Contribution of Kv1.2 Voltage-gated Potassium Channel to D2 Autoreceptor Regulation of Axonal Dopamine Overflow*

Impairments in axonal dopamine release are associated with neurological disorders such as schizophrenia and attention deficit hyperactivity disorder and pathophysiological conditions promoting drug abuse and obesity. The D2 dopamine autoreceptor (D2-AR) exerts tight regulatory control of axonal dopamine (DA) release through a mechanism suggested to involve K+ channels. To evaluate the contribution of Kv1 voltage-gated potassium channels of the Shaker gene family to the regulation of axonal DA release by the D2-AR, the present study employed expression analyses, real time measurements of striatal DA overflow, K+ current measurements and immunoprecipitation assays. Kv1.1, -1.2, -1.3, and -1.6 mRNA and protein were detected in midbrain DA neurons purified by fluorescence-activated cell sorting and in primary DA neuron cultures. In addition, Kv1.1, -1.2, and -1.6 were localized to DA axonal processes in the dorsal striatum. By means of fast scan cyclic voltammetry in striatal slice preparations, we found that the inhibition of stimulation-evoked DA overflow by a D2 agonist was attenuated by Kv1.1, -1.2, and -1.6 toxin blockers. A particular role for the Kv1.2 subunit in the process whereby axonal D2-AR inhibits DA overflow was established with the use of a selective Kv1.2 blocker and Kv1.2 knock-out mice. Moreover, we demonstrate the ability of D2-AR activation to increase Kv1.2 currents in co-transfected cells and its reliance on Gβγ subunit signaling along with the physical coupling of D2-AR and Kv1.2-containing channels in striatal tissue. These findings underline the contribution of Kv1.2 in the regulation of nigrostriatal DA release by the D2-AR and thereby offer a novel mechanism by which DA release is regulated.

Dopamine (DA) 4 neurons originating in the midbrain ventral tegmental area and substantia nigra and projecting to limbic and cortical sites are important for motor control, motivation, and cognition, and impairment in their function contributes to neurological and behavioral disorders. Diminished striatal DA concentrations are linked to drug abuse, overeating, and obesity (1)(2)(3), whereas the pattern of DA neurotransmission is critical for encoding the predictability and salience of rewarding stimuli (4 -6). The dynamics of DA release in mesolimbic and nigrostriatal pathways are regulated in part by D2-autoreceptors (D2-AR). Located on somatodendritic and axonal compartments of DA neurons, the D2-AR provides negative feedback by rapidly inhibiting DA release (7). At the somatodendritric level, D2-AR opens type-2 G-protein gated inward-rectifying K ϩ channels (GIRK-2) to reduce DA neuron firing (8,9), however, little is known about the mechanisms linking axonal D2-AR activation to reduced DA release.
The D2-AR belongs to the D2-like family of G-protein coupled receptors that inhibit adenylate cyclase via G i/o . The short splice variant of the D2 receptor (D2S), acting as the presynaptic D2-AR, is both molecularly and functionally distinct from its postsynaptic, long isoform (D2L) (10 -12). It has been proposed that the D2-AR inhibits DA release by increasing potassium conductance (13,14). The somatodendritic D2-AR directly couples to G-protein gated inward-rectifying K ϩ channels via G-protein ␤␥ subunits (15). In contrast, the axonal D2-AR appears to diminish DA release via the opening of unidentified voltage-gated K ϩ channels (14,16).
Voltage-dependent K ϩ channels play a fundamental role in the generation of action potentials and in regulating neuronal excitability. In addition, Kv1-family members have been shown to be present on axon terminals and to regulate neurotransmitter release (16 -20). Kv1 channels of the Shaker gene family consist of eight members of which Kv1.1 through 1.6 ␣ subunits are expressed in the mammalian brain (19,21,22). The Kv1 channel pore is composed of four ␣ subunits that can co-assem- ble in different combinations to form an array of heteromultimeric channels (23,24). The present work identifies the contribution of Kv1.1, -1.2, and -1.6 subunit-containing channels in the dorsolateral striatum by means of mRNA and protein expression analyses as well as real time measurements of axonal DA overflow using cyclic voltammetry during perfusions of novel Kv1 toxin blockers. We establish a particular role for Kv1.2 in the process whereby axonal D2-AR inhibits DA overflow in studies employing a toxin blocker selective for the Kv1.2 subunit and Kv1.2 knock-out mice, and by demonstrating functional and physical coupling between D2-AR and Kv1.2.

EXPERIMENTAL PROCEDURES
Subjects-A transgenic mouse line in which the tyrosine hydroxylase (TH) promoter drives green fluorescent protein expression (TH-eGFP/21-31) on a C57BL/6 background was used for most experiments. TH-eGFP mice between the ages of P0 -1 or P25-35 days were used for DA cultures and fluorescent-activated cell sorting, respectively. Male TH-eGFP heterozygous mice or wild type littermates, 4 to 6 weeks of age, were used for slice immunohistochemistry and cyclic voltammetry experiments. For the remaining cyclic voltammetry experiments, male Kv1.2 knock-out mice (Kcna2 tm1Tem , referred to here as Kv1.2 Ϫ/Ϫ ) and wild type littermates (Kv1.2 ϩ/ϩ ) on a C3HeB/FeJ background were used. Kv1.2 Ϫ/Ϫ mice completely lacking Kv1.2 channel subunit protein have increased seizure susceptibility and reduced lifespan (mean age ϭ 17 Ϯ 0.2 days) and thus were used between the ages of P15 and 17 (25). For immunoprecipitation experiments we used adult D2 knock-out (26), D2L knock-out (11), and wild type mice littermates on a C57BL/6 background. Genotyping for TH-eGFP and Kv1.2 Ϫ/Ϫ,ϩ/ϩ mice was performed on tail DNA according to published protocols (25,27). All procedures involving the use of animals were approved by the Université de Montréal Animal Ethics Committee and are in accordance with the guidelines on the care and use of experimental animals outlined by the Canadian Council on Animal Care.
Fluorescent-activated Cell Sorting-TH-eGFP mice (25-35 days old) were anesthetized with halothane and decapitated. Cubes of tissue containing the substantia nigra (SN) and ventral tegmental area (VTA) were dissected from 2 to 3 mice for each FACS sample. The tissue treatment and cell dissociation procedures used have been described elsewhere (27). Cell suspensions were sorted using a BD FACSAria flow cytometer (BD BioSciences) according to GFP fluorescence profiles. The threshold for GFP detection was set beforehand according to cell suspensions derived from wild type mice.
Multiplex PCR-Total RNA was isolated from FACS-purified cells and brain tissue controls using TRIzol (Invitrogen). PCR were performed using 20% of the RT reaction, 1.5 mM MgCl 2 , 0.5 mM dNTPs mixture, 10 pmol of each primer (see below), 5 units of Taq DNA polymerase (Qiagen) in PCR buffer (20 mM Tris-HCl, 50 mM KCl, pH 8.3). Multiplex PCR were resolved in 1.5% agarose gels and imaged on a Kodak DC290 system. Multiplex PCR and electrophoresis on cDNA derived from FACS-purified cells was performed five times using separate samples.
Confocal Microscopy-All images were captured with a point-scanning confocal microscope (Prairie Technologies, Middleton, WI) equipped with an argon (wavelength 488 nm) and helium/neon (633 nm) laser. For the analysis of Kv1.x and DAT co-localization, all image acquisition parameters were kept constant across Kv1 subtypes. For colocalization analyses in striatal slices, confocal images of the dorsal striatum were acquired from two pre-selected rostrocaudal levels: 1) ϩ1.10 -1.34 and 2) 1.42-1.54 mm from bregma. Z-stack confocal images were acquired at 1-m intervals and 10 -15 images were acquired per stack. Co-localization analyses were performed on 5 images derived from the middle of each Z-stack series. Using ImageJ software (31), DAT and Kv1.x signal thresholds were set according to negative (no primary antibody) controls. To obtain the colocalization signal values we used the colocalize function of ImageJ (additional Plugin), which isolates the overlapping signals. Colocalized images and DAT alone images were then thresholded using ImageJ (Image Ϫ Adjust) to create binary images. Particles in each DAT and DAT ϩ Kv1 binary image were measured using the Analyze Particles command. This analysis generated values for the total signal area of DAT and Kv1.xϩDAT colocalization. Values were obtained for each of the 5 stacked images and then averaged. Values representing the proportion of total DAT signal area occupied by Kv1.x were calculated (signal area of Kv1.x ϩ DAT/DAT).
Fast Scan Cyclic Voltammetry-Mice were anesthetized with halothane prior to decapitation. Coronal slices, 300-m thick, were cut using a vibratome at the level of the dorsal striatum (ϩ1.10 -1.70 mm from Bregma) in ice-cold carbogenated artificial cerebrospinal fluid (ACSF) containing 125 mM NaCl, 26 mM NaHCO 3 , 2.5 mM KCl, 2.4 mM CaCl 2 , 1.3 mM MgSO 4 , 0.3 mM KH 2 PO 4 , and 10 mM D-glucose. Slices were cut in half and maintained in carbogenated ACSF at room temperature for at least 1 h prior to recordings. Once in the recording chamber, slices were superfused with ACSF maintained at 32°C at a rate of 1 ml/min. Carbon-fiber electrodes were constructed according to procedures described elsewhere (32) and were cut to obtain basal currents between 100 and 180 nA. To promote greater sensitivity, electrodes were soaked in carbon-purified isopropyl alcohol for at least 10 min before use (33). Electrodes were calibrated in vitro using known concentrations of DA. The carbon fiber electrode was positioned in the dorsolateral striatum, ϳ100 m into the slice and ϳ100 m from the center of the tips of a bipolar stimulating electrode that rested on the surface of the slice. Upon positioning the stimulation and carbon fiber electrodes in the dorsolateral striatum, slices were left to stabilize for 20 min before electrochemical recordings. One recording and stimulation site per slice was tested and each slice served as its own precondition control. Fast scan cyclic voltammetry was used to measure electrically evoked DA overflow (34,35). The potential of the carbon fiber electrode was held at Ϫ0.4 V versus Ag/AgCl between scans and was linearly ramped from Ϫ0.4 to ϩ1.0 V and then back to Ϫ0.4 V at a rate of 300 V/s. Each triangular waveform scan lasted 10 ms and was repeated every 100 ms. Data acquisition was carried out with an Axopatch 200B amplifier (Axon Instruments) connected to a Compaq Pentium III computer and pClamp 9 software through a DigiData 1200B analog to digital interface (Molecular Devices). DA overflow was evoked at 2-min intervals by a single 400-A rectangular electrical pulse (1 ms) delivered by an S-900 stimulator (Dagan Corporation, Minneapolis, MN) through a bipolar stimulating electrode (Plastics One, Roanoke, VI). Background current waveforms obtained immediately before the stimulation were subtracted from current waveforms obtained after stimulation to generate DA cyclic voltammograms (Fig. 4A). Current waveforms remained stable over the course of testing. Oxidation current peaks for DA were obtained at potentials of ϳ300 -500 mV (versus Ag/AgCl) corresponding to 3.5-4.5 ms in the voltage waveform (Fig. 4A).
Statistical Analyses-The peak oxidation currents derived from voltammograms were converted to concentration by electrode calibration and used to generate current-time plots (Fig.  4B). Each slice served as its own precondition control, thus maximal DA concentrations (Fig. 4B) were normalized to a respective baseline period: the 8-min period immediately prior to drug or toxin application. Drugs and toxins were applied until maximal effects on DA overflow were observed, ranging between 15 and 30 min, with the exception of quinpirole, which was applied for 4 min. Thus, the effect of a drug/toxin on evoked DA overflow was evaluated during a period after which stable effects were achieved, represented by the last 8 min of drug/toxin application. Mean values from this 8-min period were used for statistical comparisons. The ability of quinpirole to inhibit DA overflow was determined at its peak effect. The effect of drug/toxin alone on DA overflow was assessed by comparing to overflow values obtained in separate slices perfused with ACSF alone for the same duration (basal overflow). Similarly, the influence of drug/toxin preincubation on the inhibitory effects of quinpirole on DA overflow was assessed by comparing to separate slices in which quinpirole alone was applied 30 min after the baseline control period (i.e. the same duration that drug/toxin was applied). The influence of 4-AP on DA was assessed using an unpaired t test. The remaining drug and toxin analyses were carried out using a one-way ANOVA and Dunnett test for multiple comparisons. DA overflow in slices derived from Kv1.2 Ϫ/Ϫ mice and wild type littermates (Kv1.2 ϩ/ϩ ) were evaluated by unpaired t tests. To evaluate the influence of TAT-GRK peptides on DA overflow, a oneway ANOVA with a Newman-Keuls test was used. Values that met Chauvenet's criterion for statistical outliers (Ͼ2 standard deviations) were excluded (6 of a total of 129 values). All data are shown as mean Ϯ S.E. The sample size is indicated as n ϭ (x;y), x referring to the number of slices and y the number of mice. An ␣ level of p Յ 0.05 was set as the criterion for statistical significance.

Transactivator of Transcription (TAT)-G Protein-coupled Receptor Kinase-Carboxyl Terminus (GRKct) Peptide Construction and Validation-See supplemental "Experimental Procedures."
Kv1.2 and D2-AR Transfections for K ϩ Current Measurements-HEK293 cells were plated on collagen/poly-L-lysine precoated 1.5-cm round glass coverslips and incubated in 100 ϫ 15-mm Petri dishes (2 coverslips per dish) containing 2.5 ml of DMEM enriched with 1% penicillin-streptomycin, 1% glutamine, and 10% FBS. Upon reaching Ϯ80% confluence, the coverslips were transferred to 12-well plates containing 1 ml/well of basic DMEM, and the cells were transiently co-transfected with Kv1.2-YFP and D2S-myc or D2L-myc plasmids or Kv1.2-YFP ϩ D2S-myc (or the FLAG epitope for immunocytochemistry) ϩ GRKct-myc (36) plasmids (in the case of triple transfections) using Lipofectamine (Invitrogen) according to the manufacturer's instructions. The coverslips were inserted in a recording chamber affixed to the stage of an inverted Nikon Eclipse TE-200 fluorescent microscope and gravitationally perfused with saline. Whole cell patch clamp recordings were performed on YFP (Kv1.2) express-ing cells using borosilicate pipettes (5.5-6.5 M⍀) filled with an intra-pipette solution composed of (in mM): KMeSO 4 , 145; KCl, 20; NaCl, 10; EGTA, 0.1; ATP (Mg salt), 2; GTP (Tris salt), 0.6; HEPES, 10; phosphocreatine, 10 (pH 7.35). The signal was amplified by a Warner PC-505 amplifier, filtered at 1 kHz, digitized at 10 kHz, and analyzed with pClamp8 software (Molecular Devices). Series resistance was in the range of 10 M⍀ and was compensated to ϳ75%. Membrane voltage was stepped sequentially from Ϫ80 to ϩ30 mV using 200-ms depolarizing steps and leak-subtracted membrane currents were recorded at a rate of one series per minute. K ϩ current amplitudes were normalized by cell capacitance. As a control, K ϩ currents were also recorded in non-transfected cells. Quinpirole (1 M) and 4-AP (100 M) were applied for 3 min. Current-voltage values between the quinpirole and saline group were compared using a two-way ANOVA. The last minute of drug or saline application was normalized to the average of the 5-min baseline control period and these values are described as percent (%) of baseline under "Results." The voltage sensitivity of currents was estimated by measuring the membrane potential at which 50% of the maximal current was recorded and was compared between quinpirole and saline groups using a t test. To compare the effects of treatment (quinpirole, 4-AP, MTX versus saline) on normalized (% baseline) peak current values, a one-way ANOVA and Dunnett test for multiple comparisons was used. To explore the influence of Kv1.2 blockade on the ability of quinpirole to increase K ϩ current, MTX was applied for 9 min and quinpirole was co-applied from minute 4 to 6 of the MTX incubation. To compare the effects of MTX preincubation on the influence of quinpirole versus saline, MTX, or quinpirole alone a one-way ANOVA with post hoc Newman Keuls pairwise comparisons was performed.
Immunoprecipitation and Western Blotting-COS-7 cells were cultured in Dulbecco's modified Eagle's medium/F-12 (Invitrogen) supplemented with 10% fetal bovine serum, 100 g/ml of streptomycin, 100 IU/ml of penicillin, cells were grown at 37°C in 5% CO 2 . Cells were transfected with D2S, D2L, or Kv1.2-YFP (37) expression plasmids using FuGENE HD (Roche Applied Science) according to the manufacturer's instructions. 24 h after transfections cells were rinsed with cold PBS and mechanically disrupted using a cell scraper and a glass homogenizer and recovered in membrane extraction buffer (MEB; 125 mM KCl, 50 mM Tris-HCl, pH 7.5, 5 mM EDTA, pH 8.0) and protease and phosphatase inhibitor mixtures. Immunoprecipitation (IP) and Western blotting were performed as described elsewhere (38). Briefly, protein samples (400 g) were pre-cleared via incubation with protein A/G-agarose (Invitrogen) for 2 h. Upon removal of agarose beads, cell lysates were probed using a mouse monoclonal antibody (4H6-7-3) directed against D2R (5 g) overnight at 4°C with agitation. Control samples were incubated with normal mouse IgG (5 g). IP was performed by adding 20 l of protein A/G-agarose slurry followed by a 3-h incubation at 4°C with agitation. The beads were collected via centrifugation, washed three times with MEB, and then resuspended in Laemmli sample buffer. For Western blotting, proteins separated by denaturing 10% SDS-PAGE were transferred onto polyvinylidene difluoride membranes (Millipore). Membranes were blocked in 5% nonfat dry milk in PBS, 0.05% Tween 20 and incubated with either anti-Kv1.2 antibody (Alomone Labs) or anti-D2 dopamine receptor antibody (made in-house and previously characterized (38) and anti-actin antibody. For detection, membranes were incubated with HRP-conjugated secondary antibodies (Jackson ImmunoResearch Laboratories). Immunoblots were visualized using a chemiluminescent HRP substrate (Immobilon Western, Millipore).
For experiments with striatal tissue, mice were killed by cervical dislocation, after which the left and right hemistriata were rapidly dissected on an ice-cold surface and homogenized in MEB buffer with protease inhibitors (as above). Striatal synaptosomes were prepared according to a previously described procedure (39). The rest of the IP procedure was as for COS-7 cell extracts. As evidence that GFP-sorted cells were dopaminergic and did not contain GABA neurons, we detected TH but not glutamic acid decarboxylase (GAD) mRNA in FACS-purified samples (Fig. 1A). We still detected GAD in mRNA extracted from whole midbrain (Fig. 1B) and in cells excluded from the sorted pool by their lack of GFP fluorescence (Fig. 1C). As shown in Fig. 1, Kv1.1, -1.2, -1.3, -1.4, and -1.6 but not Kv1.5 mRNA was expressed in FACS-purified DA neurons. The same mRNA expression profiles for Kv1.1-1.6, TH, and GAD were found in the five independent experiments using different cell samples. As a negative control we also assessed the presence of mRNA for Kv4.2, a member of the Shal voltage-gated K ϩ channel subfamily with ubiquitous expression in brain (40), but absent from midbrain DA neurons (41). In 4 of 5 experiments, Kv4.2 could not be detected in FACS-purified DA neurons, whereas a weak signal for Kv4.2 was observed in one experiment performed with FACS samples derived from the youngest mice (P25; data not shown). As a positive control, the PCR product for all primers sets, including Kv1.5 and Kv4.2, were observed in midbrain (Fig. 1B) and mixed brain samples (data not shown). Kv1.5 and Kv4.2 were also detected in FACS-negative samples (Fig. 1C).

RESULTS
Immunocytochemical Expression of Kv1.1, -1.2, -1.3, -1.4, and -1.6 in Cultured DA Neurons-Immunolabeling was used to confirm Kv1 subunit expression in cultured DA neurons of the SN/VTA. Double immunofluorescence on primary DA cultures was performed using monoclonal antibodies against the DAT, to identify dopaminergic processes, and Kv1.1-1.6. As illustrated in Fig. 2, Kv1.1, -1.2, -1.3, -1.6, and, to a lesser extent, Kv1.4 subunits are localized to DA neurons of the SN/VTA, with the signal being particularly concentrated in axonal-like processes. In contrast, only a nonspecific signal (not above background level; supplemental 4-Aminopyridine Blocks the Ability of a D2 Agonist to Inhibit DA Overflow-To determine the role of Kv1 channels in D2-AR-mediated regulation of axonal DA overflow, we measured stimulation-evoked DA overflow in striatal slice preparations using fast scan cyclic voltammetry. Peak DA current values obtained from background-subtracted voltammograms (Fig. 4A) were plotted over time and show the time course of DA release and reuptake (Fig. 4, B and C). All subsequent voltammetry data are presented as normalized concentration values (normalized to the respective baseline control period). Under the conditions used, DA overflow was sensitive to tetrodotoxin treatment and thus activity dependent (data not shown).
Block of D2-AR-induced Inhibition of DA Overflow by Kv1 Subunit-specific Toxins-To characterize the Kv1 subtypes responsible for D2-AR regulation of DA overflow, we exploited several specific toxins for Kv1 channels that have been isolated from scorpion and snake venoms (44,45). ␣-Dendrotoxin (␣DTX) (from Dendroaspis angusticeps) blocks K ϩ channels containing Kv1.1, Kv1.2, or Kv1.6 subunits, whereas dendrotoxin-K (DTX-K) (from Dendroaspis polylepis) specifically blocks Kv1.1-containing channels (45)(46)(47). We also tested two more recently described toxin blockers: maurotoxin (MTX) (from scorpio Maurus palmatus) and Osk-1 (from Orthochirus scrobiculosus). At the 10 nM dose used, MTX potently blocks Kv1.2-containing channels (48), whereas Osk-1 at 2 nM is a specific blocker of most Kv1.3 channels (49). Only one subunit of the channel is required for the binding of toxins, thus these toxins are able to block heteromeric channels containing at least one toxin-sensitive subunit (50). We investigated bath pre-application of each toxin for 25-30 min on the ability of quinpirole to inhibit DA overflow. At time-matched control conditions, the effect of quinpirole alone was determined 30 min after baseline DA overflow measurements.    (Fig. 5, C and D). These results corroborate our toxin data and provide evidence for a physiologically relevant effect of Kv1.2 activity in axonal D2-AR regulation of DA overflow.
Co-immunoprecipitation of D2-AR and Kv1.2-To begin exploring a potential interaction between D2-AR and Kv1.2, we evaluated whether Kv1.2 could co-immunoprecipitate with the D2 receptor. We first co-expressed YFP-tagged Kv1.2 with D2S and D2L expression vectors in COS cells. Kv1.2 could indeed be immunoprecipitated with a D2 receptor antibody in both D2Sand D2L-expressing cells (Fig. 6A). To validate this finding, we also examined mouse striatal tissue and found that Kv1.2 was again reliably immunoprecipitated (Fig. 6B). Furthermore, the signal was absent in tissue extracts obtained from D2 KO mice (26) but present in extracts prepared from D2L KO mice, thus confirming the interaction with D2S, the isoform previously suggested to act as the main D2-AR (11). To further distinguish between pre-and postsynaptic D2 receptors, we also prepared striatal synaptosomes, a preparation highly enriched in axon terminals. We found that Kv1.2 could also be immunoprecipitated from this material (Fig. 6C). These findings suggest that Kv1.2 and D2-AR are components of a common signaling complex at the membrane.
Quinpirole Increases K ϩ Conductance in D2-AR/Kv1.2 Cotransfected Cells-We next focused our attention on the functional interaction of the D2-AR and Kv1.2. As we were unable to measure a Kv1.2, MTX-sensitive current in the cell body of cultured midbrain DA neurons (data not shown), we assessed the functional coupling of D2-AR and Kv1.2 in transfected HEK293 cells. We measured the regulation of K ϩ currents by the D2-AR in cells co-transfected with YFP-tagged Kv1.2 and myc-tagged D2S (short isoform of the D2 receptor). Visualization of YFP fluorescence guided the selection of cells for whole cell patch clamp recordings. Counting of YFP (Kv1.2) and myc  Kv1.1, -1.2, -1.3, -1.6, and, to a lesser extent, Kv1.4  (D2S) immunolabeled cells revealed that the vast majority (84.9%) of Kv1.2-positive cells were indeed co-transfected with D2S (Fig. 7A). Cells were maintained at Ϫ80 mV and currents were recorded under voltage-clamp in response to sequences of depolarizing steps between Ϫ80 and ϩ30 mV (Fig. 7B). As illustrated by the conductance-voltage curves in Fig. 7B, depolariza-tions to values more positive than Ϫ40 mV were required to generate detectable currents in Kv1.2-and D2S-transfected cells. In contrast, currents only started to appear at 0 mV in non-transfected cells and with more than 10-fold lower amplitudes (data not shown). In both quinpirole-and saline-treated groups, the conductance amplitude increased in a voltage-de- pendent manner. Activation of D2-AR (D2S) with quinpirole increased K ϩ currents relative to cells treated with saline (F (1,11) ϭ 26.5; p Ͻ 0.0001). A shift in voltage dependence was also induced by quinpirole: voltage at 50% maximal conductance was Ϫ0.82 Ϯ 1.7 mV (mean Ϯ S.E.) in control cells and Ϫ16.1 Ϯ 1.9 mV in quinpirole-treated cells (p Ͻ 0.0001) (Fig. 7C). As shown in Fig. 7D, the wide spectrum Kv1 blocker, 4-AP (100 M), reduced K ϩ currents relative to saline-treated control (  cultures by assessing the ability of the peptide to attenuate D2-AR-induced inhibition of DA cell firing, a process known to rely on G␤␥ coupling and activation of somatodendritic G-protein gated inward-rectifying K ϩ channels (supplemental Fig. S1A) (51). Furthermore, penetration of active and control peptides in striatal slices was validated by immunolabeling with anti-FLAG and anti-DAT antibodies (supplemental Fig. S1B). As shown in Fig. 8A, preincubation of slices with TAT-GRK peptide (1 M) partially blocked the ability of quinpirole (1 M) to inhibit stimulation-evoked DA overflow by 23.6% relative to the inactive peptide (1 M) (54.9 Ϯ 1.1 versus 71.78 Ϯ 5.5; p Ͻ 0.01). NEM, a sulfhydryl alkalating agent, selectively reduces responses mediated by pertussis toxin-sensitive G proteins (G i/o ) (52). NEM was used to block G i/o because pertussis toxin requires long incubation times that are incompatible with the use of acute slice preparations. Pre-application of NEM (200 nM) blocked the inhibitory actions of quinpirole by 71.8% (23.6 Ϯ 13.4 versus 83.6 Ϯ 1.2; p Ͻ 0.01) (Fig. 8B). Finally, to determine the contribution of G␤␥ signaling in the activation of Kv1.2 by D2-AR, we triple transfected HEK293 cells with Kv1.2, D2S, and GRKct (Fig. 8C) and recorded K ϩ currents following saline or quinpirole (1 M) application. As illustrated in Fig.  8D, the presence of GRKct completely prevented the previously described increase in K ϩ conductance induced by quinpirole. These findings are consistent with the involve-

DISCUSSION
Identifying the mechanisms by which D2-AR regulates axonal DA release in the striatum has clear implications for understanding the motor, cognitive, and emotional functions tied to striatal DA signaling. In the present study, we demonstrate the importance of voltage-gated K ϩ channels of the Kv1 subfamily in the actions of presynaptic D2-AR to regulate axonal DA overflow. In particular, by using Kv1.2 Ϫ/Ϫ mice and a Kv1.2 toxin blocker during real time measures of DA release along with K ϩ conductance measurements in Kv1.2-transfected cells, we reveal a specific role for Kv1.2 in D2-AR func-tion. Collectively, our observations uncover a mechanism by which axonal D2-AR attenuates DA overflow by regulating the function of Kv1 channels and isolate the importance of Kv1.2 in this process by demonstrating the coupling of D2-AR to Kv1.2 channels.
Kv1.1 through 1.6 subunits are widely expressed in the mammalian brain where they act to regulate excitability and secretion. This study is the first to demonstrate Kv1 subunit expression in midbrain DA neurons. FACS-purified DA neuron preparations were probed for Kv1 subunit expression to reveal Kv1.1, -1.2, -1.3, -1.4, and -1.6 mRNA in DA neurons of the SN/VTA. Although Kv1.5 was observed in whole midbrain samples, it was not detected in purified DA samples, a finding that is consistent with the relatively low abundance of this subunit in neurons (54). Expression analyses at the RNA level largely match those obtained at the protein level as Kv1.1, -1.2, B, application of the D2 agonist, quinpirole (1 M; n ϭ 10) to co-transfected cells increased K ϩ conductance as compared with saline (n ϭ 7). C, the membrane potential at 50% maximal conductance was substantially reduced by quinpirole versus saline. D, quinpirole increased maximum K ϩ conductance, whereas the wide spectrum Kv1 blocker, 4-AP (100 M; n ϭ 9) and the Kv1.2 blocker, MTX (50 nM; n ϭ 8), blocked a large proportion of the K ϩ current versus saline control (n ϭ 7). E, blocking Kv1.2 by preincubation with MTX (n ϭ 8) prevented the increase in K ϩ current by quinpirole. F, quinpirole (n ϭ 6) potentiated the K ϩ current in HEK cells co-transfected with D2L and Kv1.2 as compared with saline (n ϭ 6). Mean Ϯ S.E.: *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001. -1.3, -1.4, and -1.6, but not Kv1.5, immunofluorescence was detected in cultured DA neurons. In all cases, Kv1 staining was observed along somatodendritic and axonal-like domains of DA neurons, which is consistent with Kv1 expression patterns described elsewhere and in the role of these channels in diverse cellular functions (19). Expression of each Kv1 subunit was apparent in all DAT-positive neurons, thus suggesting that these subunits are co-expressed in DA neurons. The Kv1 channel pore is composed of four ␣ subunits that often co-assemble to form heteromultimeric channels (23,24) and thus Kv1 subunits may combine in such a manner in DA neurons, although this remains to be specifically examined.
Immunohistochemical examination of striatal slices revealed Kv1.1, -1.2, and -1.6 along DA-positive axons and/or terminals. In contrast, Kv1.3 immunolabeling was near absent in DA processes. Kv1.2 is the most abundant subunit in the mammalian brain and is present in the great majority of Kv1 channel complexes and in all Kv1.1 and -1.6-containing channel complexes (55). The relative difference in the abundance of Kv1 subunits on DA-positive fibers that was measured is likely a close representation rather than a reflection of variations in antibody staining because immunolabeling of tubular microvasculaturelike structures was similar across subunits (not shown). Moreover, the specific binding and labeling of the Kv1 antibodies used has been validated (56). Kv1 channels are reported to be predominantly localized to the immediate presynaptic region of axon terminals (54). It may be considered likely that Kv1.1, -1.2, and -1.6 subunits are thus also preferentially localized to the immediate presynaptic region of DA neurons. However, the amount of DAT and Kv1 co-localization we observed in striatal slices was relatively small: our images may have underestimated the true level of co-localization, because in a given confocal image plane, we would expect the number of DA axons coursing through and terminating at sites more rostral or caudal to exceed the number of presynaptic DA terminals regions. The precise location of Kv1 channels on DA axons will require additional experimentation including electron microscopy. Taken together, the expression data provide strong evidence for the presence of Kv1.1, -1.2, and -1.6 subunit-containing channels in DA neurons and their localization on nigrostriatal dopaminergic axons and/or axon terminals.
Real time measurement of DA overflow with cyclic voltammetry provides an ideal tool to directly assess the functional contribution of Kv1 channels in D2-AR regulation of DA overflow. Moreover, the use of acute slice preparations for DA concentration measurements permits precise spatial and temporal control for the application of multiple drugs/toxins. Preincubation of slices with ␣DTX, a blocker of Kv1.1, -1.2, and -1.6 subunit-containing channels, blocked approximately half of the effect of quinpirole on evoked DA overflow. Interestingly, the influence of ␣DTX on D2-AR-mediated inhibition (48.8% block) was similar in magnitude to that obtained with the broad spectrum Kv1 blocker, 4-AP (44.2% block; Fig. 4D). Although 4-AP can also block Kv3-containing channels and a recent report implicates voltage-activated Ca 2ϩ channel facilitation (57), our results with ␣DTX suggest that the effects of 4-AP specifically on D2-AR function may be mostly due to its actions to block Kv1.1, -1.2, and -1.6.
Despite evidence of Kv1 channel involvement, 4-AP did not fully block the actions of the quinpirole to reduce DA overflow, suggesting that additional mechanisms mediate the ability of the D2-AR to inhibit DA overflow. One possibility is that D2-AR-mediated inhibition of axonal overflow is also mediated in part by other mechanisms such as direct inhibition of exocytosis or inhibition of voltage-dependent Ca 2ϩ channels in terminals. In support of the first of these mechanisms, previous work in cultured DA neurons suggests that D2-AR can negatively regulate the secretory process at a step that is downstream to calcium influx in axon terminals (16). However, DA release was not directly measured in that study and the possible implication of calcium channel inhibition was also not directly examined.
Our results establish a particular role for Kv1.2 channels in the regulation of DA release as MTX had the greatest impact of any of the Kv1 subunit-specific toxins to block the inhibiting actions of quinpirole on DA overflow. Additionally, we found that blockade of Kv1.1 with DTX-K diminished the influence of quinpirole on evoked DA overflow. In contrast, pre-treatment of slices with Osk-1, a Kv1.3 blocker, was without effect on DA overflow. Interestingly, both ␣DTX and DTX-K did not increase basal DA overflow, whereas MTX slightly enhanced it. These results may be taken to suggest a more defined role of Kv1.1, -1.2, and -1.6 channel activation in the regulation of DA release by D2-AR. On the contrary, 4-AP substantially increased basal DA overflow. It is not clear why this effect was not mimicked by ␣DTX, however, we reason that this could be due to the ability of 4-AP to additionally facilitate the opening of voltage-activated Ca 2ϩ channels (57). Altogether, these data support our immunohistochemical findings suggesting that Kv1.1, Kv1.2, and possibly Kv1.6 subunit-containing channels mediate the regulation of DA overflow operated by D2-AR.
The role of Kv1.2 in striatal DA overflow and D2-AR regulation of axonal DA overflow was substantiated in Kv1.2 Ϫ/Ϫ mouse knock-out experiments. Basal DA overflow was significantly elevated in Kv1.2 Ϫ/Ϫ mice, whereas the effect of quinpirole on DA overflow was significantly attenuated in Kv1.2 Ϫ/Ϫ mice relative to controls. These data corroborate those obtained with pharmacological blockade of Kv1.2. That functional deficits were detected in the Kv1.2 Ϫ/Ϫ mice imply that Kv1.2 is a critical regulator of DA release and D2-AR regulation of DA release, perhaps through D2R-mediated regulation of the voltage sensitivity of presynaptic Kv1 channels.
To determine whether D2-AR and Kv1.2 might be part of the same complex, we carried out IP experiments and patch clamp recordings of K ϩ currents. Our IP data from transfected cells and striatal membrane and synaptosomal preparations suggest that D2-AR and Kv1.2 proteins are part of a common signaling complex implicating D2-AR in dopaminergic axon terminals. Furthermore, we measured the ability of quinpirole to regulate K ϩ channel conductance in cells co-transfected with Kv1.2 and D2-AR (D2S). D2-AR activation substantially increased K ϩ conductance in a voltage-dependent manner. Moreover, the increase in conductance by quinpirole was fully blocked by the Kv1.2 channel blocker, MTX. Collectively, these data directly demonstrate the contribution of Kv1.2 to D2-AR activation. The voltage sensitivity of Kv1.2 was enhanced in response to D2-AR activation, thus identifying a possible mechanism underlying the enhancement of K ϩ currents.
Providing insight into the G protein signals that may mediate the interaction between D2-AR and Kv1.2, we found that a peptide inhibitor of G␤␥ diminished the inhibitory effect of quinpirole on striatal DA overflow. In addition, our K ϩ current measurements in cells show that blocking G␤␥ by introducing a GRKct construct prevents the increase in K ϩ current produced by quinpirole. Although further work will be required to elucidate the details of the function and organization of this complex, these experiments suggest that G␤␥ and Kv1.2 interact. Nonetheless, whether the regulation of Kv1.2 by D2-AR involves other effectors (58,59), or other direct or indirect interactions between D2-AR, Kv1.2, and G-proteins remains to be determined. The modest efficacy of the GRKct in reducing the inhibitory effect of quinpirole on DA overflow stands in contrast with the complete block of Kv1.2 channel regulation with this same peptide. This could result from insufficient accumulation of the TAT-delivered peptide in the brain slice preparation. In comparison, high levels of the peptide were produced in HEK293 cells following transfection. To summarize, the expression of Kv1.1, -1.2, and -1.6 in DA neurons and localization to axon terminals, along with the ability of subunitselective channel blockers to attenuate D2-AR inhibition of DA overflow implicate Kv1.1, -1.2, and -1.6 in D2-AR regulation of striatal DA release. By means of both pharmacological and genetic Kv1.2 loss-of-function, a particular role was underlined for Kv1.2 in the modulation of axonal DA overflow by D2-AR. Furthermore, we show that direct activation of D2-AR increases Kv1.2 conductance in co-transfected cells and that a Kv1.2 blocker prevents this effect, an effect that may be due to heightened voltage sensitivity of Kv1.2 channels. Co-IP experiments argue that D2-AR and Kv1.2 are part of a common signaling complex, compatible with the physiological interaction that we observed. We hypothesize that D2-AR activation decreases DA release in part by facilitating the voltage-dependent opening of Kv1.2-containing channels by a G␤␥-dependent mechanism, thus potentially leading to a shunting of axon terminal depolarization and secondary reduction in Ca 2ϩ influx and DA exocytosis. Collectively, this work adds Kv1.2 activation to the processes that mediate the plasticity of DA release and offers a novel mechanism by which DA release may be impaired in pathophysiological and neurological conditions.