Neuron, Vol. 43, 401–413, August 5, 2004, Copyright 2004 by Cell Press

Homer Proteins Regulate Sensitivity to Cocaine

arise, in part, from cocaine-induced plasticity in gluta-Karen K. Szumlinski,1,* Marlin H. Dehoff,2

mate neurotransmission in the nucleus accumbensShin H. Kang,2 Kelly A. Frys,1 Kevin D. Lominac,1

(Pierce et al., 1996; Thomas et al., 2001; Tzschentke andMatthias Klugmann,3 Jason Rohrer,1

Schmidt, 2003; Wolf, 1998), a brain region known to beWilliam Griffin III,1,4 Shigenobu Toda,1

critically involved in the addictive properties of cocaineNicolas P. Champtiaux,1 Thomas Berry,1

(Everitt and Wolf, 2002). Among the enduring changesJian C. Tu,2 Stephanie E. Shealy,1

produced in the accumbens by withdrawal from re-Matthew J. During,3 Lawrence D. Middaugh,1,4

peated cocaine administration is a reduction in Hom-Paul F. Worley,2 and Peter W. Kalivas1

er1b/c expression (Swanson et al., 2001). Moreover,1Department of Physiology and Neurosciencemimicking the cocaine-induced reduction in Homer1Medical University of South Carolinagene expression by infusing antisense oligonucleotidesCharleston, South Carolina 29425into the accumbens sensitizes drug-naive rats to the2 Department of Neurosciencemotor stimulant effects of cocaine (Ghasemzadeh et al.,The Johns Hopkins University School of Medicine2003). Also, cocaine induces a rapid induction of theBaltimore, Maryland 21205immediate early gene form of Homer1, and this gene3 Department of Molecular Medicine and Pathologyis remarkably dynamic during behavioral responses toUniversity of Aucklandenvironmental stimuli (Brakeman et al., 1997; Vazdarja-Auckland 1020nova et al., 2002). These findings pose a hypothesis thatNew Zealandgenetic or environmental alterations in the expression4 Department of Psychiatryor function of one of the three Homer genes may affectCenter for Drug and Alcohol Programsaddiction to cocaine.Medical University of South Carolina

Homer proteins are encoded by three genes (Homer1–3)Charleston, South Carolina 29425and act to coordinate synaptic proteins for a variety offunctions, including calcium signaling (Shin et al., 2003;Yuan et al., 2003), glutamate receptor signaling/traffick-Summarying (Shiraishi et al., 2003; Xiao et al., 2000), and activity-dependent synaptic remodeling (Sala et al., 2003; UsuiDrug addiction involves complex interactions betweenet al., 2003). The multiple contributions to cellular ho-pharmacology and learning in genetically susceptiblemeostasis and plasticity by Homer proteins appear to beindividuals. Members of the Homer gene family aremediated by a simple bimodal structure. The N-terminalregulated by acute and chronic cocaine adminis-Ena/VASP homology 1 (EVH1) domain mediates selec-tration. Here, we report that deletion of Homer1 ortive binding to other synaptic proteins, while the C-ter-Homer2 in mice caused the same increase in sensitiv-minal coiled-coil domain mediates self-association (Xiaoity to cocaine-induced locomotion, conditioned re-et al., 2000). Among the proteins that bind to Homerward, and augmented extracellular glutamate in nu-are group I metabotropic glutamate receptors (mGluR1/5),cleus accumbens as that elicited by withdrawal frominositol-1,4,5-triphosphate (IP3) receptors, TRP cationrepeated cocaine administration. Moreover, adeno-channels, and Shank, which cluster with many proteinsassociated virus-mediated restoration of Homer2 inregulating synaptic plasticity, such as F-actin, dynamin3,the accumbens of Homer2 KO mice reversed the co-cortactin, and NMDA (NR2B) and AMPA (GluR1) recep-caine-sensitized phenotype. Further analysis of Homer2tors (Naisbitt et al., 1999; Shiraishi et al., 2003; Usui etKO mice revealed extensive additional behavioral andal., 2003; Yuan et al., 2003). The formation of Homer-

neurochemical similarities to cocaine-sensitized ani-dependent protein complexes is regulated in part by an

mals, including accelerated acquisition of cocaine alternative transcript of Homer1 (Homer1a) that lacksself-administration and altered regulation of gluta- the coiled-coil domain and the ability to multimerizemate by metabotropic glutamate receptors and cys- (Bottai et al., 2002; Xiao et al., 2000). Homer1a is inducedtine/glutamate exchange. These data show that Homer by synaptic activity as an immediate early gene product,deletion mimics the behavioral and neurochemical functions as a natural dominant negative of Homerphenotype produced by repeated cocaine administra- multimers, and has been shown to influence synaptiction and implicate Homer in regulating addiction to co- plasticity (Brakeman et al., 1997; Hennou et al., 2003;caine. Sala et al., 2003).

In order to test the hypothesis that the regulation ofIntroduction Homer by cocaine is an important adaptation underlying

addiction-related behavior, the Homer genes were indi-Addiction to cocaine involves enduring behavioral and vidually deleted, and the knockout (KO) mice were stud-neurochemical changes, some of which are modeled by ied for differences in behavioral sensitivity to cocaine.sensitization to the reinforcing and psychomotor stimu- The role of Homer in cocaine sensitivity was also exam-lant effects of the drug (Vanderschuren and Kalivas, ined by infusing an adeno-associated virus carrying the2000). Behavioral sensitization to cocaine is thought to Homer2b gene into the nucleus accumbens of Homer2

KO mice. We found that Homer1 and Homer2 KO miceexhibited behavioral and neurochemical phenotypes*Correspondence: szumlink@musc.edu

Neuron402

Figure 1. Homer1 and Homer2 Deletion Sen-sitizes Mice to the Behavioral Effects of Co-caine

(A) Homer1 deletion enhanced place condi-tioning in KO mice relative to wt controls(n � 6 at each dose). Data points in (A) and(B) represent the mean � SEM difference intime spent in the cocaine-paired chamber be-fore and after four conditioning sessions. (B)Homer2 deletion enhanced place condition-ing in KO mice relative to wt controls (n � 6at each dose). (C) Homer1 deletion enhancedthe capacity of acute cocaine to elicit loco-motor activation during the first place condi-tioning session (n � 6). (D) Homer2 deletionenhanced the capacity of acute cocaine toelicit locomotor activation (n � 6). Data werestatistically analyzed using a two-wayANOVA. �p � 0.05 versus wt.

that mimicked the neuroadaptations produced in wild- relative to their respective wt animals. Thus, deletion ofeither Homer1 or Homer2 enhanced both sensitivity andtype rodents following withdrawal from repeated co-

caine (Vanderschuren and Kalivas, 2000; Baker et al., the maximum response to cocaine-conditioned reward.The place conditioning apparatus contained pho-2003; Bowers et al., 2004). Moreover, aspects of the

Homer2 KO phenotype were reversed by adeno-associ- tobeams that enabled simultaneous measurement oflocomotor activity, and locomotor activity was moni-ated virus (AAV)-mediated restoration of Homer2b to the

accumbens. These observations provide mechanistic tored when animals were confined to the cocaine-pairedcompartment. Similar to the data for place conditioning,insight into the neurochemical and molecular basis of

cocaine addiction and suggest that alterations in the the capacity of acute cocaine to elicit locomotor hyper-activity was enhanced in both Homer1 KO (Figure 1C;function of Homer-dependent pathways increase sus-

ceptibility to cocaine addiction. Also, since Homer path- genotype effect, p � 0.0001; genotype � dose interac-tion, p � 0.01, n � 6–7) and Homer2 KO (Figure 1D;ways are dynamically responsive to environmental stim-

uli (Vazdarjanova et al., 2002), these studies point to a genotype effect, p � 0.002; genotype � dose interaction,p � 0.04, n � 7–8) mice compared to their respective wtpotential link between environmental stress and addic-

tion (Koob, 2003; Shalev et al., 2002). controls. The increased locomotor response to cocainewas not attributable to differences in basal locomotoractivity, as no genotypic differences were observed forResultssaline-injected mice in either experiment. The differencein cocaine-induced place preference and locomotor ac-Homer1 and Homer2 but Not Homer3 Deletion

Sensitized Behavioral Responding to Cocaine tivity between the wt subjects in the Homer1 andHomer2 groups is likely due to each deletion being gen-To determine whether deletion of the Homer1, Homer2,

or Homer3 genes altered the conditioned rewarding and erated on different genetic backgrounds (see Experi-mental Procedures).locomotor stimulant effects of cocaine, a place condi-

tioning study was conducted and the locomotor re- In contrast to Homer1 and Homer2 KO mice, wheninjected with a dose of cocaine (15 mg/kg, i.p.) that wassponse to an acute injection of cocaine was assessed.

Both Homer1 KO (Figure 1A; genotype effect, p � 0.008; intermediate to doses eliciting a maximum behavioralresponse in the Homer1 and Homer2 studies, thegenotype � dose interaction, p � 0.06, n � 6–8) and

Homer2 KO (Figure 1B; genotype effect, p � 0.0006; Homer3 KO mice exhibited equivalent cocaine placeconditioning (occupancy difference, wt � 240 � 46 s;genotype � dose interaction, p � 0.04, n � 6–8) mice

demonstrated significant cocaine-induced place condi- KO � 266 � 45 s; p � 0.64, n � 6) and cocaine locomotorhyperactivity (beam breaks, wt � 471 � 68; KO � 473 �tioning at doses that did not elicit place conditioning

in their respective wt controls (for Homer1, 10 mg/kg 98, p � 0.99) as wt subjects. These data suggest that,unlike deletion of Homer1 or Homer2, deletion ofcocaine; for Homer2, 0.5 mg/kg cocaine). Also, at the

cocaine doses that elicited maximum place conditioning Homer3 does not convey increased sensitivity to co-caine and is consistent with a lack of Homer3 expression(for Homer1, 30 mg/kg; for Homer2, 10 mg/kg), the time

spent on the cocaine-paired side was greater in KO mice in the nucleus accumbens (Xiao et al., 2000).

Homer, Glutamate, and Cocaine Addiction403

Table 1. Basal Levels of Extracellular Glutamate Determined by Conventional In Vivo Microdialysis in the Nucleus Accumbens and PrefrontalCortex in Homer KO and WT Mice

Glutamate (pmol) Dopamine (fmol)

Genotype Nucleus WT KO WT KO

Homer1 accumbens 114.0 � 20.8 68.1 � 3.6* 10.9 � 1.2 10.9 � 0.9Homer2 accumbens 72.9 � 13.5 25.8 � 3.6* 19.8 � 0.3 23.1 � 3.8Homer2 PFC 27.9 � 10.8 17.4 � 3.9 ND ND

ND, not done; PFC, prefrontal cortex. *p � 0.05, comparing KO to wt using a two-tailed Student’s t test, n � 5–8.

Homer1 and Homer2 Deletion Facilitates In addition to the nucleus accumbens, glutamate lev-els in the frontal cortex of Homer2 KO mice were exam-the Effect of Cocaine upon Accumbens

Glutamate Levels ined, and no genotypic differences were observed forThe parallel in behavioral sensitivity to cocaine betweenrats withdrawn from repeated cocaine (Vanderschurenand Kalivas, 2000) and cocaine-naive Homer1 andHomer2 KO mice prompted an evaluation of neurochem-ical parallels in the regulation of extracellular glutamateand dopamine in the nucleus accumbens. Akin to theeffect of repeated cocaine (Pierce et al., 1996; Baker etal., 2003), basal extracellular glutamate was �50% lowerin Homer1 KO (p � 0.03, n � 6–8) and Homer2 KO (p �0.0008, n � 6–8) mice compared to their wt controls(Table 1). Verification of basal extracellular content ofglutamate using the no net flux microdialysis techniqueconfirmed the reduction in accumbens basal extracellu-lar glutamate content in Homer2 KO mice versus wtcontrols (Figure 2; wt, 7.51 � 1.61 �M, n � 6; KO, 4.33 �1.19, n � 5; p � 0.05). Also akin to rats withdrawn fromrepeated cocaine (Baker et al., 2003; Reid and Berger,1996), in both the Homer1 and Homer2 KO animals, anacute cocaine injection (15 mg/kg, i.p.) produced anelevation (above baseline) in extracellular glutamate inthe nucleus accumbens (Figures 3A and 3D). In contrast,acute cocaine injection either did not alter or decreasedextracellular levels of glutamate in the nucleus accum-bens of Homer1 or Homer2 wt controls, respectively.Normalizing for genotypic differences in basal extracel-lular glutamate clearly reveals the cocaine-induced in-crease in accumbens glutamate in both Homer1 KOmice (Figure 3B; genotype � time, p � 0.02) and Homer2KO mice (Figure 3E; genotype � time, p � 0.006).

Figure 3. Homer1 and Homer2 Deletion Alters Cocaine-Induced In-creases in Nucleus Accumbens Extracellular Levels of Glutamatebut Not Dopamine

(A) Cocaine (15 mg/kg, i.p.) increases extracellular glutamate in thenucleus accumbens of Homer1 KO (n � 6). (B) Data from (A) normal-ized to percent change from the average of the three baseline sam-ples. (C) Homer1 genotypic differences in accumbens dopaminewere not observed. Data were normalized to percent change fromthe average baseline value (n � 6). (D) Cocaine (15 mg/kg, i.p.)increases extracellular glutamate in the nucleus accumbens ofHomer2 KO (n � 6). (E) Data from (D) normalized to percent change

Figure 2. Homer2 Deletion Reduces Basal Extracellular Levels of from the average of the three baseline samples. (F) No Homer2Glutamate in the Nucleus Accumbens genotypic differences in accumbens dopamine were observed (n �

6). Data for conventional microdialysis were evaluated using a two-No net flux microdialysis demonstrated a signficant reduction inaccumbens glutamate in Homer2 KO mice (y � 0). No genotypic way ANOVA with repeated measures over time, followed by an LSD

post hoc comparison. �p � 0.05, KO versus wt; *p � 0.05, versusdifference was observed in the clearance of glutamate from theprobe site (slope) (n � 5). baseline in the KO group; #p � 0.05, versus baseline in the wt group.

Neuron404

either basal extracellular glutamate levels (Table 1, p �0.29, n � 5–8) or cocaine-induced increases in glutamate(genotype � time, p � 0.99; data not shown). The lackof effect by Homer2 gene deletion in the frontal cortexmimics the previously observed effect of withdrawalfrom repeated cocaine administration in rats wherebasal levels of glutamate are lower in the nucleus ac-cumbens but not in the prefrontal cortex (Baker et al.,2003).

After withdrawal from repeated cocaine administra-tion, a subsequent injection of cocaine elicits a sensi-tized increase in extracellular dopamine levels (Van-derschuren and Kalivas, 2000). However, microdialysisin the nucleus accumbens revealed no genotypic differ-ences between KO and wt mice in either basal (Table1; Homer1, p � 0.98, n � 5; Homer2, p � 0.41, n � 6–8)or cocaine-induced increases in dopamine (Homer1,Figure 3C, genotype � time, p � 0.99; Homer2, Figure3F, genotype � time, p � 0.94). Thus, in contrast toextracellular glutamate, Homer deletion did not mimicthe effects of repeated cocaine administration on extra-cellular dopamine that were previously observed in rats(Vanderschuren and Kalivas, 2000).

Intra-Accumbens Overexpression of Homer2bReverses the Cocaine Phenotype of Homer2 KO MiceTo confirm an active role for Homer protein in the KOphenotype, wt and Homer2 KO mice were infused intothe nucleus accumbens with an AAV carrying the Hom-er2b splice variant (Homer2b-AAV). Five weeks followingAAV microinjection, substantial transduction up to 1 mmbeyond the site of microinjection was revealed (Figure4A). Relative to eGFP-AAV-treated mice, Homer2b-AAVinfusion reversed the genotypic differences betweenHomer2 KO and wt mice in the capacity of 5 mg/kgcocaine to induce place conditioning (genotype �

Figure 4. Restoration of Homer2b in the Accumbens Normalizes thevirus � dose, p � 0.04; Figure 4B) or locomotor hyperac-Cocaine Phenotype of Homer2 KO Mice

tivity (genotype � virus � dose, p � 0.003; Figure 4C).(A) Cellular transfection by GFP-AAV at 5 weeks after infusion into

Consistent with the findings in Figure 1, decomposition the nucleus accumbens. Scale bar, 1 mm; ac, anterior commissure.of the three-way interaction revealed significant geno- (A�, A″, and A�) High-power micrograph showing cellular transfec-typic differences between eGFP-AAV-treated groups for tion up to 1 mm beyond the injection site (arrow in [A]). (B) Bilateral

infusion of Homer2b-AAV (H2) into the accumbens prevented theboth place conditioning (genotype effect, p � 0.06;development of place conditioning produced by 5 mg/kg cocainegenotype � dose, p � 0.02) and locomotor hyperactivityin Homer2 KO mice (mean � SEM difference in time spent in the(genotype effect, p � 0.01; genotype � dose, p � 0.001).paired environment before and after four conditioning trials; n � 6).

In contrast, restoration of Homer2b in the accumbens (C) Homer2b-AAV infusion blocked the acute locomotor responseby Homer2b-AAV eliminated the genotypic differences to 5 mg/kg cocaine in Homer2 KO mice (mean � SEM photocellbetween KO and wt mice for both place conditioning counts during the first place conditioning trial; n � 6). (D) Homer2b-

AAV prevented the rise in extracellular glutamate in the nucleus(genotype effect, p � 0.19; genotype � dose, p � 0.68)accumbens elicited by 15 mg/kg cocaine in Homer2 KO mice (n �and locomotor hyperactivity (genotype effect, p � 0.87;6). (E) Data in (D) normalized to percent change from the averagegenotype � dose, p � 0.32). AAV-mediated restorationbaseline value. All data are shown as mean � SEM. Behavioral data

of Homer2b in the accumbens also eliminated the rise were evaluated using a two-way ANOVA, and neurochemical datain extracellular glutamate produced in Homer2 KO mice were evaluated using a three-way ANOVA with repeated measuresby cocaine (Figures 4D and 4E; genotype � virus � time, over time. *p � 0.05, KO-H2 versus KO-eGFP; �p � 0.05, KO-eGFP

versus wt-eGFP.p � 0.001). In contrast, Homer2b-AAV infusion did notappear to alter the genotypic difference in basal extra-cellular levels of glutamate in the nucleus accumbens,although a no net flux study is necessary to validate this the lack of an effect of Homer gene deletion upon ac-

cumbens dopamine (Figure 3), Homer2b-AAV infusionconclusion (Figure 4D and Table 2; genotype effect, p �0.0005; genotype � virus, p � 0.97). This lack of effect altered neither the basal extracellular levels of dopamine

(Table 2; genotype effect, p � 0.22; genotype � virus,may be explained by the fact that AAV transfection waslocalized to the accumbens, and reduced basal levels p � 0.46) nor the cocaine-induced rise in extracellular

dopamine (time, p � 0.0001; genotype � time, p � 0.95;of glutamate may arise, in part, from the effects of genedeletion outside of the accumbens. As expected from genotype � virus �x time, p � 0.81; data not shown).

Homer, Glutamate, and Cocaine Addiction405

Table 2. Basal Levels of Extracellular Glutamate and Dopamine Determined by Conventional In Vivo Microdialysis in the Nucleus Accumbensof AAV-Treated Homer2 KO and WT Mice

Glutamate (pmol) Dopamine (fmol)

AAV WT KO WT KO

eGFP 100.1 � 19.6 33.1 � 7.45* 24.3 � 6.9 13.9 � 1.4Homer2b 100.2 � 26.9 31.8 � 4.2* 15.6 � 3.4 12.9 � 4.6

*p � 0.05, comparing KO to wt using a two-tailed Student’s t test, n � 6–9.

Homer2 Deletion Selectively Enhances Operant Homer2 Deletion Reduces the Regulationof Extracellular Glutamate by mGluR1Responding for Cocaineand Cystine/Glutamate ExchangeAs both Homer1 and Homer2 KOs exhibited increasesHomer proteins modulate various aspects of group1in cocaine-induced extracellular glutamate as well asmGluR signaling by cross-linking between synaptic pro-enhanced cocaine place conditioning and locomotorteins such as shank, IP3 receptors, and TPR channelsactivity, an operant paradigm was used to assess the(Brakeman et al., 1997; Kato et al., 1998; Tu et al., 1998;role of Homer in the motivation to self-administer intra-Xiao et al., 2000; Yuan et al., 2003). Increased behavioralvenous cocaine. In the first experiment, the effects ofsensitivity to cocaine is associated with reduced Hom-Homer gene deletion on lever pressing for the naturaler1b/c content and the ability of the mGluR1/5 agonistreinforcer sucrose (15% w/v) were examined. WhileDHPG to increase extracellular glutamate (Swanson etHomer2 KO mice exhibited wt levels of motivation (rein-al., 2001). Consistent with the parallel behavioral andforcers earned) for sucrose during training (Figure 5A;neurochemical sensitivity between cocaine-withdrawngenotype effect, p � 0.74; genotype � session, p �animals and Homer2 KO mice, the capacity of DHPG to0.34, n � 7–10), Homer1 deletion significantly reducedincrease extracellular glutamate in the nucleus accum-responding (genotype effect, p � 0.001; genotype �bens was nearly abolished in Homer2 KO mice (Figuresession interaction, p � 0.002, n � 12). Given the appear-6A; genotype � time, p � 0.0001, n � 7–8).ance of motivational deficits for sucrose reward, further

Following withdrawal from repeated cocaine adminis-experimentation with Homer1 KO mice was discon-tration, the reduction in basal glutamate levels results

tinued, and only Homer2 KO and wt mice were trainedfrom decreased activity of the cystine-glutamate ex-

to acquire intravenous cocaine self-administration (0.1 changer (Baker et al., 2003). The cystine/glutamate ex-and 0.3 mg/kg). When trained on an FR1 schedule of changer is a heterodimer that exchanges extracellularreinforcement, Homer2 KO mice acquired operant self- cystine for intracellular glutamate and serves as bothadministration of cocaine more readily than wt mice the primary source of cystine for glutathione synthesis(Figure 5B; genotype effect, p � 0.01, n � 8–12). (Sato et al., 1999; Shih and Murphy, 2001) and the pri-

Importantly, the genotypic distinctions in behavioral mary source of nonsynaptic extracellular glutamate inresponding to cocaine were not associated with differ- the brain (Baker et al., 2002). Since the exchange ofences in cocaine levels in either brain (ng cocaine/g extracellular cystine for intracellular glutamate is nottissue: wt, 410.5 � 125.3, n � 6; Homer2 KO, 326.1 � energy dependent and relies upon substrate concentra-26.4, n � 7; p � 0.48) or plasma (ng cocaine/ml plasma: tions (Shih and Murphy, 2001; Baker et al., 2002), thewt, 1511.2 � 315.9; Homer2 KO, 1205.6 � 260.1; p � reduced activity of the exchanger after repeated cocaine0.46) at 15 min after acute administration of cocaine (7.5 administration can be overcome by supplying extracel-

lular cystine (Baker et al., 2003). Accordingly, cystinemg/kg, i.p.).

Figure 5. Homer1 Deletion Reduces Respond-ing for 15% Sucrose and Homer2 DeletionPromotes Acquisition of Intravenous CocaineSelf-Administration

(A) Mice were trained to lever press for su-crose (15%) using an FR1 schedule of rein-forcement. Data were normalized to the aver-age total number of reinforcements achievedby wt in sessions 5 through 7, and each geno-type was analyzed with a separate two-wayANOVA with repeated measures over ses-sion, followed by an LSD post hoc compari-son. n � 6 for each group. (B) Homer2 dele-tion caused a more rapid acquisition ofcocaine self-administration. Data are shownas mean � SEM number of days to reachcriterion for stable self-administration (seeExperimental Procedures) (n � 6 at eachdose). Data were analyzed using a two-wayANOVA. �p � 0.05, KO versus wt.

Neuron406

Figure 6. Homer2 Gene Deletion Reduces the Increases in Extracellular Glutamate Produced in the Accumbens by mGluR1 and Cystine/Glutamate Exchange

(A) Relative to wt mice, Homer2 KO mice displayed an �50% reduction in basal extracellular glutamate in the nucleus accumbens, whichwas accompanied by a blunted capacity of reverse microdialysis of DHPG to increase glutamate (n � 7). (B) The capacity of reverse microdialysisof cystine to increase accumbens extracellular glutamate was enhanced in Homer2 KO mice relative to wt animals, an effect that was reversedby blocking the cystine-glutamate antiporter with CPG (n � 7). (C) Homer2 deletion reduced total protein for mGluR1 and xCT and elevatedAGS3, without altering mGluR5, NR2A, or NR2B content in the nucleus accumbens. Data are representative immunoblots from wt and KOmice (n � 6–8). Behavioral and dialysis data were evaluated using a two-way ANOVA with repeated measures over time, followed by an LSDpost hoc comparison. Immunoblots were compared using a two-tailed Students’s t test. *p � 0.05 versus average basal levels of glutamate;�p � 0.05 versus wt.

administered via the microdialysis probe dose-depen- with the Homer-shank protein complex (Naisbitt et al.,1999). In addition, it was recently shown that withdrawaldently increased extracellular glutamate in the nucleus

accumbens of Homer2 KO mice (Figure 6B; cystine ef- from cocaine elevated the level of activator of G proteinsignaling 3 (AGS3), a protein that regulates signalingfect, p � 0.002; cystine � geneotype, p � 0.0001, n �

7). Similar to the effects of cystine in repeated saline- through group II mGluR (Bowers et al., 2004). Consistentwith Homer2 deletion producing a cocaine-sensitizedtreated rats (Baker et al., 2003), cystine perfusion did

not alter levels of extracellular glutamate in wt mice phenotype, the content of AGS3 was elevated in theaccumbens of Homer2 KO mice (wt, 100 � 9.7, n � 7;(Figure 6B). The cystine-induced increase in glutamate

was reversed by (s)-4-carboxyphenylglycine (CPG), a KO, 136.7 � 6.7, n � 7, p � 0.021).cystine-glutamate antiporter antagonist (Warr et al.,1999). Moreover, as was shown in rats (Baker et al., Selectivity of Homer2 KO Behavioral

Phenotype for Cocaine2002), CPG significantly reduced glutamate below basallevels in wt mice (Figure 6B). Thus, similar to withdrawal Of primary importance in behavioral studies employing

constitutive KO mice is the need to verify the behavioralfrom repeated cocaine administration (Baker et al.,2003), the reduced basal levels of glutamate following specificity and selectivity of the effects of gene deletion.

The deficiencies shown after Homer1 deletion in operantHomer2 deletion may arise from reduced activity of thecystine-glutamate antiporter. responding for sucrose (Figure 5A) raised concerns that

learning and memory deficits may accompany Homer2Immunoblotting for the mGluR5 dimer, mGluR1 di-mer, and xCT (catalytic protein of the cystine/glutamate deletion. Thus, radial arm and Morris water mazes were

used to examine genotypic distinctions in working andexchanger) was conducted in nucleus accumbens ho-mogenates to determine if protein content may contrib- reference memory, respectively, between Homer2 KO

and wt mice. No genotypic differences were observedute to the genotypic distinctions in DHPG and cystine-stimulated increase in extracellular glutamate (Figure for the latency to locate a hidden platform in a Morris

water maze (Figure 7A; genotype � trial, p � 0.86, n �6C). While no genotypic difference in total protein con-tent for the mGluR5 dimer was measured (wt, 100 � 6–8) or for the recall of the platform location at 24 hr

(wt, 21.0 � 8.5; KO, 25.3 � 6.2 s; p � 0.80) or 1 week6.5; KO, 109.0 � 13.9, mean percent change from wt �SEM; p � 0.57, n � 8–9), the amount of mGluR1 dimer (wt, 18.3 � 4.9; KO, 21.2 � 6.6 s; p � 0.67) following

acquisition of the task. Also, no genotypic differenceswas reduced by �25% in KO mice (wt, 100 � 8.3; KO,76.7 � 7.3; p � 0.05, n � 8–9). Immunoblotting also were observed in the number of working or reference

memory errors in the radial arm maze (working memoryrevealed reduced levels of xCT in Homer2 KO versuswt mice (wt, 100 � 11.4, n � 6; KO, 66.5 � 11.2, n � 7; errors: genotype � session � trial, p � 0.51; reference

memory errors: p � 0.22, n � 12; data not shown). Asp � 0.034). Given that two proteins related to excitatoryneurotransmission were affected by Homer2 deletion, additional evidence for a lack of genotypic differences in

learning, both KO and wt mice demonstrated equivalentimmunoblotting of additional glutamate receptor pro-teins in the accumbens was conducted. No significant acquisition of cocaine discrimination (days to reach dis-

crimination criterion: wt, 39.9 � 7.3; KO, 39.8 � 11.1;genotypic effect was measured in GluR1 (wt, 100 � 9.6,n � 9; KO, 82.3 � 8.4, n � 8; p � 0.77), NR1 (wt, 100 � p � 0.99, n � 9). A dose of 7.5 mg/kg cocaine was

chosen as a dose midway between two doses showing12.1, n � 9; KO, 97.4 � 10.4, n � 8, p � 0.87), NR2A(wt, 100 � 9.6, n � 9; KO, 82.3 � 8.4, n � 8, p � 0.19), augmented motor activity in Homer2 KO mice (Figure

1D). Once animals were trained to discriminate cocaineor NR2B (wt, 100 � 9.8, n � 8; KO, 89.7 � 11.7, n � 8).Of note, NR2 has been shown to coimmunoprecipitate from saline, the shape of the cocaine dose discrimina-

Homer, Glutamate, and Cocaine Addiction407

fect, p � 0.02), the levels of experimenter-directed ag-gression in a reactivity to handling test were the samebetween Homer2 KO and wt mice (reactivity scores: wtfemale, 1.57 � 0.53; KO female, 1.00 � 0.37; wt male,2.00 � 0.37; KO male, 3.44 � 0.80; genotype effect, p �0.46; genotype � gender, p � 0.09, n � 12). The lackof distinction between KO and wt mice in the precondi-tioning and experimenter-directed aggression tests indi-cates no effect by Homer2 deletion in responding tostressful stimuli.

To determine if the effects of Homer2 deletion maygeneralize to other drugs of abuse, genotypic differ-ences in the locomotor response to heroin (100 �g/kg),phencyclidine (3 and 6 mg/kg), and caffeine (10 mg/kg)were assessed. Behavior was analyzed for 2 hr using a16 photocell activity apparatus, and as found in theconditioned place preference apparatus (Figure 1D), co-caine (5 mg/kg) elicited a greater locomotor responsein Homer2 KO than wt mice (Figure 7C; saline versuscocaine, genotype � drug: p � 0.04, n � 6–8). The higherdose of phencyclidine (6 mg/kg) also produced a greaterlocomotor response in Homer2 KO mice (p � 0.001,n � 6–8). No genotypic difference in locomotion wasobserved following an injection of saline, nor were differ-ences observed following injections of 100 �g/kg heroin(p � 0.78, n � 6–8) or 10 mg/kg caffeine (p � 0.86, n �7–8), indicating that Homer may not regulate responsesto all drugs of abuse. The relatively selective effect ofHomer2 deletion on phencyclidine-induced motor activ-ity is consistent with the facts that phencyclidine is anoncompetitive antagonist at NMDA glutamate recep-tors (Lodge and Johnson, 1990) and that Homer proteinshave been shown to coimmunoprecipitate with NMDAreceptor subunits (Naisbitt et al., 1999) and to regulatethe trafficking of NMDA receptors to the synapse (Ebi-hara et al., 2003; Sala et al., 2003; Shiraishi et al., 2003).The lack of effect of Homer2 deletion on motor re-sponding to heroin indicates that the role of Homer pro-teins in addiction may not generalize to all addictivedrugs.Figure 7. Homer2 Deletion Does Not Alter Learning and Memory

Processes but Sensitizes the Locomotor Stimulant Effect of Phency-clidine Discussion(A) No genotypic difference was observed for performance duringthe acquisition of a Morris water maze task (n � 8). (B) Homer2 Deletion of Homer2 elicited changes in extracellular glu-deletion did not alter the dose-response function for cocaine dis-

tamate and behavior that were remarkably parallel tocrimination (n � 5). (C) Genotypic differences in locomotion (dis-the neuroadaptations produced by withdrawal from re-tance traveled) were revealed by cocaine and 6 mg/kg phencycli-peated cocaine. Changes included (1) augmented ac-dine, but not for saline, heroin, or caffeine (n � 6). Dose (mg/kg) is

indicated in parenthesis after drug name. Data were evaluated using quisition of cocaine self-administration (Horger et al.,a two-way ANOVA, with repeated measures over session (A) or dose 1990), (2) enhanced cocaine place conditioning (Ship-(B). *p � 0.05 versus saline; �p � 0.05 versus wt. penberg et al., 1998), (3) augmented cocaine-induced

motor activity (Vanderschuren and Kalivas, 2000), (4)reduced basal levels of extracellular glutamate in thetion function revealed no genotypic distinctions (Figure

7B; genotype � dose, p � 0.51, n � 5–8), and the time accumbens (Baker et al., 2003), (5) increased extracellu-lar glutamate by cocaine (Hotsenpiller et al., 2001; Piercecourses of cocaine discrimination were similar between

wt and Homer2 KO mice (genotype � time, p � 0.32; et al., 1996; Reid and Berger, 1996), (6) increased sensi-tivity of cystine/glutamate exchange to extracellular cys-data not shown). In addition to finding no genotypic

differences in learning, memory, and cocaine discrimi- tine (Baker et al., 2003), (7) reduced rise in extracellularglutamate by stimulating mGluR1/5 (Swanson et al.,nation tasks, wt and Homer2 KO mice showed equiva-

lent unconditioned avoidance of the white compartment 2001), and (8) increased and decreased content of AGS3and xCT, respectively, in the accumbens. The sensitiza-on the preconditioning test conducted prior to beginning

the cocaine place conditioning study in Figure 1C (wt, tion-like phenotype in Homer2 KO mice occurred in theabsence of prior cocaine exposure and could be re-235.8 � 6.2 s; KO, 255.5 � 5.7 s; p � 0.94, n � 6). Also,

while the gender differences were observed (gender ef- versed by Homer2b-AAV infusion into the nucleus ac-

Neuron408

cumbens. These findings are consistent with an emerg- accumbens of Homer2 KO mice restored the augmentedbehavioral and neurochemical responses to wt levels.ing model of addiction whereby sensitization to cocaine

results from changes in glutamate signaling to the ac- The active evolution in our understanding of the diverseroles for Homer proteins in excitatory cell signaling andcumbens (Baker et al., 2003; Kelz et al., 1999; Sutton et

al., 2003; Tzschentke and Schmidt, 2003; Wolf, 1998). neuroplasticity does not reveal a simple mechanism me-diating sensitization to the rewarding and motor stimu-Importantly, the sensitization-like consequences of

Homer2 deletion were not associated with effects on lant effects of cocaine. Nonetheless, the present datapoint to alterations in excitatory signaling, and foremostmeasures of spontaneous motor activity, anxiety, or

learning and memory. The fact that Homer1 or Homer2 among the influences of Homer proteins in excitatorytransmission are effects on calcium homeostasis (Shira-KO mice show augmented responses to cocaine poses

the possibility that Homer proteins and/or synaptic pro- ishi et al., 2003; Usui et al., 2003; Yuan et al., 2003).Homer effects on calcium signaling could contributeteins functionally linked to Homer contribute to co-

caine addiction. to alterations in cystine-glutamate exchange and xCTsynthesis that are known to contribute to cocaine-induced drug seeking and synaptic glutamate releaseExtracellular Glutamate and Sensitization(Baker et al., 2002, 2003). xCT transcription is regulatedto Cocaineby binding of the basic leucine zipper transcription fac-The hypothesis that neuroadaptions in accumbens glu-tor Nrf2 to an antioxidant response element (ARE), andtamate transmission produced by repeated cocaine ad-the nuclear translocation of Nrf2 occurs in response toministration underlie behaviors associated with addic-oxidative stress (Sasaki et al., 2002). Efficient activationtion emerged from neuroimaging cocaine addictsof ARE-driven transcription by Nrf2 requires calcium-(Goldstein and Volkow, 2002) as well as from animaldependent protein kinase (PKC) phosphorylation of Nrf2models of cocaine-induced behaviors (Baker et al.,(Huang et al., 2002). Another potential mechanism is2003; Cornish and Kalivas, 2000; Di Ciano and Everitt,suggested by the observation that Homer2 modulates2001; Pierce et al., 1996; Park et al., 2002). In animalG protein receptor signaling in pancreatic acinar cellsmodels, in vivo microdialysis reveals a marked reductionby recruiting RGS proteins and PLC GAP activitiesof extracellular glutamate in the accumbens following(Shin et al., 2003). This action of Homer2 is manifestwithdrawal from repeated cocaine administration thatwith G protein receptors that do not appear to directlyappears to result from reduced activity of the cystine-bind Homer, including M1 muscarinic and bradykininglutamate exchanger (Baker et al., 2003). Stimulation ofreceptors, and suggests that Homer may influence sig-cystine-glutamate exchange normalizes glutamate lev-naling by multiple G protein receptors, in addition toels and prevents behavioral sensitization and cocainemGluR1/5.seeking in animal models of relapse (Baker et al., 2003).

Homer2 deletion elicited the same changes in extracel-lular glutamate that have been demonstrated to be criti- Gene Deletion and Cocaine Behaviorscal mediators of cocaine-seeking, including reduced The effect of deleting other genes on cocaine-inducedbasal extracellular glutamate, augmented increase in behaviors has been examined. Deleting RGS9-2 orextracellular glutamate by acute cocaine administration, 5HT1B produces augmented behavioral responses toand altered cystine-glutamate exchange. cocaine that are akin to those elicited by Homer1 or

It is notable that cocaine-naive Homer2 KO mice did Homer2 deletion (Gingrich and Hen, 2001; Rahman etnot manifest augmented extracellular dopamine in the al., 2003), while -arrestin and DARP-32 deletions areaccumbens, which is characteristic of sensitization to without effect (Bohn et al., 2003; Hiroi et al., 1999) andrepeated cocaine administration in wild-type rodents mGluR5 deletion results in mice that are apparently in-(Vanderschuren and Kalivas, 2000). This dissociation sensitive to the behavioral effects of cocaine (Chiamul-highlights the importance of adaptations in extracellular era et al., 2001). The sensitization that is associated withglutamate in the long-lasting behavioral changes pro- the reduced capacity of mGluR1/5 stimulation to induceduced by cocaine. In contrast, alterations in extracellular extracellular glutamate in Homer2 KO mice (Figure 6A)dopamine in the accumbens appear to be more critical would seem to contradict the lack of cocaine respon-for the drug’s acute rewarding and locomotor stimulant siveness produced by mGluR5 deletion. However, theeffects (Cornish and Kalivas, 2000; Wolf, 1998). increase in extracellular glutamate by DHPG is mediated

by mGluR1, not mGluR5 (Cochilla and Alford, 1998;Swanson et al., 2001). In contrast with Homer, RGS9-2Homer Proteins and Addiction

The observation that Homer1a is rapidly induced by is elevated by repeated cocaine administration and ispostulated to serve a compensatory rather than facilita-acute exposure to cocaine first suggested a role for

Homer in addiction (Brakeman et al., 1997). Subse- tory role in regulating cocaine behaviors (Rahman et al.,2003). Deletion of Homer2 elicited effects that appearquently, Homer1b/c was shown to be reduced in the

accumbens of rats withdrawn from repeated cocaine to be selective for cocaine behaviors and apparentlydid not impact natural reward, aggression, anxiety, oradministration (Swanson et al., 2001). This adaptation

appears to be important for cocaine-induced sensitiza- learning and memory, while 5HT1B KO mice show ele-vated levels of aggression (Gingrich and Hen, 2001). Thetion, since, similar to Homer1 or Homer2 deletion, reduc-

ing Homer1 expression in the nucleus accumbens using detailed neurochemical adaptations that we describe inthe Homer2 KO mice have not been reported in thesean antisense oligonucleotide strategy augmented co-

caine-induced motor stimulation (Ghasemzadeh et al., other genetic models, but the convergence of Homer2(Shin et al., 2003) and RGS signaling (Rahman et al.,2003), and restoring Homer2b expression in the nucleus

Homer, Glutamate, and Cocaine Addiction409

phosphate transfection methods. Forty-eight hours after transfec-2003) focuses attention on G protein signaling. A roletion, cells were harvested and the vector purified using heparinfor altered G protein signaling is further implicated byaffinity columns as described (During et al., 2003). Genomic titersthe similar elevation in AGS3 produced by Homer2 dele-were determined using the Prism 7700 sequence detector system

tion and withdrawal from repeated cocaine administra- (Applied Biosystems, Foster City, CA) with primers designed totion and that inactivation of Gi by elevated AGS3 sensi- WPRE as previously described (Clark et al., 1999).tizes cocaine-induced behaviors and increases inextracellular glutamate in the accumbens (Bowers et Place Conditioning and Cocaine-Induced Locomotion

Cocaine reward was assessed using a place conditioning procedureal., 2004).and apparatus similar to that described previously (Szumlinski etWhile it can be anticipated that additional genetical., 2003). Place conditioning chambers were constructed of plexi-models may be discovered that mimic or block behav-glas (46 cm long � 24 cm high � 22 cm wide) and were dividediors associated with cocaine addiction, the striking con-into two distinct compartments by a removable divider. The com-

cordant neurochemical phenotype between Homer2 de- partments were designed to provide a “biased” place conditioningletion and withdrawal from chronic cocaine treatment paradigm, which is considered to be a sensitive indicator of condi-

tioned drug reward (Shimosato and Ohkuma, 2000). One of theindicates that Homer is a particularly good candidatecompartments was painted black and contained bedding; the otherto play a central role in cocaine addiction. Moreover,was painted white and contained no bedding. The black compart-Homer gene products are regulated by acute and re-ment was covered with an opaque lid, and a translucent white lidpeated cocaine administration (Brakeman et al., 1997;covered the white compartment. Photocells along the base of the

Swanson et al., 2001), and Homer1a is dynamically re- chamber detected entrance into and movements within each com-sponsive to environmental cues and stress (Vazdarja- partment.

Place conditioning proceeded in the following four sequentialnova et al., 2002). Thus, not only does Homer providephases: habituation, preconditioning test (Pre-Test), conditioning,an important window to understand cocaine-inducedand postconditioning test (Post-test). Tests for each phase (includ-neuroplasticity and addiction, but also to study the mo-ing habituation) were 15 min in duration. The habituation phaselecular basis of the important link between environmen-enabled free access to both compartments, and the amount of time

tal stress and cocaine addiction. spent in the white compartment served as an index of anxiety. ThePre-Test was given the day following habituation in which the timespent in each compartment was recorded to establish compartmentExperimental Procedurespreference, and the time in the nonpreferred compartment servedas a measure of reward-induced place conditioning.Animals

Subjects were adult (20–35 g) wt and Homer1 [CMV-Cre (BALB/ Cocaine conditioning was conducted over an 8 day period. Forcocaine place conditioning, mice were brought into the conditioningcJ) � Flpe (C57BL/6 � 129sv) � C57BL/6; Yuan et al., 2003], Homer2

(129sv � C57BL/6 F5-F10; Shin et al., 2003), and Homer3 (129sv � room and were injected (0.01 ml/g body weight, i.p.) with one ofseveral doses of cocaine (NIDA, Bethesda, MD) and returned toC57BL/6 F5-F10; Shin et al., 2003) KO mice. As the background

strains differed between the different Homer KO mice, KOs were their home cage for 5 min prior to being placed in the conditioningchamber. The 5 min delay between injection and conditioning wascompared to their respective wt littermates. Mice were housed indi-

vidually in an AAALAC-approved animal facility in standard mouse based on a study indicating greater cocaine place conditioning formice using this delay than for mice placed immediately in the cham-cages (lights on, 0800 hr; 25�C). All testing was conducted during

the light cycle. Unless otherwise indicated, food and water were ber following injection (Cunningham et al., 1999). Mice were placedin their preferred compartment following saline injections and onavailable ad libitum. All experiments were approved by the Institu-

tional Animal Care and Use Committee of the Medical University the nonpreferred compartment following cocaine. For Homer1 KOmice, the cocaine doses employed were 3, 10, and 30 mg/kg. Forof South Carolina and conducted in accordance with the National

Institutes of Health (NIH) Principles of Laboratory Animal Care Homer2 KO mice, the doses were 0.5, 2.5, 5, 10, and 50 mg/kg.To screen for genotypic differences in cocaine-induced behavior(1985). Unless indicated, naive mice were employed in each experi-

ment. As insufficient numbers of one gender were available at the between wt and Homer3 KO mice, 15 mg/kg cocaine was adminis-tered, a dose intermediate to those which elicited maximal placetime of study, both male and female mice were employed. One

exception was the cocaine self-administration experiment, which conditioning and locomotion in the Homer1 and Homer2 studies(see Results). For all cocaine experiments, control groups receivedincluded male mice only. For all the experiments employing both

genders, the numbers of male and female wt and KO subjects em- saline (0 mg/kg cocaine) in both compartments. The photocellcounts in the cocaine-paired compartment during conditioningployed in each experiment were equal (�1), and gender was in-

cluded as a factor in all initial statistical analyses. With the exception served to index cocaine-induced locomotor hyperactivity. For placeconditioning studies, a Post-Test was administered the day follow-of experimenter-directed aggression, statistical analyses of the data

in this report revealed no main effects of, or interactions with, the ing the last cocaine conditioning session, which was identical tothe Pre-Test. The difference in the time spent in the reward-pairedgender factor (p � 0.05).compartment between the Pre- and Post-Tests served to indexconditioned reward.Generation of Recombinant Adeno-Associated Virus

The rat Homer2b coding sequence was amplified using whole-braincDNA, and the PCR was product expressed as an N-terminal fusion Operant Responding for Sucrose

To assess for genotypic differences in reinforcement by a naturalprotein with the hemagglutinin (HA)-tag in a rAAV backbone con-taining the 1.1 kb CMV enhancer/chicken -actin (CBA) promoter, reward, Homer1 KO, Homer2 KO, and their respective wt mice were

trained in an operant paradigm to lever press for a palatable 15% w/v800 bp human interferon- scaffold attachment region (SAR) in-serted 5� of the promoter, the woodchuck posttranscriptional regu- sucrose solution. The apparatus employed in the operant studies

consisted of standard mouse operant conditioning chambers (Med-latory element (WPRE), and the bovine growth hormone polyAflanked by inverted terminal repeats (Homer2b-AAV). The same Associates, St. Albans, VT) located within ventilated sound-attenu-

ated chambers. The chambers were equipped with two levers lo-rAAV-CBA-WPRE-bGH backbone encoding the enhanced green flu-orescent protein (EGFP) was used as control (GFP-AAV). AAV pseu- cated on one side of the chamber (located 6.5 cm above the floor),

amber stimulus lights located above each lever, an amber housedotyped vectors (virions containing a 1:1 ratio of AAV1 and AAV2capsid proteins with AAV2 ITRs were generated as described (Hauck light located on the wall opposite to the levers, and a tone generator

(ENV323A, MedAssociates) located behind the house light. Traininget al., 2003). Briefly, HEK 293 cells were transfected with the AAVcis plasmid, the AAV1 (pH21) and AAV2 (pRV1) helper plasmids sessions were conducted during daily 30 min sessions during the

light cycle beginning at 1300 hr. For these sessions, sucrose was(pF 6) and the adenovirus helper plasmid by standard calcium

Neuron410

delivered via a dipper arm up to a nose-poke hole located equidis- Cocaine DiscriminationGenotypic differences in the discriminative stimulus (“subjective”)tant from both levers. Mice were trained to press the left lever for

sucrose delivery on an FR1 schedule of reinforcement with a 20 s effects of cocaine were assessed in Homer2 KO and wt mice usingprocedures similar to those described previously for C57BL/6 micetimeout period. The house light was illuminated throughout the ses-

sion. Responses on the active lever resulted in simultaneous activa- (Middaugh et al., 1998; also see the Supplemental ExperimentalProcedures [http://www.neuron.org/cgi/content/full/43/3/401/DC1]).tion of the dipper arm, the stimulus light above the active lever, and

the tone generator for 2 s. Responses on the inactive lever andresponses during the timeout period had no programmed conse-

Radial Arm and Morris Water Mazesquences. Mice were restricted to 80% of their ad libitum body weightThe apparatus and procedures that were employed to train mice inthroughout testing.a water version of the radial arm maze and the Morris water mazeare similar to those described previously Bimonte et al. (2003) and

Operant Responding for Cocaine Avgeropoulos et al. (1998), respectively (also see the SupplementalTo assess for genotypic differences in cocaine self-administration, Experimental Procedures [http://www.neuron.org/cgi/content/full/Homer2 KO and wt mice were trained in an operant paradigm to lever 43/3/401/DC1]).press for an intravenous infusion of cocaine. Under pentobarbitalanesthesia (50 mg/kg, vol � 0.01 ml/g body weight, i.p.), mice were

Cocaine Concentration in Plasma and Brainfitted with in-dwelling intravenous catheters that were externalizedCocaine concentration was determined in plasma and brain fromto the top of the skull as described previously (Griffin and Middaugh,Homer2 KO and wt mice immediately following the 15 min cocaine2003). Catheters were flushed with 0.02 ml cefazolin (0.1 g/ml) anddiscrimination test. For this, blood was sampled from the intraocularheparinized saline (70 U/ml) immediately after surgery, on the morn-sinus, and the plasma was prepared for analysis as described pre-ing and evening of the following day, and immediately after eachviously (Patrick et al., 1993).self-administration session. Catheter patency was verified by ad-

ministering 0.02 ml of methohexital sodium (6.3 mg/ml, Eli Lilly,Indianapolis, IN) on the day after the surgery and every 4–5 days of

In Vivo Microdialysis-Cocaine Studyself-administration. Mice were given at least 48 hr to recover priorTo relate the behavioral phenotype of Homer1 and Homer2 KO miceto the start of self-administration training.to gene deletion-induced alterations in neurotransmission, in vivoThe apparatus employed in the cocaine self-administration studymicrodialysis was performed prior to, and following, a challengewas identical to that described for the sucrose study above, with theinjection of cocaine (15 mg/kg, i.p.) in Homer KO and wt mice. Underexception that the nose-poke hole was covered. Self-administrationpentobarbitol anesthesia (50 mg/kg, i.p.), naive mice were implantedwas conducted during daily 2 hr sessions during the light cycleunilaterally with a 20 gauge stainless steel guide cannula (7 mmbeginning at 1300 hr. For these sessions, intravenous drug infusionslong) aimed 3 mm above the nucleus accumbens (AP, �0.5 mm;were delivered through an infusion apparatus that was housedML, � 0.5 mm; DV, �3.5 mm, relative to bregma) or the frontalwithin the sound-attenuated chamber, which consisted of a liquidcortex (AP, �1.34 mm; ML, � 0.6 mm; DV, �2.0 mm, relative toswivel (Instech, Plymouth Meeting, PA) and a syringe mounted onbregma) (Franklin and Paxinos, 1997), and the guide cannula wasan infusion pump (MedAssociates). Tygon microbore tubing (Nortonfixed to the skull using dental resin as previously described (Grose-Performance Plastics, Akron, OH) connected the components ofclose et al., 1998). Following 5–7 days recovery, a microdialysisthe infusion apparatus to each other and to the exit port of theprobe (24 gauge; 10 mm in length, including 1–1.5 mm of activemouse’s catheter.membrane) was lowered into the guide cannula and perfused witha microdialysis buffer (5 mM glucose, 2.5 mM KCl, 140 mM NaCl,1.4 mM CaCl2, 1.2 mM MgCl2, 0.15% phosphate-buffered saline [pHAcquisition of Cocaine Self-Administration7.4]) at a rate of 2 �l/min as described previously for C57BL/6 miceMice were trained to press the right lever for infusions of cocaine(Middaugh et al., 2003). Four hours following probe implantation,(0.1 or 0.3 mg/kg, 19�l/infusion; NIDA, Bethesda, MD) on an FR1dialysis samples were collected every 20 min, beginning 80 minschedule of reinforcement with a 20 s timeout period. The houseprior to an i.p. injection of 15 mg/kg cocaine (vol � 0.01 ml/g) andlight was illuminated throughout the session. Responses on thethen for 120 min thereafter. Samples were frozen at �80�C untilactive lever resulted in simultaneous activation of the infusion pump,analysis (see below).the stimulus light above the active lever, and the tone generator for

2 s. Responses on the inactive lever and responses during thetimeout period had no programmed consequences. To expedite the

In Vivo Microdialysis Regulation of Accumbensacquisition of self-administration, mice were restricted to 80% ofExtracellular Glutamatetheir ad libitum body weight throughout testing. Mice were limitedGenotypic differences in basal extracellular glutamate were as-to 120 infusions per 2 hr session and were allowed 14 days to reachsessed between Homer2 KO and wt mice using no net flux microdial-self-administration criterion which was defined as obtaining ten rein-ysis. Four hrs after probe implantation, the concentration of extracel-forcers within 2 hr with 80% of total responses on the active leverlular glutamate was measured for 60 min, and at 60 min intervalsfor 4 consecutive days.thereafter, different concentrations of glutamate were passedthrough the probe. The resulting linear regression provides a quanti-tative estimate of basal extracellular concentrations (y intercept)Drug-Induced Locomotion

In an experiment designed to compare genotypic differences in and elimination of glutamate (slope; Baker et al., 2003). Genotypicdifferences in mGluR and cystine-glutamate antiporter function weresensitivity to the psychomotor-activating effects of other drugs of

abuse, Homer2 KO and wt mice were injected i.p. (vol � 0.01 ml/g), also assessed using in vivo microdialysis. Increasing concentrationsof the group1 mGluR agonist (S)-3,5-dihydroxyphenylglycineevery third day, with one of the following: saline, cocaine (5 mg/kg;

NIDA), heroin (100 �g/kg; NIDA), caffeine (10 mg/kg; Sigma/RBI, St. (DHPG; 3, 30 and 300 �M/side; Tocris Cookson, Ballwin, MO) or thecystine-glutamate antiporter agonist cystine (10, 30, and 100 nM/Louis, MO), or phencyclidine (3 or 6 mg/kg; Sigma/RBI). Each mouse

received a maximum of four injections, and the order of drug delivery side; Sigma/RBI) was infused directly into the nucleus accumbensof mice via the microdialysis probe. In the cystine experiment, thewas random across the four injection sessions. Locomotor activity

was monitored in plexiglas activity chambers (22 � 43 � 33 cm) antiporter antagonist (s)-4-carboxyphenylglycine (CPG; 50 �M;Tocris Cookson) (Ye et al., 1999) was also coinfused with 100 nM(Omnitech Electronics, Columbus, OH). To reduce the contribution

of context-dependent factors upon the expression of drug-induced cystine. Samples were collected every 20 min, and each drug con-centration/combination was infused for 60 min, as previously de-hyperactivity, mice were exposed to different testing chambers on

each injection session. For these sessions, animals were allowed scribed for rats (Swanson et al., 2001; Baker et al., 2003). The lasttwo samples of each drug concentration were averaged and usedto habituate to the activity chambers for 30 min prior to injection,

drug was injected, and then behavior was recorded for 120 min. in the statistical analysis of the data.

Homer, Glutamate, and Cocaine Addiction411

Behavioral and Neurochemical Studies ReferencesFollowing AAV TreatmentTo confirm an active role for Homer proteins in the behavioral and Avgeropoulos, N., Kelley, B., Middaugh, L.D., Arrigo, S., Persidsky,neurochemical phenotype produced by Homer gene deletion, Y., Gendelman, H., and Tyor, W.R. (1998). SCID mice with HIV en-Homer2 KO and wt mice were implanted with guide cannulae, as cephalitis develop behavioral abnormalities. J. Acquir. Immune De-described above. Following 7 days recovery, mice were infused fic. Syndr. Hum. Retrovirol. 18, 13–20.bilaterally at a rate of 0.05 �l/min for 5 min (total volume � 0.25 �l/ Baker, D.A., Xi, Z.X., Shen, H., Swanson, C.J., and Kalivas, P.W.side) with either an AAV carrying the Homer2b splice variant of (2002). The origin and neuronal function of in vivo nonsynaptic gluta-the Homer2 gene (genomic titre � 3.6 � 1011) or an AAV carrying mate. J. Neurosci. 22, 9134–9141.enhanced green fluorescent protein (eGFP; genomic titre � 4.9 �

Baker, D.A., McFarland, K., Lake, R.W., Shen, H., Tang, X.C., Toda,1011) as a control. Three weeks following AAV infusion, animals wereS., and Kalivas, P.W. (2003). Neuroadaptations in cystine-glutamateassessed for genotypic differences in cocaine-induced changes inexchange underlie cocaine relapse. Nat. Neurosci. 6, 743–749.accumbens glutamate using conventional in vivo microdialysis asBimonte, H., Nelson, M., and Granholm, A. (2003). Age-releateddescribed above. Five days following the microdialysis session, ge-deficits as working memory load increases: relationships withnotypic differences in the capacity of 5 mg/kg cocaine to inducegrowth factors. Neurobiol. Aging 24, 37–48.locomotion and place conditioning were assessed using procedures

identical to those described above. Homer2b-AAV-mediated trans- Bohn, L.M., Gainetdinov, R.R., Sotnikova, T.D., Medvedev, I.O., Lef-duction was verified using a mouse anti-HA primary antibody (Co- kowitz, R.J., Dykstra, L.A., and Caron, M.G. (2003). Enhanced re-vance; 1:1,000 dilution). Labeled proteins were detected using a warding properties of morphine, but not cocaine, in beta(arrestin)-2biotinylated anti-mouse secondary IgG and visualized with DAB. knock-out mice. J. Neurosci. 23, 10265–10273.Fluorescence microscopy was used for transgene detection of Bottai, D., Guzowski, J.F., Schwarz, M.K., Kang, S.H., Xiao, B., Lana-GFP-AAV. han, A., Worley, P.F., and Seeburg, P.H. (2002). Synaptic activity-

induced conversion or intronic to exonic sequence in Homer 1 imme-diate early gene expression. J. Neurosci. 22, 167–175.HPLC Analysis of Glutamate and DopamineBowers, M.S., McFarland, K., Lake, R.W., Peterson, Y.K., Lapish,Glutamate and dopamine in the dialysis samples were measuredC.C., Lanier, S.M., and Kalivas, P.W. (2004). Activator of G-proteinusing a high-pressure liquid chromatography (HPLC) system withsignaling 3 (AGS-3): a gatekeeper of cocaine sensitization and drug-fluorescent or electrochemical detection, respectively, as describedseeking. Neuron 42, 269–281.in detail elsewhere (Baker et al., 2003; also see the Supplemental

Experimental Procedures [http://www.neuron.org/cgi/content/full/ Brakeman, P.R., Lanahan, A.A., O’Brien, R., Roche, K., Barnes, C.A.,43/3/401/DC1]). Huganir, R.L., and Worley, P.F. (1997). Homer: a protein that selec-

tively binds metabotropic glutamate receptors. Nature 386, 221–223.

Chiamulera, C., Epping-Jordan, M., Zocchi, A., Marcon, C., Cottiny,ImmunoblottingC., Tacconi, S., Corsi, M., Orzi, F., and Conquiet, F. (2001). Reinforc-Tissue was collected from wt and Homer2 KO mice and was sub-ing and locomotor stimulant effects of cocaine are absent in mGluR5jected to immunoblotting for AGS3, xCT, mGluR1, mGluR5,null mutant mice. Nat. Neurosci. 4, 873–874.NMDAR1, NMDAR2a, NMDAR2b, and GluR1 as described for ratsClark, K.R., Liu, X., McGrath, J.P., and Johnson, P.R. (1999). Highly(Ghasemzadeh et al., 2003; also see the Supplemental Experimentalpurified recombinant adeno-associated virus vectors are biologi-Procedures [http://www.neuron.org/cgi/content/full/43/3/401/DC1]).cally active and free of detectable helper and wild-type viruses.Hum. Gene Ther. 10, 1031–1039.

Histology and Statistical Analysis Cochilla, A.J., and Alford, S. (1998). Metabotropic glutamate recep-The placements of the microdialysis probes within the nucleus ac- tor-mediated control of neurotransmitter release. Neuron 20, 1007–cumbens and frontal cortex were verified in cresyl violet-stained 1016.tissue sections (50 �m) from wt and KO mice, as described pre-

Cornish, J., and Kalivas, P. (2000). Glutamate transmission in theviously for rats (Swanson et al., 2001). Only those mice whose cannu-nucleus accumbens mediates relapse in cocaine addiction. J. Neu-lae were located within the boundaries of the region of interest wererosci. 20, RC89.included in the statistical analysis. All neurochemical and behavioralCunningham, C.L., Dickinson, S.D., Grahame, N.J., Okorn, D.M.,data were statistically evaluated using analyses of variance, andand McMullin, C.S. (1999). Genetic differences in cocaine-inducedwhen appropriate, the data were decomposed for simple effects,conditioned place preference in mice depend on conditioning trialand post hoc comparisons were made using a Fischer’s least signifi-duration. Psychopharmacology (Berl.) 146, 73–80.cant difference test. Immunoblotting data were analyzed using a

two-tailed Student’s t test. � 0.05 for all analyses. Di Ciano, P., and Everitt, B.J. (2001). Dissociable effects of antago-nism of NMDA and AMPA/KA receptors in the nucleus accumbenscore and shell on cocaine-seeking behavior. Neuropsychopharma-

Acknowledgments cology 25, 341–360.

During, M.J., Young, D., Baer, K., Lawlor, P., and Klugmann, M.The authors wish to acknowledge NIDA for the generous gift of(2003). Development and optimization of adeno-associated viruscocaine. The authors would like to thank the following individualsvector transfer into the central nervous system. Methods Mol. Med.for their efforts: A.L. Marlowe and Y. Van Patten for their technical76, 221–236.advice; Drs. Andrenyak and Foltz (University of Utah) for the GC-Ebihara, T., Kawabata, I., Usui, S., Sobue, K., and Okabe, S. (2003).MS analysis of cocaine; Drs. Timothy Murphy and Andrew ShihSynchronized formation and remodeling of postsynaptic densities:(University of British Columbia) for helping characterize the xCTlong-term visualization of hippocampal neurons expressing post-antibody; and G. Harris for genotyping and maintenance of thesynaptic density proteins tagged with green fluorescent protein. J.mouse colony. This work was supported, in part, by grants fromNeurosci. 23, 2170–2181.NIDA and NIMH to P.F.W. (KO2-MH01152, DA-11742, DA-10309),

P.W.K. (DA-03906, DA-14185, MH-40817) and L.D.M. (DA-14185- Everitt, B.J., and Wolf, M.E. (2002). Psychomotor stimulant addic-01). K.K.S. is supported by a postdoctoral fellowship from the CIHR. tion: a neural systems perspective. J. Neurosci. 22, 3312–3320.M.K. was supported by an EMBO fellowship.

Franklin, B.J., and Paxinos, G. (1997). The Mouse Brain in Stereo-taxic Coordinates (San Diego: Academic Press).

Ghasemzadeh, M.B., Permenter, L.K., Lake, R., Worley, P.F., andReceived: February 13, 2004Revised: May 10, 2004 Kalivas, P.W. (2003). Homer1 proteins and AMPA receptors modu-

late cocaine-induced behavioural plasticity. Eur. J. Neurosci. 18,Accepted: July 14, 2004Published: August 4, 2004 1645–1651.

Neuron412

Gingrich, J.A., and Hen, R. (2001). Dissecting the role of the serotonin accumbens only in rats having developed behavioral sensitization.J. Neurosci. 16, 1550–1560.system in neuropsychiatric disorders using knockout mice. Psycho-

pharmacology (Berl.) 155, 1–10. Rahman, Z., Schwarz, J., Gold, S.J., Zachariou, V., Wein, M.N., Choi,K.H., Kovoor, A., Chen, C.K., DiLeone, R.J., Schwarz, S.C., et al.Goldstein, R.A., and Volkow, N.D. (2002). Drug addiction and its(2003). RGS9 modulates dopamine signaling in the basal ganglia.underlying neurobiological basis: neuroimaging evidence for theNeuron 38, 941–952.involvement of the frontal cortex. Am. J. Psychiatry 159, 1642–1652.

Reid, M.S., and Berger, S.P. (1996). Evidence for sensitization ofGriffin, W.C., and Middaugh, L.D. (2003). Acquisition of lever press-cocaine-induced nucleus accumbens glutamate release. Neurore-ing for cocaine in C57BL/6J mice: effects of prior Pavlovian condi-port 7, 1325–1329.tioning. Pharmacol. Biochem. Behav. 76, 543–549.

Sala, C., Futai, K., Yamamoto, K., Worley, P.F., Hayashi, Y., andGroseclose, C.H., Draughn, R.A., Tyor, W.R., Sallee, F.R., and Mid-Sheng, M. (2003). Inhibition of dendritic spine morphogenesis anddaugh, L.D. (1998). Long-term intracranial cannula stabilization insynaptic transmission by activity-inducible protein Homer1a. J. Neu-mice with light cured resin composites. J. Neurosci. Methods 79,rosci. 23, 6327–6337.31–36.

Sasaki, H., Sato, H., Kuriyama-Matsumura, K., Sato, K., Maebara,Hauck, B., Chen, L., and Xiao, W. (2003). Generation and character-K., Wang, H., Tamba, M., Itoh, K., Yamamoto, M., and Bannai, S.ization of chimeric recombinant AAV vectors. Mol. Ther. 7, 419–425.(2002). Electrophile response element-mediated induction of theHennou, S., Kato, A., Schneider, E.M., Lundstrom, K., Gahwiler, B.H.,cystine/glutamate exchange transporter gene expression. J. Biol.Inokuchi, K., Gerber, U., and Ehrengruber, M.U. (2003). Homer-1a/Chem. 277, 44765–44771.Vesl-1S enhances hippocampal synaptic transmission. Eur. J. Neu-Sato, H., Tamba, M., Ishii, T., and Bannai, S. (1999). Cloning androsci. 18, 811–819.expression of a plasma membrane cystine/glutamate exchange

Hiroi, N., Fienberg, A.A., Haile, C.N., Alburges, M., Hanson, G.R.,transporter composed of two distinct proteins. J. Biol. Chem.

Greengard, P., and Nestler, E.J. (1999). Neuronal and behavioural274, 11455–11458.

abnormalities in striatal function in DARPP-32-mutant mice. Eur. J.Shalev, U., Grimm, J.W., and Shaham, Y. (2002). Neurobiology ofNeurosci. 11, 1114–1118.relapse to heroin and cocaine seeking: a review. Pharmacol. Rev.

Horger, B.A., Shelton, K., and Schenk, S. (1990). Preexposure sensi-54, 1–42.

tizes rats to the rewarding effects of cocaine. Pharmacol. Biochem.Shih, A.Y., and Murphy, T.H. (2001). xCt cystine transporter expres-Behav. 37, 707–711.sion in HEK293 cells: pharmacology and localization. Biochem. Bio-

Hotsenpiller, G., Giorgetti, M., and Wolf, M.E. (2001). Alterations phys. Res. Commun. 282, 1132–1137.in behaviour and glutamate transmission following presentation of

Shimosato, K., and Ohkuma, S. (2000). Simultaneous monitoring ofstimuli previously associated with cocaine exposure. Eur. J. Neu-conditioned place preference and locomotor sensitization followingrosci. 14, 1843–1855.repeated administration of cocaine and methamphetamine. Phar-

Huang, H.C., Nguyen, T., and Pickett, C.B. (2002). Phosphorylation macol. Biochem. Behav. 66, 285–292.of Nrf2 at Ser-40 by protein kinase C regulates antioxidant response

Shin, D.M., Dehoff, M., Luo, X., Kang, S.H., Tu, J., Nayak, S.K., Ross,element-mediated transcription. J. Biol. Chem. 277, 42769–42774.E.M., Worley, P.F., and Muallem, S. (2003). Homer 2 tunes G protein-

Kato, A., Ozawa, F., Saitoh, Y., Fukazawa, Y., Sugiyama, H., and coupled receptors stimulus intensity by regulating RGS proteinsInoduchi, K. (1998). Novel members of the Vesl/Homer family of PDZ and PLCbeta GAP activities. J. Cell Biol. 162, 293–303.proteins that bind metabotropic glutamate receptors. J. Biol. Chem.

Shippenberg, T.S., LeFevour, A., and Thompson, A.C. (1998). Sensi-273, 23969–23975.tization to the conditioned rewarding effects of morphine and co-

Kelz, M.B., Chen, J., Carlezon, W.A., Jr., Whisler, K., Gilden, L., caine: differential effects of the kappa-opioid receptor agonistBeckmann, A.M., Steffen, C., Zhang, Y.J., Marotti, L., Self, D.W., et U69593. Eur. J. Pharmacol. 345, 27–34.al. (1999). Expression of the transcription factor deltaFosB in the

Shiraishi, Y., Mizutani, A., Mikoshiba, K., and Furuichi, T. (2003).brain controls sensitivity to cocaine. Nature 401, 272–276.Coincidence in dendritic clustering and synaptic targeting of homer

Koob, G.F. (2003). Alcoholism: allostasis and beyond. Alcohol. Clin. proteins and NMDA receptor complex proteins NR2B and PSD95Exp. Res. 27, 232–243. during development of cultured hippocampal neurons. Mol. Cell.

Neurosci. 22, 188–201.Lodge, D., and Johnson, K.M. (1990). Noncompetitive excitatoryamino acid receptor antagonists. Trends Pharmacol. Sci. 11, 81–86. Sutton, M.A., Schmidt, E.F., Choi, K.H., Schad, C.A., Whisler, K.,

Simmons, D., Karanian, D.A., Monteggia, L.M., Neve, R.L., and Self,Middaugh, L.D., McGroarty, K.K., Groseclose, C.H., and Adinoff, B.D.W. (2003). Extinction-induced upregulation in AMPA receptors(1998). Cocaine discrimination: relationship to local anesthetics andreduces cocaine-seeking behavior. Nature 421, 70–75.monoamine uptake inhibitors in C57BL/6 mice. Psychopharmacol-

ogy (Berl.) 136, 44–49. Swanson, C., Baker, D., Carson, D., Worley, P., and Kalivas, P.(2001). Repeated cocaine administration attenuates group I metabo-Middaugh, L.D., Szumlinski, K.K., Van Patten, Y., Marlowe, A.L., andtropic glutamate receptor-mediated glutamate release and behav-Kalivas, P.W. (2003). Chronic ethanol consumption by C57BL/6 miceioral activation: A potential role for Homer 1b/c. J. Neurosci. 21,promotes tolerance to its interoceptive cues and increases extracel-9043–9052.lular dopamine, an effect blocked by naltrexone. Alcohol. Clin. Exp.

Res. 27, 1892–1900. Szumlinski, K.K., Frys, K.A., and Kalivas, P.W. (2003). Pretreatmentwith serotonin 5-HT(3) receptor antagonists produces no observableNaisbitt, S., Kim, E., Tu, J.C., Xiao, B., Sala, C., Valtschanoff, J.,blockade of long-term motor sensitization to cocaine in rats. Psy-Weinberg, R.J., Worley, P.F., and Sheng, M. (1999). Shank, a novelchopharmacology (Berl.) 165, 329–336.family of postsynaptic density proteins that binds to the NMDA

receptor/PSD-95/GKAP complex and cortactin. Neuron 23, 569–582. Thomas, M.J., Beurrier, C., Bonci, A., and Malenka, R.C. (2001).Long-term depression in the nucleus accumbens: a neural correlatePark, W.K., Bari, A.A., Jey, A.R., Anderson, S.M., Spealman, R.D.,of behavioral sensitization to cocaine. Nat. Neurosci. 4, 1217–1223.Rowlett, J.K., and Pierce, R.C. (2002). Cocaine administered into

the medial prefrontal cortex reinstates cocaine-seeking behavior by Tu, J.C., Xiao, B., Yuan, J.P., Lanahan, A.A., Leoffert, K., Li, M.,increasing AMPA receptor-mediated glutamate transmission in the Linden, D.J., and Worley, P.F. (1998). Homer binds a novel proline-nucleus accumbens. J. Neurosci. 22, 2916–2925. rich motif and links group 1 metabotropic glutamate receptors with

IP3 receptors. Neuron 21, 717–726.Patrick, K.S., Boggan, W.O., Miller, S.R., and Middaugh, L.D. (1993).Gas chromatographic-mass spectrometric determination of plasma Tzschentke, T.M., and Schmidt, W.J. (2003). Glutamatergic mecha-and brain cocaine in mice. J. Chromatogr. 621, 89–94. nisms in addiction. Mol. Psychiatry 8, 373–382.

Usui, S., Konno, D., Hori, K., Maruoka, H., Okabe, S., Fujikado, T.,Pierce, R.C., Bell, K., Duffy, P., and Kalivas, P.W. (1996). Repeatedcocaine augments excitatory amino acid transmission in the nucleus Tano, Y., and Sobue, K. (2003). Synaptic targeting of PSD-Zip45

Homer, Glutamate, and Cocaine Addiction413

(Homer 1c) and its involvement in the synaptic accumulation ofF-actin. J. Biol. Chem. 278, 10619–10628.

Vanderschuren, L.J., and Kalivas, P.W. (2000). Alterations in dopa-minergic and glutamatergic transmission in the induction and ex-pression of behavioral sensitization: a critical review of preclinicalstudies. Psychopharmacology (Berl.) 151, 99–120.

Vazdarjanova, A., McNaughton, B.L., Barnes, C.A., Worley, P.F., andGuzowski, J.F. (2002). Experience-dependent coincident expressionof the effector immediate-early genes arc and Homer 1a in hippo-campal and neocortical neuronal networks. J. Neurosci. 22, 10067–10071.

Warr, O., Takahashi, M., and Atwell, D. (1999) Modulation of extracel-lular glutamate concentration in rat brain slices by cystine-glutamateexchange. J. Physiol. 514, 783–793.

Wolf, M.E. (1998). The role of excitatory amino acids in behavioralsensitization to psychomotor stimulants. Prog. Neurobiol. 54,679–720.

Xiao, B., Tu, J.C., and Worley, P.F. (2000). Homer: a link betweenneural activity and glutamate receptor function. Curr. Opin. Neuro-biol. 10, 370–374.

Ye, Z.-C., Rothstein, J.D., and Sontheimer, H. (1999). Compromisedglutamate transport in human glioma cells: reduction-mislocaliza-tion of sodium-dependent glutamate transporters and enhancedactivity of cystine-glutamate exchange. J. Neuroscience 19, 10767–10777.

Yuan, J.P., Kiselyov, K., Shin, D.M., Chen, J., Shcheynikov, N., Kang,S.H., Dehoff, M.H., Schwarz, M.K., Seeburg, P.H., Muallem, S., andWorley, P.F. (2003). Homer binds TRPC family channels and is re-quired for gating of TRPC1 by IP3 receptors. Cell 114, 777–789.