The “Orphan” Na1/Cl2-Dependent Transporter, Rxt1, IsPrimarily Localized Within Nerve Endings of Cortical

Origin in the Rat Striatum

P. Kachidian, *J. Masson, *Z. Aı¨douni, †P. Gaspar, *M. Hamon,*S. El Mestikawy, and L. Kerkerian-Le Goff

CNRS-UPR 9013, Marseille; and* INSERM U288 and†INSERM U106, Institut Fe´deratif des Neurosciencesde la Pitie-Salpetriere, Paris, France

Abstract: Previous studies have shown that the striatumexpresses very low levels of Na1/Cl2-dependent “or-phan” transporter Rxt1 transcripts but contains high lev-els of protein. This study investigated the origin of Rxt1expression in rat striatum. Striatal Rxt1 contents as-sessed by immunocytochemistry or western blottingwere found to be significantly reduced after corticostriataldenervation but not after striatal or thalamic lesion withkainic acid or selective 6-hydroxydopamine-induced ni-grostriatal deafferentation. Corticostriatal neurons retro-gradely labeled by intrastriatal fluorogold injections wereshown to express Rxt1 mRNA. Combination of antero-grade biotin–dextran amine labeling of the corticostriatalpathway with Rxt1 immunogold detection at the ultra-structural level demonstrated the presence of Rxt1 inabout one-third of the corticostriatal synaptic terminalsand in numerous unidentified synaptic terminals. All theRxt1-positive terminals formed asymmetrical contacts onspines. These data provide evidence that striatal Rxt1immunoreactivity is mainly of extrinsic origin and morespecifically associated with the corticostriatal pathway.Rxt1 appears as a selective presynaptic marker of syn-apses formed by presumably excitatory amino acid af-ferents, but it segregates a subclass of these synapses,thereby revealing a functional heterogeneity among exci-tatory amino acid systems. Key Words: Na1/Cl2-depen-dent transporter—Excitatory amino acids—Corticostria-tal pathway—Asymmetrical synapses.J. Neurochem. 73, 623–632 (1999).

Thanks to molecular cloning, the superfamily of Na1/Cl2-dependent transporters has been unravelled (Amara,1992; Uhl and Hartig, 1992; Amara and Kuhar, 1993;Worral and Williams, 1994). These transporters are incharge of the clearance of various neurotransmitters (do-pamine, serotonin, noradrenaline, GABA, and glycine)and other neuroactive molecules (proline and taurine)from the synaptic cleft in the CNS, as well as solutes andosmolytes (betaine and creatine) in the periphery (Guim-bal and Kilimann, 1993; Nash et al., 1994). These pro-teins are monomers sharing common structural as well as

functional characteristics (Amara and Kuhar, 1993).Four additional members of this protein superfamilyhave been isolated by homology cloning in the rat andhave in common the fact that their transported substrateshave not yet been identified. These “orphan” transportersare Rxt1 [also named NTT4 (Liu et al., 1993; El Mes-tikawy et al., 1994)], V-7-3-2 (Uhl et al., 1992), ROSIT(Yamauchi et al., 1992; Wasserman et al., 1994), andrB21a (Smith et al., 1995). It is surprising that recentevidence has been provided that Rxt1 is a vesicularmembrane-bound protein, in contrast to Na1/Cl2-depen-dent neurotransmitter transporters, which are targeted tothe plasma membrane (Masson et al., 1999b).

Anatomical studies have established that Rxt1 mRNAis essentially found in the CNS of the adult rodent andhuman, where it is exclusively distributed in neurons(Liu et al., 1993; El Mestikawy et al., 1994; Luque et al.,1996; Masson et al., 1999a,b). The highest levels of Rxt1mRNA in the rat CNS were found in the cerebral cortex,hippocampus, thalamus, and cerebellum (Purkinje andgranular cells) (El Mestikawy et al., 1994; Luque et al.,1996). The localization of the protein was shown tocorrelate mainly with the distribution pattern of the tran-script. At the cellular level, immunoreactivity wasmostly confined to neuronal processes, so that the dif-ferences between the distribution of the protein and itsmRNA observed in some regions could be ascribed tothe transport of the protein into efferent projections ordendritic arborization of the neurons expressing the tran-script (Masson et al., 1995, 1999b; Luque et al., 1996; El

Received December 23, 1998; revised manuscript received March15, 1999; accepted March 17, 1999.

Address correspondence and reprint requests to Dr. L. Kerkerian-Le Goffat Laboratoire de Neurobiologie Cellulaire et Fonctionnelle, CNRS-UPR9013, 31, chemin Joseph Aiguier, 13402 Marseille Cedex 20, France.

Abbreviations used:BDA, biotin–dextran amine; BSA, bovine se-rum albumin; DIG, digoxigenin; DPL, days postlesion; GAD, glutamicacid decarboxylase; 6-OHDA, 6-hydroxydopamine; PB, phosphatebuffer; PBS, phosphate-buffered saline; SDS, sodium dodecyl sulfate.

623

Journal of NeurochemistryLippincott Williams & Wilkins, Inc., Philadelphia© 1999 International Society for Neurochemistry

Mestikawy et al., 1997). Furthermore, in all these stud-ies, Rxt1 appeared to be expressed mostly in glu-tamatergic terminals and in some subpopulations ofGABAergic nerve endings.

Among the structures showing a marked mismatchbetween Rxt1 mRNA and protein expression is the stri-atum, which contains very low levels of transcript buthigh levels of the protein (El Mestikawy et al., 1994,1997; Masson et al., 1995; Luque et al., 1996). At theultrastructural level, striatal Rxt1 immunostaining hasbeen recently shown to be confined to axonal processesand terminals forming asymmetrical synapses, a featurecharacteristic of excitatory amino acid-containing striatalprojections (Masson et al., 1999b). Accordingly, striatalRxt1 could be associated mainly with afferents of extrin-sic origin. Indeed, the striatum receives two major pre-sumably glutamatergic pathways originating from thecerebral cortex and the thalamus (Parent and Hazrati,1995), two brain regions showing high levels of Rxt1mRNA (El Mestikawy et al., 1994; Luque et al., 1996).The substantia nigra from which the third main striatalinput pathway originates also contains neurons express-ing Rxt1 mRNA (El Mestikawy et al., 1994; Luque et al.,1996), and evidence has been provided that the nigro-striatal dopaminergic system may include an excitatoryamino acid-containing component (Hattori et al., 1991;Shiroyama et al., 1996). Whether or not Rxt1 is a markerfor these various excitatory amino acid afferents is aninteresting matter of concern, inasmuch as no specificmarker for these systems, which play a major role in thecontrol of sensorimotor function, has still been identified.For instance, it is noteworthy that among the five high-affinity excitatory amino acid membrane transportersisolated to date (Kanai and Hediger, 1992; Pines et al.,1992; Arriza et al., 1994, 1997; Fairman et al., 1995), theneuronal transporter EAAT3/EAAC1 expressed in thestriatum is not associated with glutamatergic nerve end-ings (Rothstein et al., 1994; Ginsberg et al., 1995).

The primary aim of the present study was to specifythe origin of Rxt1 immunostaining in the rat striatum. Ina first set of experiments, the effects of (a) kainic acid-induced striatal lesion, (b) corticostriatal deafferentationby large cortical thermocoagulation, (c) kainic acid-induced thalamostriatal denervation, and (d) 6-hydroxy-dopamine (6-OHDA)-induced nigrostriatal lesion on stri-atal Rxt1 contents were determined by immunocyto-chemistry and western blotting. In a second set of exper-iments, the hypothesis that the corticostriatal pathwaymay be a major source of Rxt1 in the striatum has beenfurther explored by combining tract-tracing procedureswith Rxt1 mRNA or protein detection. For this purpose,retrograde fluorogold labeling of the corticostriatal path-way was combined with Rxt1 mRNA detection by in situhybridization at the cellular level, and, conversely, an-terograde biotin–dextran amine (BDA) labeling of thecorticostriatal pathway was combined with Rxt1 immu-nogold detection at the electron microscopic level.

MATERIALS AND METHODS

Experiments were performed using adult male Wistar andSprague–Dawley rats (weighing 200–300 g; Iffa-Credo,France). The animal experimental protocols performed in thisstudy strictly conformed to the guidelines of the French Agri-culture and Forestry Ministry (decree 87-849, license no.01499) and were approved by the CNRS.

Surgical proceduresSurgery was performed on rats anesthetized with sodium

pentobarbital (50 mg/kg, i.p.) and placed in a stereotaxic ap-paratus.

Lesions.Bilateral lesions of the frontoparietal cortex wereperformed by superficial thermocoagulation, as previously de-scribed by Errami and Nieoullon (1986). In brief, the skullcovering the top of the frontoparietal cortex was removed, andthe tip of the thermocoagulation probe was applied onto themeninges covering the whole rostrocaudal extent of the fron-toparietal cortex. After completion of the thermocoagulation,the cortical surface was covered with Gelfoam, and the skinsutured. Animals were killed by decapitation on the 12th daypostlesion (DPL).

Lesions of thalamostriatal afferents or of striatal intrinsicneurons were made by stereotaxic injection of 1ml of a kainicacid solution (1 mg/ml in 25 mM phosphate buffer, pH 7) intothe right thalamus or striatum [A20.3, L 3.4, H 5.6 and A22.8, L 2, H 6 relative to bregma, respectively (Paxinos andWatson, 1986)]. For lesioning dopaminergic neurons of thesubstantia nigra (pars compacta), 2ml of a 6-OHDA solution [2mg/ml in saline containing 0.05% (wt/vol) of ascorbic acid]was slowly injected into the right substantia nigra [A25.3, L2.6, H 8.4] (Paxinos and Watson, 1986)]. Animals were al-lowed to recover for 7 days before they were killed. Anesthe-tized but unoperated animals served as controls.

Tract-tracing procedures.For anterograde tracing of thecorticostriatal projection, animals received unilateral multiple(10 sites) pressure injections of BDA (10,000 mol wt; Molec-ular Probes) as a 5% solution in saline, into the frontal andparietal cortices, as described in details elsewhere (Kachidian etal., 1998). Animals were perfused transcardially 9–10 daysafter BDA injection.

Retrograde tracing of the corticostriatal projection neuronswas performed by fluorogold injections (4% in saline; Fluoro-chrome) in the striatum as previously described (Gaspar et al.,1995). Animals were allowed to recover for 5–7 days beforeperfusion.

Biochemical measurementsHigh-affinity glutamate uptake assay (Bloc et al., 1995) and

western blotting experiments (El Mestikawy et al., 1997) wereperformed on 10 animals with cortical lesioning at 12 DPL andfive controls. The striata were rapidly dissected out at 4°C.Each striatum was immediately homogenized in an isotonicsucrose solution (0.32M) and diluted 1:20 (wt/vol) in a chilledbuffered physiological medium containing 10 mM glucose, 5mM KCl, 140 mM NaCl, 1.2 mM CaCl2, 1.2 mM MgSO4, 1.2mM NaH2PO4/Na2HPO4, and 15 mM Tris, pH 7.4. Each ho-mogenate suspension was divided in two parts. One sample wasdiluted 1:4 (vol/vol) in the same chilled buffered physiologicalmedium for immediate high-affinity glutamate uptake assay.The second sample was stored at280°C until western blotting.

For high-affinity glutamate uptake measurement, homoge-nates were centrifuged at 10,000g for 10 min. The supernatantswere discarded, and the pellets were resuspended in twice the

J. Neurochem., Vol. 73, No. 2, 1999

624 P. KACHIDIAN ET AL.

initial volume of physiological medium (final dilution, 1:160wt/vol). Uptake assays were started by adding 20-ml aliquots ofthese crude synaptosomal suspensions to 80ml of bufferedphysiological medium containingL-[3H]glutamate (specific ac-tivity, 45 Ci/mmol; Amersham, U.K.) at a final concentration of1026 M. Incubations were carried out at 25°C for 2 min. Thereaction was stopped by filtration through Millipore Immo-bilon-P membranes (pore size, 0.45mm). The filters were thencut up and placed in 100ml of 1% Triton X-100. The radio-activity was measured by liquid scintillation counting. Assayswere performed in quadruplicate for each striatum. Proteinswere quantified by the method of Lowry et al. (1951) withbovine serum albumin (BSA) as the standard. The results areexpressed as picomoles of [3H]glutamate taken up per minuteper microgram of protein.

For western blotting experiments, aliquots (5mg of protein)of homogenates were electrophoresed in a 12% polyacryl-amide/bisacrylamide slab gel in the presence of sodium dode-cyl sulfate (SDS). Proteins were then electrically transferredfrom the gel to a nitrocellulose sheet (Bio-Rad, France), and theresulting blot was subsequently processed at room temperature.It was first incubated overnight with purified rabbit anti-Rxt1antibodies [1:400 dilution (Masson et al., 1999)] in phosphate-buffered saline (PBS; pH 7.4) containing Tween 20 (0.5%vol/vol) and powdered milk (1% wt/vol). After extensive wash-ing, the blot was incubated with peroxidase-linked goat anti-rabbit antibodies (Amersham) at a 1:1,000 dilution for 1 h.Immunoreactive bands were visualized by chemiluminescentreaction following the manufacturer’s instructions (ECL; Am-ersham). Relative quantification of western blots was per-formed by computer analysis of gel images in an Imagestore5000 System (Ultra Violet Products). In these conditions, thewestern blot yielded an ECL signal whose intensity increasedlinearly with the amount of protein loaded on the SDS-poly-acrylamide gel for electrophoresis in the range of 0.25–20mgper lane. Proteins were quantified as above. Results are ex-pressed as relative optical density values (arbitrary unit).

Results are mean6 SEM values of the data obtained fromboth brain sides in the n animals in each experimental group.Statistical comparison was performed using two-tailed Stu-dent’s t test (unpaired). A significance level ofp , 0.05 wasrequired for rejection of the null hypothesis.

Morphological studiesImmunoautoradiography.Animals were anesthetized with

sodium pentobarbital (50 mg/kg, i.p.) and perfused for 20 min(10–20 ml/min) through the aortic arch with saline supple-mented with 1 g/L sodium nitrite. Brains were carefully dis-sected out and frozen in isopentane at230°C. Cryostat-cutsections (10mm) were thaw-mounted onto Superfrost Plusslides, and radioimmunocytochemical labeling was performedas described in detail elsewhere (Masson et al., 1995). In brief,sections were fixed with 4% paraformaldehyde for 20 min andincubated first overnight with purified rabbit anti-Rxt1 or anti-glutamic acid decarboxylase (anti-GAD; Chemicon, U.S.A.)antibodies (1:2,000 and 1:400 dilution, respectively) and sec-ond with anti-rabbit35S-IgG (1 mCi/ml; Amersham) for 2 h.Sections were dried and apposed to x-ray films (bmax; Amer-sham) for 5 days. Optical density of the immunoautoradio-grams was measured using a computerized image analysissystem (Biocom, Les Ulis, France). The linearity of the signalwas confirmed with14C standards (Amersham). For each ani-mal, the mean optical density was determined from at least foursections. Data are mean6 SEM values of the results deter-mined from three or four animals for each condition.

In situ hybridization histochemistry: (a) Probes.Rxt1 anti-sense cRNA probe was transcribed with the T7 RNA polymer-ase (using the T7 Ampliscribe kit; Tebu, France) from 50 ng ofa 1-kb XhoI/KpnI cDNA fragment, inserted in pBluescript(Stratagene) previously linearized with anXhoI cut. This por-tion of the “orphan” transporter corresponds to its N-terminal/EL4 region (El Mestikawy et al., 1994).35S-labeled antisenseprobe was generated as already described (Masson et al., 1996).The digoxigenin (DIG)-UTP antisense cRNA probe was ob-tained using the same protocol except that35S-UTP was re-placed by a DIG-UTP labeling mix (Boehringer, France).

(b) Tissue preparation.Brains were frozen in isopentane at230°C, and cryostat-cut sections (10mm thick) were thaw-mounted onto Superfrost Plus glass slides. Sections werewarmed for 10 min at 60°C, fixed with 3.5% formaldehyde inPBS for 1 h atroom temperature, rinsed twice in PBS, washedfor 10 s in distilled water, dehydrated in 50% ethanol for 30 s,and stored in 70% ethanol at 4°C.

(c) Hybridization procedure.Hybridization was performedessentially as described by Masson et al. (1996). In brief,sections were incubated overnight (60°C) with antisense35S- orDIG-cRNA (53 106 cpm or 50 ng per slide) in the presence of50% formamide. Nonhybridized probe was eliminated byRNase treatment (50mg/ml) and by extensive washes (finalstringency, 0.13 saline–sodium citrate at 60°C). Radiolabeledsections were exposed to x-ray films (bmax; Amersham) for4–15 days. For detection of the DIG-labeled probes, the sec-tions were exposed to anti-DIG-alkaline phosphatase conju-gates. Colorimetric detection of alkaline phosphate was ob-tained by incubation in 0.4 mM 5-bromo-4-chloro-3-indolylphosphate and nitroblue tetrazolium chloride (Boehringer,France).

Retrograde fluorogold labeling of the corticostriatal path-way combined with Rxt1 mRNA detection.Animals with flu-orogold injection were anesthetized with chloral hydrate (400mg/kg, i.p.) and perfused transcardially with 4% paraformal-dehyde in 0.12M phosphate buffer. The brains were postfixedin the same fixative for 1 h, cryoprotected in phosphate bufferwith 10% sucrose, and frozen in isopentane at between250and 280°C. Cryostat-cut sections (10mm thick) were pro-cessed for in situ hybridization as described above, then coatedwith Ilford K5 liquid emulsion (diluted 1:1 in distilled water),exposed for 4 days, developed, and mounted without counter-staining. Neurons retrogradely labeled with fluorogold wereidentified using an Olympus fluorescence microscope with aUGI excitation filter. Dual fluorogold–Rxt1 mRNA labelingwas analyzed by combining epifluorescence and dark-field il-lumination with varying transmitted light intensity. The fluoro-gold-labeled neurons showing a density of silver grains at leasttwofold higher than the background were considered to bedually labeled.

Anterograde BDA labeling of the corticostriatal pathwaycombined with Rxt1 immunogold detection: electron micro-scopic analysis.At 9–10 days after cortical BDA injection, ratswere transcardially perfused under chloral hydrate anesthesia(400 mg/kg, i.p.) with (a) 50 ml of cold saline containing 5,000IU of heparin and 100 mg of sodium nitrite followed by (b) 300ml of 4% paraformaldehyde and 0.1% glutaraldehyde in 0.12Mphosphate buffer (PB; pH 7.4) and then (c) 300 ml of 4%paraformaldehyde in PB. The brains were removed from theskull 1 h later and postfixed overnight in 4% paraformaldehydeat 4°C. After four or five washes in PB (10 min each), 40-mmcoronal sections were cut using a Lancer vibratome. Sectionswere washed in PB and preincubated for 30 min in PB con-

J. Neurochem., Vol. 73, No. 2, 1999

625Rxt1 IN STRIATAL AFFERENTS

taining 0.1M lysine and 0.3% hydrogen peroxide and then inPB containing 3% BSA and 1% normal goat serum for 30 min.Sections were extensively washed in 0.05M Tris-bufferedsaline (pH 7.6) and incubated for 48 h at 4°C with purifiedanti-Rxt1 antibodies (1:500) in the same buffer. After severalrinses, sections were first processed for BDA detection throughsuccessive incubations with (a) avidin–biotin–peroxidase com-plex (ABC; Vector) solution for 90 min, (b) diaminobenzidinedevelopment solution without hydrogen peroxide for 20 min,and then (c) diaminobenzidine solution with hydrogen peroxidefor 6–7 min. All incubations and rinses after each step werecarried out in 0.05M Tris-buffered saline (pH 7.6). Rxt1immunogold labeling was then performed as follows: Sectionswere sequentially incubated in (a) 0.3% gelatin and 1% BSAfor 5 min, (b) goat anti-rabbit immunoglobulins bound to 1-nmcolloidal gold (Amersham) diluted at 1:50 in 0.3% gelatin and1% BSA for 2 h, (c) 2% glutaraldehyde for 10 min, and (d)IntenSE M (Amersham) for silver intensification for 20–25min. All incubations and washes were carried out in PB, exceptthose before and after silver intensification, which were per-formed in citrate buffer (pH 7.4).

After dual labeling, sections were postfixed in 2% osmiumtetroxide, dehydrated in graded ethanol solutions, and flat em-bedded in Epon between two siliconized glass slides. For eachrat, two blocks of the dorsomedial area of the striatum showingthe optimal labeling were selected under the light microscope,cut out, and glued on the top of prepolymerized Epon blocks.Serial ultrathin (80–100-nm-thick) sections were cut out fromthe most superficial part of each block using an LKB ultrami-

crotome. Sections were collected on grids and counterstainedwith uranyl acetate and lead citrate to be examined with a JEOL1200 electron microscope. Observations were performed at amagnification of320,000 and were mainly focused on BDA-labeled axonal elements. Axonal profiles were segregated intosynaptic terminals or fibers depending on whether or not bothpre- and postsynaptic membranes as well as the synaptic cleftwere visible. For each block, BDA-labeled axonal elementswere randomly sampled from the most superficial portions offour or five ultrathin sections distant each other from at least3,000–5,000 nm (to minimize recounting of the same immu-nolabeled element) until 100 BDA-labeled synaptic terminalswere identified. A given BDA-labeled structure was consideredas Rxt1-immunopositive when the density of superimposedsilver gold particles was at least twice the background observedin the surrounding neuropil. Results representing the proportionof BDA-labeled synaptic terminals or fibers exhibiting Rxt1immunoreactivity are expressed as mean6 SEM values of thedata obtained in the blocks.

RESULTS

Rxt1 expression in striatumAs shown in Fig. 1A, C, and D, high levels of Rxt1

transcript are found in cortex, hippocampus, and thala-mus. In contrast, striatum contains only low levels ofRxt1 mRNA (Fig. 1A). However, a higher-resolution insitu hybridization technique, i.e., DIG-labeled cRNA

FIG. 1. Autoradiograms showing the distribution of Rxt1 mRNA (A) and protein (B) in coronal rat brain sections at the level of the striatum(St). C: In situ hybridization, using DIG-labeled cRNA probe and photomicroscopic detection of Rxt1 mRNA on a parasagittal section.D and E: Higher magnifications of the cortex and St, respectively. cc, corpus callosum; Cx, cerebral cortex; Hi, hippocampus; Sept,septum; Thal, thalamus. Bars 5 2 mm for C and 250 mm for D and E.

J. Neurochem., Vol. 73, No. 2, 1999

626 P. KACHIDIAN ET AL.

probe and optic microscope detection, allowed the visu-alization of a large number of neurons expressing theRxt1 transcript in the striatum (Fig. 1E). This labeling isdifficult to detect because the amount of Rxt1 mRNAwithin these neurons is much less than in cortical neu-rons (see Fig. 1D and E).

In contrast with the low abundance of Rxt1 mRNA,very high amounts of the protein are present in thestriatum (see Fig. 1B; Masson et al., 1995; Luque et al.,1996; El Mestikawy et al., 1997). This mismatch be-tween the respective concentrations of Rxt1 mRNA andprotein in the striatum strongly suggests that most of the“orphan” transporter content in this area has an exoge-nous origin. To investigate this hypothesis further, weanalyzed the striatal Rxt1 contents after lesion of intrin-sic striatal neurons or of each of the main striatal inputs.

Striatal lesion.Unilateral intrastriatal kainic acid in-jection resulted in a dramatic loss of striatal intrinsic

neurons, as shown by the marked decrease of GADimmunoautoradiographic labeling in the injected stria-tum (Fig. 2B). In contrast, no significant change in Rxt1immunolabeling was concomitantly detected in the le-sioned striatum (Fig. 2A). Quantitative analysis of theimmunoautoradiograms showed a 42% (p , 0.001) de-crease in the GAD content but no change in the Rxt1labeling in the injected striatum compared with the con-tralateral one (Table 1).

Kainic acid-induced thalamostriatal deafferentation.Animals with unilateral intrathalamic kainic acid injec-tion showed a large lesion of the thalamus, including theintralaminar nuclei (paracentral, parafascicular, and cen-tral lateral nuclei), which are known to project ipsilater-ally to the striatum (Berendse and Groenewegen, 1990).Whereas this lesion was associated with the completedisappearance of Rxt1 mRNA in the injected thalamus(data not shown), only a slight (29%) nonsignificantdecrease in Rxt1 immunostaining was detected in theipsilateral versus the contralateral striatum (Table 1).

6-OHDA-induced nigrostriatal deafferentation.Ani-mals with unilateral intranigral 6-OHDA injectionshowed a large decrease in immunostaining by anti-tyrosine hydroxylase antibodies in the ipsilateral striatum(285%; data not shown), indicating a massive degener-ation of nigrostriatal dopaminergic projections. Theseanimals did not exhibit any significant change in Rxt1immunostaining in the deafferented versus the contralat-eral striatum (Table 1).

Thermocoagulation-induced corticostriatal deafferen-tation. The animals with bilateral superficial thermoco-agulation of the frontoparietal cortex showed a large andsymmetrical lesion of this area with no damage to thecorpus callosum. As shown in Table 2, [3H]glutamateuptake in striatal homogenates was decreased by 42% inthe operated animals versus controls, indicating a signif-icant corticostriatal glutamatergic denervation. StriatalRxt1 contents of aliquots of the same homogenates sol-ubilized in SDS were measured by western blotting andsemiquantitative analysis. A large diffuse (probably gly-cosylated) band with a molecular mass of around 100

FIG. 2. Immunoautoradiograms showing the effects of a unilat-eral intrastriatal kainic acid injection on Rxt1 (A) and GAD (B)immunostaining in coronal rat brain sections at the striatal level.Solid arrowheads indicate the injected side.

TABLE 1. Effects of various lesions on Rxt1 and/or GAD content in striatum

Type of lesion

Optical density (arbitrary units)

Rxt1 (%) GAD (%)

Control Lesioned Control Lesioned

Kainic acid/striatum 3.256 0.21 (100) 3.006 0.15 (92)a 2.006 0.04 (100) 1.176 0.11 (58)b

Kainic acid/thalamus 1.576 0.04 (100) 1.436 0.04 (91)a ND ND6-OHDA/substantia nigra 0.896 0.03 (100) 0.846 0.02 (94)a ND ND

Each lesion was performed unilaterally, on the right side (see Materials and Methods). The amounts of Rxt1 or GADwere estimated by measuring the optical density of immunoautoradiograms. In all cases, control experiments with14C-standards performed in parallel confirmed that optical density values were in the appropriate range for which a linearrelationship existed between these values and corresponding radioactivities. Data are mean6 SEM values of resultsobtained in three or four animals (percentage of control); from 10 to 20 adjacent sections were analyzed for each animal.ND, not determined.

a Nonsignificant,b p , 0.01 compared with the control (contralateral) striatum by Student’st test.

J. Neurochem., Vol. 73, No. 2, 1999

627Rxt1 IN STRIATAL AFFERENTS

kDa as well as a lighter (unglycosylated) band could bedetected on western blots (Fig. 3). We have previouslydemonstrated that the 100-kDa diffuse band is in factRxt1 (Masson et al., 1995). In the example illustrated inFig. 3, a clear-cut, but limited, decrease in the immuno-labeling of this band is observed in both the left and theright striata of rats with bilateral thermocoagulation ofthe frontoparietal cortex, as compared with intact controlanimals. Quantitative analysis of western blots showedthat thermocoagulation-induced corticostriatal deafferen-tation was associated with a significant 21% decrease inRxt1 immunoreactivity in both striata versus controlvalues (Table 2).

Rxt1 expression in cerebral cortexHigh levels of the Rxt1 protein and its encoding

mRNA were found in the cerebral cortex (Figs. 1 and4A). In situ hybridization of Rxt1 mRNA using a DIG-labeled antisense cRNA probe showed that the “orphan”transporter is synthesized in all layers of the neocorticalareas, except in molecular layer I (Fig. 4B). When com-pared with histological staining (Fig. 4A), a large pro-portion of cortical neurons appeared Rxt1-positive. It isinteresting that Rxt1 expression was particularly high inthe large pyramidal neurons of layer V (Fig. 4C and D),which are known to project into various subcorticalstructures, including the striatum.

Rxt1 expression in corticostriatal pathwayThe hypothesis that Rxt1 mRNA may be expressed in

corticostriatal projection neurons has been further ex-plored in the experiments illustrated in Fig. 4C and D. Inthese experiments, efferent corticostriatal neurons wereretrogradely labeled by intrastriatal injections of fluoro-gold and then processed for emulsion-coated in situhybridization using the35S-labeled antisense cRNAprobe. Numerous large neurons in layer V were retro-gradely labeled by the fluorescent dye; many of theseneurons also showed Rxt1 mRNA labeling.

To ascertain the presence of the Rxt1 protein in striatalterminals of cortical origin, anterograde BDA labeling ofthe corticostriatal pathway was combined with Rxt1 im-munogold detection at the electron microscope level. Theultrastructural features of the corticostriatal BDA-labeled

synaptic terminals observed here (Fig. 5) were similar tothose previously described (Kachidian et al., 1998). Thesynaptic contacts were of the asymmetrical type, and thepostsynaptic elements were mostly dendritic spines (Fig.5). Dense Rxt1 immunolabeling was found in numerousanterogradely labeled axonal profiles and synaptic termi-nals (Fig. 5), but numerous anterogradely labeled axonalelements without Rxt1 immunolabeling were also ob-served in the striatal neuropil (Fig. 5D and F). Semiquan-titative analysis showed that 34.26 2.5% of the BDA-labeled synaptic terminals also exhibited Rxt1 immuno-gold labeling. In addition to the 600 BDA-labeledterminals examined, 306 BDA-positive fibers were alsoseen. It is interesting that only 15.56 3.5% of theseelements showed Rxt1 immunolabeling.

Numerous Rxt1-positive axonal profiles without BDAlabeling were also observed in the striatal neuropil at theelectron microscope level (Fig. 5G and H). These pro-files were mostly synaptic terminals forming asymmet-rical contacts (Fig. 5G); all the postsynaptic elementsobserved were unlabeled and were, for the most part,dendritic spines.

DISCUSSION

The striatum has been shown to express very lowlevels of Rxt1 transcript but high levels of the protein (ElMestikawy et al., 1994; Masson et al., 1995; Luque et al.,1996). This mismatch suggests that most of the striatalRxt1 contents may be of extrinsic origin. The presentstudy confirms this proposal by showing that (a) excita-tory amino acid-induced striatal lesions do not signifi-cantly affect striatal Rxt1 immunostaining and (b) Rxt1-positive terminals in the striatum form asymmetricalsynapses, a feature characteristic of excitatory aminoacid-containing striatal afferents. The cerebral cortex andthe thalamus (including the intralaminar nuclei), the twomain sources of excitatory amino acid afferents to thestriatum (see Parent and Hazrati, 1995), express highlevels of Rxt1 transcript (El Mestikawy et al., 1994;Luque et al., 1996) and could thus be considered as themajor potential sources of striatal Rxt1 contents.

The present dual labeling experiments allowed us toconfirm that Rxt1 transcripts are located in the vastmajority of corticostriatal projection neurons and to dem-onstrate the presence of the Rxt1 protein in striatal ter-minals of cortical origin; accordingly, corticostriatal

FIG. 3. Western blot analysis of Rxt1 contents in the striatum ofa rat subjected to bilateral cortical thermocoagulation (lesioned)as compared with an intact (control) rat. L, left striatum; R, rightstriatum. Similar experiments repeated in 10 lesioned ratsyielded the quantitative data of Table 2.

TABLE 2. Effects of bilateral cortical lesion by superficialthermocoagulation on [3H]glutamate uptake and Rxt1 levels

in striatal homogenates

[3H]Glutamate uptake(pmol/min/mg of protein)

Rxt1 western blot(arbitrary units)

Control(n 5 5) 0.3336 0.027 1.1076 0.032

Cortical lesion(n 5 10) 0.1926 0.007 (57.6%)a 0.8796 0.045 (79.4%)a

The amounts of Rxt1 were estimated by measuring optical densitieson western blots. Data are mean6 SEM values of results measuredfrom both striata in n animals per condition.

a p , 0.01 compared with the respective control values by two-tailed(unpaired) Student’st test.

J. Neurochem., Vol. 73, No. 2, 1999

628 P. KACHIDIAN ET AL.

deafferentation was found here to result in a significantreduction in striatal Rxt1 contents. We further show thatRxt1 immunolabeling occurs more frequently in synapticterminals than within axonal fibers of the corticostriatalsystem. Together with the previous observation that onlyrare somatal immunostaining occurs in cortical neurons(Luque et al., 1996; El Mestikawy et al., 1997), thesedata indicate that Rxt1 is located almost exclusively atthe terminal level in the corticostriatal system. This pref-erential localization of the protein is consistent with itspresumed function as a vesicular transporter in excitatoryterminals (Masson et al., 1999b). It is interesting that wealso showed that only about one-third of the antero-gradely labeled corticostriatal synaptic terminals exhib-ited Rxt1 immunoreactivity. Because analysis was per-formed at the most superficial portion of sections con-taining both BDA and Rxt1 reaction products andconcerned a relatively large number of neuronal pro-cesses, dual labeling limitations, e.g., differential pene-tration of reagents in the tissue or use of a particulatemarker, should have been largely overcome (see Kachid-ian and Pickel, 1993). Hence, although the proportion ofdually labeled terminals is presumably underestimated, itcan be inferred that Rxt1 expression is probably confinedto a portion of the corticostriatal projections. It is inter-esting that although virtually all corticostriatal neuronshave been shown to contain excitatory amino acids,

recent evidence has been provided that these neuronsmay be segregated in at least two subpopulations on thebasis of their respective glutamate or aspartate contents(Bellomo et al., 1998). Thus, it is tempting to speculatethat the Rxt1 protein may be present in synaptic termi-nals belonging to only one of these two neuronal sub-populations.

Regarding the possible contribution of the thalamostriatalinput to striatal Rxt1 contents, thalamostriatal deafferenta-tion was shown here to have no significant effect on striatalRxt1 immunostaining. On the other hand, our ultrastructuralobservations indicated that the recipient structures for Rxt1-positive synaptic terminals are almost exclusively dendriticspines. In contrast, thalamostriatal synaptic terminals areknown to contact mainly dendritic shafts (Dube´ et al., 1988;for review, see Parent and Hazrati, 1995). Consequently, itcan be assumed that most of the thalamostriatal projectionscontain either none or only very low levels of the Rxt1protein.

Altogether, our data support the idea that Rxt1 maynot be a general marker of the presynaptic compartmentof excitatory amino acid synapses but may identify aspecific subclass of excitatory terminals. This furtherreveals a heterogeneity among excitatory amino acid-containing striatal afferents, which may be related to therespective levels of synaptic activity and/or to the iden-tity of the associated neurotransmitter (glutamate vs.

FIG. 4. Rxt1 mRNA expression in cortical neurons. A and B: Laminar distribution of Rxt1 mRNA-expressing neurons in the cerebralcortex, as shown by comparison of (A) cresyl violet staining and (B) Rxt1 mRNA labeling using a DIG-labeled probe. Cortical layers areindicated with roman numerals. wm, white matter. C and D: Retrograde labeling of neurons in layer V of the cortex after injection offluorogold in the ipsilateral striatum combined with Rxt1 mRNA detection using a 35S-cRNA probe. The arrowhead shows a neuronlabeled by fluorogold only; arrows point at dually labeled neurons.

J. Neurochem., Vol. 73, No. 2, 1999

629Rxt1 IN STRIATAL AFFERENTS

FIG. 5. Electron micrographs ofdually labeled sections showingBDA peroxidase-labeled cortico-striatal afferents and Rxt1 immuno-gold-labeled elements. A–C: Ex-amples of BDA/Rxt1 dually labeledaxon terminals. D–F: Rxt1 immuno-labeling occurs in a contingent ofthe corticostriatal afferents. An-terogradely labeled corticostriatalaxonal elements with or withoutRxt1 immunolabeling are observedin the same field. G and H: Rxt1immunolabeling is not confined toidentified corticostriatal afferents;Rxt1-immunopositive axonal ele-ments with or without anterogradeBDA peroxidase labeling are ob-served in close vicinity. Note thatall the anterogradely labeled and/orRxt1-positive synaptic terminalsform contact of the asymmetricaltype with unlabeled dendritic ele-ments (d), which are mostly spines(s). Bars 5 500 nm.

J. Neurochem., Vol. 73, No. 2, 1999

630 P. KACHIDIAN ET AL.

aspartate?). Moreover, because high levels of Rxt1 tran-scripts are present both in the thalamic nuclei from whichthe thalamostriatal input originates and in virtually all thecorticostriatal neurons (El Mestikawy et al., 1994; Luqueet al., 1996), it may be postulated that the intraneuronaldistribution of the protein, or even its expression, maydepend on posttranscriptional regulatory mechanismsthat differ from one neuronal type of another.

Whether or not the cerebral cortex is the only majorextrinsic source of Rxt1 immunoreactivity in the striatumis a matter of concern. Numerous Rxt1-positive terminalsdevoid of BDA anterograde labeling from the homolat-eral frontoparietal cortices were observed here in thestriatal neuropil. These terminals may well belong toprojections from the other ipsilateral cortical areas orfrom the contralateral cortex (McGeorge and Faull,1989). However, we found that a high degree of bilateralcorticostriatal deafferentation reduced striatal Rxt1 con-tents by only;20%, so that other sources of striatal Rxt1have to be considered. The main candidates are thenondopaminergic, presumably glutamatergic, componentof the nigrostriatal pathway and the raphe–striatal pro-jections; indeed, these systems, like those expressingRxt1, make asymmetrical synapses in the striatum (Sog-homonian et al., 1989; Hattori et al., 1991). In addition,Rxt1 transcripts are present in the substantia nigra andthe raphe (Luque et al., 1996). We showed here that6-OHDA-induced lesion of the nigrostriatal pathway didnot significantly affect striatal Rxt1 immunostaining,suggesting that striatal Rxt1 expression may not be re-lated to this afferent system. However, whether or notthese nondopaminergic neurons in the substantia nigra(pars compacta) are sensitive to 6-OHDA lesioning re-mains an open question. Moreover, several lines of evi-dence indicate that this lesion results in a reactive in-crease of corticostriatal glutamatergic transmission (Lin-defors and Ungerstedt, 1990; Hajji et al., 1996), whichmay mask the actual contribution of the nigral afferentsto striatal Rxt1 expression.

As shown here, with a DIG-labeled antisense cRNAprobe, the striatum contains low but detectable levels ofRxt1 transcripts (see also El Mestikawy et al., 1994; Luqueet al., 1996), indicating a local synthesis of this orphantransporter. Previous studies have suggested that, in addi-tion to excitatory amino acid nerve endings, some popula-tions of GABAergic neurons express Rxt1 (Masson et al.,1995, 1999a; Luque et al., 1996; El Mestikawy et al., 1997).It is interesting that the great majority of striatal neurons,including the efferent spiny neuronal subsets and a subclassof aspiny interneurons, are GABAergic (see Parent andHazrati, 1995). Furthermore, the recurrent axons of striataloutput neurons and the terminals of interneurons are usuallydescribed as forming symmetrical synaptic differentiations(see Parent and Hazrati, 1995). As Rxt1 immunolabelingwas never observed in such synaptic terminals or in thesomatodendritic compartment of striatal neurons (Massonet al., 1999b), it can be assumed that Rxt1 immunoreactivityis not associated with GABAergic neurons in the striatum.In agreement with this inference, the present kainic acid

striatal lesion, which markedly decreased GAD immuno-staining, did not significantly affect Rxt1 immunolabeling.On the other hand, recent evidence has been provided forthe presence of a few aspartate-containing interneurons inthe striatum (Pettersson et al., 1996). Given their excitatoryamino acid phenotype, these interneurons are likely to formasymmetrical synapses and as such may represent an intrin-sic source of Rxt1-immunopositive synaptic terminals. Thishypothesis cannot be ruled out only on the basis of ourstriatal lesion experiments, because of the paucity of thisinterneuronal population (Pettersson et al., 1996). More-over, the contribution of intrinsic neurons to the striatalcontent of Rxt1 might well vary after the different lesionsperformed in the present study. For instance, some up-regulation of Rxt1 expression by striatal cells might havepartly compensated for the loss of the orphan transporterafter thermocoagulation of the frontoparietal cortex, therebyexplaining the lower decrease in Rxt1 contents than in[3H]glutamate uptake in the striatum of lesioned rats. Thishypothesis is currently under investigation.

Previous studies on the distributions of Rxt1 and itssynthesis sites have indicated that this protein couldmainly be associated presynaptically with excitatorypathways and postsynaptically with neurons in specificglutamate-innervated areas. Our data not only confirmedthat Rxt1 is presynaptically associated with excitatoryamino acid afferents but also showed that it segregates asubclass of these afferents. At the postsynaptic level, theRxt1 labeling was never found associated with the recip-ient structures of excitatory synaptic terminals (eitherRxt1-positive or not). Therefore, Rxt1 is very probablynot a postsynaptic marker of excitatory synapses. Addi-tional tract-tracing experiments and studies of the sub-cellular distribution of Rxt1 in the striatum after selectivedeafferentation conditions are in progress to investigatefurther the localization and possible physiological in-volvements of Rxt1 within the striatal network.

Acknowledgment: This research was supported by grantsfrom INSERM and CNRS. We are grateful to Claude Sais forher excellent secretarial assistance. J.M. was the recipient of aMENRT fellowship during the performance of this work.

REFERENCES

Amara S. G. (1992) Neurotransmitter transporters: a tale of two fam-ilies. Nature360,420–421.

Amara S. G. and Kuhar M. J. (1993) Neurotransmitter transporters:recent progress.Annu. Rev. Neurosci.16, 63–93.

Arriza J. L., Fairman W. A., Wadiche J. I., Murdoch G. H., KavanaughM. P., and Amara S. G. (1994) Functional comparisons of threeglutamate transporter subtypes cloned from human motor cortex.J. Neurosci.14, 5559–5569.

Arriza J. L., Eliasof S., Kavanaugh M. P., and Amara S. G. (1997)Excitatory amino acid transporter 5, a retinal glutamate transportercoupled to a chloride conductance.Proc. Natl. Acad. Sci. USA94,4155–4160.

Bellomo M., Giuffrida R., Palmeri A., and Sapienza S. (1998) Excita-tory amino acids as neurotransmitters of corticostriatal projec-tions: immunocytochemical evidence in the rat.Arch. Ital. Biol.136,215–223.

J. Neurochem., Vol. 73, No. 2, 1999

631Rxt1 IN STRIATAL AFFERENTS

Berendse H. W. and Groenewegen H. J. (1990) Organization of thethalamostriatal projections in the rat, with special emphasis on theventral striatum.J. Comp. Neurol.299,187–228.

Bloc A., Samuel D., Forni C., Dusticier N., and Kerkerian-Le Goff L.(1995) Effects of ionotropic excitatory amino acid receptor antag-onists on glutamate transport and transport-mediated changes inextracellular excitatory amino acids in the rat striatum.J. Neuro-chem.64, 1598–1604.

DubeL., Smith A. D., and Bolam J. P. (1988) Identification of synapticterminals of thalamic or cortical origin in contact with distinctmedium size spiny neurons in the rat neostriatum.J. Comp.Neurol.267,455–471.

El Mestikawy S., Giros B., Pohl M., Hamon M., Kingsmore S. F.,Seldin M. F., and Caron M. G. (1994) Characterization of anatypical member of the Na1/Cl2-dependent transporter family:chromosomal localization and distribution in GABAergic andglutamatergic neurons in the rat brain.J. Neurochem.62, 445–455.

El Mestikawy S., Werhle´ R., Masson J., Lombard M., Hamon M., andSotelo C. (1997) Distribution pattern and ultrastructural localiza-tion of Rxt1, an orphan Na1/Cl2-dependent transporter, in thecentral nervous system of rats and mice.Neuroscience77, 319–333.

Errami M. and Nieoullon A. (1986) Development of a micromethod tostudy the Na1-independentL-3H-glutamic acid binding to ratstriatal membranes. II. Effects of selective striatal lesions anddeafferentations.Brain Res.366,178–186.

Fairman W. A., Vanderberg R. J., Arriza J. L., Kavanaugh M. P., andAmara S. G. (1995) An excitatory amino-acid transporter withproperties of a ligand-gated chloride channel.Nature 375, 599–603.

Gaspar P., Bloch B., and Le Moine C. (1995) D1 and D2 receptor geneexpression in the rat frontal cortex: cellular localization in differ-ent classes of efferent neurons.Eur. J. Neurosci.7, 1050–1063.

Ginsberg S. D., Martin L. J., and Rothstein J. D. (1995) Regionaldeafferentation down-regulates subtypes of glutamate transporterproteins.J. Neurochem.65, 2800–2803.

Guimbal C. and Kilimann M. W. (1993) A Na1-dependent creatininetransporter in rabbit brain, muscle, heart and kidney: cDNA clon-ing and functional expression.J. Biol. Chem.268,8418–8421.

Hajji M. D., Salin P., and Kerkerian-Le Goff L. (1996) Chronicdizocilpine maleate (MK-801) treatment suppresses the effects ofnigrostriatal dopamine deafferentation on enkephalin but not onsubstance P expression in the rat striatum.Eur. J. Neurosci.8,917–926.

Hattori T., Takada M., Morizumi T., and Van der Kooy D. (1991)Single dopaminergic nigrostriatal neurons form two chemicallydistinct synaptic types: possible transmitter segregation withinneurons.J. Comp. Neurol.329,337–353.

Kachidian P. and Pickel V. M. (1993) Localization of tyrosine hydrox-ylase in neuronal targets and efferents of the area postrema in thenucleus tractus solitarii of the rat.J. Comp. Neurol.329,337–353.

Kachidian P., Vuillet J., Bartoli M., Castets F., Nieoullon A., andKerkerian-Le Goff L. (1998) Relationships between striatin-containing neurons and cortical or thalamic afferent fibres in therat striatum. An ultrastructural study by dual labelling.Neuro-science85, 111–122.

Kanai Y. and Hediger M. A. (1992) Primary structure and functionalcharacterization of a high-affinity glutamate transporter.Nature360,467–471.

Lindefors N. and Ungerstedt U. (1990) Bilateral regulation of gluta-mate tissue and extracellular levels in caudate-putamen by mid-brain dopamine neurons.Neurosci. Lett.115,248–252.

Liu Q. L., Mandiyan S., Lopez-Corcuera B., Nelson A., and Nelson N.(1993) A rat brain cDNA encoding the neurotransmitter trans-porter with an unusual structure.FEBS Lett.315,114–118.

Lowry O. H., Rosebrough N. J., Farr A. L., and Randall R. J. (1951)Protein measurement with the Folin phenol reagent.J. Biol. Chem.193,265–275.

Luque J. M., Jursky F., Nelson N., and Richards J. G. (1996) Distri-bution and sites of synthesis of NTT4, an orphan member of the

Na1/Cl2-dependent neurotransmitter transporter family in the ratCNS.Eur. J. Neurosci.8, 127–137.

Masson J., Langlois X., Lanfumey L., Ge´rard C., Aıdouni Z., Giros B.,Hamon H., and El Mestikawy S. (1995) Immunolabeling of theNa1/Cl2-dependent “orphan” transporter Rxt1 in the rat centralnervous system.J. Neurosci. Res.42, 423–432.

Masson J., Pohl M., Aı¨douni Z., Giros B., Hamon M., and El Mes-tikawy S. (1996) The two orphan Na1/Cl2-dependent transportersRxt1 and V-7-3-2 have an overlapping expression pattern in therat central nervous system.Receptors Channels4, 227–242.

Masson J., Cervera P., Coˆte S., Morissette J., Aı¨douni Z., Giros B.,Hamon M., Falardeau P., and El Mestikawy S. (1999a) Charac-terization and distribution of Hxt1, a Na1/Cl2-dependent orphantransporter, in the human brain.J. Neurosci. Res.(in press).

Masson J., Riad M., Chaudry F., Darmon M., Aı¨douni Z., Conrath M.,Giros B., Hamon M., Storm-Mathisen J., Descarries L., and ElMestikawy S. (1999b) Unexpected localization of the Na1/Cl2-dependent orphan transporter, Rxt1, on synaptic vesicles in the ratcentral nervous system.Eur. J. Neurosci.(in press).

McGeorge A. J. and Faull R. L. M. (1989) The organization of theprojection from the cerebral cortex to the striatum in the rat.Neuroscience29, 503–537.

Nash S. R., Giros B., Kingsmore S. F., Rochelle J. M., Suter S. T.,Gregor P., Seldin M. F., and Caron M. G. (1994) Cloning, phar-macological characterization, and genomic localization of thehuman creatinine transporter.Receptors Channels2, 165–174.

Parent A. and Hazrati L. N. (1995) Functional anatomy of the basalganglia. I. The cortico-basal ganglia-thalamo-cortical loop.BrainRes. Rev.20, 91–127.

Paxinos G. and Watson C. (1986)The Rat Brain in Stereotaxic Coor-dinates.Academic Press, London.

Pettersson E., Herrera-Marschitz M., Rodriguez-Puertas R., Xu Z. Q.,You Z. B., Hughes J., Elde R. P., Ungerstedt U., and Ho¨kfelt T.(1996) Evidence for aspartate-immunoreactive neurons in theneostriatum of the rat: modulation by the mesencephalic dopaminepathway via D1-subtype of receptor.Neuroscience74, 51–66.

Pines G., Danbolt N. C., Bjoras M., Zhang Y., Bendahan A., Eide L.,Koepsell H., Storm-Mathisen J., Seeberg J. S., and Kanner B. I.(1992) Cloning and expression of a ratL-glutamate transporter.Nature360,464–467.

Rothstein J. D., Martin L., Levey A. I., Dykes-Hoberg M., Jin L., WuD., Nash N., and Kuncl R. W. (1994) Localization of neuronal andglial glutamate transporters.Neuron13, 713–725.

Shiroyama T., Richards C. D., and Kitai S. K. (1996) Dopamine andglutamate co-localized in substantia nigra neurons: immunocyto-chemical evidence.Soc. Neurosci. Abstr.22, 892.

Smith K. E., Fried S. G., Durkin M. M., Gustafson E. L., Borden L. A.,Branchek T. A., and Weinshank R. L. (1995) Molecular cloning ofan orphan transporter, a new member of the neurotransmittertransporter family.FEBS Lett.357,86–92.

Soghomonian J. J., Descarries L., and Watkins K. C. (1989) Serotonininnervation in adult rat neostriatum. II. Ultrastructural features: aradioautographic and immunocytochemical study.Brain Res.481,67–86.

Uhl G. R. and Hartig P. R. (1992) Transporter explosion: update onuptake.Trends Pharmacol. Sci.13, 421–425.

Uhl G. R., Kitayama S., Gregor P., Nanthakumar E., Persico A., andShimada S. (1992) Neurotransmitter transporter family cDNAs inrat midbrain library: “orphan transporter” suggests sizable struc-tural variations.Mol. Brain Res.16, 343–359.

Wasserman J. C., Delpire E., Tonidandel W., Kojima R., and GullansR. S. (1994) Molecular characterization of ROSIT, a renal osmoticstress-induced Na1/Cl2-organic solute cotransporter.Am. J.Physiol.267,F688–F694.

Worral D. M. and Williams D. C. (1994) Sodium ion-dependenttransporters for neurotransmitters: a review of recent develop-ments.Biochem. J.297,425–436.

Yamauchi A., Uchida S., Kwon H. M., Preston A., Robey R. B.,Garcia-Perez A., Burg M. B., and Handler J. S. (1992) Cloning ofa Na1- and Cl2-dependent transporter that is regulated by hyper-tonicity. J. Biol. Chem.267,649–652.

J. Neurochem., Vol. 73, No. 2, 1999

632 P. KACHIDIAN ET AL.