NEUROPSYCHOPHARMACOLOGY 1994-VOL. 11, NO.2 77

Molecular Neurobiology of Drug Eric J. Nestler, M.D., Ph.D.

The purpose of this review is to illustrate the ways in in the brain that lead to addiction; and (2) mechanisms of which molecular neurobiological investigations will individual risk, identification of specific genetic and contribute to an improved understanding of drug environmental factors that increase or decrease an addiction and, ultimately, to the development of more individual's vulnerability for addiction. This information effective treatments. Such molecular studies of drug will one day lead to fundamentally new approaches to the addiction are needed to establish two general types of treatment and prevention of addictive disorders. information: (1) mechanisms of pathophysiology, [Neuropsychopharmacology 11:77-87, 1994J identification of the changes that drugs of abuse produce

KEY WORDS: Synaptic transmission; Drug addiction; extracellular aspects of synaptic transmission, namely, G proteins; Ventral tegmental area; Cyclic AMP; on and their interactions with reup­ Nucleus accumbens take sites, receptors, and ion channels located through­ out the brain. More recent molecular studies have shed critical new light on this level of understanding of syn­ EVOLVING MODELS OF SYNAPTIC aptic transmission. The cloning of numerous subtypes TRANSMISSION of reuptake proteins, receptors, and channels has refinedour understanding of the mechanisms of action The most important area where molecular neurobiolog­ of psychotropic drugs. This information is also being ical investigations have contributed to drug addiction, used to develop drugs that interact with various pro­ and to biological in general, is the increas­ tein subtypes with ever-increasing specifIcity. Such ingly complete and sophisticated picture of synaptic drugs will provide invaluable tools in basic studies of transmission these studies have provided. Figure 1 drug addiction and could also represent improved phar­ shows a working model of synaptic transmissionaround macotherapeutic agents, although this latter possibil­ the time (1968-1976) that the frrst two volumes of the ity remains conjectural. American College of Neuropsychopharmacology's By the time the Third Generation of Progress was pub­ Generation of Progress series were published. During this lished in 1987 (Meltzer 1987), there was a much greater time, most studies in biological psychiatry focused on appreciation for the complexity of synaptic transmis­ sion (Figure 2). It was known that ­ receptor regulation of ion channels represents only a Based on a presentation at the President's Plenary Session, An­ small fraction of a neurotransmitter's effects on its tar­ nual Meeting of the American College of Neuropsychopharmacol­ get neurons. In addition to regulation of ion channels, ogy, Honolulu, Hawaii, December 14, 1993. From the Laboratory of Molecular Psychiatry, Departments of Psy­ it was becoming increasingly clear that neurotransmit­ chiatry and Pharmacology, Yale University School of Medicine, New ters regulated virtually all processes that occurred Haven, Connecticut. Address correspondence to: Eric J. Nestler, M.D., Ph.D., Labora­ within neurons, including gene expression. It was tory of Molecular Psychiatry, Departments of Psychiatry and Phar­ known further that many of the effects of neurotrans­ macology, Yale University School of Medicine, and Connecticut Men­ mitters on ion channels, and all of the other effects of tal Health Center, Park Street, New Haven, CT 34 06508. neurotransmitters on target neurons, are achieved Received January 12, 1994; revised February 18, 1994; accepted March 18, 1994. through biochemical cascades of intracellular mes-

© 1994 American College of Neuropsychopharmacology Published by Elsevier Science Inc. 655 Avenue of the Americas, New York, NY 10010 0893-133X/94/$7.oo 78 NEUROPSYCHOPHARMACOLOGY 1994-VOL. 11, E.J. Nestler NO.2

sengers. These intracellular messengers consist of G proteins (GTP-binding membrane proteins), second messengers (such as cAMP and Ca2+), and protein phosphorylation whereby the addition or removal of phosphate groups from all types of neural proteins alters their function and leads to the myriad of biologi­ cal responses in question (see Nestler and Duman 1994; -- Nestler and Greengard 1994). receptors As the Fourth Generationof Progress (Bloom and Kup­ neurotransmitters fer, 1994) approaches publication, we have gained an channels / even greater appreciation of the complexity of synap­ \ /" / tic transmission. As shown in Figure 3, we now know that neurotransmitter-receptor regulation of G proteins and second messenger-dependent protein phosphory­ Figure 1. A working model of synaptic transmission, circa lation pathways represents a small part of a neuron's 1968-1976. During this period of time, synaptic transmission intracellular regulatory machinery (see Duman and was conceived as the release of neurotransmitter from a nerve Nestler 1994). Neurons express high levels of trans- terminal, the binding of the neurotransmitter to specmc recep­ tor sites on target neurons, and the resulting alterations in the conductances of specmc ion channels. The action of the neurotransmitter is then terminated by its reuptake into the ing coupling factors (termed G proteins), second messengers 2 nerve terminal. [e.g., cAMP, cGMP, Ca +, nitric oxide (NO), and the metab­ olites of phosphatidylinositol (PI) and arachidonic acid (AA)], and protein phosphorylation (involving the phosphorylation of phosphoproteins by protein kinases and their dephosphor­ ylation by protein phosphatases), in mediating multiple ac­ )� tions of neurotransmitters on their target neurons. Second messenger-dependent protein kinases (e.g., those activated � by cAMP or Ca2+) are classmed as protein serine/threonine kinases, because they phosphorylate substrate proteins on o serine or threonine residues. Each second messenger­ channels � dependent protein kinase phosphorylcites a specmc array of �-I G-proteins I substrate proteins (that can be considered third messengers) and thereby leads to multiple biological responses of the neu­ � rotransmitter. Second Messengers: The multiple biological responses to a neurotransmitter cAMP cGMP caz· PI NO AA can be divided into three main categories. In some cases, in­ tracellular messengers mediate the actions of some neu­ I I I I rotransmitters in opening or inhibiting particular ion chan­ PROTEIN PHOSPHORYLATION nels. However, intracellular messengers mediate most of the protein kinase many other actions of neurotransmitters on their target neu­ -----.. Depho�phO- Phospho- rons. Some are relatively short-lived and involve modulation ... proteins proteins of the general metabolic state of the neurons, their ability to � "------' :::J protein phosphatase synthesize or release neurotransmitter, and the functional sen­ Q) CJ sitivity of their various receptors and ion channels to various I'll + + + + ... synaptic inputs. Others are relatively long-lived and are - c:: I Diverse Biological ResponsesI achieved through the regulation of gene expression in the tar­ get neurons. Thus, neurotransmitters, through the regula­ / � � tion of intracellular messenger pathways and alterations in Rapid Short-term Long-term gene transcription and protein synthesis, alter the numbers mediatory modulatory modulatory and types of receptors and ion channels in target neurons, processes processes processes the functional activity of the intracellular messenger systems (Regulation of in those neurons, and even the shape and numbers of syn­ gene expression) apses the neurons form. The fIgure is drawn to illustrate the amplmcation that intracellular messenger systems can give Figure 2. A working model of synaptic transmission, circa to neurotransmitter action: the single event of a neurotrans­ 1987. Studies in molecular from 1976 to 1987 mitter binding to its receptor (the fIrst messenger level) can provided a much more complex view of synaptic transmis­ act through the second, third, fourth, etc. messenger levels sion than that shown in Figure 1. These studies focused on to produce an increasingly wider array of physiological effects. the involvement of intracellular messenger systems, involv- Modmed from Hyman and Nestler 1993. NEUROPSYCHOPHARMACOLOGY 1994-VOL. 11, NO.2 Molecular Neurobiology of Drug Addiction 79

membrane protein tyrosine kinases (such as the trk pro­ teins) that serve as plasma membrane receptors for neu­ 1� rotrophins and many other growth factors. In addition DRUG � to their role in development, these growth factors are now known to produce profound effects on fully differ­ Neurotrophins entiated, adult neurons. In fact, the traditional distinc­ � tion between a neurotransmitter and a neurotrophin "0 is becoming increasingly arbitrary. Neurons also ex­ press large numbers of cytoplasmic protein tyrosine ki­ nases (such as src kinase) and protein serine/threonine kinases (such as the ERKs or MAP-kinases). Although 2nd messenger these cytoplasmic enzymes are not regulated directly dependent protein by extracellular signals, they are under the control of kinases extracellular signals via second messenger-dependent and growth factor-dependent pathways. 2nd messenge protein The scheme shown in Figure 3 is still a simplifIed independent tyrosine version of our current knowledge of synaptic transmis­ protein ser/thr kinases kinases sion. The fIguredoes not include the many types of pro­ (e.g., src) (e.g., ERK) tein phosphatases associated with each of these pro­ � � tein kinase systems that are also subject to regulation I Diverse Biological Responsesl by extracellular and intracellular signals. The fIgurealso does not adequately depict the numerous interactions t � that exist among the various intracellular messenger / Rapid Short-term Long-term pathways. This high degree of interactions means that mediatory modulatory modulatory a primary perturbation in one particular pathway pro­ processes processes processes (Regulation of duces secondary, tertiary, etc., changes in other path­ gene expression) ways, with regulation of these other pathways con­

tributing eventually to the many biological responses Figure 3. A working model of synaptic transmission, circa of the original perturbation. 1994. It is now apparent that the brain contains many impor­ This information means that although most psy­ tant intracellular regulatory pathways in addition to those chotropic drugs interact initially with proteins located regulated directly by G proteins and second messengers. Brain at the extracellular aspect of the (e.g., cocaine contains numerous protein serine/threonine kinases that are not regulated directly by second messengers (e.g, the ERKs inhibits monoamine reuptake proteins and opiates ac­ or MAP kinases). In addition, the brain contains numerous tivate opioid receptors), the numerous acute actions of types of protein tyrosine kinases (that phosphorylate substrate these drugs are produced ultimately through the many proteins on tyrosine residues), some of which reside in the intracellular messenger pathways that mediate these receptors for neurotrophins and most other growth factors extracellular signals. Moreover, these intracellular mes­ (e.g., the trk proteins), and others that are not associated with senger pathways play a central role in mediating the growth factor receptors (e.g., src kinase). Each of these vari­ long-term effects that these drugs exert on brain func­ ous protein kinases are highly regulated by extracellular stim­ tion. This information, then, provides a conceptual uli. The second messenger-dependent protein kinases are framework within which highly specifIc and detailed regulated by receptor--second messenger pathways hypotheses can be formulated concerning the precise as shown in Figure 2. The receptor-associated protein tyro­ mechanisms by which psychotropic drugs regulate sine kinases are activated upon growth factor binding to the receptor. The second messenger-independent protein ser­ brain function both acutely and chronically. ine/threonine kinases and the protein tyrosine kinases that are not receptor associated seem to be regulated indirectly via the second messenger-dependent and growth factor­ HYPOTHETICAL ROLE OF G PROTEINS dependent pathways as depicted in the figure. The brain also AND THE cAMP PATHWAY IN DRUG contains numerous types of protein serine/threonine and pro­ REINFORCEMENT AND CRAVING tein tyrosine phosphatases, not shown in the figure, which are also subject to regulation by extracellular and intracellu­ A large behavioral and pharmacological literature has lar stimuli. Thus, the binding of neurotransmitter to its recep­ revealed considerable information concerning the ex­ tor extracellularly results in numerous short- and long-term tracellular actions of drugs of abuse (see Wise 1990; Ku­ biological responses through the complex regulation of mul­ har et a1. 1991; K.oob 1992). For example, as shown in tiple intracellular regulatory pathways and the phosphoryla­ tion or dephosphorylation of numerous substrate proteins. Figure 4, cocaine, via increasing synaptic levels of dopa­ ModifIed from Duman and Nestler 1994. mine, would activate all known subtypes of 80 E.J. Nestler NEUROPSYCHOPHARMACOLOGY 1994-VOL. 11, NO. 2

cocaine Based on our knowledge of intracellular messenger t pathways (see Duman and Nestler 1994), speciflc hy­ potheses can be formulated concerning the post­ eel-+ DA receptor mechanisms underlying the acute and chronic

01,5 /"" D2-4 actions of drugs of abuse on the mesolimbic dopamine system. A small portion of such hypotheses, focused on regulation of the cAMP pathway, is shown in Fig­ ure 4. Cocaine activation of D1 and D5 dopamine recep­ tors would be expected to activate the cAMP pathway via the stimulation of Gs, the activation of adenylyl cy­ clase, the generation of cAMP, and the activation of cAMP-dependent protein kinase. The protein kinase (protein kinase A) would then phosphorylate numerous substrate pro­ teins, which would mediate many of the acute actions /I�\" of cocaine on mesolimbic dopamine function. Cocaine substrate proteins (multiple ) would also activate D2, D3, and D4 dopamine recep­ /I�\" tors, which may be located on a partially differentsub­ set of neuronal elements in the mesolimbic dopamine multiple physiological effects ( ) system, although this remains to be established with certainty. Activation of these receptors would exert the Figure4. Hypothetical role of G proteins and the cAMPpath­ way in mediating some of the acute effectsof heroin, cocaine, opposite effect on the cAMP pathway, through the and alcohol on mesolimbic dopamine function. This model recruitment of Gi and Go and the inhibition of adenylyl is based on the known extracellular mechanisms of action of cyclase. Heroin activation of most types of opioid recep­ these drugs and the known receptor and post-receptor mech­ tors, which are also coupled to Gi and Go, would pro­ anisms that would be expected to follow. The fIgure focuses duce similar effects. Even alcohol, through direct ac­ on G proteins and the cAMP pathway and therefore repre­ tions on NMDA glutamate receptors and indirect sents only a small portion of the many receptor and postrecep­ regulation of the state of deplorarization of neurons, tor mechanisms that likely mediate acute drug effects on this would be expected to regulate the cAMPpathway. This and other neural pathways. is based on the knowledge that cellular levels of Ca2+ can exert powerful stimulatory or inhibitory effects on , depending on the type of adenylyl receptors throughout the brain. Heroin would activate cyclase expressed in neurons of the mesolimbic dopa­ the growing number of opioid receptors now known mine system (see Nestler and Duman 1994). It should to be expressed in the brain. Recent work has even be emphasized that each of these various actions of co­ provided insight into some of the proteins with which caine opiates, and alcohol could potentially occur both alcohol interacts initially, such as the NMDA glutamate in presynaptic and postsynaptic elements in these brain receptor (as depicted in the flgure), as well as the regions. Clearly, the information provided in Figure 4 GABAA receptor and L-type calcium channels (not represents only a small portion of the expected conse­ shown) (Wafford et al. 1991; Morrisett and Swartz­ quences of activation of the various receptors and G welder 1993). Clearly, cocaine, heroin, and alcohol pro­ proteins indicated; the flgure fails, for example, to show duce very differenteffects in different brain regions and expected regulation of the Ca2+ and phosphatidylino­ peripheral tissues, but growing evidence has shown sitol pathways (see Agranoff and Fisher 1994; Nestler that the actions of these drugs converge on the mesolim­ and Greengard 1994). As information is obtained con­ bic dopamine system (See Wise 1990; Kuhar et al. 1991; cerning the role of the cAMP pathway in drug addic­ Koob 1992). This neural pathway consists of doparniner­ tion, the possible involvement of these other intracel­ gic neurons in the ventral tegmental area (VT A) and lular messenger pathways must also be investigated. their numerous projections, particularly those to the nucleus accumbens (NAc) and prefrontal cortex. The DIRECT EVIDENCE FOR A ROLE OF G PROTEINS mesolimbic dopamine system appears to play a critical AND THE cAMP PATHWAY IN DRUG role in mediating the acute reinforcing actions of these REINFORCEMENT AND CRAVING and other drugs of abuse. In addition, the mesolimbic dopamine system may be one of the sites where drugs The tools of molecular neurobiology coupled with more of abuse produce chronic adaptations that underlie the traditional neuropharmacological disciplines can now long-term changes in drug reinforcement mechanisms directly test the validity of the hypotheses shown in Fig­ (e. g., craving) that are the core feature of addiction clin­ ure 4. For example, growing evidence supports a role ically (Nestler 1992; Nestler et al. 1993). for Gi and Go in the acute and chronic actions of drugs NEUROPSYCHOPHARMACOLOGY 1994-VOL. NO. 2 11, Molecular Neurobiology of Drug Addiction 81

Table 1. Evidence Supporting a Role for Gi and Go in Table 2. Evidence Supporting a Role for the cAMP the Acute and Chronic Actions of Drugs of Abuse on Pathway in the Acute and Chronic Actions of the Mesolimbic Dopamine System Drugs of Abuse on the Mesolimbic Dopamine System

1. Direct inactivation of Gi/Go in the NAc regulates 1. Direct activation or inactivation of the cAMP cocaine and heroin self-administration behavior. pathway in the NAc exerts opposite effects on 2. Chronic administration of cocaine or opiates cocaine and heroin self-administration behavior. decreases levels of Gi/Go in the NAc. 2. Chronic administration of cocaine, opiates, or alcohol 3. Chronic administration of cocaine decreases levels of upregulates the cAMP pathway in the NAc. Go in the VTA, an effect that parallels the 3. Direct activation of the cAMP pathway in the NAc development of locomotor sensitization, and direct facilitates cocaine-induced locomotor activity and inactivation of Gi/Go in the VTA mimics locomotor locomotor sensitization. sensitization. Based on data from Terwilliger et aI. 1991; Cunningham and Kel­ Based on data from Nestler et aL 1990; Steketee et aL 1991; Ter­ ley 1993; Self et al. 1993; Miserendino and Nestler 1994; Nestler et williger et aL 1991; Striplin and Kalivas 1992; Self et al. 1994. aL 1994.

of abuse on the rnesolirnbic dopamine systern (Table tations in the G proteins rnay represent part of the 1). First, direct rnanipulationof these G proteins in the pathophysiological rnechanisrns underlying drug ad­ NAc has been shown to regulate cocaine and heroin diction. self-adrninistration behavior (Self et al. 1994). In these Chronic cocaine is al�o known to decrease levels experirnents, pertussis toxin, which inactivates Gi and of Go in the VTA (Nestler et al. 1990), changes that Go, was administered bilaterally into the NAc. It was correlate ternporally with the induction of locornotor found that this treatrnent produces a rightward shift sensitization (Striplin and Kalivas 1992) (see Table 1). in the dose-response functions for both cocaine and Moreover, local adrninistration of pertussis toxin into heroin self-administration (Figure 5). It was further the VTA has been shown to mimic aspects of locorno­ found that the tirne course by which pertussis toxin ad­ tor sensitization behavior (Steketee et al. 1991). These ministration produces these changes in drug self­ data indicate that Go rnay play a role in the acute and administration closely parallels the time course by chronic effectsof cocaine on locornotoractivity, another which the toxin inactivates Gi and Go in this brain re­ behavior known to be regulated by the rnesolirnbic gion as determined by biochernical rneasures (Self et dopamine systern (see Kalivas and Stewart 1991). al., 1994). In addition, in separate studies, chronic ad­ Similar types of inforrnation have been obtained to ministration of opiates or cocaine has been shown to support a role of the cAMP pathway in drug reinforce­ reduce levels of Gi and Go selectively in the NAc (Nes­ rnent rnechanisrns (Table 2). These studies have rnade tler et al. 1990; Terwilliger et al. 1991). Together, these use of cholera toxin, which activates Gs, and analogs studies provide increasing evidence for the hypothesis of cAMP that regulate cAMP-dependentprotein kinase that these G proteins mediate sorne of the acute rein­ (Self et al. 1993). For exarnple, as shown in Figure 6, forcing actions of drugs of abuse, and that chronic adap- direct administration of activators or inhibitors of the

Cocaine Heroin

120 - 60 PTX ---+- PTX ---0--- c: co xPTX 0 0 -0- xPTX -;;; 'u; '" en '" '" '" 80 en 40 Uic Uic: .S! .S! c::;'" c::;'" 20 � 40 � 'c '"

.375 0.75 1.50 .015 .03 .06 Dose (mg/kg/inj) Dose (mg/kg/inj)

Figure 5. Effectsof bilateral pertussis toxin (PTX) injections into the NAc on the dose-response relationship of intravenous cocaine or heroin self-administration. The data are expressed as the mean ± SEM of self-administration totals as a function of injection dose for daily 3-hour sessions conducted at the end of maintenance testing. Control rats received similar injec­ tions of inactive pertussis toxin (xPTX) . Asterisks indicate that values differ from xPTX-treated controls; • p < .01, student's t-test. From Self et al. 1994. NEUROPSYCHOPHARMACOLOGY 1994-VOL. 11, NO. 2 82 E.]. Nestler

c: 120 .2 .E"i" gene expression. There are two general types of evi­ i3 Qi 80 dence to suggest that alterations ingene expression may Q) III1/1 ·c .c Sp-cAMPS contribute importantly to drug addiction. First, drug . - E 40 Rp-cAMPS regulation of many of the target proteins identifIed to Qj 0 1/1 � date in the mesolimbic dopamine system (e.g., G pro­ Q) Q) 01 c: c: teins and the cAMP pathway) occurs at the messenger .cIII uIII u ·40 RNA level (see Nestler et al. 1993; Vrana et al. 1993). 0 ::.e Chronic cocaine treatment has also been reported re­ 0 L ·80 Sal 40 80 cently to regulate mRNA expression of the dopamine nmoi/ill transporter in the VTA-NAc pathway (Cerruti et al. 1994). Second, it has been known for many years that Figure Effectsof bilateralcAMP analog injections into the 6. many prominent features of drug addiction in people NAc on the maintenance of preestablished intravenous self­ administration of cocaine (0.5 mg/kg/injection). The cAMP and laboratory animals can persist for a long time, per­ analogs used included: Sp-cAMPS (an activator of cAMP­ haps a lifetime, after discontinuation of drug exposure. dependent protein kinase) and Rp-cAMPS (an inhibitor of These considerations have stimulated many investiga­ cAMP-dependent protein kinase). The data are expressed as tors to examine directly the regulation of gene expres­ the mean ± SEM of self-administration totals from daily sion by drugs of abuse. 3-hour test sessions following the injection of the cAMP ana­ These studies have focused on classes of nuclear log or vehicle. Asterisks indicate that values differ from regulatory proteins, termed transcription factors, that vehicle-treated control; • p < .05 by student's t-test. Data bind to specifIc sequences of DNA located in the pro­ from Self et al. 1993. moter regions of certain genes and thereby increase or decrease the rate at which those genes are transcribed protein kinase into the NAc produces opposite effects (Hymanand Nestler 1993). Prominentexamples of tran­ 6), on cocaine self-administration(Figure and more pre­ scription factors in the brain include CREB (cAMP re­ liminary data indicate a similar action on heroin self­ sponse element binding protein), Fos, and related pro­ administration. In addition, chronic treatment of rats teins. Figure 7 illustrates a general scheme by which with opiates, cocaine, or alcohol has been shown to up­ drug regulation of gene expression may contribute to regulate the cAMP pathway in the NAc, with drug­ addiction. Drug-induced alterations in intracellular induced increases seen in levels of adenylyl cyclase and messenger pathways would result in alterations in tran­ cAMP-dependentprotein kinase (Terwilliger et al. 1991; scription factor function, either through the directphos­ Nestler et al. 1994). Together, these fIndings provide phorylation of transcription factors or through regula­ the fIrst direct evidence for the hypothesis (Nestler 1992; tion of their expression. Such regulation of transcription Nestler et al. 1993) that the cAMP protein phosphory­ factors would then alter the expression of specifIc tar­ lation pathway does indeed regulate the acute reinforc­ get proteins, which would result in the adaptive ing properties of drugs of abuse, and that chronic adap­ changes in brain function that are responsible for ad­ tations in this intracellular pathway may represent part diction. Although there is a plethora of reports that of the pathophysiological mechanisms involved in drug acute drug treatment or other acute stimuli can result addiction. Similar types of studies support a role of the in regulation of transcription factors in the brain, what cAMP pathway in the NAc in cocaine-induced locomo­ would appear to be particularly relevant to addiction tor activity and sensitization (Cunningham and Kelley per se are long-term changes in nuclear regulatory 1993; Miserendino and Nestler 1994). mechanisms, which have been more diffIcultto identify. Ultimately, acute and chronic drug regulation of G Recent work has made progress in this area by proteins and the cAMP pathway will result in changes delineating long-term effects of cocaine on Fos-like tran­ in mesolirnbic dopamine function through the phos­ scription factors. Several laboratories have reported that phorylation, by cAMP-dependent protein kinase, of acute administration of cocaine or other stimulants numerous types of substrate proteins. A major focus results in the induction of Fos and several related im­ of current research, then, is on identifying the specifIc mediate early gene products (e.g., FosB, Fra2, Jun, and substrates whose phosphorylation mediates drug rein­ JunB) in the NAc and caudate/putamen (Graybiel et al. forcement mechanisms in the VTA-NAc pathway 1990; Young et al. 1991; Hope et al. 1992; Cole et al. acutely and chronically. 1992; Nguyen et al. 1992). Induction of these proteins is associated, as would be expected, with an increase A ROLE FOR ALTERATIONS IN GENE in AP-1 DNA binding activity (Young et al. 1991; Hope EXPRESSION IN DRUG ADDICTION et al. 1992). AP-1 refers to the sequence of DNA to which Fos and related proteins bind. Among the many possible substrate proteins involved In contrast, chronic cocaineadministration produces in drug addiction mechanisms are those that regulate very different actions on this transcription factor fam- NEUROPSYCHOPHARMACOLOGY 1994-VOL. 11, NO. 2 Molecular Neurobiology of Drug Addiction 83

First messengers .. /Drug recognizes a moiety in the Fos molecule that is shared r----A ReceptorsN by numerous Fos-like proteins. These blot immuno­ � IG.�sl labeling studies have shown that acute and chronic co­ caine treatments are associated with the induction of � different Fos-like proteins, and that some of the Fos-like ---, "'---::-:-""""'-;-1 1 - 2nd mes engers 1 Mult p e . ; proteins induced under chronic cocaine-treated condi­ Phosphoproteins .----- (Third Me;sengers) tions do not correspond to any known proteins. These _I Protein kinases 1 MULTIPLE novel proteins, designated chronic Fra's (Fos-related an­ PHYSIOLOGICAL tigens), show different functional properties compared RESPONSES to c-Fos and the other acute Fra's. First, the chronic Fra's nucleus exhibit a relatively long half-life: whereas the acute Fra's (and the associated acu'te to control levels within 12 to 24 hours following acute cocaine administration, the chronic Fra's (and the as­ sociated chronic AP-1 binding complex) remain at max­ imal levels 24 hours following the last chronic dose of cocaine and are at half-maximal levels one week later (Hope et al. 1994). The chronic and acute AP-1 com­ plex also show different binding affmity toward certain AP-1 sites, consistent with the possibility that the ADAPTIVE CHANGES chronic and acute Fra's may differ in their interactions IN NEURONAL FUNCTION with certain target genes. Figure 8 shows a working hypothesis concerning Figure 7. Schematic illustration of the hypothetical role played by gene expression in drug addiction. According to the physiological signifIcance of these observations this scheme, an initial extracellular effectof a drug of abuse (Nestler et al. 1993). According to this scheme, acute would trigger changes in multiple intracellular messenger cocaine exposure leads to the induction of a transient pathways in target neurons. Changes in the intracellular mes­ AP-1 complex composed of Fos and related immediate sengers would result in numerous physiological responses early gene products. This transient complex presum­ to the drug (as shown in Figure 3), including alterations in ably mediates some of the short-term effects of cocaine gene expression. The latter types of alterations would occur on gene expression. In contrast, chronic cocaine ex­ through the regulation of many classes of nuclear, DNA-bind­ posure leads to the induction of a persistent AP-1 com­ ing proteins, termed transcription factors, such as CREB and plex composed of different Fos-like proteins. An AP-1 Fos. CREB-like transcription factors refer to those that are regulated by extracellular agents primarily through changes complex of different half-life and composition would be expected to produce some different effects on gene intheir degree of phosphorylation. Fos-like transcription fac­ tors refer to those that are expressed at very low levels under expression, including some unique to the chronic­ basal conditions and regulated by extracellular agents primar­ treated state. The induction of such altered AP-1 com­ ily through induction of their expression (presumably via plexes highlights the fact that in addition to the tradi­ CREB-like proteins). Both types of transcription factors would tional quantitative characterizations of chronic drug then result in altered levels of expression of specific target effects on the brain as representing either up- or down­ proteins that underlie the adaptive changes in brain function regulation (i.e., sensitization or tolerance), chronic drug associated with addiction. Modified from Nestler 1992. treatments also result in qualitative changes in brain function, that is, that the chronic-treated state is sim­ ply different from the control or acute-treated state in important respects. ily (Hope et al. 1992). A course of chronic cocaine dra­ The isolation and cloning of the chronic Fos-like matically reduces the ability of a subsequent acute proteins will make it possible to study their role in al­ exposure to the drug to induce Fos and the other im­ tering the expression of '!arious putative target proteins mediate early gene products. However, chronic cocaine (e.g., G proteins and the cAMP pathway) identifIed in treatment is paradoxically associated with a robust in­ the mesolimbic dopamine system. Of course, one at­ duction of AP-1 DNA binding activity. These fIndings tractive feature of the chronic Fos-likeproteins, and the suggest that chronic cocaine treatment may induce differ­ long-term changes in gene expression they suggest, is ent types of Fos-likeproteins compared to the acute sit­ that they could account for the long-lasting nature of uation. Direct support for this possibility has been ob­ the biochemical adaptations that have been observed tained recently (Hope et al., 1994) through the analysis in the meso limbic dopamine system following chronic of Fos-like proteins by one- and two-dimensional blot drug exposure and for the long-lasting changes in brain immunolabeling procedures utilizing an antibody that function associated with addiction. NEUROPSYCHOPHARMACOLOGY 1994-VOL. 11, 84 E.J. Nestler NO.2

Acute Chronic The importance of genetic factors in establishing Cocaine Cocaine individual responsiveness to drugs of abuse is indicated by differencesin drug-related behaviors among differ· � � ent inbred animal strains (e.g., Lewis and Fischer rats). Transient Persistent The Lewis rat self-administers opiates, cocaine, and al­ AP-1 Complex AP-1 Complex cohol to a greater extent than the Fischer rat (George and Goldberg 1989), and develops a greater degree of conditioned place preference to cocaine and morphine (Guitart et al. 1992; Kosten et al. 1994). The Lewis rat also exhibits a greater facilitation in brain self-stimu· ffi lation thresholds in response to cannabinoids than the Fischer rat (Gardner and Lowinson 1991). These findings raise the possibility that the Lewis rat is inher­ ently more"drug-preferring" than the Fischer rat. How­ ever, the observation that Lewis and Fischer rats also show different locomotor responses to cocaine after + acute and chronic administration (George et al. 1991; Kosten et al. 1994), raises the alternativepossibility that, rather than a difference in drug preference per se, the 0®®®® two rat strains may differ in their inherent sensitivity to certain of the effects of drugs of abuse.

Short-term Long-term Interestingly, Lewis and Fischer rats show marked changes in inherent differencesin several biochemical parameters changes in ___.... �� . gene expression gene expression in the mesolimbic dopamine system (see Table 3 for summary). The NAc of the drug-naive Lewis rat Figure Scheme illustrating the regulation of AP-1 bind­ con­ 8. tains lower levels of Gi and higher levels of adenylyl ing proteins by acute and chronic cocaine. Acute cocaine ad­ ministrationtransiently induces several AP-1 binding proteins cyclase and cAMP-dependentprotein kinase compared in the NAc and caudate/putamen, including Fos, FosB, Fra2, to the NAc of the drug-naive Fischer rat (Guitart et al. c-Jun, and JunB. Induction of these transcription factors could 1993). In this manner, the Lewis rat (as compared to

mediate some of the rapid stimulatory ( + ) and inhibitory ( - ) the Fischer rat) resembles outbred Sprague-Dawley rats effects of cocaine on the expression of neural genes (A to E) treated chronically with opiates, cocaine, or alcohol. that contain AP-1 binding sites. In contrast, chronic cocaine Related studies, also summarized in Table 3, have simi· administrationresults in the persistent induction of AP-1 bind­ larly shown that the VTA of the drug-naive Lewis rat ing activity that is accounted for by different Fos-like (and pos­ (as compared to the VTA of the drug-naive Fischer rat) sibly Jun-like)proteins, including recently identifIedproteins shows different levels of a number of phosphoproteins designated chronic Fra's (Fos-related antigens). Such chronic, known to be altered in Sprague-Dawley rats in the persistent AP-1 complexes of altered composition would be chronic drug-treated state (see Nestler 1992; Nestler et expected to exert different transcriptional effects compared al. 1993). These studies support the possibility that the to the acute AP-1 complexes and could mediate some of the long-term effects of cocaine on gene expression. From Nes­ various intracellular messenger proteins identifiedcould tler et al. 1993. not only contribute to the pathophysiology of drug ad­ diction but also represent partof the biochemical mech­ anisms that determine inherent responsiveness to drugs of abuse, including drug preference. CONTRIBUTION OF INTRACELLULAR The importance of environmental factors in drug MESSENGER PROTEINS TO INDIVIDUAL RISK addiction vulnerability is demonstrated by the well­ FOR DRUG ADDICTION established ability of many exogenous treatments to al­ ter drug-related behaviors. An animal's responsiveness As more is learned of the pathophysiology of drug ad­ to a drug of abuse can be altered by prior exposure to diction through the investigation of intracellular mes­ that drug itself, to other drugs, or to environmental senger pathways, it will be possible to identify specific stimuli. For example, glucocorticoid treatment and en­ genetic and environmental factors that determine an vironmental stress have been reported to augment the individual's predilection for drug addiction. Such fac­ reinforcing and locomotor-activating properties of opi­ tors presumably determine drug addiction vulnerabil­ ates, cocaine, and amphetamine, and prior exposure ity by influencing an individual's responsiveness to the to these drugs of abuse can exert the same effects(e.g., multiple positively and negatively reinforcing effects see Cole et al. 1990; Kalivas and Stewart 1991; Piazza of a drug of abuse after acute or chronic exposure. et a1. 1991). NEUROPSYCHOPHARMACOLOGY 1994-VOL. 11, NO. 2 Molecular Neurobiology of Drug Addiction 85

Table 3. Summary of Lewis-Fischer Strain Differences in Biochemical Parameters in the Mesolimbic Dopamine System

Chronic Morphine-Treated Drug Naive Conditions Conditions Lewis vs Fischer � in S-Dt NAc cAMP kinase L>F t Adenylate cyclase L>F t Gia LF Neurohlaments LF

The symbols used in this table «, >, t, +) refer to differences in levels of enzyme activity (for adenylyl cyclase and cAMP-dependent protein kinase) or levels of immunoreactivity (for Ciu, tyrosine hydroxy­ lase, neurohlaments, and glial fibrillary acidic protein). * In this case, the downward arrow refers to a decrease in phosphorylation state that would be expected to have a similar functional effect as a decrease in total amount. t S-D, Sprague-Dawley rats. For references, see Nestler 1992; Nestler et al. 1993.

Genetic and environmental factors cannotbe viewed leads in the development of new therapeutic agents, as separate, independent variables; rather, it is the in­ revolutionize diagnostic capabilities, and enable the tar­ teraction between the two that determines an individ­ geting of preventive measures to particularly vulnera­ ual's predilection for drug addiction. An illustrationof ble individuals. such interactions are the fmdingsthat chronic exposure to drugs of abuse or to glucocorticoids produces differ­ ent behavioral effects, and differentbiochemical adap­ CONTRIBUTIONS OF INTRACELLULAR tations in the mesolimbic dopamine system, in Lewis MESSENGER PROTEINS TO DRUG and Fischer rats (Guitart et al. 1992, 1993; Ortizand Nes­ DEVELOPMENT EFFORTS tler 1993). In this way, Lewis and Fischer rats provide a novel experimental system to investigate the specifIc As an increasing number of intracellular messenger genetic and environmental factors that combine to de­ proteins are identifIedas contributing to the pathophys­ termine the biochemical phenotype in the mesolimbic iology or individual risk of drug addiction, these pro­ dopamine system and the various drug-related be­ teins could be used to develop novel pharmacothera­ haviors regulated via this neural pathway. The fact that peutic agents for the treatment and prevention of drug Lewis and Fischer rats are genetically divergent, such addiction. At the very least, measures of the intracellu­ that strain differences probably exist for a large num­ lar messenger proteins in rats can be used as a novel ber of genes, does not limit their usefulness in the study procedure by which to screen test compounds preclin­ of drug addiction. This situation is, in fact, clinically ically for their potential antiaddiction properties. For relevant in terms of the known differencesin suscepti­ example, does drug X prevent or reverse cocaine's abil­ bility to the various effectsof drugs of abuse observed ity to produce adaptations in G proteins, the cAMP among people, who also differ greatlyin their individual pathway, etc., in the mesolimbic dopamine system? In genetic makeup. Lewis and Fischer rats could be used contrast to all current drug development effortsin the to identify specifIc genes related to drug addiction by area of drug addiction, which are focused on test agents cross-breeding the two strains through several genera­ with a specifIc pharmacological property (e.g., selec­ tions to identify biochemical and behavioral traits that tive antagonism of a dopamine receptor subtype), this cosegregate, followed by genetic linkage analysis. Var­ new screen would enable the rational testing of novel ious drug-regulatedproteins (e.g., those shown in Fig­ drugs with diverse pharmacological actions. ure 7) could serve as candidate genes in these studies. It is even conceivable that the intracellular mes­ Theseinvestigations inrats could be followed by genetic senger proteins themselves could be targeted by new studies in clinical populations to determine whether the therapeutic agents. After all, some of the most effec­ same or related genes may contribute to drug addic­ tive and widely used drugs are directed at intracellular tion vulnerability in people. IdentifIcationof genes that messenger pathways. Aspirin and all other nonsteroi­ predispose individuals for addiction would provide dal antiinflammatory agents inhibit cyclo-oxygenase, NEUROPSYCHOPHARMACOLOGY NO.2 86 E.J. Nestler 1994 -VOL. 11,

the fIrstenzyme in the generation of prostaglandin, leu­ Bloom FE, Kupfer DJ (1994): : Fourth kotriene, and endoperoxide intracellular messengers Generation of Progress, New York: Raven Press, in press (Piomelli and Greengard 1990). FK-506 and related im­ Cerruti C, Pilotte NS, Uhl G, Kuhar MJ (1994): Reduction of munosuppressive agents bind to a newly discovered dopamine transporter mRNA after cessation of repeated class of protein, the immunophilins (also present at high cocaine administration. Mol Brain Res, in press levels in brain), that regulate the activity of specificpro­ Cole BJ, Cador M, Stinus L, Rivier C, Rivier I, Vale W, Le tein phosphatases (see Steiner et al. 1992). Lithium, one Moal M, Koob GF (1990): Critical role of the hypotha­ lamic pituitary adrenal axisin amphetamine-induced sen­ of the most effective psychotherapeutic agents avail­ sitization of behavior. Life Science 47:1715-1720 able, directly regulates the activity of adenylyl cyclase, Cole AI, Bhat RB, Patt C, Worley PF, Baraban JM (1992): DJ G proteins, and certain inositol phosphatases, actions dopamine receptor activation of multiple transcription that presumably result in the drug's antimanic and an­ factor genes in rat striatum. J Neurochem 58:1420-1426 tidepressant effects(see Hyman and Nestler 1993). The Cunningham ST, Kelley AE (1993): Hyperactivity and sen­ targetingof intracellular messenger proteins would rep­ sitization to psychostimulantsfollowing cholera toxin in­ resent a dramatic departure for drug development fusion into the nucleus accumbens. J Neurosci 13:2342- effortsin psychiatry and would clearly be of high risk. 2350 On the other hand, such efforts offer the potential of Duman RS, Nestler EJ (1994): pathways tremendous breakthroughs; drug development efforts for receptors. In Psychopharmacology: Fourth Generation of Progress, Bloom FE, Kupfer D}, in psychiatry over four generations have failed to re­ (eds), New York: Raven Press, in press sult in fundamentally more efficaciousagents, with fun­ damentally new mechanisms of action, in the treatment Gardner EL, LowinsonJH (1991):Marijuana's interaction with brain reward systems: Update 1991. Pharmacol Biochern of drug addiction or other neuropsychiatric disorders. Behav 40:571-580 George FR, Goldberg SR (1989): Genetic approaches to the CONCLUSIONS analysis of addiction processes. Trends Pharmacol Sci 10:78-83 The fact that many features of drug addiction in people George FR, Porrino LJ, Ritz MC, Goldberg SR (1991): Inbred rat strain comparisons indicate different sites of action can be reproduced accurately in laboratory animals for cocaine and amphetamine locomotor stimulant means that studies of drug addiction offer a unique ad­ effects. Psychopharmacology 104:457-462 vantage in the investigation of the neurobiological Graybiel AM, Moratalla R, Robertson HA (1990): Ampheta­ mechanisms underlying a complex and clinically im­ mine and cocaine induce drug-specifIcactivation of the portant behavioral abnormality. Molecular adaptations c-fos gene in striosome-matrix compartments and lim· that occur in the brain in association with drug addic­ bic subdivisions of the striatum. Proc Natl Acad Sci USA tion in animal models can be related to behavioral 87:6912-6916 aspects of addiction in the animals, and eventually to Guitart X, Beitner-Johnson D, Nestler EJ (1992): Fischer and the clinical situation. This information will lead to an Lewis rat strains differ in basal levels of neuro&lament improved understanding of drug addiction and to the proteins and in their regulation by chronic morphine. Synapse 12:242-253 development of more effectivetreatments and preven­ tive measures. It is also anticipated that advances made Guitart X, Kogan JH, Berhow M, Terwilliger RZ,Aghajanian in the fIeld of drug addiction will have important im­ GK, Nestler EJ (1993): Lewis and Fischer rat strains dis­ play differencesin biochemical, electrophysiological, and plications for how we approach the pathophysiology behavioral parameters: Studies in the nucleus accumbens and treatment of other major neuropsychiatric dis­ and locus coeruleus of drug naive and morphine-treated orders. animals. Brain Res 611:7-17 Hayward MD, Duman RS, Nestler EJ (1990): Induction of the c-fos proto-oncogene during opiate withdrawal in the lo­ ACKNOWLEDGMENT cus coeruleus and other regions of rat brain. Brain Res 525:256-266 This work was supported by DA07359, DA08227, and Hope BT, Kosofsky B, Hyman SE, Nestler EJ (1992): Regula­ DA00203, and by the Abraham RibicoffResearch Facilities of tion of lEG expression and AP-1 binding by chronic co­ the Connecticut Mental Health Center, State of Connecticut caine in the rat nucleus accumbens. Proc Natl Acad Sci Department of Mental Health. USA 89:5764-5768 Hope BT, Kelz MB, Duman RS, Nestler EJ (1994): Chronic REFERENCES electroconvulsive seizures induce expression of a long­ lasting AP-1 complex in rat brain with altered composi­ tion and characteristics. J Neurosci 14:4318-4328 AgranoffBW, Fisher SK (1994): Phosphoinositides. In Siegel GJ, Albers RW, Agranoff BW, Molinoff P (eds), Basic Hyman SE, Nestler EJ (1993): The Molecular Foundations of : Molecular, Cellular, and Medical As­ Psychiatry. Washington, DC: American Psychiatric Press pects, 5th ed., Boston: Little, Brown, pp 417-428 Kalivas PW, Stewart J (1991): Dopamine transmission in the NEUROPSYCHOPHARMACOLOGY 1994-VOL. NO. 2 11, Molecular Neurobiology of Drug Addiction 87

initiation and expression of drug- and stress-induced sen­ Ortiz J, Nestler EJ (1993): Fischer and Lewis rat strains differ sitization of motor activity. Brain Res Rev 16:223-244 in corticosterone regulation of tyrosine hydroxylase and Koob GF (1992): Drugs of abuse: Anatomy, pharmacology other phosphoproteins in the ventral tegmental area. Soc and function of reward pathways. Trends Pharmacol Sci Neurosci Abs 19:942 13:177-184 Piazza PV, Maccari S, Deminiere JM, Le Moal M, Mormede Kosten TA, Miserendino MJD, Chi S, Nestler EJ (1993): Fischer P, Simon H (1991): Corticosterone levels determine in­ and Lewis rat strains show differentialcocaine effects in dividual vulnerability to amphetamine self-administra­ conditioned place preference and behavioral sensitiza­ tion. Proc Natl Acad Sci USA 88:2088-2092 tion, but not in locomotor activity or conditioned taste Piomelli D, Greengard P (1990): Lipoxygenase metabolites aversion. J Pharmacol Exp Ther 269:137-144 of arachidonic acid in neuronal transmembrane signal­ Kuhar MJ, Ritz MC, Boja JW (1991): The dopamine hypothe­ ing. Trends Pharmacol Sci 11:367-373 sis of the reinforcing properties of cocaine. Trends Neu­ Self DW, Chi S, Nestler EJ (1993): Modulation of cocaine self­ rosci 14:299-302 administration by cyclic AMP analogues in the nucleus Meltzer HY (ed) (1987): Psychopharmacology: Third Gener­ accumbens. Soc Neurosci Abs 19:1858 ation of Progress, New York: Raven Press Self DW, Nestler EJ, Stein L (1994): Role of G-proteins in Miserendino MJD, Nestler EJ (1994): Behavioral sensitization: heroin and cocaine reinforcement. J Neurosci, 14:6239- role of the cyclic AMPsystem in nucleus accumbens. Col­ 6247 lege on Problems of Drug Dependence Abstracts, in press Steiner JP, Dawson TM, Fotuhi M, Glatt CE, Snowman AM, Morrisett RA, Swartzwelder HS (1993): Attenuation of hip­ Cohen N, Snyder SH (1992): High brain densities of the pocampal long-term potentiation by ethanol: A patch­ imrnunophilin FKBPcolocalized with calcineurin. Nature clamp analysis of glutamatergic and GABAergic mecha­ 358:584-587 nisms. J Neurosci 13:2264-2272 Steketee JD, Striplin CD, Murray TF, Kalivas PW (1991): Pos­ Nestler EJ (1992): Molecular mechanisms of drug addiction. sible role for G-proteins in behavioral sensitization to co­ J Neurosci 12:2439-2450 caine. Brain Res 545 :287-291 Nestler EJ, Duman R (1994): G-proteins and cyclic nucleo­ Striplin CD, Kalivas RW (1992): Correlation between be­ tides in the . In Siegel GJ, Albers RW, havioral sensitization to cocaine and G protein ADP­ Agranoff BW, MolinoffP (eds), Basic Neurochemistry: ribosylation in the ventral tegmental area. Brain Res Molecular, Cellular, and Medical Aspects, 5th ed., Bos­ 579:181-186 ton: Little, Brown, pp 429-448 Terwilliger RZ, Beitner-Johnson D, Sevarino KA, Crain SM, Nestler EJ, Greengard P (1994) : Protein phosphorylation and Nestler EJ (1991): A general role for adaptations in G-pro­ the regulation of neuronal function. In Siegel GJ, Albers teins and the cyclic AMP system in mediating the chronic RW, Agranoff BW, Molinoff P (eds), Basic Neurochemis­ actions of morphine and cocaine on neuronal function. try: Molecular, Cellular, and Medical Aspects, 5th ed., Brain Res 548:100-110 Boston: Little, Brown, pp 449-474 Vrana SL, Vrana KE, Koves TR, Smith JE, Dworkin SI (1993): Chronic cocaine administration increases CNS tyrosine Nestler EJ, Terwilliger Walker JR, Sevarino Duman RZ, KA, hydroxylase enzyme activity and rnRNAlevels and tryp­ RS (1990): Chronic cocaine treatment decreases levels of tophan hydroxylase enzyme activity levels. J Neurochem the G-protein subunits Gia and Goa in discrete regions 61:2262-2268 of rat brain. J Neurochem 55: 1079-1082 Wafford KA,Burnett DM, Leidenheimer NJ, Burt DR, Wang Nestler EJ, Hope BT, Widnell KL (1993): Drug addiction: A JB, Kofuji P, Dunwiddie TV, Harris RA, and Sikela JM model for the molecular basis of neural plasticity. Neu­ (1991): Ethanol sensitivity of the GABAA receptor ex­ ron 11:995-1006 pressed in xenopus oocytes requires 8 amino acids con­ Nestler EJ, Guitart X, Ortiz J, Trevisan L (1994): Second mes­ tained in the y2L subunit. Neuron 7:27-33 senger and protein phosphorylation mechanisms under­ Wise RA (1990): The role of reward pathways in the develop­ lying possible genetic vulnerability to alcoholism. Ann ment of drug dependence. In Balfour DJK (ed), Psy­ NY Acad Sci 708:108-118 chotropic Drugs of Abuse. Oxford: Pergamon Press, pp 23-57 Nguyen TV, Kosofsky BE, Birnbaum R, Cohen BM, Hyman SE (1992): Differential expression of c-Fos and Zif268 in Young ST, Porrino LJ, ladarola MJ (1991): Cocaine induces rat striatum after haloperidol, clozapine, and ampheta­ striatal c-Fos-immunoreactive proteins via dopaminer­ mine. Proc Natl Acad Sci USA 89:4270-4274 gic D1 receptors. Proc Natl Acad Sci USA 88:1291-1295