitude distribution (inset). ‫ء‬pϽ0.05, Bonferroni t test, amphetamine compared with either saline or amphetamine with nicotine. I, The distribution (left) and the average peak frequency (right) of mEPSCs from amphetamine self-administering mice show that amphetamine reduces 1-2 Hz activity (arrow). The addition of nicotine moderated the effect of amphetamine by increasing 1-2 Hz activity (double arrows). ‫ء‬pϽ0.05, paired t test. J, The distribution (left) and average peak frequency (right) show that mEPSCs s from saline-treated mice were similar to those from amphetamine self-administering mice exposed to amphetamine. K, Both amphetamine and amphetamine with nicotine reduced the mEPSC frequency in MSNs from nonresponding mice. ‫ء‬pϽ0.05, ‫ءء‬pϽ0.01, paired t test. L, Amphetamine and nicotine reduced low-amplitude mEPSCs, but had no effect on their amplitude distribution (inset). ‫ء‬pϽ0.05, amphetamine compared with either vehicle or amphetamine with nicotine; & pϽ0.01, vehicle compared with either amphetamine or amphetamine with nicotine; Bonferroni t test. M, The frequency distribution (left) and average peak frequency (right) of mEPSCs from nonresponding mice show that amphetamine with (double arrow) or without nicotine (arrow) increases low-frequency events. ‫ء‬pϽ0.05, paired t test.

Figure 7.
Nicotine in vitro blocks CPD through ‫-ء7␣‬ and ␣4␤2‫-ء‬type nicotinic receptors. A, Responses of individual MSNs show that nicotine inhibited corticostriatal activity in saline-treated mice by (B) reducing high-frequency, low-amplitude mEPSCs, but had no effect on their amplitude distribution (inset). For B, E, J, ‫ء‬pϽ0.05, Bonferroni t test; for all other panels, ‫ء‬pϽ0.05, paired t test and $$ pϽ0.01, Student's t test. C, The frequency distribution (left) and the average peak frequency (right) of mEPSCs. Nicotine increased low-frequency mEPSCs in saline-treated mice. D, Nicotine increased the frequency of mEPSCs in MSNs from amphetamine self-administering mice by (E) augmenting high-frequency, low-amplitude mEPSCs, while having no effect on their amplitude distribution (inset). F, The frequency distribution (left) and average peak frequency (right) of mEPSCs. Nicotine reduced low-frequency mEPSCs in amphetamine self-administering mice. G, MeanϮSE frequencies of mEPSCs from saline-treated and amphetamine self-administering mice in response to nicotine. H, The frequency distribution (left) and average peak frequencies (right) compares mEPSCs in MSNs from saline-treated mice with those from amphetamine self-administering mice exposed to nicotine in vitro. I, The ␣4␤2‫-ء‬type receptor antagonist DH␤E blocked the synaptic depression generated by nicotine and (J) increased high-frequency, low-amplitude synaptic events. K, The ␣7‫-ء‬type nicotinic receptor antagonist MLA reduced (L) high-frequency, low-amplitude mEPSCs. J, L, Insets show similar mEPSC amplitude distributions. tain this drug. As this experiment is technically challenging, we then used a within-subject design (Ma et al., 2013), to assess the effect of nicotine on amphetamine-seeking behaviors following abstinence. When combined with the electrophysiological analysis of MSNs and ChIs, amphetamine self-administration provides a valuable tool to explore the striatal circuit-and ChI-adaptations induced by contingent drug exposure (Fig. 9).
ChIs in the dorsal striatum respond to conditioned sensory stimuli that signal reward delivery and elicit behavioral reactions . Measures of ChI frequencies ex vivo revealed a reduction in tonic ChI activity from amphetamine self-administering mice. Although having no effect on pausing, an amphetamine challenge following abstinence decreased the frequency of discrete bursts and burstiness, thereby dissociating or uncoupling burst activity from pausing in the absence of DA-releasing stimuli. Nicotine reduced tonic firing, discrete bursts, burstiness, and discrete pauses in ChIs from both saline-treated and amphetamine self-administering mice. When nicotine suppressed already low tonic firing in amphetamine self-administering mice, the discrete pauses coalesced into pause strings. As ChIs are widely distributed throughout the striatum, the uncoupling of burst and pause activity could serve to modulate the activity of surrounding neurons and mediate rewarding behaviors.
The literature on where these forms of plasticity and their related behaviors occur is unclear. However, much evidence indicates that the circuitry underlying am- Figure 8. Nicotine modifies locomotor sensitization and corticostriatal activity. A, Protocol for testing the effect of amphetamine and nicotine on locomotor sensitization. The amphetamine treatment group received saline injections for the first 2 d, whereas the saline treatment group received saline on all days. B, Average locomotor ambulations over 90 min in mice treated with either saline or amphetamine. C, Interval locomotor ambulations over 90 min; compare the ambulations following the first injection of amphetamine with those following an amphetamine challenge in withdrawal. Also shown are ambulations in saline-treated mice following a saline challenge. D, Interval locomotor ambulations over 90 min in saline-treated mice that follow an injection of low-dose nicotine, high-dose nicotine, or saline. E, Interval locomotor ambulations over 90 min in amphetamine-treated mice. Ambulations that follow the first injection of amphetamine are compared with ambulations that follow an amphetamine challenge in withdrawal, with and without simultaneous treatment with nicotine in vivo. Low-dose nicotine had little effect on ambulations, whereas the higher dose of nicotine reduced ambulations to those observed during the last daily treatment of amphetamine. F, Interval locomotor ambulations over 90 min in saline-treated mice following an amphetamine challenge, with and without high-dose nicotine. G, Interval locomotor ambulations over 90 min in amphetamine-treated mice following a challenge with saline or high-dose nicotine are compared to ambulations in saline-treated mice challenged with saline. H, In MSNs from saline-treated mice, nicotine in vitro decreased the amplitude of the first eEPSC and increased the PPR. For all panels, ‫ء‬pϽ0.05, ‫ءء‬pϽ0.01, ‫,100.0‪pϽ‬ءءء‬ paired t test. I, In MSNs from amphetamine-treated mice, nicotine increased the amplitude of the first eEPSC and decreased the PPR. phetamine-induced locomotion is mediated by NAc, whereas stereotypies are mediated by dorsal striatum (Asher and Aghajanian, 1974;Pijnenburg et al., 1976;Segal, 1976;Castall et al., 1977). Furthermore, the development of sensitization is associated with prefrontal cortex and VTA, whereas expression is associated with the prefrontal cortex and the NAc (McFarland and Kalivas, 2001;). The data presented here suggests that the potentiation of locomotor sensitization is also associated with long-term plasticity in the dorsal striatum. The role of ChIs in the dorsal striatum has been associated with cue-dependent behaviors and the expression of locomotor responses to repeated amphetamine (Wang et al., 2013a), thus agreeing with the results of our selfadministration studies. Repeated psychostimulant use produces similar forms of glutamatergic plasticity in the . Proposed mechanism for nicotinic modulation of striatal synaptic plasticity following amphetamine self-administration. A, The simplified striatal circuit is composed of MSNs, ChIs, and glia. Glutamate released from cortical and thalamic afferents excites MSNs through ionotropic glutamate receptors (iGluRs). In the drug naive condition, ACh efflux from ChIs binds to excitatory ‫-ء2␤4␣‬ and ␣7‫-ء‬type nicotinic receptors and modulates ChI firing and glutamate release from corticostriatal terminals (Bamford et al., 2008;Wang et al., 2013a). The glial glutamate transporter (GLT1) limits the access of synaptic glutamate to the extracellular space . B, Synaptic depression develops during abstinence from repeated amphetamine. Repeated DA release by amphetamine stimulates inhibitory D2-class receptors on ChIs (Yan et al., 1997;Wang et al., 2013a). The reduction in tonic excitation at ‫-ء2␤4␣‬ and ␣7‫-ء‬type nicotinic receptors depresses corticostriatal activity. Low tonic levels of ACh are supported during drug withdrawal by nicotinic autoreceptors that lower the set point of ChI firing (Wang et al., 2013a). C, A drug cue promotes transient synaptic potentiation. Because of the lowered set point of ChI activity, DA released by a drug cue stimulates ACh efflux via excitatory D1-class receptors on ChIs (Yan and Surmeier, 1997;Wang et al., 2013a). The increase in ACh boosts glutamate release, causing PPP. D, Nicotine prevents PPP. The concentration of nicotine achieved in smokers (100 nM) desensitizes excitatory ␣4␤2‫-ء‬type nicotinic receptors (Dani and Harris, 2005). The lack of nicotinic receptor excitation reduces ChI spiking and corticostriatal activity in the drug naive condition and during abstinence. The drug cue enhances ACh tone but desensitization prevents PPP in response to drug reinstatement. Activation of ␣7‫-ء‬type nicotinic receptors expressed on astrocytes (Pirttimaki et al., 2013) recruit AMPA receptors to glutamatergic terminals (Wang et al., 2013b), have a protective effect on oxidative stress (Liu et al., 2015), and increase activity of GLT1 to clear glutamate from the extracellular space  ventral dorsal striatum (Bamford et al., 2008), suggesting that both structures play a role in the behavioral potentiation seen in locomotor sensitization and self-administration.
Psychostimulant-released DA reduces ChI firing through D2-class DA receptors on ChIs (Yan et al., 1997), and the firing rate of ChIs remained suppressed following amphetamine self-administration. Data support the hypothesis that a lasting inhibition of ChI activity in withdrawal occurs through a reduction in tonic excitation at autoregulatory receptors, because nicotine reduced spontaneous firing and concentered the spectral signatures in ChIs from both saline-treated and amphetamine self-administering mice. Therefore, a decrease in tonic excitation at nicotinic autoreceptors on ChIs (Azam et al., 2003) in amphetamine self-administering mice may maintain the depression of ChI firing in withdrawal.
The differential action of excitatory D1-and inhibitory D2-class DA receptors located on ChIs establish a set point for ChI activity (Abercrombie and DeBoer, 1997;Wang et al., 2013a). An amphetamine challenge reduced tonic firing and discrete bursting in ChIs from amphetamine naive mice, but increased the already low tonic firing in ChIs from amphetamine self-administering mice. Amphetamine regulates ChI activity through DA receptors on ChIs, where the reduction in tonic firing in salinetreated mice occurs through D2 receptors, whereas the excitation during amphetamine withdrawal are generated through heightened responses mediated through D1 receptors (Wang et al., 2013a). Thus, repeated DA release during acquisition followed by abstinence may create a shift in the activity set point of ChIs and reverse their response to an amphetamine challenge or to cues paired with the drug. Nicotine suppressed these two divergent DA-dependent changes in ChI activity, suggesting that nicotinic autoreceptors participate in DA-dependent plasticity. Although amphetamine modulated ChI firing at frequencies centered on 1 Hz, nicotine shifted ChI activity toward a pre-existing spectral baseline by modulating frequencies centered on 3 Hz. Nicotine may oppose DAdependent changes in ChI activity by stabilizing membrane potentials required for GABA receptor modulation of ChIs (DeBoer and Westerink, 1994) or hyperpolarization-activated cation channels that encode their autonomous firing (Bennett et al., 2000).
Data indicate that ChI firing (and ACh availability) promotes parallel changes in corticostriatal activity. Similar to responses in ChIs, amphetamine in vitro reduced corticostriatal activity in saline-treated mice and repeated amphetamine promoted CPD in withdrawal. Nicotine blocked both CPD and amphetamine-induced PPP, signifying that nicotinic receptors are both necessary and sufficient to promote long-lasting corticostriatal plasticity. PPP was absent in nonresponding mice even though they were exposed to small doses of amphetamine in vivo. Although the responses to amphetamine and nicotine were similar to saline-treated mice, nonresponding mice had a much higher frequency of mEPSCs at baseline and consequently, amphetamine produced a robust depression in activity. Future investigations of both the behav-ioral responding to different doses of amphetamine and cellular adaptations will be required to understand the basis of these observations. Nicotine reduced amphetamine-seeking behaviors and the expression of locomotor sensitization. Although nicotinic receptors are located throughout the brain, nicotine may modify ChI and corticostriatal activity through ‫-ء2␤4␣‬ and ␣7‫-ء‬type nicotinic receptors that are located on ChIs (Azam et al., 2003) and presynaptic corticostriatal terminals (Marchi et al., 2002;Pakkanen et al., 2005). The level of nicotine (100 nM) has been shown to desensitize ␣4␤2‫-ء‬type nicotinic receptors (Lester and Dani, 1995;Wooltorton et al., 2003) and may therefore act to lessen ChI excitation and corticostriatal activity. In saline-treated mice, nicotine reduced corticostriatal excitation caused by synaptic ACh, likely by desensitizing and internalizing excitatory ‫ء2␤4␣‬ receptors (Zhou et al., 2002;Pakkanen et al., 2005). In amphetamine self-administering mice, the decrease in ACh availability may promote CPD by reducing excitation at both ‫-ء2␤4␣‬ and ␣7‫-ء‬type nicotinic receptors, whereas the increase in ChI activity following an amphetamine challenge supports PPP. Nicotine may therefore supplement the low availability of ACh to increase corticostriatal activity, but desensitizes the ‫-ء2␤4␣‬ type nicotinic receptor and prevents PPP when given in combination with amphetamine. Thus, nicotine may offset corticostriatal plasticity by preventing the change in ACh availability that occurs during the incubation of drugseeking behavior. However, the modulation of ChI and corticostriatal activity by ACh and nicotine is likely complex and involves a host of nicotine receptor subtypes with different sensitivities and levels of expression.
Increasing evidence suggests that ChI firing and ACh availability promote an imbalance in glutamate release from corticostriatal terminals (Bamford et al., 2008;Wang et al., 2013a) to produce motor and neuropsychological symptoms of drug dependence (Robinson and Camp, 1987;. Although psychostimulant-releasing DA reduces ChI and corticostriatal activity in amphetamine naive mice, the increase in DA and ACh availability following a drug challenge in sensitized mice promotes a PPP of corticostriatal activity that correlates with the increase in locomotor activity and promotes sensitization by specifically activating D1R-expressing MSNs (Gass and Olive, 2008;Wang et al., 2013a). Together, CPD and PPP provide a mechanism by which a drug challenge promotes locomotor sensitization and allostasis (Ahmed and Koob, 2005). The much higher basal level of corticostriatal activity in nonresponding mice may provide some protection against PPP, because an amphetamine challenge reduced the frequency of mEPSCs to that found in saline-treated mice.
These data extend to observations of cocaine-induced glutamatergic neuroadaptations, where drug withdrawal and reinstatement accompany a reversible reduction in extracellular  and synaptic glutamate (Kozell and Meshul, 2003). Sensitization models of incentive salience (Robinson and Camp, 1987) suggest that a supra-physiological glutamatergic drive promotes compulsive drug-seeking in addicts by decreasing the value of natural rewards . While demonstrating that CPD and PPP play key roles in sensitization, these experiments show that alterations in ACh may overlap with the plasticity and learning associated with self-administration and incubation. Thus, DA modulation of ACh release may provide a set point for glutamate availability (Ahmed and Koob, 2005) that extends to the escalation of drug intake (Hansen and Mark, 2007).
ChIs appear to support motor coordination, cognitive flexibility, and locomotion through ACh interactions with striatal projection neurons (Bamford et al., 2008;Witten et al., 2010;Wang et al., 2013a). Because DA provokes state changes in corticostriatal activity, our findings may explain how salient experiences encode behaviors and automatic movements. Changes in ACh and glutamate availability are promoted by repeated DA release. Our findings are consistent with human trials, which demonstrate that nicotinic receptors and changes in glutamate transmission participate in the pathogenesis of many neuropsychological disorders (Mihailescu and Drucker-Colín , 2000;Gass and Olive, 2008), including Parkinson's disease where nicotinic receptor ligands prevent the disruption of striatal and cholinergic activity following DA depletion (Quik, 2004). Similarly, therapeutic approaches using nicotinic receptor ligands that prevent PPP may facilitate the extinction of drug seeking behavior. If provided during the process of sensitization, these drugs would prevent CPD and disrupt the reinforcing effects of drugs in addicts (Gass and Olive, 2008;Reissner and Kalivas, 2010). Together, these observations suggest the need for additional circuit-level and behavioral experiments in conjunction with epidemiological studies to determine whether pharmacological targeting of nicotinic receptors can modify the development of drug seeking and reduce relapse.