Embed Size (px)
Citation preview
ttrtecs
nasatTpicon
map
ARCHIVAL REPORT
Context-Induced Relapse to Alcohol Seeking AfterPunishment in a Rat ModelNathan J. Marchant, Thi N. Khuc, Charles L. Pickens, Antonello Bonci, and Yavin Shaham
Background: Rat studies have demonstrated that exposure to environments associated with alcohol intake reinstates alcohol seeking afterextinction of alcohol-reinforced responding in a different context. However, extinction is limited as an abstinence model, because humanstypically abstain because of negative consequences associated with excessive drinking. It is currently unknown whether alcohol-associatedcontexts can provoke relapse to alcohol seeking after alcohol-taking behavior is suppressed by adverse consequences in a different context.
Methods: Alcohol-preferring P rats were first given home-cage access to 20% ethanol. Next, they were trained to self-administer 20%ethanol in one context (context A). Subsequently, all rats continued to self-administer alcohol in a different context (context B). For onegroup, 50% of alcohol-reinforced responses were punished by mild footshock; two other groups either received noncontingent shocks or noshock. A fourth group was given extinction training in context B. All rats were then tested for relapse to alcohol seeking under extinctionconditions in contexts A and B.
Results: In Context B, alcohol-taking behavior was suppressed by contingent shock (punishment) and extinction training but not bynoncontingent shock. In Context A, relapse to alcohol seeking was reliably observed in the punished and extinction groups; a context switchhad no effect on alcohol seeking in the no-shock or noncontingent shock groups.
Conclusions: Our data indicate that punishment-induced suppression of alcohol-taking behavior is context-dependent. We propose thatour procedure can be used to explore mechanisms of context-induced relapse to alcohol seeking after alcohol-taking behavior is suppressed
by adverse consequences.araapc
dsmstsSnm(araesTcl
wwhFsWbcb
Key Words: Alcohol, context, extinction, intermittent access, P rats,punishment, relapse
I n humans, places or contexts previously associated with alcoholuse often provoke relapse during abstinence (1–5). In rats, stud-ies using the ABA renewal procedure (6,7) have demonstrated
hat exposure to the alcohol self-administration context after ex-inction of the alcohol-reinforced responding in a different contexteinstates alcohol seeking (8 –11). In this procedure, rats are trainedo self-administer alcohol in one context. Next, lever-responding isxtinguished in a different (nonalcohol) context. Subsequently,ontext-induced reinstatement (renewal) of alcohol seeking is as-essed by testing rats in the alcohol-associated context.
In the context-induced reinstatement procedure, alcohol absti-ence is achieved by operant extinction training. During extinction,lcohol is unavailable and the previously alcohol-reinforced re-ponding declines over time. In humans, however, alcohol is oftenvailable and abstinence is typically initiated because of the nega-ive consequences associated with excessive drinking (12,13).herefore, one limitation of the context-induced reinstatementrocedure (14) and other extinction-reinstatement procedures (15)
s the lack of homology between the animal model and the humanondition (16 –20). Our goal here was to develop an animal modelf context-induced relapse after alcohol intake is suppressed byegative consequences.
One way to impose negative consequences on drug use in ani-al models is to introduce a punishment contingency during self-
dministration training. In most rats, response-contingent shockunishment suppresses operant responding for both drug (21–25)
From the Intramural Research Program, National Institute on Drug Abuse,Baltimore, Maryland.
Address correspondence to Nathan Marchant, Ph.D., Intramural ResearchProgram, National Institute on Drug Abuse, Baltimore, Maryland; E-mail:[email protected].
tReceived Mar 30, 2012; revised Jun 15, 2012; accepted Jul 5, 2012.
0006-3223/$36.00http://dx.doi.org/10.1016/j.biopsych.2012.07.007
nd nondrug (26,27) rewards. Response-contingent shock causesapid suppression of morphine, amphetamine, and cocaine self-dministration (23,28). The magnitude of suppression of drug self-dministration is correlated with shock intensity (28,29), and non-unished cocaine-reinforced responses are chosen over punishedocaine-reinforced responses (30,31).
It is well known that punishment of food-reinforced respondingoes not permanently suppress responding (26,27). Furthermore,tudies using the Geller-Seifert procedure indicate that punish-
ent-induced response suppression during food-reinforced ses-ions is under discriminative stimulus control (27,32). However,here are surprisingly few studies on recovery of drug or alcoholeeking after punishment-induced abstinence. In an early study,mith and Davis (23) found that morphine priming injections doot provoke “relapse” to morphine seeking after severe punish-ent (high intensity shock). However, more recently, Panlilio et al.
22) reported that priming injections of remifentanil (an opiate)fter punishment-induced abstinence led to resumption ofemifentanil self-administration (relapse). Subsequently, Panlilio etl. (33) reported that priming injections of lorazepam (a benzodiaz-pine) or heroin provoke relapse to remifentanil seeking, as as-essed in extinction tests, after punishment-induced abstinence.his finding demonstrates that punished drug seeking can be re-overed in the absence of drug-reinforced responding during “re-
apse” testing.In previous studies relapse to drug seeking after punishment
as assessed after drug priming injections. Here we assessedhether alcohol-associated contexts can provoke relapse to alco-ol seeking after alcohol seeking is punished in a different context.or this purpose, we adapted the renewal procedure (6,7,34) anduppressed alcohol seeking by punishment in a different context.
e then assessed relapse to alcohol seeking in extinction tests inoth the punishment and the original alcohol self-administrationontexts. We used the selectively bred alcohol- preferring P rats,ecause these rats are a useful model for voluntary self-administra-
ion of alcohol (35). Finally, for comparison purposes, we also exam-
BIOL PSYCHIATRY 2013;73:256–262© 2013 Society of Biological Psychiatry
Etme
P
Ud1fhadbthuest
P
sfimdS.rt
G2sa ession
N.J. Marchant et al. BIOL PSYCHIATRY 2013;73:256–262 257
ined context-induced reinstatement of alcohol seeking after extinc-tion, with the established ABA renewal procedure (14,34).
Methods and Materials
SubjectsMale alcohol-preferring P rats (approximately 30 days old, total
n � 44) were obtained from Indiana University Medical Center. Ratswere quarantined for approximately 3 weeks. The rats were housedsingly under a reversed 12-hour light/dark cycle (8:00 AM light off).The experimental procedures followed the Guide for the Care andUse of Laboratory Animals (eighth edition) and were approved bythe Animal Care and Use Committee.
ApparatusSee Supplement 1.
ProcedureIn Experiment (Exp.) 1, three groups were distinguished by the
manipulation during the context B training phase. These groupseither received footshock that was contingent on lever-presses(Punished, n � 15), footshock that was not contingent on lever-presses (Noncontingent, n � 8), or no shock (Unpunished, n � 11).
xp. 1 timeline is provided in Figure 1A. In Exp. 2, one group (Extinc-ion, n � 8) was used to demonstrate context-induced reinstate-
ent of alcohol seeking after extinction of lever-presses in a differ-
B. Home-cage intermittent access
Alcohol intake Alcohol preference
A. Experim
2-bo�le choice, 20% alcohol vs. water
Every other day
12 x 24 h sessions 1: Home cage
3 x FR-1 sessions 4 x FR-5 sessions 4 x V2: Context A
20% alcohol Every other day
20% alcohol Every other day
20Eve
2 4 6 8 10 12
Day
0
5
10
15
20
g/kg
/24
h
AlcoholWater
2 4 6 8 10 120
25
50
75
100
Day
Per
cent
Figure 1. Intermittent free-choice home-cage access to 20% alcohol leadsraphical representation of the experimental design, see Methods for details0% alcohol over water (right panel) during the 12 sessions of free-choicemaller than the symbol size. (C) Mean � SEM alcohol intake in grams/kilogralcohol self-administration schedules: fixed-ratio (FR)-1, 3 sessions, FR-5, 4 s
nt context. a
hase 1: Intermittent Home-Cage AccessWe used the intermittent access 20% alcohol procedure (36,37).
pon arrival to the animal facility from quarantine, rats were han-led for 2 days. All rats received 12 � 24-hour sessions of access tobottle of 20% ethanol v/v in water and 1 bottle of water. Food was
reely available. The session started approximately 9:00 AM. After 24ours, the alcohol bottle was replaced with a second water bottle,nd the no-alcohol session also lasted for 24 hours. The followingay, the second water bottle was replaced with a 20% ethanolottle, and the location of the alcohol bottle was alternated from
he previous session. The majority of no-alcohol sessions were 24ours; however, some no-alcohol sessions were 48 hours. The vol-me consumed in each session was calculated as the weight differ-nce between the start and end of the sessions, minus 2 g forpillage, multiplied by .97 (density of 20% ethanol). Two rats withhe lowest g/kg consumption were excluded.
hase 2: Operant Self-Administration in Context AThis phase began 48 hours after the final home-cage access
ession. Sessions were 2 hours long and run every other day. Therst 2 sessions were 2-hour “autoshaping” sessions during which .1L of alcohol was delivered noncontingently every 5 min. Alcohol
elivery was accompanied by a 2-sec compound tone-light cue.ubsequently, the rats were trained for 3 sessions to self-administer
1-mL deliveries of alcohol on a fixed-ratio-1 (FR-1) 2-sec timeouteinforcement schedule. Lever-presses on one lever (an active, re-ractable lever) resulted in presentation of the 2-sec tone-light cue
C. Operant self-administration
Alcohol intake Lever presses
g/kg
/2 h
Leve
r pre
sses
(2 h
)
l procedure
ssions
hol r day
Ex�nc�on condi�ons in both context A and B
Every other day
30 min sessions 4: Alcohol-free test
0
0.6
1.2
1.8
2.4
FR-1 FR-5 VI-30 FR-1 FR-5 VI-300
60
120
180
240
Reinforcement schedule
20% alcohol Every other day
3 - 7 x VI-30 sessions 3: Context B
Punished: Response con�ngent shock
Non-con�ngent: 10 x random shock
Unpunished: no shock
Ex�nc�on: no shock, no alcohol
* * * *
h alcohol intake and robust acquisition of operant self-administration. (A)ean � SEM alcohol intake in grams/kilogram (left panel) and preference for
-cage access to 20% alcohol. The standard errors for these data points aret panel) and active lever-presses (right panel) during the three different 20%s, variable-interval (VI) VI-30, 4 sessions (n � 42).
enta
I-30 se
% alcory othe
to hig. (B) M
homem (lef
nd activated the infusion pump for 2 sec. Lever-presses on the
www.sobp.org/journal
h(tagl(t
R
HC
hfwaoa
1ass[3
E
ptSlcswbUsdt9cPw
tpfcUslp
vbp
258 BIOL PSYCHIATRY 2013;73:256–262 N.J. Marchant et al.
other lever (an inactive, stationary lever) were recorded but had noprogrammed consequences. Subsequently, rats were trained on anFR-5 2-sec timeout reinforcement schedule for four sessions. Afterthis, rats were trained on a variable-interval 30-sec (VI-30) reinforce-ment schedule for four sessions. During these sessions, alcoholdelivery was available after an active lever press at random intervals(range: 1–59 sec) after the preceding alcohol delivery. The variableinterval began immediately after an alcohol-reinforced response.Responses during the timeout interval had no consequences.
Phase 3: Punishment (Exp. 1) or Extinction (Exp. 2) in Context BExp. 1. During this phase, the rats continued alcohol self-admin-
istration every other day (2-hour sessions) under the VI-30 rein-forcement schedule. For group Punished, 50% of the reinforcedlever-presses, which occurred after the VI-30 response requirementwas met, delivered a .5-sec .45-mA footshock through the grid floor.Responses during the timeout interval were recorded but had noconsequences. During the punished responses, the tone-light cuewas presented and .1 mL of 20% alcohol was delivered. There was aminimum of three punishment sessions; subsequently, rats ingroup Punished that made more than 20 responses/session weregiven additional sessions with increasing shock intensity until thiscriterion was met. The final shock intensity was 1.09 mA on theseventh session. For rats in group Noncontingent Shock, 10 .5-sec.45-mA footshocks were delivered randomly throughout the2-hour sessions; we chose to deliver 10 footshocks, because thiswas the approximate mean number of shocks/session given to therats in group Punished during the first three sessions. For rats ingroup Unpunished, conditions were identical to the previous ses-sions in context A. Rats in groups Unpunished and Noncontingentwere given three to seven sessions in context B, to match theircontext B exposure to group Punished. The number for each con-text B session for each group was: “Punished”: session 1–3 � 15,session 4 � 9, session 5 � 7, session 6 � 7, and session 7 � 3;“Unpunished”: sessions 1–3 � 11, session 4 � 6, session 5 � 4,session 6 � 3, and session 7 � 2; “Noncontingent”: sessions 1–3 � 8,session 4 � 5, session 5 � 5, session 6 � 2, and session 7 � 2.
Exp. 2. Rats were placed into context B, and each response thatmet the VI-30 reinforcement schedule resulted in the presentationof the tone-light cue but no alcohol. There were 13 (2-hour) extinc-tion sessions; the first 10 sessions were every other day, and the final3 extinction sessions were given daily to minimize the influence ofspontaneous recovery—the recovery of extinguished conditionedresponses after the passage of time (34)— on lever-presses.
Phase 4: Relapse TestsAll rats were tested for alcohol seeking (operationally defined as
active lever responding under extinction conditions) with no foot-shock punishment, in both context A and context B in 30-minextinction sessions. In Exp. 1, the test order in each context wascounterbalanced between and within groups. The first test was 48hours after the final context B training session of each rat, and testswere separated by 48 hours. In Exp. 2, the rats were tested 24 hoursafter the last extinction session, and the tests were 24 hours apart.The order of testing in contexts A and B was counterbalanced.
Statistical AnalysisData were analyzed separately for the four phases: home-cage
access, context A training, context B punishment or extinction, andrelapse tests. For the home-cage access phase, the dependent mea-sures were 20% alcohol preference and grams/kilogram alcoholintake. For the training and punishment phases, the dependent
measures were total active lever-presses and grams/kilogram alco- cwww.sobp.org/journal
ol intake. For the relapse test, the dependent measures were totalnonreinforced) responses on the previously active lever and inac-ive lever responses. Group differences in these measures werenalyzed by analyses of variance and Tukey post hoc tests. Theroups were matched for their home-cage alcohol intake context A
ever-presses during training. Inactive lever-presses were very lowmean of �four presses session) (Table S1 in Supplement 1); thus,hese data were only used as covariates in the analyses.
esults
ome-Cage Alcohol Intake and Initial Operant Training inontext A (Exp. 1–2)
Figure 1B shows alcohol intake (g/kg) and preference during theome-cage access phase (see Figure S1 in Supplement 1 for data
rom each group in Exp. 1). Mean alcohol intake during this phaseas 12.8 � .2 g/kg/24 hours. The rats increased their alcohol intake
nd preference over time. The analyses showed significant effectsf Session for alcohol intake (g/kg) and preference [F (1,41) � 41.5nd 42.6, p � .01].
All rats acquired alcohol self-administration in context A (FigureC), as indicated by a reliable (�90%) discrimination between thective and inactive levers, and mean intake �.5 g/kg/session. Thewitch from the FR-1 schedule to the intermittent reinforcementchedules (FR-5 and VI-30) caused decreased alcohol intake (g/kg)F (1,41) � 70.9, p � .01] and increased active lever-presses [F (1,38) �8.1, p � .01].
xp. 1: Relapse to Alcohol Seeking After PunishmentFigure 2A shows the mean (� SEM) number of active lever-
resses in the final self-administration session and the context Braining sessions. The Punished group but not the Noncontingenthock or Unpunished (no shock) groups decreased their active
ever-presses over time. All rats were given at least three sessions inontext B. After the third session, rats in group Punished that re-ponded above the suppression criteria (�20 responses/2 hours)ere given additional sessions with footshock intensity increasedy .15 mA/session for up to seven sessions. Some rats from groupsnpunished and Noncontingent were also given additional ses-
ions in context B. Analysis of covariance of active lever-pressesuring the first 3 punishment days (with the final self-administra-
ion day as a covariate) showed a main effect of Group [F (2,30) �.0; p � .05] but no Group � Day interaction (p � .05). Duringontext B training, no alcohol remained in the receptacles of theunished rats, indicating that alcohol earned during these sessionsas consumed.
Figure 2B shows the mean (� SEM) number of lever-presses onhe previously active lever during the relapse test in context B (leftanel) and context A (right panel). In the Punished group, nonrein-
orced active lever responding was higher in context A than inontext B and similar to that of the Noncontingent Shock andnpunished Groups. Lever-presses in the latter two groups was
imilar in contexts A and B. Analysis of covariance of total activeever-presses showed a Group � Context interaction [F (2,29) � 7.0;� .05]. Post hoc group differences are shown in Figure 2.
Figure 2C shows the lever-presses during testing for each indi-idual rat in group Punished. There were no significant correlationsetween the number of context B punishment sessions and lever-resses during testing in either context A (r � .105; p � .05) or
ontext B (r � .284; p � .05).
EA
fisippttC
D
tsvNomcscadpst
cU
p � .05.
Fts(dtT8
N.J. Marchant et al. BIOL PSYCHIATRY 2013;73:256–262 259
xp. 2: Context-Induced Reinstatement of Alcohol Seekingfter Extinction
Figure 3A shows the mean (� SEM) active lever-presses in thenal self-administration session and the context B extinction ses-ions. The rats decreased their active lever-presses over time, asndicated by a main effect of Extinction Session [F (12,84) � 12.1,
� .01]. Figure 3B shows the mean (� SEM) number of lever-resses on the previously active lever during the reinstatement
ests in contexts A and B. The rats reinstated lever-presses in con-ext A after extinction in context B, as indicated by a main effect ofontext [F (1,7) � 18.6; p � .05].
iscussion
We found that exposure to the original alcohol self-administra-ion environment (context A) after punishment-induced suppres-ion of alcohol seeking in a different environment (context B) pro-oked relapse to alcohol seeking in alcohol-preferring P rats.onreinforced lever-presses in context A (the operational measuref alcohol seeking) of the previously punished rats was similar inagnitude to that of nonpunished rats or rats that received non-
ontingent shock in context B, which had no effect on alcoholelf-administration. Additionally, although comparison across pro-edures should be done with caution, context-induced relapse tolcohol seeking after punishment seems as robust as context-in-uced reinstatement after extinction in the traditional ABA renewalrocedure (14). We propose that our procedure can be used totudy mechanisms of relapse to alcohol seeking under conditions
C. Reinstatement test: individual rats
Rat
A. Extinction responding
Leve
r pre
sses
(2 h
)
B. Reinstatement test
0
30
60
90
120
Context A
Context B
SA 2 4 6 8 10 120
60
120
180
240
Leve
r pre
sses
(30
min
)
Context BContext A
Leve
r pre
sses
(30
min
)
0
30
60
90
120
1 2 3 4 5 6 7 8
Extinction
*
igure 3. Context-induced reinstatement of alcohol seeking after extinctionraining. (A) Mean � SEM reinforced active lever-presses during the finalelf-administration (SA) session (context A) and nonreinforced lever-pressesextinction) in context B. (B) Mean � SEM nonreinforced active lever-pressesuring the reinstatement tests in context A and context B. All rats were
ested in both contexts, and the order of testing was counterbalanced. (C)otal active lever-presses during test in each context for individual rats (n �). *Different from Context B, p � .05.
C. Relapse tests: individual rats Punished group
A. Context B training
Unpunished Non-contingent Punished
Session
B. Relapse tests Context A
0
30
60
90
120
Leve
r pre
sses
(30
min
)
Context B
*
Rat
120
0
30
60
90
Context B Context A
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Leve
r pre
sses
(30
min
)
SA 0
60
120
180
240
Leve
r pre
sses
(2 h
)
1 2 3 4 5 6 7
15 15 15 9 7
7 3
6 4
3 2 11
11 11
8 5 2 2 2 8
6 3 3 4 3 4 7 6 6 7 3 7 3 6 3 Punishment days
2 4 6 8 0
30
60
90
120
Punishment days
Leve
r pre
sses
(30
min
)
2 4 6 8
Punishment days
Context B Context A
150
150
r = 0.284
r = 0.105
Figure 2. Relapse to alcohol seeking when rats returned to the originaltraining context after response-contingent punishment in an alternativecontext. (A) Mean � SEM reinforced active lever-presses during the finalself-administration (SA) session in the initial training context (context A) andall self-administration sessions in the punishment context (context B). Thenumbers within the symbol represent n/group for each session. After thethird session, rats in group Punished with more than 20 active lever-pressesin the 2-hour session were given additional sessions with increased shockintensity; rats from the other groups were given additional sessions tomatch this. (B) Mean � SEM nonreinforced active lever-lever-presses duringthe relapse tests in context B (left panel) and context A (right panel). Lever-presses of the Punished rats (n � 15) were higher in context A than in
ontext B. Context switch had no effect on nonreinforced responding in thenpunished (n � 11) and Noncontingent Shock (n � 8) groups. (C) The top
panel shows active lever-presses during testing in each context for individ-ual rats in group Punished. The bottom panel shows correlation plots ofactive lever-presses during testing as a function of the number of context Bpunishment sessions. *Different from the other experimental conditions,
hat more closely mimic the human condition.
www.sobp.org/journal
dluFbsussefoptrwsstfmmsv(pUt
msprbptPltcoumaeosIt
spcdelipso
atpodauvabc
P
isrpcb(ea
ocit“sliir
iqaitiapoms“t5ta
260 BIOL PSYCHIATRY 2013;73:256–262 N.J. Marchant et al.
Context-Induced Relapse to Alcohol Seeking AfterPunishment
Our aim was to develop a procedure in which relapse occursafter active alcohol-taking behavior is suppressed by adverse con-sequences. We have developed this alternative procedure becausein current animal models of relapse abstinence is achieved by ex-tinction training and/or by imposing a forced abstinence period(15,38 – 40). These relapse models, however, do not fully capturethe human condition where abstinence is often initiated because ofa desire to avoid the negative consequences associated with exces-sive use (12,13,19,20). This lack of homology between the animalmodels and the human condition has been identified as a potentiallimitation of current relapse models (17,19,20).
One issue to consider with regard to the validity of our proce-ure is that response-contingent punishment is also somewhat
imited, because time-locked contingent punishment of alcoholse is likely not a common reason for abstinence in alcoholics.urthermore, we do not propose that human abstinence, imposedy punishment or negative consequences of alcohol use, exclu-ively occurs within a specific physical context. Although we havesed a physical environment to signal punishment or adverse con-equences of alcohol seeking, contextual control over the expres-ion of learned behavior is not limited to physical environments. Forxample, internal states, induced by drug effects, mood, or otheractors can lead to state-dependent learning. Consequently, mem-ry retrieval and expression of learned behaviors can become de-endent on those internal states (41– 43). Furthermore, passage of
ime can also act as a “temporal context,” with learned behaviorse-emerging after an extended time period (44). More importantly,e have shown that punishment-induced suppression of alcohol
eeking can be encoded as a context-specific memory, with alcoholeeking recovered by exposure to the original training context. Athe very least, we argue that our procedure more closely mimicsactors leading to human abstinence before relapse or, more for-
ally, it has adequate face validity—the degree to which the ani-al model seems to reflect the human condition it has been de-
igned to model (45– 47). However, as argued elsewhere, facealidity alone is not a sufficient criterion for a valid animal model48). Thus, a future research question is the predictive validity of ourrocedure, which can be determined by assessing the effect of theS Food and Drug Administration-approved alcoholism medica-
ion naltrexone (49) on context-induced relapse after punishment.What psychological mechanisms mediate relapse after punish-
ent in an alternative context? Our results demonstrate that re-ponse-contingent punishment does not permanently erase orrevent the expression of the previously established action (lever
esponse)– outcome (alcohol delivery) associations. Indeed, it haseen known for some time that punishment training does notermanently reduce the strength of the action– outcome associa-
ions after the termination of the punishment contingencies (26).unishment might share similar features to other procedures that
ead to suppression of instrumental responding, such as extinctionraining. Several lines of evidence suggest that extinction trainingauses the formation of new associations that inhibit or “mask” theriginal associations (44,46,50). Furthermore, other proceduressed to suppress learned responses, such as differential-reinforce-ent-of-other-behavior and noncontingent reward delivery, are
lso context-dependent (51,52). Thus, a common feature of differ-nt response suppression procedures is that they do not erase theriginal learned associations, and the context-dependent suppres-ion is thought to be mediated by new inhibitory associations (53).t is tempting to speculate that punishment training, like extinction
raining, causes the formation of new inhibitory associations that twww.sobp.org/journal
erve to suppress alcohol seeking in the punishment context. Oneotential test of this hypothesis is the demonstration of “relapse” inontext B after training and punishment in context A (an AAB con-ition). The outcome of such experiment is difficult to predict, how-ver, because although AAB renewal is commonly observed in Pav-
ovian conditioning (53), evidence for such renewal of extinguishednstrumental responding is weak (7,54,55). Finally, an alternativeotential learning mechanism is that contexts A and B act as “occa-ion setters” (14,56) and serve as retrieval cues that signal whetherr not alcohol-reinforced responding will be punished.
Finally, we believe that it is unlikely that motivational effects oflcohol dependence (withdrawal relief) (57,58) played a role in con-ext-induced relapse after punishment under our experimentalrocedure. This is because alcohol intake of our rats (1–2 g/kg everyther day) during training is not sufficient to induce alcohol depen-ence and, consequently, to allow rats to associate alcohol self-dministration with withdrawal relief, as demonstrated in studiessing alcohol vapor chambers (59). Future studies, combining theapor chamber procedure with our punishment-relapse procedure,re needed to determine the impact of alcohol dependence onoth suppression of alcohol self-administration by punishment inontext B and relapse in context A.
unishment-Based Models of AddictionTo our knowledge, our study is the first to demonstrate context-
nduced relapse to alcohol seeking after punishment-induced ab-tinence. As mentioned in the preceding text, Panlilio et al., (22,33)eported drug-priming-induced relapse to opiate seeking afterunishment of the opiate-reinforced response. Additionally, with aonflict procedure (60,61) in which the rats had to cross a shockarrier to gain access to the cocaine-reinforced lever, Cooper et al.
17) reported cue-induced relapse to drug seeking when rats werexposed to a discrete cue previously paired with cocaine infusionsfter suppression of drug seeking by shock exposure.
Recent studies have identified an “addiction” phenotype in rats,perationally defined as resistance to punishment after extendedocaine self-administration training (21,24,25). We also observed
ndividual differences in sensitivity to footshock punishment. Al-hough this suggests that a subset of the punished rats has anaddicted” phenotype, we found that increased shock intensityuppressed alcohol self-administration in all rats. The lack of corre-ation between number of punishment sessions and test respond-ng in both contexts (Figure 2C) suggests that individual differencesn shock-induced suppression of alcohol seeking do not predictelapse in our procedure.
Our procedure is conceptually similar to addiction models thatnvolve punishment or the introduction of an aversive conse-uence associated with drug intake. The primary focus of previouslcohol models has been to assess whether alcohol consumption
tself is resistant to taste adulteration as an experimental procedureo study “compulsive” alcohol use. Extended access to alcoholn the home-cage results in alcohol intake that is insensitive todulteration with quinine (62– 65). Other investigators have pro-osed that continuation of drug self-administration in the presencef response-contingent shock punishment can serve as an experi-ental procedure to study of compulsive drug use (21,24,25). These
tudies show that extended cocaine self-administration results incompulsive” cocaine use in a sub-group (approximately 20%) ofhe rats. We found that a larger proportion of rats (approximately0%) continued alcohol self-administration during punishment inhe first three sessions with a mild shock intensity (Figure 2A). Were hesitant to describe this as “compulsive” alcohol self-adminis-
ration, because of variable shock intensities across our operant
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
N.J. Marchant et al. BIOL PSYCHIATRY 2013;73:256–262 261
chambers. However, we did find that different rats displayed differ-ent thresholds of shock intensity to suppress alcohol self-adminis-tration below our suppression criterion. As such, in future studies, itwill be of interest to determine whether a sub-group of alcohol-preferring P rats would demonstrate resistance to punishment un-der more controlled experimental conditions and could be labeledas “compulsive.”
Finally, one advantage of the context-induced relapse after pun-ishment procedure is that the majority of rats are sensitive to theexperimental manipulation (Figure 2C). As previously mentioned,“compulsive” cocaine self-administration is typically observed inonly a small sub-group of punishment-resistant rats (approximately20%) (21,24,25). In the conflict procedure (17,66), presentation ofthe cocaine-associated discrete cue causes highly variable relapseresponding in approximately 50% of the rats and no relapse-likeresponding in other rats. Therefore, these procedures are moresuitable to study individual differences in drug-taking behavior. Bycontrast, our procedure enables more straightforward studies onneuropharmacological mechanisms of relapse after punishment-induced abstinence, because most rats demonstrate reliable re-lapse behavior during testing.
Conclusions and Future DirectionsIn humans, short- or long-term abstinence often occurs because
the negative consequences associated with excessive alcohol use out-weigh the drug rewarding effects, and during abstinence, relapse of-ten occurs after exposure to alcohol-associated contexts. We describea rat model of context-induced relapse after punishment that mimics,to some degree, this clinical scenario. An attractive feature of our pro-cedure is the reliability of the relapse responding across subjects,which seems as robust as the relapse/reinstatement response in theestablished context-induced reinstatement after extinction proce-dure.
Finally, like with any new relapse procedure, there are manyoutstanding questions that warrant further investigation, some ofwhich we list in the following text. One question that has implica-tions for medication development is whether similar or differentneuropharmacological mechanisms mediate context-induced re-lapse after punishment versus context-induced reinstatement afterextinction. Another question is the extent to which our findingsgeneralize to other drugs and other manipulations that cause rein-statement of extinguished drug and alcohol seeking. On the basisof previous research with different drugs on context-induced rein-statement after extinction (8,14), it is likely that context-inducedrelapse after punishment would occur with other abused drugs. It isalso likely that context-induced relapse after punishment wouldgeneralize to food rewards. This is because phenomena observed indrug relapse models such as priming-induced, cue-induced, andstress-induced reinstatement (15) as well as incubation of craving(67) and deprivation effects (40,68) are also observed with foodrewards (69 –72). Additionally, although we have used a physicalcontext to signal punishment, our procedure can also likely beadapted for the use of discriminative stimuli (73,74) to signal pun-ishment or nonpunishment. Finally, it will be interesting to deter-mine whether punished alcohol seeking recovers after the passageof time (spontaneous recovery) or by alcohol priming (40,46).
This work was supported by the Intramural Research Program ofthe National Institute on Drug Abuse. We thank Drs. Leigh Panlilio andMarkus Heilig for helpful comments.
The authors are employed by the United States federal governmentand report no biomedical financial interests or potential conflicts of
interest.Supplementary material cited in this article is available online.
1. O’Brien CP, Childress AR, McLellan AT, Ehrman R (1992): Classical condi-tioning in drug-dependent humans. Ann N Y Acad Sci 654:400 – 415.
2. Staiger PK, White JM (1991): Cue reactivity in alcohol abusers: Stimulusspecificity and extinction of the responses. Addict Behav 16:211–221.
3. Wikler A (1973): Dynamics of drug dependence: Implications of a con-ditioning theory for research and treatment. Arch Gen Psychiatry 28:611– 616.
4. Collins BN, Brandon TH (2002): Effects of extinction context and retrievalcues on alcohol cue reactivity among nonalcoholic drinkers. J ConsultClin Psychol 70:390 –397.
5. McCusker CG, Brown K (1990): Alcohol-predictive cues enhance toler-ance to and precipitate “craving” for alcohol in social drinkers. J StudAlcohol 51:494 – 499.
6. Bouton ME, Bolles RC (1979): Contextual control of the extinction ofconditioned fear. Learn Motiv 10:445– 466.
7. Crombag HS, Shaham Y (2002): Renewal of drug seeking by contextualcues after prolonged extinction in rats. Behav Neurosci 116:169 –173.
8. Janak PH, Chaudhri N (2010): The potent effect of environmentalcontext on relapse to alcohol-seeking after extinction. Open Addict J3:76 – 87.
9. Zironi I, Burattini C, Aicardi G, Janak PH (2006): Context is a trigger forrelapse to alcohol. Behav Brain Res 167:150 –155.
0. Hamlin AS, Blatchford KE, McNally GP (2006): Renewal of an extin-guished instrumental response: Neural correlates and the role of D1dopamine receptors. Neuroscience 143:25–38.
1. Burattini C, Gill TM, Aicardi G, Janak PH (2006): The ethanol self-admin-istration context as a reinstatement cue: Acute effects of naltrexone.Neuroscience 139:877– 887.
2. Burman S (1997): The challenge of sobriety: Natural recovery withouttreatment and self-help groups. J Subst Abuse 9:41– 61.
3. Klingemann HK (1991): The motivation for change from problem alco-hol and heroin use. Br J Addict 86:727–744.
4. Crombag H, Bossert JM, Koya E, Shaham Y (2008): Context-inducedrelapse to drug seeking: A review. Trans R Soc Lond B Biol Sci 363:3233–3243.
5. Shaham Y, Shalev U, Lu L, de Wit H, Stewart J (2003): The reinstatementmodel of drug relapse: History, methodology and major findings. Psy-chopharmacology 168:3–20.
6. Ahmed SH (2010): Validation crisis in animal models of drug addiction:Beyond non-disordered drug use toward drug addiction. NeurosciBiobehav Rev 35:172–184.
7. Cooper A, Barnea-Ygael N, Levy D, Shaham Y, Zangen A (2007): A conflictrat model of cue-induced relapse to cocaine seeking. Psychopharmacol-ogy 194:117–125.
8. Epstein D, Preston K (2003): The reinstatement model and relapse pre-vention: A clinical perspective. Psychopharmacology 168:31– 41.
9. Epstein D, Preston K, Stewart J, Shaham Y (2006): Toward a model ofdrug relapse: An assessment of the validity of the reinstatement proce-dure. Psychopharmacology 189:1–16.
0. Katz JL, Higgins ST (2003): The validity of the reinstatement model ofcraving and relapse to drug use. Psychopharmacology 168:21–30.
1. Pelloux Y, Everitt BJ, Dickinson A (2007): Compulsive drug seeking byrats under punishment: Effects of drug taking history. Psychopharma-cology 194:127–137.
2. Panlilio L, Thorndike E, Schindler C (2003): Reinstatement of punish-ment-suppressed opioid self-administration in rats: An alternativemodel of relapse to drug abuse. Psychopharmacology 168:229 –235.
3. Smith SG, Davis WM (1974): Punishment of amphetamine and mor-phine self-administration behavior. The Psychological Record 24:477–480.
4. Deroche-Gamonet V, Belin D, Piazza PV (2004): Evidence for addiction-like behavior in the rat. Science 305:1014 –1017.
5. Belin D, Balado E, Piazza PV, Deroche-Gamonet V (2009): Pattern ofintake and drug craving predict the development of cocaine addiction-like behavior in rats. Biol Psychiatry 65:863– 868.
6. Estes WK (1944): An experimental study of punishment. PsychologicalMonographs 57:1– 40.
7. Azrin NH, Holz WC (1966): Punishment. In: Honig WK, editor. Operant
Behavior: Areas of Research and Application. Englewood Cliffs, New Jer-sey: Prentice-Hall, 380 – 447.www.sobp.org/journal
2
5
5
5
5
5
5
5
5
6
6
6
6
6
6
6
6
6
6
7
7
7
7
262 BIOL PSYCHIATRY 2013;73:256–262 N.J. Marchant et al.
28. Grove RN, Schuster CR (1974): Suppression of cocaine self-administra-tion by extinction and punishment. Pharmacol Biochem Behav 2:199 –208.
9. Bergman J, Johanson CE (1981): The effects of electric shock on re-sponding maintained by cocaine in rhesus monkeys. PharmacolBiochem Behav 14:423– 426.
30. Johanson CE (1977): The effects of electric shock on responding main-tained by cocaine injections in a choice procedure in the rhesus mon-key. Psychopharmacology 53:277–282.
31. Johanson CE (1975): Pharmacological and environmental variables af-fecting drug preference in rhesus monkeys. Pharmacol Rev 27:343–355.
32. Geller I, Seifter J (1960): The effects of meprobamate, barbiturates, d-amphetamine and promazine on experimentally induced conflict in therat. Psychopharmacology 1:482– 492.
33. Panlilio LV, Thorndike EB, Schindler CW (2005): Lorazepam reinstatespunishment-suppressed remifentanil self-administration in rats. Psy-chopharmacology 179:374 –382.
34. Bouton ME, Swartzentruber D (1991): Sources of relapse after extinctionin Pavlovian and instrumental learning. Clin Psychol Rev 11:123–140.
35. Li TK, Lumeng L, McBride WJ, Waller MB, Hawkins TD (1979): Progresstoward a voluntary oral consumption model of alcoholism. Drug AlcoholDepend 4:43– 60.
36. Simms JA, Steensland P, Medina B, Abernathy KE, Chandler LJ, Wise R, etal. (2008): Intermittent access to 20% ethanol induces high ethanolconsumption in Long–Evans and Wistar rats. Alcohol Clin Exp Res 32:1816 –1823.
37. Wise RA (1973): Voluntary ethanol intake in rats following exposure toethanol on various schedules. Psychopharmacology 29:203–210.
38. Tran-Nguyen TL, Fuchs RA, Coffey GP, O’Dell LE, Baker DA, NeisewanderJL (1998): Time-dependent changes in cocaine-seeking behavior anddopamine overflow in the amygdala during cocaine withdrawal. Neuro-psychopharmacology 19:48 –59.
39. Adams CL, Short JL, Lawrence AJ (2010): Cue-conditioned alcohol seek-ing in rats following abstinence: Involvement of metabotropic gluta-mate 5 receptors. Br J Pharmacol 159:534 –542.
40. Le A, Shaham Y (2002): Neurobiology of relapse to alcohol in rats. Phar-macol Ther 94:137–156.
41. Cunningham C (1979): Alcohol as a cue for extinction: State depen-dency produced by conditioned inhibition. Learn Behav 7:45–52.
42. Bouton ME, Kenney FA, Rosengard C (1990): State-dependent fearextinction with two benzodiazepine tranquilizers. Behav Neurosci104:44 –55.
43. Overton DA (1974): Experimental methods for the study of state-depen-dent learning. Fed Proc 33:1800 –1813.
44. Bouton ME, Westbrook RF, Corcoran KA, Maren S (2006): Contextual andtemporal modulation of extinction: Behavioral and biological mecha-nisms Biol Psychiatry 60:352–360.
45. Willner P (1984): The validity of animal models of depression. Psychop-harmacology 83:1–16.
46. Bouton ME (2004): Context and behavioral processes in extinction.Learn Mem 11:485– 494.
47. Markou A, Weiss F, Gold LH, Caine B, Schulteis G, Koob GF (1993): Animalmodels of drug craving. Psychopharmacology 112:163–182.
48. Sarter M, Bruno JP (2002): Animal models in biological psychiatry. In:D’haenen H, den Boer JA, Willner P, editors. Biological Psychiatry. Hobo-ken, NJ: John Wiley and Sons, 1– 8.
49. O’Brien CP (1997): A range of research-based pharmacotherapies foraddiction. Science 278:66 –70.
50. Rescorla RA (2001): Experimental extinction. In: Mowrer RM, Klein SB,editors. Handbook of Contemporary Learning Theories. Mahwah, NJ: Erl-baum, 119 –154.
51. Kearns DN, Weiss SJ (2007): Contextual renewal of cocaine seeking inrats and its attenuation by the conditioned effects of an alternative
reinforcer. Drug Alcohol Depend 90:193–202.7
www.sobp.org/journal
2. Nakajima S, Urushihara K, Masaki T (2002): Renewal of operant perfor-mance formerly eliminated by omission or noncontingency trainingupon return to the acquisition context. Learn Motiv 33:510 –525.
3. Bouton ME (2002): Context, ambiguity, and unlearning: Sources of re-lapse after behavioral extinction. Biol Psychiatry 52:976 –986.
4. Bossert JM, Liu SY, Lu L, Shaham Y (2004): A role of ventral tegmentalarea glutamate in contextual cue-induced relapse to heroin seeking.J Neurosci 24:10726 –10730.
5. Nakajima S, Tanaka S, Urushihara K, Imada H (2000): Renewal of extin-guished lever-press responses upon return to the training context.Learn Motiv 31:416 – 431.
6. Holland PC (1992): Occasion setting in Pavlovian conditioning. In: Doug-las LM, editor. Psychology of Learning and Motivation. New York: Aca-demic Press, 69 –125.
7. Heilig M, Egli M, Crabbe JC, Becker HC (2010): Acute withdrawal, pro-tracted abstinence and negative affect in alcoholism: Are they linked?Addiction Biol 15:169 –184.
8. Koob GF, Le Moal M (2008): Addiction and the brain antireward system.Annu Rev Psychol 59:29 –53.
9. Roberts AJ, Heyser CJ, Cole M, Griffin P, Koob GF (2000): Excessiveethanol drinking following a history of dependence: Animal model ofallostasis. Neuropsychopharmacology 22:581–594.
0. Warden CJ (1931): Animal Motivation: Experimental Studies on the AlbinoRat. New York: Columbia University Press.
1. Jenkins TN, Warner LH, Warden CJ (1926): Standard apparatus for thestudy of animal motivation. J Comp Psychol 6:361–382.
2. Wolffgramm J (1991): An ethopharmacological approach to the devel-opment of drug addiction. Neurosci Biobehav Rev 15:515–519.
3. Wolffgramm J, Heyne A (1995): From controlled drug intake to loss ofcontrol: The irreversible development of drug addiction in the rat. BehavBr Res 70:77–94.
4. Lesscher HM, Van Kerkhof LW, Vanderschuren LJ (2010): Inflexible andindifferent alcohol drinking in male mice. Alcohol Clin Exp Res 34:1219 –1225.
5. Hopf FW, Chang SJ, Sparta DR, Bowers MS, Bonci A (2010): Motivation foralcohol becomes resistant to quinine adulteration after 3 to 4 months ofintermittent alcohol self-administration. Alcohol Clin Exp Res 34:1565–1573.
6. Barnea-Ygael N, Yadid G, Yaka R, Ben-Shahar O, Zangen A (2012): Cue-induced reinstatement of cocaine seeking in the rat “conflict model”:Effect of prolonged home-cage confinement. Psychopharmacology 219:875– 883.
7. Pickens CL, Airavaara M, Theberge F, Fanous S, Hope BT, Shaham Y(2011): Neurobiology of incubation of drug craving. Trends Neurosci34:411– 420.
8. Sinclair JD, Senter RJ (1968): Development of an alcohol-deprivationeffect in rats. Q J Stud Alcohol 29:863– 867.
9. Pickens C, Cifani C, Navarre B, Eichenbaum H, Theberge F, Baumann M,et al. (2012): Effect of fenfluramine on reinstatement of food seeking infemale and male rats: Implications for the predictive validity of thereinstatement model. Psychopharmacology 221:341–353.
0. Nair SG, Navarre BM, Cifani C, Pickens CL, Bossert JM, Shaham Y (2011):Role of dorsal medial prefrontal cortex dopamine D1-family receptors inrelapse to high-fat food seeking induced by the anxiogenic drug yohim-bine. Neuropsychopharmacology 36:497–510.
1. de Wit H (1996): Priming effects with drugs and other reinforcers. ExpClin Psychopharmacol 4:5–10.
2. Pinel JPJ, Rovner LI (1977): Saccharin elation effect. Bull Psychonom Soc9:275–278.
3. Katner SN, Magalong JG, Weiss F (1999): Reinstatement of alcohol-seeking behavior by drug-associated discriminative stimuli after pro-longed extinction in the rat. Neuropsychopharmacology 20:471– 479.
4. Martin-Fardon R, Weiss F (2013): Modeling relapse in animals. Curr Top
Behav Neurosci 13:403– 432.
