The ability of addictive drugs to produce enduring changes in learning about cue-reward associations and developing habits, suggest that drugs may cause long-lasting synaptic plasticity in cortico-limbic-striatal circuitry. Drugs may produce these effects via altered gene and protein expression/activity, changes in electrophysiological activity, and by modifying neural morphology. A wealth of research has clearly demonstrated that chronic use of addictive substances causes long-term changes to brain structure and function in many brain regions, including the PFC, dorsal and ventral striatum, hippocampus, hypothalamus, amygdala, and ventral tegmental area (VTA). The exact behavioral consequence of all the neurobiological changes associated with chronic drug use have not yet been determined, but certain aspects of drug-induced neuroplasticity in learning and memory circuits have been well-characterized and will be the focus of our discussion here.
Among the strongest evidence for neurobiological alterations in systems associated with reward-related learning and memory comes from a series of reports showing that chronic drug exposure increases activity of the dopamine-regulated cAMP/protein kinase A (PKA) pathway in cortico-limbic-striatal circuits (Nestler, 2004). Chronic morphine and cocaine exposure has been found to increase PKA activity in the nucleus accumbens and to some extent in the amygdala (Terwilliger, Beitner-Johnson, Sevarino, Crain, & Nestler, 1991
; Pollandt et al., 2006
). Likewise, stimulation of PKA within the amygdala facilitates stimulus-reward learning and mimics the facilitation of stimulus-reward learning found after prior chronic cocaine, amphetamine or nicotine exposure in rodents (Olausson, Jentsch, & Taylor, 2003
; Harmer, & Phillips, 1998
; Jentsch, Olausson, Nestler, & Taylor, 2002
; Taylor, & Jentsch, 2001
). Moreover, our laboratory has demonstrated that inhibition of amygdala PKA activity impairs the acquisition of appetitive stimulus-reward learning using the Pavlovian approach paradigm (Jentsch, Olausson, Nestler, & Taylor, 2002
). Therefore, because normal appetitive learning requires PKA activity, the increased PKA activity, particularly in the amygdala, produced by chronic drug exposure is likely to be one mechanism by which drug-associated memories become abnormally strong. Indeed, we have shown that inhibiting PKA activity in the amygdala after a drug-memory reactivation (discussed below) can inhibit cue-induced reinstatement, suggesting that manipulations of PKA activity might help treat addiction (Sanchez, Quinn, Torregrossa, & Taylor, 2010
PKA activation leads to the downstream phosphorylation and activation of CREB, which leads to the increased expression of several genes. Chronic cocaine and amphetamine exposure increase CREB phosphorylation and activity in the nucleus accumbens and the amygdala (Konradi, Cole, Heckers, & Hyman, 1994
; Shaw-Lutchman, Impey, Storm, & Nestler, 2003
; Mattson et al., 2005
; Brenhouse, Howe, & Stellar, 2007
), while chronic morphine and nicotine exposure decrease levels and activity of CREB in these regions (Pandey, Roy, Xu, & Mittal, 2001
; Brunzell, Russell, & Picciotto, 2003
; Widnell et al., 1996
). Consequently, the specific role CREB activity plays in the development of addiction is complicated. CREB has been shown to produce opposing roles on drug-motivated behavior depending on the brain region studied. Overexpression or activation of CREB in the nucleus accumbens has been found to reduce, while reductions in CREB activity have been found to enhance, the rewarding effects of psychostimulants (Carlezon et al., 1998
; Walters, & Blendy, 2001
). CREB, however, is required for the rewarding properties of nicotine and morphine (Walters, Cleck, Kuo, & Blendy, 2005
; Walters, & Blendy, 2001
), suggesting an important role for CREB depending on the pharmacological mechanism of action and the brain region studied. However, because CREB-regulated gene transcription is essential for virtually all forms of memory consolidation, including amygdala-dependent memory (Kida et al., 2002
; Josselyn et al., 2001
; Lonze, & Ginty, 2002
; Josselyn, & Nguyen, 2005
; Lamprecht, Hazvi, & Dudai, 1997
; Josselyn, Kida, & Silva, 2004
; Carlezon, Duman, & Nestler, 2005
), it is possible that even short-term drug-induced increases in PKA/CREB signaling in the amygdala or nucleus accumbens could contribute to the ability of stimuli to acquire enhanced conditioned reinforcing properties and increase the ability of drug-associated stimuli to drive behavior. In support of this hypothesis, our laboratory has found that nucleus accumbens infusion of BDNF, a growth factor known to activate CREB through stimulation of TrkB receptors, potentiates responding for a conditioned reinforcer, and this effect is augmented by cocaine administration (Horger et al., 1999
). On the other hand, cocaine-induced increases in CREB and BDNF activity in the medial PFC appear to put a brake on non-cued progressive ratio responding for cocaine (Sadri-Vakili et al., 2010
). Therefore, while CREB may have differential effects on the direct reinforcing properties of drugs, CREB activity may be important for stimuli associated with drugs to become conditioned reinforcers.
However, few studies have identified a direct link between persistent drug-induced neuroadaptive changes and altered processing of reward-related stimuli. One exception is the association between a time-dependent increase in the reinforcing effects of drug cues (i.e., incubation) and increases in extracellular signal regulated kinase (ERK) activity in the central nucleus of the amygdala and an increase in BDNF in the VTA, accumbens, and amygdala (Grimm et al., 2003
; Lu et al., 2005
; Li et al., 2008
). Activation of ERK also leads to the activation of CREB, and CREB phosphorylation is enhanced in the central nucleus of the amygdala during incubation of morphine conditioned place preference (Li et al., 2008
). In addition, blockade of ERK and CREB activation in this region can block incubation of the incentive motivational value of drug stimuli (Li et al., 2008
; Lu et al., 2005
). Moreover, in unpublished work from our laboratory, we have found that inhibition of PKA and CREB activity in the central amygdala can prevent the enhanced responding for conditioned reinforcement produced by cocaine, while activation of CREB mimics the effect of cocaine (Olausson et al., submitted
). Therefore, CREB activity, either through activation by BDNF and ERK, or through PKA, may cause long-lasting increases in the ability of drug cues to motivate behavior, particularly through actions in the amygdala, and possibly the nucleus accumbens.
Another transcription factor that is altered in a long-lasting manner by chronic drug exposure is DeltaFosB (Nestler, Barrot, & Self, 2001
). The effects of chronic drug exposure on DeltaFosB expression and function have been studied extensively, but the role DeltaFosB plays in mediating incentive learning associated with drugs is less well known. However, we have found that overexpression of DeltaFosB in the nucleus accumbens of transgenic mice, or by viral-mediated gene transfer in rats, increases instrumental responding and progressive ratio breakpoints for food in a similar manner to chronic exposure to several drugs of abuse (Olausson et al., 2006
). However, mice with deletions in striatal FosB show greater locomotor activity in response to cocaine and form place preferences at lower doses (Hiroi et al., 1997
). Therefore, DeltaFosB, like CREB, may affect different aspects of addiction and be temporally dependent (McClung, & Nestler, 2003
), with high levels of DeltaFosB increasing normal learning and memory processes, while also counteracting some of the intrinsic pharmacological and conditioned reinforcing effects of drugs.
One mechanism by which DeltaFosB may affect addiction-related behaviors is through increasing the expression of cyclin-dependent kinase 5 (Cdk5). Both overexpression of DeltaFosB and chronic cocaine exposure increase Cdk5 mRNA, protein, and activity in the striatum (Bibb et al., 2001
). Our laboratory has found that pharmacological inhibition of Cdk5 enhances the cocaine-induced increase in responding for conditioned reinforcement. In addition, Cdk5 inhibition increases cocaine-induced locomotor activity and break points for self-administered cocaine (Taylor et al., 2007
). Likewise, genetic and viral-mediated reduction in Cdk5 activity increased incentive motivation for food and facilitated a cocaine conditioned place preference (Benavides et al., 2007
). Therefore, chronic drug-induced increases in DeltaFosB and Cdk5 appear to be part of a feedback system that moderates the effects of drugs on behavior.
Further evidence supporting this hypothesis comes from studies of cocaine's effects on dendritic spine density in the nucleus accumbens. Chronic cocaine and amphetamine increase the density of dendritic spines and spine head size in the accumbens (Robinson, & Kolb, 1997
; Robinson, & Kolb, 1999
; Robinson, Gorny, Mitton, & Kolb, 2001
; Shen et al., 2009
). In addition, the Cdk5 inhibitor roscovitine prevents the increase in accumbens spine density produced by chronic cocaine exposure (Norrholm et al., 2003
). If the increase in accumbens spine density and Cdk5 activity produced by cocaine are indeed protective against the addictive properties of cocaine, then one would expect that preventing the drug-induced changes in spine density would produce an increase
in addiction-like behaviors. Accordingly, an inhibitor of actin cycling, latrunculin A, which prevents dynamic changes in spine morphology, actually increases cocaine-primed reinstatement of drug seeking (Toda, Shen, Peters, Cagle, & Kalivas, 2006
), providing evidence that psychostimulant-induced changes in DeltaFosB, Cdk5, and spine morphology in the accumbens are protective against some of the effects of drugs. However, it is not clear how these molecular and morphological changes influence other learning and memory processes.
In addition to the nucleus accumbens, drug-induced changes in spine morphology have been reported in the orbital frontal cortex (OFC). However, unlike the accumbens, chronic amphetamine self-administration actually decreases spine density (Crombag, Gorny, Li, Kolb, & Robinson, 2005
), while chronic morphine increases spine density (Robinson, Gorny, Savage, & Kolb, 2002
) in the OFC. In addition, chronic cocaine administration disrupts reversal learning, much like lesions of the OFC (Schoenbaum, Saddoris, Ramus, Shaham, & Setlow, 2004
), suggesting that the psychostimulant-induced reduction in spine density in this region may have detrimental effects on certain forms of learning. Indeed, neurons in the OFC of cocaine-treated rats failed to signal adverse outcomes in a decision-making task and these neurons lost the ability to reverse their cue-selectivity during behavioral reversals. Again, these data suggest that chronic drug use causes impairments in synaptic plasticity in learning and memory systems (Stalnaker, Roesch, Franz, Burke, & Schoenbaum, 2006
). On the other hand, it is unclear whether the ability of chronic morphine to increase spine density in this region has behavioral consequences. Though, one study has reported that cue-induced reinstatement of heroin, but not sucrose seeking, activates immediate early gene expression in the OFC, which indicates a possible increase in the engagement of this region that is specific for heroin-seeking behavior (Koya et al., 2006
The molecular and morphological changes produced by drugs of abuse can result in alterations in the physiology of learning and memory circuits (as discussed briefly above), which likely underlies changes in behavior. One area that has well-documented changes in physiology after drug exposure is the VTA. Dopaminergic activity in the VTA is not only critical for the pharmacological effects of most drugs of abuse, but its projections to the nucleus accumbens, PFC, and amygdala also underlie learning about rewarding and aversive events in the environment. Acute and chronic cocaine can induce long-term potentiation (LTP) at excitatory synapses on dopaminergic cells in the VTA that lasts several days (Ungless, Whistler, Malenka, & Bonci, 2001
; Borgland, Malenka, & Bonci, 2004
). LTP is a putative mechanism underlying learning and memory; however, the increase in LTP induced by cocaine in the VTA (indexed by the AMPAR/NMDAR ratio) has to date, not been correlated with specific learning effects. The AMPAR/NMDAR ratio has been shown to correlate with acute cocaine-induced locomotor activity, but not with locomotor activity after 7 days of cocaine or to a challenge dose of cocaine in a sensitization paradigm (Borgland, Malenka, & Bonci, 2004
). Therefore, chronic cocaine may alter the normal association between LTP at VTA synapses and locomotor output. In contrast to experimenter-administered cocaine, self-administered cocaine has been shown to produce an enduring (at least 3 month) increase in LTP in the VTA that is resistant to disruption by extinction training. Self-administration of natural rewards also increased LTP, but only transiently (Chen et al., 2008
). Therefore, chronic drug use may increase the drive to take drugs even after long periods of abstinence due to an abnormal enhancement of synaptic strength in the VTA.
In conclusion, chronic drug exposure not only produces long-lasting changes in behaviors relevant to learning and memory, but also produces enduring plasticity in the circuits and molecules that underlie learning and memory processes. Many of these changes occur in PKA, ERK, and CREB signaling cascades. In addition, some changes appear to occur as a negative feedback system to limit the effects of chronic drug use on behavior, including the observed increases in DeltaFosB and Cdk5 signaling. Interestingly, many published studies actually find opposite effects of psychostimulants and other classes of drugs, like opiates, on spine density and transcription factor manipulations. Therefore, the precise relevance of some of these molecules and morphological changes to the development of addiction still needs to be determined. Further study into the relationship between these molecular changes and effects on learning and memory processes associated with addiction is warranted because molecular changes that limit drug effects on locomotor activity and primary reward may still promote long-lasting increases in incentive learning and possibly the formation of drug-associated habits and compulsions.