LETTER doi:10.1038/nature12024 Rescuing cocaine-induced prefrontal cortex hypoactivity prevents compulsive cocaine seeking Billy T. Chen1, Hau-Jie Yau1, Christina Hatch1, Ikue Kusumoto-Yoshida1, Saemi L. Cho2, F. Woodward Hopf2,3 & Antonello Bonci1,3,4 Loss of control over harmful drug seeking is one of the most intract- Clinical evidence implicates prefrontal cortex (PFC) hypofunction2–4 able aspects of addiction, as human substance abusers continue to in the loss of inhibitory control over drug seeking, but a causal link pursue drugs despite incurring significant negative consequences1. between drug-induced neuroadaptations and compulsive seeking beha- Human studies have suggested that deficits in prefrontal cortical viour remains elusive. We focused on the prelimbic cortex, because a function and consequential loss of inhibitory control2–4 could be homologous region in the human PFC9,10 regulates many cognitive crucial in promoting compulsive drug use. However, it remains functions, including decision making and inhibitory response con- unknown whether chronic drug use compromises cortical activity trol11–15, which are compromised in human addicts3. We propose that and, equally important, whether this deficit promotes compulsive chronic cocaine use induces prelimbic cortex hypoactivity, and that cocaine seeking. Here we use a rat model of compulsive drug seek- compromised prelimbic cortex function in turn impairs inhibitory ing5–8 in which cocaine seeking persists in a subgroup of rats despite a delivery of noxious foot shocks. We show that prolonged cocaine 4–5 weeks 2–3 weeks 1–2 weeks self-administration decreases ex vivo intrinsic excitability of deep- layer pyramidal neurons in the prelimbic cortex, which was sig- Seek-take Long-access Seek-take Contingent nificantly more pronounced in compulsive drug-seeking animals. task (30 inf) (80 inf) (baseline) foot shocks Furthermore, compensating for hypoactive prelimbic cortex neurons b c with in vivo optogenetic prelimbic cortex stimulation significantly Baseline Shock NS 500 prevented compulsive cocaine seeking, whereas optogenetic prelimbic Sensitive 500 NS *** cortex inhibition significantly increased compulsive cocaine seeking. 400 Resistant 400 † Our results show a marked reduction in prelimbic cortex excitability 300 300 in compulsive cocaine-seeking rats, and that in vivo optogenetic pre- 200 limbic cortex stimulation decreased compulsive drug-seeking beha- 200 viours. Thus, targeted stimulation of the prefrontal cortex could serve 100 100 Seek lever presses Seek lever presses as a promising therapy for treating compulsive drug use. 0 0 To investigate the neurobiological dysfunctions that underlie com- –3 –2 –1 0 1 2 3 4 Base Base pulsive drug use, we used an animal model of addiction5 previously Day Shock Shock e shown to engender a subpopulation of rats with compulsive cocaine d Baseline Shock seeking. Rats underwent extended cocaine self-administration training 40 40 NS (Fig. 1a) under a heterogeneous seeking-taking chain schedule5 (details *** *** 30 30 in the Supplementary Information). Daily self-administration sessions † consisted of 30 trials that began with insertion of the seek lever. After 20 20 one lever press and a variable interval (5 to 75 s), one additional press on the seek lever resulted in seek lever retraction and take lever 10 10 Earned cocaine (inf) insertion. One take lever press triggered a single intravenous cocaine Earned cocaine (inf) 0 0 infusion and retraction of the take lever. After .8 weeks of self- –3 –2 –1 0 1 2 3 4 Base Base administration training, rats received four days of shock sessions in Day Shock Shock which 30% of the seek link ended with delivery of a noxious foot shock Figure 1 | Prolonged cocaine self-administration induces compulsive $ (0.4 mA, 0.5 s). Rats earning 10 cocaine infusions (14 of 46 rats, 30%, cocaine seeking in a subpopulation of rats. a, Experimental timeline showing 24.1 6 1.9 infusions) in the fourth shock session were considered the seek-take task and long-access sessions before the beginning of the foot shock-resistant, whereas rats earning ,10 infusions (32 of 46 rats, shock sessions. After approximately two months of cocaine self-administration, 70%, 1.3 6 0.3 infusions) were considered shock-sensitive (Supplemen- rats received four days of shock sessions, in which a foot shock was tary Fig. 1). Previous works have observed a relatively similar propor- administered in 30% of trials. b, Foot shocks decreased seek-lever presses in tion of shock-resistant rats5–8 and suggested that these rats exhibit shock-sensitive rats (n 5 32) but not shock-resistant rats (n 5 14). c, Seek-lever greater basal impulsivity but not higher reactivity to novelty6.Seeklever presses on last day of baseline and last day of shock (arrowheads in b) presses (Fig. 1b, c) and earned cocaine infusions (Fig. 1d, e) were sig- (F(3,68) 5 27.89, P , 0.001; Tukey’s post-hoc: ***P , 0.001 baseline versus shock sessions; {P , 0.001 shock-sensitive versus shock-resistant shock nificantly inhibited by foot shocks in shock-sensitive rats, whereas session). d, Foot shocks greatly reduced cocaine infusions in shock-sensitive shock-resistant rats were much less affected (Fig. 1b–e). Baseline rats. e, Earned cocaine infusions (inf) on last day of baseline and last day of cocaine-seeking behaviours and earned cocaine infusions were similar shock (F(3,68) 5 246.8, P , 0.001; Tukey’s post-hoc: ***P , 0.001 baseline in both groups (Fig. 1c, e), suggesting that divergent responses to con- versus shock session; {P , 0.001 shock-sensitive versus shock-resistant shock tingent foot shocks did not reflect cocaine self-administration history8. sessions). NS, not significant. Error bars indicate s.e.m. 1Intramural Research Program, National Institute on Drug Abuse, Baltimore, Maryland 21224, USA. 2Ernest Gallo Clinic and Research Center, Department of Neurology, University of California San Francisco, San Francisco, California 94608, USA. 3Department of Neurology, University of California San Francisco, California 94110, USA. 4Solomon H. Snyder Neuroscience Institute, Johns Hopkins School of Medicine, Baltimore, Maryland 21205, USA. 18APRIL2013|VOL496|NATURE|359 ©2013 Macmillan Publishers Limited. All rights reserved RESEARCH LETTER control over compulsive drug seeking. Thus, we performed ex vivo (Fig. 2b) and exhibited significantly smaller input resistance electrophysiology to determine whether cocaine self-administration (Fig. 2c). Thus, the ability of prelimbic cortex neurons to rapidly reduced prelimbic cortex intrinsic excitability, because action potentials generate an action potential was compromised in shock-resistant rats. are the main mechanism of neuronal communication and altered spike In addition, prelimbic cortex neurons from shock-resistant and shock- firing contributes to neuronal dysfunction16 and central nervous system sensitive rats showed impairments in repetitive firing induced by disorders17. a more moderate, sustained depolarizing currents (1 s, 0–500 pA, Intrinsic excitability of deep-layer, morphologically confirmed 50 pA steps), with a significantly greater impairment in shock-resistant (Fig. 2a)18 pyramidal prelimbic cortex neurons from shock-sensitive versus shock-sensitive rats (Fig. 2d, e). Also, cocaine intake under and shock-resistant rats was evaluated 24 h after the fourth shock shock correlated with decreased input resistance and increased current session, and was compared with naive rats. We targeted deep-layer required for firing across all rats tested (Supplementary Fig. 2). pyramidal prelimbic cortex neurons because they project to brain Analysis of the action potential shape during whole-cell recordings structures implicated in drug-seeking behaviours, including the nuc- and nucleated patch electrophysiological recordings revealed no dif- leus accumbens, dorsal striatum and amygdala15,19,20. We first mea- ferences in Na1 channel activation, or in the composition of somatic sured the amount of current required to induce an action potential in K1 currents or barium-sensitive K1 conductances (Supplementary response to brief depolarization (2 ms, 0–2,500 pA, 10 pA steps). Table 1 and Supplementary Figs 3 and 4) among naive, shock-sensitive Relative to naive and shock-sensitive neurons, shock-resistant neurons and shock-resistant rats. Control experiments in which shock- required almost twice as much current to elicit an action potential sensitive rats self-administered additional cocaine after each shock session suggested that the quantity of cocaine per se was not respon- a sible for firing adaptations in shock-resistant rats (Supplementary Fig. 5). Thus, our results demonstrate that long-term cocaine self- administration reduced prelimbic cortex excitability, with a much more pronounced effect in compulsive rats. If compromised spike firing was causally related to compulsive cocaine seeking, then increas- ing prelimbic cortex neuronal activity could decrease compulsive b c behaviour. One way to compensate for reduced excitability is to take advantage of in vivo optogenetic stimulation21,22, in which prelimbic 10 mV cortex neurons can be activated in a temporally precise manner during 500 ms cocaine self-administration trials. Deep-layer prelimbic cortex neurons 2,500 200 *** expressing channelrhodopsin-2 (ChR2) (Supplementary Fig. 6a) 2,000 150 ) showed minimal loss of spike fidelity at photostimulation frequencies (pA) 1,500 Ω †** up to 20 Hz both ex vivo and in vivo (Supplementary Fig. 6b, c). Also, (M 100 in