Voltage-gated L-type Ca2+ channels are key determinants of synaptic integration and plasticity, dendritic electrogenesis, and activity-dependent gene expression in neurons. Fulfilling these functions requires appropriate channel gating, perisynaptic targeting, and linkage to intracellular signaling cascades controlled by G-protein-coupled receptors (GPCRs). Surprisingly, little is known about how these requirements are met in neurons. The studies described here shed new light on how this is accomplished. We show that D2 dopaminergic and M1 muscarinic receptors selectively modulate a biophysically distinctive subtype of L-type Ca2+ channels (CaV1.3) in striatal medium spiny neurons. The splice variant of these channels expressed in medium spiny neurons contains cytoplasmic Src homology 3 and PDZ (postsynaptic density-95 (PSD-95)/Discs large/zona occludens-1) domains that bind the synaptic scaffolding protein Shank. Medium spiny neurons coexpressed CaV1.3-interacting Shank isoforms that colocalized with PSD-95 and CaV1.3a channels in puncta resembling spines on which glutamatergic corticostriatal synapses are formed. The modulation of CaV1.3 channels by D2 and M1 receptors was disrupted by intracellular dialysis of a peptide designed to compete for the CaV1.3 PDZ domain but not with one targeting a related PDZ domain. The modulation also was disrupted by application of peptides targeting the Shank interaction with Homer. Upstate transitions in medium spiny neurons driven by activation of glutamatergic receptors were suppressed by genetic deletion of CaV1.3 channels or by activation of D2 dopaminergic receptors. Together, these results suggest that Shank promotes the assembly of a signaling complex at corticostriatal synapses that enables key GPCRs to regulate L-type Ca2+ channels and the integration of glutamatergic synaptic events.
- L-type channels
- patch clamp
- PDZ domain
- state transitions
- basal ganglia
Voltage-dependent Ca2+ channels perform an astonishing array of specialized functions in neurons. In perisomatic and dendritic regions, L-type Ca2+ channels are particularly important in translating synaptic activity into alterations in gene expression and neuronal function. Synaptically driven activation of transcription factors, like cAMP response element-binding protein (CREB) and nuclear factor of activated T-cells, depend on these channels (Murphy et al., 1991; Bading et al., 1993; Graef et al., 1999; Mermelstein et al., 2000; Dolmetsch et al., 2001). L-type Ca2+ channels also are key participants in the regulation of long-term alterations in synaptic strength (Bolshakov and Siegelbaum, 1994; Calabresi et al., 1994; Kapur et al., 1998; Yasuda et al., 2003) as well as short-term dendritic excitability, the active processing of synaptic events and repetitive spiking (Avery and Johnston, 1996; Hernandez-Lopez et al., 2000; Bowden et al., 2001; Lo and Erzurumlu, 2002; Vergara et al., 2003).
The engagement of L-type Ca2+ channels is regulated by G-protein-coupled receptors (GPCRs). In a variety of excitable cells, GPCRs modulate L-type channel opening by controlling channel phosphorylation state (Hell et al., 1993a; Surmeier et al., 1995; Gao et al., 1997; Kamp and Hell, 2000). In striatal medium spiny neurons, both D2 dopaminergic and M1 muscarinic receptors suppress L-type Ca2+ channel currents by activating the Ca2+/calmodulin-dependent protein phosphatase calcineurin (Howe and Surmeier, 1995; Hernandez-Lopez et al., 2000). However, it isn't clear whether GPCRs modulate perisynaptic L-type Ca2+ channels. Both GPCRs are found in abundance near corticostriatal glutamatergic synapses that are formed preferentially on spine heads in medium spiny neurons (Bolam et al., 2000), where functional studies suggest L-type channels are located (Calabresi et al., 1994; Cepeda et al., 1998).
The physical association that would be necessary for synaptic GPCR regulation of L-type channels could be created by synaptic scaffolding or adaptor proteins (Kennedy, 1998; Sheng and Kim, 2000). Recent work has shown that L-type Ca2+ channels possessing a pore-forming CaV1.2 subunit bind to several adaptor proteins, and this interaction enables phosphorylated pCREB signaling (Kurschner et al., 1998; Kurschner and Yuzaki, 1999; Weick et al., 2003). This terminal region also associates the channel with signaling enzymes and GPCRs (Davare et al., 2000, 2001). Much less is known about targeting of the other major class of L-type Ca2+ channel found in neurons in which the pore is formed by a CaV1.3 subunit. Unlike CaV1.2 channels, these channels open at membrane potentials likely to be achieved during modest synaptic stimulation (Koschak et al., 2001; Scholze et al., 2001; Xu and Lipscombe, 2001). The C-terminal region of this subunit is alternatively spliced, yielding long (CaV1.3a) and short (CaV1.3b) variants (Safa et al., 2001; Xu and Lipscombe, 2001). The long splice variant contains an Src homology 3 (SH3) and a class I PDZ (postsynaptic density-95 (PSD-95)/Discs large/zona occludens-1) binding domain that selectively binds Shank proteins, which are thought to be key scaffolding elements in the synaptic signaling complex (Kennedy, 1998; Sheng and Kim, 2000). The results presented here suggest that, indeed, Shank targeting of CaV1.3a channels enables their selective modulation by dopaminergic and cholinergic GPCRs.
Materials and Methods
Animals. C57BL/6 mice were obtained from Harlan (Indianapolis, IN). CaV1.3 knock-out mice were obtained from Joerg Striessnig (Institut für Biochemische Pharmakologie, Innsbruck, Austria) (Platzer et al., 2000), rederived, and backcrossed on a C57Bl/6 background in the Northwestern University barrier facility. All experiments followed protocols approved by the Northwestern University Center for Comparative Medicine, an Association for Assessment and Accreditation of Laboratory Animal Care accredited facility, and followed guidelines issued by the National Institutes of Health and the Society for Neuroscience.
Tissue preparation for single neurons. The brain was removed from young (>3-6 weeks of age) male C57BL/6 mice, chilled, sliced (300 μm) coronally in sucrose, and immediately placed in NaHCO3-buffered Earl's balanced salt solution saturated with 95% O2 and 5% CO2 where it was held from 20 min to 6 h. The striatum was dissected from the slices immediately before dissociation and placed in HBSS saturated with O2 and containing type XIV bacterial protease (1 mg/ml) for 25 min. EDTA (0.5 mm) and l-cysteine (1 mm) was added to papain (final concentration, 40 U/ml, pH 7.2) (Raman et al., 2000) for experiments involving D2, M1, or 5-HT2 receptor agonists. Penicillin-streptomycin was added to the papain solution to inhibit bacterial growth. After enzyme digestion, the tissue was triturated for physical dissociation of the medium spiny neurons and then placed in a suspension dish. In some experiments, CaV1.3 (Platzer et al., 2000) or CaV2.3 (Wilson et al., 2000) knock-out mice were used for experiments.
Whole-cell voltage-clamp recordings in acutely isolated neurons. Whole-cell recordings used standard techniques (Bargas et al., 1994). The internal solution contained the following (in mm): 180 N-methyl-d-glucamine, 40 HEPES, 4 MgCl2, 12 phosphocreatine, 2 Na2ATP, 0.5 Na3GTP, and 0.1 leupeptin, pH 7.2-3, with H2SO4, 265-270 mOsm/L. A calcium chelator was often omitted; however, for the experiments with dialysis of CaV1.2 binding peptide into the cell, EGTA (100 μm) was added to the internal. A 2-5 mm Ba2+ NaCl external recording solution was used (in mm: 110 NaCl, 1 MgCl2, 2-5 BaCl2, 10 HEPES, 10 glucose, pH 7.35-7.4, 300 mOsm/L). In experiments determining dihydropyridine affinity or voltage dependence of activation, the major salt was tetraethyl-ammonium chloride (135 mm). All drugs were prepared according to the specifications of the manufacturers and applied with a gravity-fed “sewer pipe” capillary array (Surmeier et al., 1995). Electrodes were pulled and fire-polished to 1-2 μm for sufficient resistance in the bath (2-6 MΩ). For acutely isolated recordings, medium spiny neuron morphological identification was supported by a whole-cell capacitance ranging from 2-10 pF. Isolation of calcium currents was possible with the use of tetrodotoxin (TTX; 200 μm) to block sodium currents and CsCl (20 mm) to block potassium currents. All reagents were obtained from Sigma (St. Louis, MO) except GTP (Roche Diagnostic, Indianapolis, IN), TTX (Alomone Labs, Jerusalem, Israel), leupeptin, ω-conotoxin GVIA (Bachem, Torrance, CA), papain (Worthington, Lakewood, NJ), penicillin-streptomycin (Invitrogen, Carlsbad, CA), and ω-agatoxin TK (Peptides International, Louisville, KY). CaV1.2 COOH-tail and CaV1.3 Shank-PDZ binding peptides were synthesized by Biosource (Hopkinton, MA). CaV1.2 COOH-tail peptide sequence was SEEALPD-SRSYVSNL, and CaV1.3 Shank-PDZ binding peptide sequence was EE-EDLADEMICITTL. Whole-cell voltage-clamp recordings were obtained with an Axon Instruments (Foster City, CA) 200A patch-clamp amplifier and controlled and monitored with a Macintosh G4 running pulse 8.53 with a 125 KHz interface. Data were analyzed using Igor Pro 4.05, and Systat 5.2 and SigmaStat 3.0 software were used for statistical analysis. For clarity, the capacitance artifacts were suppressed in the figures containing current traces. Box plots were used to represent small sample sizes.
Whole-cell recordings in slices. Slices were obtained as described above and placed for >1 h into an artificial CSF (ACSF) containing the following (in mm): 125 NaCl, 3.5 KCl, 25 NaHCO3, 1.25 Na2HPO4, 1 MgCl2, 1.2 CaCl2, 25 glucose, pH 7.3, with NaOH, 300 mOsm/L, saturated with 95% CO2 and 5% O2. Thereafter, slices were transferred to a recording chamber and superfused with ACSF at a rate of 3-4 ml/min. Current-clamp recordings were performed on visually identified medium spiny neurons with an infrared video-microscopy system. The pipette solution consisted of the following (in mm): 119 KMeSO4, 1 MgCl2, 0.1 CaCl2, 10 HEPES, 1 EGTA, 12 phosphocreatine, 2 Na2ATP, 0.7 Na2GTP, pH 7.2-3, with KOH, 280-300 mOsm/L. Whole-cell recordings were obtained at 32°C.
Single-cell reverse transcription-PCR. Two types of reverse transcription (RT)-PCR profiling were performed. To maximize mRNA yields, some neurons were aspirated without recording. Isolated neurons were patched in the cell-attached mode and lifted into a stream of control solution. Neurons were then aspirated into an electrode containing sterile water. In other experiments, neurons were briefly subjected to recordings before aspiration. In these cases, the electrode recording solution was made nominally RNase free. RT procedure was performed using Superscript First-Strand Synthesis System (Invitrogen) as described previously (Surmeier et al., 1996). Primers for enkephalin (ENK) and upper primer for substance P (SP) were described previously (Surmeier et al., 1996). The lower primer for SP was ATG AAA GCA GAA CCA GGG GTA G (position 838; GenBank accession number D17584), which gave a PCR product of 616 bp. CaV1.2 mRNA (GenBank accession number NM009781) was detected with a pair of primers GAC AAC TGA CCT GCC CAG AG (position 6763) and GCT GTT GAG TTT CTC GCT GGA (position 7098), which gave a PCR product of 356 bp. CaV1.3 mRNA (GenBank accession number AJ437291) was detected with a pair of primers CTG ACT CGG GAC TGG TCT ATT C (position 4363) and CTG GAG GGA CAA CTT GGT CAA GCA (position 4718), which gave a PCR product of 379 bp. For detection of CaV1.3 splice variant mRNAs, a common upper primer TCC GGA CAG CTC TCA AGA TCA AG (position 4616) was used. The lower primer for the long form of CaV1.3 was GAC GGT GGG TGG TAT TGG TCT GC (position 5058), which gave a PCR product of 465 bp. The lower primer for the short form of CaV1.3 was GCG GTA GCT CAG GCA GAC AAC TC (position 4932; GenBank accession number AF370009), which gave a PCR product of 306 bp. TH mRNA (GenBank accession number NM012740) was detected with a pair of primers CAG GAC ATT GGA CTT GCA TCT (position 1415) and ATA GTT CCT GAG CTT GTC CTT G (position 1690), which gave a PCR product of 297 bp.
Immunoprecipitation. Rat brain synaptosomal fraction [postnatal day 2 (P2)] was purified as described previously (Maximov et al., 1999) and solubilized in the extraction buffer B containing 1% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate as follows (in mm): 137 NaCl, 2.7 KCl, 4.3 Na2HPO4, 1.4 KH2PO4, 5 EDTA, 5 EGTA, and protease inhibitors, pH 7.2. The lysate was clarified by centrifugation at 100,000 × g and incubated for 2 h at 4°C with protein A-Sepharose beads (Amersham Biosciences, Piscataway, NJ) coated with affinity-purified anti-CaV1.3a polyclonal antibody (pAb). Beads were pelleted by centrifugation, washed two times in the extraction buffer, and analyzed by Western blotting with guinea pig anti-Shank pAb.
Immunocytochemistry. Rabbit anti-Shank was a generous gift from Dr. Eunjoon Kim (Korea Advanced Institute of Science and Technology, Daejeon, Korea); mouse monoclonal anti-PSD-95 (clone 7E3-IB8) was purchased from Sigma-Aldrich; Alexa Fluor 488- and 568-labeled anti-rabbit and anti-mouse goat IgGs were from Molecular Probes (Eugene, OR). Eight normal C57BL/6 male mice and two CaV1.3 knock-out mice were deeply anesthetized with sodium pentobarbital (60 mg/kg body weight, i.p.) and transcardially perfused with 0.9% saline, followed by 4% freshly depolymerized paraformaldehyde and 15% saturated picric acid in 0.1 m phosphate buffer (PB), pH 7.4, at 4°C. Brains were dissected out, postfixed at room temperature for 1 h, and cryoprotected in 30% sucrose in PB. Frozen sections were cut in the sagittal and coronal planes at 20 μm on a freezing stage microtome. Free-floating sections were processed for indirect immunofluorescence. Unspecific binding was suppressed in a blocking solution containing 5% normal goat serum (NGS) in PBS with 0.05% Triton X-100 for 1 h at room temperature. Sections were then incubated with primary antibodies diluted in 2% NGS/PBS-0.05 Triton X-100 overnight at 4°C, followed by incubation with fluorescent secondary antibodies for 30 min at room temperature. After rinsing in PBS, sections were mounted with Vectashield (Vector Laboratories, Burlingame, CA). For double labeling, sections were incubated with pairs of primary antibodies (anti-PSD-95/anti-Shank, or anti-PSD-95/anti-CaV1.3a) and then with pairs of appropriate IgGs tagged with Alexa Fluor 488 and 568. Control sections incubated with normal serum in lieu of the primary antiserum were free of immunolabeling, and CaV1.3a knock-out mice did not show any immunoreactivity after incubation with antiserum to CaV1.3a.
Dissociated striatal neurons were obtained as described above, collected on a polylysine-coated glass slide, and fixed with 4% freshly depolymerized paraformaldehyde in 0.1 m PB, pH 7.4, at 4°C. Slides were immunoreacted as above with rabbit anti-CaV1.3a and rabbit anti-Shank, and binding sites were revealed with Alexa Fluor 488 and 568-labeled rabbit and goat IgGs. After coverslipping with Vectashield, slides were analyzed by fluorescence microscopy.
Laser-scanning confocal immunofluorescence microscopy was done on a Nikon Eclipse 800 microscope equipped with a Nikon PCM 2000 Confocal Microscope System (Nikon, Melville, NY). This utilizes an Argon laser (457, 488, 514 nm) and a green HeNe laser (535.5 nm). Standard emission filter 515/530 and bandpass filter 605/632 transmitting light between 589 and 621 nm were used for Argon and HeNe lasers, respectively. The system allowed simultaneous detection as well as sequential detection of orange/red (TCP-1 ring complex; Alexa 568; Texas Red) and green (FITC; Alexa 488; and green fluorescent protein) dyes. Detection parameters for sequential detection were optimized for each dye, and spillover between channels was minimized. Images for analysis were obtained with a Nikon planapochromatic 60× (numerical aperture, 1.4) oil immersion lens. Sequential optical sections were taken at 0.5-2.0 μm intervals in the Z plane to build an image volume in three dimensions on a graphics workstation using commercial software (SIMPLE PCI program; Compix Imaging Systems, Lake Oswego, OR). Software settings for optimal gain and black levels were individually adjusted for each fluoroprobe.
Spatial cross-correlations were computed from aligned, 0.5 μm thick, optical sections. The goal of this analysis was to determine whether there was a consistent spatial relationship between immunofluorescent signal intensity for CaV1.3, PSD-95, and Shank. Although our analytic approach differed somewhat from that recently described by Li et al. (2004), the approach was conceptually similar. Paired neuropil images for PSD-95/Shank or PSD-95/CaV1.3a were imported to Adobe Photoshop (Adobe Systems, San Jose, CA). Here, images were converted from color to grayscale; the files were not manipulated in any other way (e.g., no deconvolution or smoothing). Image intensity files were then imported as two-dimensional matrices into Igor Pro where they were standardized using self-reference. Conventional, spatial cross-correlations between aligned, standardized image matrices (Bendat and Piersol, 1971) were computed on a row pixel by row pixel basis for pixel lags up to the equivalent of 10 μm (column-by-column approach yielded similar results). The two-dimensional cross-correlation was then collapsed by averaging across columns. To obtain a measure of the correlation expected by chance, image rows and columns were shifted four times by a random number (generated from a uniform distribution by Igor Pro) and then the image cross-correlations were recomputed. This is presented as the “shuffled” correlation. This shuffling consistently led to “flat” cross-correlations and gave a measure of the background signal generated by the algorithm.
D2 dopaminergic and M1 muscarinic receptors suppress L-type Ca2+ channel currents in striatal medium spiny neurons
The vast majority of neurons in the rodent striatum are GABAergic medium spiny neurons (>90%). They can readily be identified by their typically bipolar shape and medium size (see Fig. 2A). Virtually all medium spiny neurons express M1 and M4 muscarinic receptors, reflecting the prominent role of cholinergic signaling in the functioning of the striatum (Graybiel, 1984; Bernard et al., 1992; Yan et al., 2001). Another key neuromodulator in the striatum is dopamine. The postsynaptic effects of dopamine are mediated mostly by D1 and D2 dopaminergic receptors, which are differentially expressed by medium spiny neurons projecting to the substantia nigra and the globus pallidus, respectively (Albin et al., 1989; Gerfen, 1992; Surmeier et al., 1996).
One of the most prominent consequences of D2 dopaminergic or M1 muscarinic receptor activation in medium spiny neurons is the suppression of L-type Ca2+ channel currents. This is easily seen in whole-cell recordings from acutely isolated medium spiny neurons in the presence of the dihydropyridine agonist S(-)-BayK 8644 (1 μm). BayK 8644 enhances the open probability of L-type channels and slows their deactivation rate, enabling them to be examined in isolation without using blockers for the other Ca2+ channels. The slowing of the deactivation kinetics can be seen clearly in Figure 1A. In the presence of BayK 8644, the application of the D2 receptor agonist R(-)-propyl-norapomorphine hydrochloride (NPA; 10 μm) reversibly reduces the amplitude of slow, L-type tail currents in approximately half of all medium spiny neurons (Fig. 1A,B) (Hernandez-Lopez et al., 2000). The tail currents are shown at higher resolution in the right panel in Figure 1A. Typically, NPA reduced tail currents measured 5 ms after the repolarizing step by ∼20%. Similarly, application of the muscarinic receptor agonist muscarine (2 μm) reversibly reduced the slow tail currents resulting from L-type channels in medium spiny neurons (Fig. 1C,D). Both GPCRs bring about their modulation by turning on phospholipase Cβ (PLCβ) isoforms, mobilizing inositol trisphosphate (IP3) receptor (IP3R) Ca2+ stores and activating the calcium-dependent protein phosphatase calcineurin (Howe and Surmeier, 1995; Hernandez-Lopez et al., 2000).
Medium spiny neurons coexpress CaV1.2 and CaV1.3 mRNAs
Although D2 and M1 receptor activation suppresses L-type Ca2+ channel currents, it is not known whether this modulation is specific to a particular channel subtype. In the brain, L-type channels have either a CaV1.2 or CaV1.3 pore-forming subunit (Catterall, 1998). To determine whether one or both of these was expressed in striatal medium spiny neurons, single-cell RT-PCR (scRT-PCR) profiling was performed. Medium spiny neurons were identified by their expression of SP or ENK mRNAs. These two peptides reliably mark the two major populations of medium spiny neurons that form so-called “direct” and “indirect” pathways (Gerfen, 1992). These two populations also differ in their expression of D1 and D2 receptors. Because the mRNAs for these peptides have high single-cell copy numbers, they are much more reliable markers in scRT-PCR protocols than low abundance GPCR mRNAs that can easily be missed (Surmeier et al., 1996). These experiments revealed that both SP and ENK neurons coexpressed CaV1.2 and CaV1.3 α subunit mRNAs (Fig. 2B). CaV1.2 and CaV1.3 mRNA were found in >70% of a large sample of medium spiny neurons (n = 53) (Fig. 2C). Given the probability of false negatives with low-abundance templates at the single-cell level, these results suggest that virtually all medium spiny neurons coexpress CaV1.2 and CaV1.3 mRNAs.
CaV1.2 and CaV1.3 channels in medium spiny neurons differ pharmacologically and biophysically
In heterologous expression systems, CaV1.2 and CaV1.3 channels differ in their affinity for dihydropyridine antagonists (Koschak et al., 2001; Xu and Lipscombe, 2001). To determine whether this distinction was maintained in a native expression system, medium spiny neurons were acutely isolated, voltage clamped using whole-cell techniques, and Ca2+ channels were pharmacologically isolated (Bargas et al., 1994). Using Ba2+ (5 mm) as a charge carrier to eliminate Ca2+-dependent inactivation, fast (150 ms) voltage ramps running from -80 to +30 mV from a holding potential of -60 mV were used to activate channels (Fig. 3A). To better isolate L-type channel currents, CaV2.1/2.2 channels were blocked (using a combination of ω-conotoxin GVIA and ω-agatoxin TK), leaving just CaV1 (L-type) and CaV2.3 (R-type) channels capable of fluxing ions (Bargas et al., 1994). In this situation, the application of the CaV1, L-type channel antagonist nimodipine blocked the evoked current in a dose-dependent manner. Selected traces from one of these experiments are shown in Figure 3A. Construction of dose-response plots from wild-type neurons were well fit only with two-site Langmuir isotherm having IC50 values near 200 nm and 2 μm (n = 8) (data not shown), values close to those obtained for CaV1.2 and CaV1.3 channels, respectively, in heterologous expression systems (Koschak et al., 2001; Xu and Lipscombe, 2001). At 10 μm, nimodipine blocked >90% of the current (Fig. 3A) with little increase in the block until concentrations reached 50 μm, at which point CaV2.3 channels began to be blocked (P. A. Olson, R. J. Miller, and D. J. Surmeier, unpublished observations). The low-affinity binding site was preserved in medium spiny neurons from CaV2.3 knock-out mice (Wilson et al., 2000) (n = 8), arguing that there was no confusion with this R-type Ca2+ channel. To verify that the high affinity site was attributable to CaV1.2 channels, medium spiny neurons from CaV1.3 knock-out mice were examined. As predicted, L-type currents in these neurons were completely blocked by 1 μm nimodipine (Fig. 3B)(n = 8). These data show that medium spiny neurons coexpress CaV1.2 and CaV1.3 Ca2+ channels that can be separated primarily on the basis of their sensitivity to dihydropyridine antagonists. In addition, we studied the voltage dependence of the dihydropyridine block (Bean, 1984). CaV1.2 sensitivity was not affected by holding potential. In contrast, holding the membrane potential at -80 mV dramatically reduced CaV1.3 channel block by nimodipine, shifting the IC50 value by a factor of ∼10 (Xu and Lipscombe, 2001).
This difference in nimodipine sensitivity was then used to determine whether CaV1.2 and CaV1.3 channels were biophysically distinguishable. To determine the voltage dependence of CaV1.2 channels, the currents that were resistant to blockade by 200 nm nimodipine were subtracted from control currents. This should yield currents that were solely attributable to CaV1.2 channels. CaV1.2 channel current-voltage plots derived from wild-type neurons peaked near 0 mV (Fig. 3C, black trace); currents isolated from CaV1.3 knock-out neurons were virtually identical (Fig. 3C, gray trace). CaV1.3 channel currents were operationally defined as those blocked by 10 μm but not 1 μm nimodipine. In contrast to CaV1.2 channels, these channels activated at relatively hyperpolarized membrane potentials (Fig. 3D). Permeability estimates were calculated to give a better picture of how channel open probability changed with voltage (Bargas et al., 1994). These estimates were readily fit with a single-site Boltzmann function (Fig. 3E). Based on this analysis, the half-activation voltage of CaV1.3 channels was ∼15 mV more negative than that for CaV1.2 channels. The data from these experiments are summarized in a nonparametric format in Figure 3F. CaV1.2 channel currents had half-activation voltages of approximately -5 mV in wild-type neurons as well as in neurons from CaV1.3 and CaV2.3 knock-outs. In contrast, CaV1.3 channel currents had half-activation voltages between -20 and -25 mV in neurons from wild-type and CaV2.3 knock-out mice, significantly more negative than CaV1.2 channels (n = 5; p < 0.05, Mann-Whitney). These results show that medium spiny neurons coexpress functional CaV1.2 and CaV1.3 L-type channels and that these channels differ in both their dihydropyridine and voltage sensitivity. We next asked whether these two channels were differentially modulated by GPCRs.
D2 dopaminergic and M1 muscarinic receptors selectively modulate CaV1.3 channels
To determine whether CaV1.2 or CaV1.3 channels were targeted by D2 receptor activation, recordings were made from medium spiny neurons in which the CaV1.3 subunit had been genetically deleted (Platzer et al., 2000). To our surprise, D2 receptor agonists had no effect on tail currents in these neurons (Fig. 4A). This was not a consequence of D2 receptor dysfunction, because the modulation of CaV2.1/2.2 channel currents evoked by the step depolarization, which relies on a distinct, membrane-delimited signaling cascade, remained. To verify that recordings were made from the subset of neurons expressing D2 receptors, scRT-PCR profiling was performed. In a sample of six CaV1.3-deficient neurons expressing ENK mRNA, a reliable marker of D2 receptor-expressing neurons (Gerfen, 1992; Surmeier et al., 1996), none exhibited a discernible reduction in tail amplitude after NPA application (Fig. 4C,D). In contrast, the modulation of the CaV2.1/2.2 (N, P/Q) currents evoked by the depolarizing step was not significantly altered (n = 4; median suppression control, 16%; CaV1.3 knock-out, 20%; p > 0.05, Kruskal-Wallis) (Hernandez-Lopez et al., 2000). To provide an additional test of the D2 receptor selectivity, ramp currents in wild-type neurons were examined. Again, concentrations of nimodipine (1 μm) that fully block CaV1.2 channels failed to significantly reduce the effects of NPA, whereas concentrations of nimodipine (10 μm) that also blocked CaV1.3 channels occluded the modulation, leaving the residual CaV2.1/2.2 channel current modulation (n = 4) (Fig. 4B).
Next, the selectivity of M1 muscarinic receptor linkage was examined. As with the D2 receptor modulation, the effect of muscarine on the slow, L-type tail currents was abolished in neurons lacking CaV1.3 channels (Fig. 4C). This result suggests that activation of the ubiquitously expressed M1 receptor, which is responsible for the suppression of L-type channel currents, selectively targets CaV1.3 channels. Despite the loss of this modulation, the M4 muscarinic receptor suppression of CaV2.1/2.2 Ca2+ channels (Howe and Surmeier, 1995; Yan et al., 2001) was left intact in the CaV1.3-deficient neurons. This can be seen in the response to the step depolarization (n = 5; median suppression control, 19%; CaV1.3 knock-outs, 17%; p > 0.05, Kruskal-Wallis). To provide another test of the selectivity of the M1 receptor modulation, ramp currents were recorded in wild-type neurons as described above. Concentrations of nimodipine (1 μm) that should fully block CaV1.2 channels failed to substantially alter the magnitude of the muscarinic modulation, whereas application of CaV1.3-inclusive concentrations of nimodipine (10 μm) occluded the effects of muscarine (Fig. 4D) (n = 5), leaving the residual M4 modulation of CaV2.1/2.2 channels. Last, the voltage dependence of the current modulated by muscarine was estimated by subtracting the currents unaffected by muscarine from control currents. This analysis yielded currents with a half-activation voltage indistinguishable from that of CaV1.3 channels isolated pharmacologically (n = 6) (data not shown) (see supplemental figure, available at www.jneurosci.org as supplemental material).
Medium spiny neurons coexpress a long CaV1.3 channel splice variant and interacting Shank isoforms
Zhang et al. (2005) have shown that the scaffolding protein Shank selectively interacts with SH3 and PDZ binding domains in the C-terminal region of a long CaV1.3 splice variant (CaV1.3a). An interaction with Shank could bring CaV1.3 channels into physical proximity to D2/M1 receptors and elements of the signaling cascade they engage (Sheng and Kim, 2000). As a first step toward testing this hypothesis, whole striatal tissue was profiled for the presence of the two C-terminal splice variants of CaV1.3 expressed in the brain (Safa et al., 2001). Serial dilution RT-PCR revealed that both long (CaV1.3a) and short (CaV1.3b) variants were present but that CaV1.3a was present at apparently higher levels (Fig. 5A). Profiling of individual medium spiny neurons confirmed this analysis, showing that CaV1.3a mRNA was easily detected in both ENK- and SP-expressing subtypes of medium spiny neurons (14/15) (Fig. 5B). In contrast, CaV1.3b was infrequently detected (2/15). This was not a limitation posed by the single-cell analysis, because CaV1.3b was readily detected in dopaminergic neurons of the substantia nigra (n = 5) (Fig. 5B, right panel).
The C-terminal region of CaV1.3a subunits binds preferentially with two of the three Shank isoforms found in the brain (Shank3/Shank1) (Zhang et al., 2005). If targeting of CaV1.3 channels was dependent on Shank in medium spiny neurons, they should express one or both of these isoforms. To determine whether this was the case, striatal tissue was assayed for Shank mRNA. At the tissue level, all three Shank isoforms were readily detected at nominally the same level of abundance (Fig. 5C). However, in medium spiny neurons, the Shank isoform with the highest affinity for the CaV1.3a subunit, Shank3, appeared to be the most abundant, being detected in seven of eight cells. Shank1, which also has a high affinity for the CaV1.3a terminal region, also was readily detected (5/8), whereas Shank2 was not detected at all (0/8) (Fig. 5D). This pattern was cell-type specific, because examination of cholinergic interneurons failed to detect Shank2 or Shank3 but readily detected Shank1 (Fig. 5E)(n = 6).
Next, to determine whether CaV1.3a channels and Shank were in close physical proximity, striatal tissue sections were studied using immunocytochemical approaches. Because both CaV1.3a and Shank antibodies were raised in the same host, colabeling of the same section was not possible. Because PSD-95 is colocalized with Shank at excitatory synapses (Sheng and Kim, 2000), it was used as a surrogate marker. As expected, PSD-95 immunoreactivity in striatal tissue had a punctate distribution (Fig. 6A, top left), reflecting the localization of excitatory synapses predominantly on dendritic spine heads (Kennedy, 1998; Bolam et al., 2000). Shank immunoreactivity was punctate as well but had an additional somatic component that presumably reflected cytoplasmic protein (Fig. 6A, top center). Color overlays from 500 nm optical sections provided evidence that these two proteins had a significant level of colocalization (Fig. 6A, top right). However, the images were not perfectly aligned, as one might expect if the proteins were slightly displaced from one another near synapses. To quantify the relationship, spatial cross-correlations were computed on aligned image fields after standardizing the optical image intensity (Li et al., 2004). To reduce spurious correlations attributable to somatic labeling, fields in which there were few visible somata were chosen for analysis. This analysis consistently revealed a strong spatial correlation between PSD-95 and Shank labeling that was strongest with no image displacement and then sharply declined within ∼1 μm, approximately the dimension of a spine (Fig. 6B). In a sample of five sections, the median peak cross-correlation was 0.35 (range, 0.71-0.23). The secondary slower decline in the cross-correlation in Figure 6B reflected somatic labeling, because it was absent in correlations from subfields lacking any obvious somatic profiles.
CaV1.3a immunolabeling was strikingly similar to that for PSD-95, being distinctly punctate (Fig. 6A, bottom center). Overlays of regions processed for both PSD-95 and CaV1.3a revealed a significant degree of colocalization (Fig. 6A, bottom right). To quantify the relationship between the two markers, cross-correlation analysis was performed as for Shank and PSD-95. Analysis of five different sections consistently found a strong relationship between PSD-95 and CaV1.3 labeling with peak values between 0.28 and 0.41 (Fig. 6C) (n = 5). As with Shank, the peak cross-correlation was found near zero displacement and then fell rapidly to baseline within the expected dimensions of a spine. These data suggest that PSD-95, Shank, and CaV1.3a subunits are colocalized in medium spiny neurons at excitatory synapses, which are predominantly found on spines.
PSD-95, Shank, and CaV1.3a channels appear to be in close physical proximity in medium spiny neurons, but are they bound to one another? To test this possibility, coimmunoprecipitation experiments were performed. Extracts of rat brain synaptosomes were prepared, and protein was immunoprecipitated using an anti-CaV1.3a rabbit pAb. A Western blot of this protein was then probed with guinea pig anti-Shank pAb, revealing Shank (Fig. 6D). Shank immunoreactivity was lost by preincubation with CaV1.3a C-terminal peptide, demonstrating the specificity of the precipitation.
Dialysis with a peptide mime of the CaV1.3a PDZ binding domain disrupts the D2 and M1 receptor modulation
If the interaction between Shank and CaV1.3a subunits is critical to the selective modulation by D2 and M1 receptors, competitive inhibitors of this binding should disrupt the modulation. Zhang et al. (2005) have shown that Shank1/3 binds to a CaV1.3a “ITTL” PDZ motif but does not bind strongly to a CaV1.2 “VSNL” PDZ motif. Therefore, peptides containing the ITTL motif should act as efficient competitive inhibitors of the Shank-CaV1.3a interaction, whereas peptides containing the VSNL motifs should not. To provide a functional test of this hypothesis, we turned back to acutely isolated medium spiny neurons in which the modulation of L-type Ca2+ channels could be examined. Although this preparation affords excellent experimental control, the principle regions of excitatory synaptic contact are removed, eliminating much of the structure hypothetically responsible for the selective modulation of CaV1.3a channels. To determine whether some of this organization remained after dissociation, neurons were probed with Shank and CaV1.3 antibodies. As shown in Figure 7A, despite the dendritic truncation, punctate labeling for Shank and CaV1.3a protein was found in the proximal dendrites of dissociated neurons. It is worth noting that D2 or M1 receptor agonists failed to modulate L-type currents in neurons lacking any visible dendrite (n > 20).
Having demonstrated that a portion of the scaffolding remained in the dissociated neurons, peptide mimes of the PDZ binding domains were introduced through the patch pipette during whole-cell recordings. The ability of D2 receptor agonists to modulate L-type Ca2+ channel currents was unaffected by dialysis with the VSNL peptide but was disrupted by dialysis with an ITTL peptide (Fig. 7C-E). The modulation of the step currents, which was mostly attributable to CaV2.1/2.2 channel currents, was not significantly altered by either peptide, arguing that the receptor signaling mechanisms were intact [median control suppression, 16% (n = 4); +VSNL, 21% (n = 4); +ITTL, 10% (n = 5); p > 0.05, Kruskal-Wallis]. Similarly, M1 receptor modulation of L-type Ca2+ channel currents was not affected by dialysis with the VSNL peptide (Fig. 7F,H) (n = 4). In contrast, dialysis with an ITTL peptide dramatically reduced the ability of M1 receptors to modulate L-type currents (Fig. 7G,H)(n = 5; p < 0.05, Mann-Whitney). As with the dopaminergic modulation, the modulation of the currents evoked by the step, which were attributable to primarily the M4 receptor modulation of Ca 2.1/2.2 Ca2+V channels, was not significantly affected by the VSNL or ITTL peptides [median control suppression, 19% (n = 5); +VSNL, 21% (n = 5); +ITTL, 10% (n = 5); p > 0.05, Kruskal-Wallis]. Moreover, neither ITTL (n = 8) nor VSNL (n = 4) peptide had any significant effect on current density or voltage dependence (data not shown) (n = 5; p > 0.05, Kruskal-Wallis). The failure of the VSNL peptide to alter the D2 and M1 receptor modulations of tail currents was not a consequence of its inability to compete for the CaV1.2 PDZ domain, because dialysis of the peptide into cortical pyramidal neurons effectively blocked the 5-HT2 receptor modulation of CaV1.2 channel currents (n = 5; control suppression, 26%; +VSNL, 10%; p < 0.05, Kruskal-Wallis) (Day et al., 2002).
The data presented thus far suggest that the ability of D2 and M1 receptors to suppress CaV1.3 Ca2+ channel currents depends on channel binding of Shank3/Shank1 proteins. However, why is this interaction important? Our previous studies have shown that the modulation of L-type channels depends on mobilization of Ca2+ from IP3R-controlled intracellular stores (Howe and Surmeier, 1995; Hernandez-Lopez et al., 2000). Shank may serve to bring CaV1.3a channels close to IP3Rs and these release sites. One way in which this could be accomplished is through Homer adaptor proteins, which are known to bind to both IP3Rs and Shank through Ena/VASP homology 1 (EVH1) domains (Xiao et al., 2000). If this were the case, medium spiny neurons should express Homer isoforms possessing coiled-coil (CC) regions (Homer 1b-d, Homer 2-3) that allow self-association and assembly of the Shank-Homer-IP3R scaffold. Indeed, in agreement with the predictions of in situ hybridization work (Shiraishi et al., 2004), scRT-PCR profiling of the GABAergic medium spiny neurons (n = 17) revealed expression of Homer 1b/c (16/17), Homer 1d (12/17), and Homer 2 (11/17) but lower levels of the Homer splice variant lacking these domains (Homer1a, 9/17) and Homer 3 (0/17) (Fig. 8A). To provide a functional test for the involvement of Homer proteins, neurons were dialyzed with a peptide targeting the EVH domain that should competitively inhibit Shank and IP3R binding. Doing so suppressed the M1 receptor modulation of CaV1.3 channel currents (n = 6), but dialysis of a similar peptide lacking the EVH binding domain (HBP-KE) failed to alter the modulation (n = 3) (Fig. 8B). These data are consistent with the hypothesis that Homer proteins are part of the Shank scaffold necessary for selective modulation of CaV1.3 channels.
D2 receptor suppression of CaV1.3 Ca2+ channel currents dramatically reduce upstate transitions
Shank is thought to act as a scaffold at glutamatergic synapses (Sheng, 2001). In medium spiny neurons, the principal glutamatergic input arises from corticostriatal axons that terminate on dendritic spines (Bolam et al., 2000). In vivo, convergent activation of this input induces depolarized episodes called upstate, during which neurons can spike (Wilson, 1993). Recently, an in vitro slice model of these state transitions has been described that utilizes repetitive cortical stimulation or low micromolar concentrations of NMDA that induce tonic firing in corticostriatal pyramidal neurons (Vergara et al., 2003). In this model, upstates are dependent not only on synaptic input but postsynaptic L-type channels that help create dendritic plateau potentials, maintaining the upstate. Positioning of low-threshold CaV1.3a channels at dendritic spines would be one way in which this glutamatergic receptor driven dendritic electrogenesis could occur. To test for the involvement of CaV1.3 channels in this phenomenon, upstates were examined in neurons recorded in slices from wild-type and CaV1.3 knock-out mice. Because medium spiny neurons express very low or undetectable levels of CaV1.3b mRNA (see above), the deletion of the CaV1.3 gene should mimic the selective deletion of CaV1.3a subunits. In wild-type neurons, spontaneous upstates were seen in most of the neurons (7 of 10) recorded after bath application of NMDA (5 μm) (Fig. 9A). In contrast, upstates were not seen in neurons from the CaV1.3 knock-out under identical recording conditions (n = 5) (Fig. 9B). This deficit did not appear to arise from an alteration in synaptic function, because local electrical stimulation evoked robust glutamatergic EPSPs, and miniature EPSPs were, if anything, elevated in frequency in CaV1.3-deficient neurons (data not shown). Additional evidence that these neurons were otherwise viable came from experiments where bath application of BayK 8644 (1 μm) enabled occasional upstate transitions in these neurons (n = 3), presumably by shifting the voltage dependence of the remaining CaV1.2 channels into the range of normal CaV1.3 channel gating (Bargas et al., 1994) (Fig. 9B, inset).
To determine whether GPCR suppression of CaV1.3 channels could transiently suppress upstate generation, medium spiny neurons were recorded as described above, and the D2 agonist NPA was applied locally through a puffer pipette. M1 receptor agonists were not examined because they also modulate K+ channels involved in state transitions (W. Shen and Surmeier, unpublished observations). Before NPA application, in the presence of NMDA (5 μm), neurons spontaneously oscillated between upstates and downstates, yielding a distinctly bimodal membrane potential distribution (n = 6) (Fig. 9C,D). In four of six neurons, the application of NPA (10 μm) reversibly suppressed the generation of upstates causing the membrane potential distribution to become unimodal (Fig. 9C,D); in the other two neurons, NPA had no discernible effect. Although there are undoubtedly other ion channels participating in the state transitions that are potentially targets of modulation, these data are consistent with the hypothesis that modulation of synaptically located CaV1.3a channels is an important component of the cellular response to D2 receptor activation.
M1 and D2 receptor signaling cascades selectively modulate CaV1.3 L-type Ca2+ channels
Previous studies have shown that D2 dopaminergic and M1 muscarinic receptors suppress L-type Ca2+ currents in striatal medium spiny neurons (Howe and Surmeier, 1995; Hernandez-Lopez et al., 2000). Although robust, the suppression of L-type Ca2+ currents by these GPCR cascades is invariably incomplete, suggesting that some channels are targeted for modulation and others are not. This differential vulnerability appears to turn in large measure on molecular heterogeneity of the L-type channels themselves. Medium spiny neurons express two major classes of L-type Ca2+ channels: one possessing a pore-forming CaV1.2 subunit and the other with a CaV1.3 subunit. Four lines of evidence support this conclusion. First, scRT-PCR profiling revealed that both major classes of medium spiny neurons coexpress CaV1.2 and CaV1.3 subunit mRNAs. Second, L-type Ca2+ channels in medium spiny neurons were heterogeneous in their affinity for dihydropyridines; one component was blocked with high affinity by the dihydropyridine L-type channel antagonist nimodipine (IC50, ∼200 nm), whereas another component was blocked only at higher antagonist concentrations (IC50, ∼2 μm). This affinity difference is just what is expected of CaV1.2 and CaV1.3 channels from work in heterologous expression systems (Safa et al., 2001; Xu and Lipscombe, 2001). More compelling perhaps was the loss of the lower affinity component in neurons from CaV1.3 knock-out mice and its preservation in CaV2.3 knock-outs. Third, as in heterologous expression systems (Xu and Lipscombe, 2001), Ca2+ channels with a low dihydropyridine affinity (CaV1.3) opened at more negative membrane potentials than did high affinity CaV1.2 channels. Nominal CaV1.3 channel half-activation voltages were 10-15 mV more negative than those for CaV1.2 channels. Fourth, immunocytochemical analysis revealed the presence of CaV1.3 protein in medium spiny neurons, complementing previous work showing CaV1.2 expression (Hell et al., 1993b).
Three lines of evidence argue that D2 and M1 receptor signaling cascades differentially regulate these two channel types. The first comes from the failure of either GPCR to modulate BayK 8644 tail currents in neurons where functional CaV1.3 subunits had been genetically deleted. The second comes from the ability of high, but not low, CaV1.2-selective concentrations of dihydropyridine antagonists to occlude the modulation. The third comes from the similarity in gating voltage dependence of the modulated channels and CaV1.3 Ca2+ channels.
The selective modulation appears to be dependent on Shank3/Shank1 binding to CaV1.3a subunits
One way in which this kind of signaling specificity could be achieved is by bringing the receptors, channels, and signaling elements to the same subcellular locale. The best-characterized signaling complex of this sort is the postsynaptic density at glutamatergic synapses, where scaffolding or anchoring proteins organize and concentrate transduction proteins (Kennedy, 1998; Sheng and Kim, 2000). Shank has been referred to as a “master” scaffolding protein at these sites because of its prominence and capacity to link a broad array of participating proteins. Zhang et al. (2005) have expanded this list to include L-type Ca2+ channels containing CaV1.3 subunits, showing that Shank proteins selectively bind to a long C-terminal splice variant of this subunit via SH3 and PDZ domains. This interaction promotes synaptic targeting of the channel in hippocampal neurons.
Targeting of this type also appears to occur in striatal medium spiny neurons. In addition to the long splice variant of the CaV1.3 subunit (CaV1.3a), these neurons expressed mRNA for interacting Shank isoforms (Shank3, Shank1). Confocal analysis of immunocytochemical labeling showed that CaV1.3a and Shank proteins colocalized with the synaptic protein PSD-95 within a spatial domain expected of a spine, the predominant site of glutamatergic synapses in medium spiny neurons (Bolam et al., 2000; Sheng, 2001). A direct physical interaction was supported by the coimmunoprecipitation of CaV1.3a and Shank proteins. Moreover, recent Ca2+ imaging studies have shown that L-type channels contribute to intraspine Ca2+ dynamics in medium spiny neurons (Carter and Sabatini, 2004), complementing work showing a similar localization elsewhere (Simon et al., 2003; Hoogland and Saggau, 2004).
The functional association between Shank and CaV1.3a Ca2+ channels also appears to be necessary for GPCR regulation. Dialysis with a competitive inhibitor of the CaV1.3a-Shank PDZ domain (ITTL) dramatically attenuated the ability of D2 and M1 GPCRs to suppress CaV1.3 L-type currents, whereas dialysis with a competitive inhibitor of the CaV1.2 PDZ binding domain (VSNL) was without effect on the modulation. Because PDZ domains are commonly used to link proteins in neurons, it is possible that effect of the ITTL peptide is attributable to disruption of some other PDZ-dependent, protein-protein interaction. However, the failure of the closely related VSNL PDZ peptide to attenuate the CaV1.3 channel modulation argues against this possibility, especially in light of the ability of this construct to disrupt the modulation of CaV1.2 channels in cortical pyramidal neurons. Nevertheless, additional study on this point is warranted.
Additional evidence for the importance of Shank in the regulation of CaV1.3 channels comes from the ability of peptides targeting the EVH1 binding site of Homer to attenuate GPCR effects. This site enables Homer isoforms to bind to the proline-rich region of Shank (Sheng and Kim, 2000). Dimerization of Homer isoforms having CC domains creates a way in which IP3-regulated intracellular Ca2+ stores could be brought into close physical proximity to Shank and CaV1.3 channels (Xiao et al., 2000). Previous studies have strongly implicated these stores in both D2 and M1 receptor modulations of L-type channels (Howe and Surmeier, 1995; Hernandez-Lopez et al., 2000). Because Ca2+ is rapidly buffered in dendrites (Sabatini et al., 2002), bringing the IP3R release sites close to the channels could be critical to the efficient activation of the remaining elements in the Ca2+-dependent modulatory cascade. Calcineurin, which is thought to be the final effector of the D2/M1 receptor cascade, is anchored by A-kinase anchoring protein 79/150, a known participant in PSD-95 scaffolds (Altier et al., 2002; Gomez et al., 2002). The recent discovery of that calcineurin-mediated modulation of CaV1.3 channels is controlled by the calmodulin binding phosphoprotein regulator of calcium/calmodulin-dependent signaling (Rakhilin et al., 2004) creates a rich intraspine regulatory network. Although GPCR positioning is less well characterized, there is growing evidence that they also can interact with synaptic scaffolding proteins (Davare et al., 2001; Kreienkamp, 2002). Such an interaction is consistent with ultrastructural studies showing that both D2 and M1 receptors are enriched at synaptic sites in medium spiny neurons (Hersch et al., 1994, 1995; Bolam et al., 2000).
This picture complements the growing recognition that CaV1.2 channels also participate in neuronal synaptic signaling complexes. In hippocampal pyramidal neurons, β2 adrenergic receptors assemble with CaV1.2 channels, protein kinase A, protein phosphatase 2A, and adenylyl cyclase in perisomatic and dendritic sites (Davare et al., 2001). Although the mechanism remains to be determined, this close association enables β2 receptor agonists to rapidly enhance CaV1.2 channel currents in a spatially restricted manner. Scaffolding that is dependent on neuronal interleukin-16 PDZ domains could be a factor in the GPCR association, as it is in CaV1.2 channel-dependent CREB phosphorylation (Weick et al., 2003). Given their biophysical differences, it is tempting to speculate that coexpression of CaV1.2 and CaV1.3 channels enables neurons to sense and signal over a much broader activity range than would be possible with either channel alone.
GPCR-Shank-CaV1.3 channel complex regulates glutamatergic receptor driven upstates in medium spiny neurons
The recruitment of CaV1.3 channels to synaptic signaling complexes by Shank proteins provides a framework in which we can begin to understand activity-dependent plasticity in medium spiny neurons. In vivo, glutamatergic synaptic input promotes the transition from a hyperpolarized downstate to a depolarized upstate (Wilson, 1993). Although the upstate transitions depend on glutamatergic receptors for their initiation, postsynaptic voltage-dependent ion channels clearly shape their duration and potential envelope. L-type channels appear to be particularly important in this phenomena, because dihydropyridine antagonists suppress upstate transitions and agonists promote them (Vergara et al., 2003). Our data extend these findings by showing the loss of upstate transitions in neurons from CaV1.3 knock-out mice or in neurons after activation of GPCRs that suppress CaV1.3 channel currents. This disruption suggests that there is a synergy between glutamatergic receptors and CaV1.3 channels in the generation of upstates that is regulated by D2 dopaminergic receptors.
The strategic positioning of CaV1.3 channels afforded by Shank also may be a critical factor in determining how afferent synaptic activity is translated into long-term alterations in neuronal function. For example, in striatal neurons, long-term synaptic depression (LTD) is dependent on opening of L-type Ca2+ channels (Calabresi et al., 1994). Synaptic scaffolding of CaV1.3 channels may be critical to LTD induction. Activity-dependent alterations in the phosphorylation state of transcription factors that signal synaptic events to the nucleus may also be depend on “privileged positioning” of L-type Ca2+ channels (Cepeda et al., 1998; Rajadhyaksha et al., 1999; Zhang et al., 2005). The inclusion of D2 and M1 GPCRs in this synaptic complex creates a mechanism by which the two most clinically important striatal modulators (dopamine and acetylcholine) could regulate this plasticity.
This work was supported by National Institutes of Health (NIH) Grants NS34696 and DA12958 and a grant from the Picower Foundation to D.J.S. and by grants from the Robert A. Welch Foundation and NIH (NS39552) to I.B. We thank Dr. Richard Miller for providing CaV2.3 knock-out mice and insight into channel biology. We also acknowledge the expert technical assistance of Karen Burnell and Yu Chen.
Correspondence should be addressed to D. James Surmeier, Department of Physiology, Feinberg School of Medicine, Northwestern University, 303 East Chicago Avenue, Chicago, IL 60611. E-mail:.
Copyright © 2005 Society for Neuroscience 0270-6474/05/251050-13$15.00/0