Differential Interactions of the C terminus and the Cytoplasmic I-II Loop of Neuronal Ca2+ Channels with G-protein α and βγ Subunits

The present study was designed to obtain evidence for direct interactions of G-protein α (Gα) and βγ subunits (Gβγ) with N- (α1B) and P/Q-type (α1A) Ca2+ channels, using synthetic peptides and fusion proteins derived from loop 1 (cytoplasmic loop between repeat I and II) and the C terminus of these channels. For N-type, prepulse facilitation as mediated by Gβγ was impaired when a synthetic loop 1 peptide was applied intracellularly. Receptor agonist-induced inhibition of N-type as mediated by Gα was also impaired by the loop 1 peptide but only when applied in combination with a C-terminal peptide. For P/Q-type channels, by contrast, the Gα-mediated inhibition was diminished by application of a C-terminal peptide alone. Moreover, in vitro binding analysis for N- and P/Q-type channels revealed direct interaction of Gα with C-terminal fusion proteins as well as direct interaction of Gβγ with loop 1 fusion proteins. These findings define loop 1 of N- and P/Q-type Ca2+ channels as an interaction site for Gβγ and the C termini for Gα.

The present study was designed to obtain evidence for direct interactions of G-protein ␣ (G␣) and ␤␥ subunits (G␤␥) with N-(␣ 1B ) and P/Q-type (␣ 1A ) Ca 2؉ channels, using synthetic peptides and fusion proteins derived from loop 1 (cytoplasmic loop between repeat I and II) and the C terminus of these channels. For N-type, prepulse facilitation as mediated by G␤␥ was impaired when a synthetic loop 1 peptide was applied intracellularly. Receptor agonist-induced inhibition of N-type as mediated by G␣ was also impaired by the loop 1 peptide but only when applied in combination with a C-terminal peptide. For P/Q-type channels, by contrast, the G␣-mediated inhibition was diminished by application of a C-terminal peptide alone. Moreover, in vitro binding analysis for N-and P/Q-type channels revealed direct interaction of G␣ with C-terminal fusion proteins as well as direct interaction of G␤␥ with loop 1 fusion proteins. These findings define loop 1 of N-and P/Q-type Ca 2؉ channels as an interaction site for G␤␥ and the C termini for G␣.
High voltage-activated (HVA) 1 Ca 2ϩ channels are negatively regulated by guanine nucleotide-binding regulatory proteins (G-proteins) in various neuronal preparations, including neuroblastoma ϫ glioma hybrid NG108-15 cells (1,2), dorsal root ganglion neurons (3,4), sympathetic neurons (5), and rat pituitary GH 3 cells (6). This response appears to be controlled by a membrane-delimited mechanism via pertussis toxin (PTX)sensitive G-proteins, in which the G o ␣ subunit has been shown to mediate an inhibitory signal to HVA Ca 2ϩ channels. The primary structures of multiple subtypes of G-protein ␣ subunits (G␣) including G o ␣ have been deduced by molecular cloning and sequencing of their cDNAs. These studies revealed very similar but distinct amino acid sequences for each subtype cloned (7). It remains to be seen, however, which subtypes of G␣ preferentially interact with HVA Ca 2ϩ channels such as Nand P/Q-types. In the previous study using mutant and chimeric channels expressed in Xenopus oocytes (8), our results provided evidence that the cytoplasmic I-II loop (referred to as "loop 1" in the present study) of N-type (␣ 1B ) Ca 2ϩ channels is a regulatory site for the G-protein ␤␥ dimer (G␤␥) and the C termini of P/Q-(␣ 1A ) and N-type Ca 2ϩ channels for G␣. However, this does not answer the question as to whether G␣, as well as G␤␥ (9,10), directly interact with these Ca 2ϩ channels.
To address these issues, we have expressed ␣ 1B and ␣ 1A HVA Ca 2ϩ channels in Xenopus oocytes, in which the effects of intracellularly applied loop 1 and C-terminal peptides derived from ␣ 1B and ␣ 1A were investigated. Furthermore, a direct association of G␣ and G␤␥ with these Ca 2ϩ channels was determined by in vitro binding using glutathione S-transferase (GST) proteins fused with the loop 1 and the C-terminal segments of ␣ 1B or ␣ 1A . These results, taken together with the findings of the previous study (8), define the interaction sites of G␣ and G␤␥ within Ca 2ϩ channels.
In some experiments, multiple single channel events in a membrane patch were recorded from oocytes as an ensemble average (or "pseudomacroscopic"), using an EPC-7 patch-clamp amplifier (List Electronics, Darmstadt, Germany). The procedure was similar to that described elsewhere (14). Cell-attached membrane patches were obtained using fire-polished borosilicate pipettes (Narishige, GD-1.5, Japan) having a resistance of 2-4 megohms (tip diameter ϳ1 m) when filled with the pipette solution. The pipette solution contained 110 mM BaCl 2 and 10 mM HEPES (pH 7.4 with tetraethylammonium-OH). Oocytes that were implanted with ␣ 1B , ␣ 2 , ␤ 1a , DOR, and G i3 ␣, and then injected with the antisense AGO, were bathed in a depolarizing solution consisting of 90 mM K ϩ , 10 mM Na ϩ , 1 mM Mg 2ϩ , 1 mM Ca 2ϩ , 104 mM Cl Ϫ , and 5 mM HEPES (pH 7.5 with KOH). Under these experimental conditions, the average membrane potential was Ϫ2 Ϯ 4 mV (n ϭ 5). To apply Leu-enkephalin (Leu-EK) inside the patch electrode, a thin polyethylene tube having a tapered tip of about 100 m in diameter was inserted into the glass pipette and used as an inlet tubing (18). Through this thin tubing, near (about 0.5 mm) the tip of the patch pipette, the pipette solution containing 50 M Leu-EK was applied slowly using a syringe microinjector. Because the volume of injection was small (0.5-1.0 l, about 1/50 volume of the solution in the pipette), no waste chamber (reservoir) was attached to the patch electrode. A negative pressure of about 30 -40 cm H 2 O was applied to the patch electrode to minimize the mechanical effects of the pressure injection. Data acquisition and analysis were done on a computer using the software, DAAD system (19).
Unless otherwise stated, statistical data were represented by the mean and S.E.
Expression of Exogenous G␣ in NG108-15 Cells-By using a HindIII linker, the G␣ cDNA fragment (15) containing the entire protein coding segment of G i1 ␣, G i2 ␣, G i3 ␣, or G o1 ␣ was inserted into the HindIII site of the pKGS␣N (20), to yield expression vector, pKGi1␣, pKGi2␣, pKGi3␣, or pKGo1␣. NG108-15 cells in culture were transfected with the above plasmids by electroporation (20); the voltage was set at 1.7 kV and the capacitance at 20 microfarads. Transformants were selected in culture medium supplemented with 800 g/ml G418 for 2-3 weeks. One hundred clones of the ϳ1 ϫ 10 7 cells transfected were transformed to G418 resistance. Among these G418-resistant transformants, G␣-transformed NG108-15 clones were selected out by blot hybridization analysis (20), using total cellular RNA prepared from each transformant and each cDNA fragment containing a protein coding segment as a probe. The presence of a hybridizable RNA species with an expected size from the construction was used to measure the transformants of NG108-15 cells by G i1 ␣, G i2 ␣, G i3 ␣, or G o1 ␣ (20).
Muscarinic acetylcholine (ACh) receptors were stimulated by a focal application of 1 mM ACh (3 l). In experiments in which PTX was used, the cells were preincubated with the toxin (500 ng/ml) for 12-14 h prior to the measurements. The electrophysiological experiments were carried out at approximately 30°C. Statistical data were represented by the mean and S.E.
G o ␣ and G␤␥ 2 Bindings-The 1.6-kilobase pair SrfI/SalI fragment (encoding amino acid residues 1912-2339 of the ␣ 1B subunit) and the 1.4-kilobase pair ScaI/BamHI fragment (encoding amino acid residues 1975-2424 of the ␣ 1A subunit) were excised from the plasmids pSPB3 (16) and pSPCBI-2 (14) and fused in-frame to the GST coding sequence in pGEX-2T (Amersham Pharmacia Biotech) to produce GST fusion proteins of C terminus (GST-B3T and GST-B1T), respectively. Similarly, to produce GST fusion proteins of loop 1 (GST-B3L1 or GST-B1L1), the 380-base pair fragment amplified by polymerase chain reaction (PCR), encoding either amino acid residues 361-483 of the ␣ 1B subunit or 365-489 of the ␣ 1A subunit, was fused in-frame. The sets of primers for PCR were GGAGAATTCGTAAGGAGCGCGAGAGA and TCTGTGCCTTCACCATGCGCC for ␣ 1B and GGGGAATTCGCAAA-GAAAGGGAGCGG and AGAAGGCCTGAGTTTTGACCATG for ␣ 1A . The plasmids pSPB3 and pSPCBI-2 were used as a template for PCR. Four kinds of fusion proteins or GST proteins were expressed in Escherichia coli DH5␣, induced with 0.5 mM isopropyl-1-thio-␤-D-galactoside, and prepared according to the procedure described previously (25). The solubilized proteins were incubated with glutathione-Sepharose 4B beads (Amersham Pharmacia Biotech) at 4°C for 5 h and washed five times with 40 volumes of phosphate-buffered saline containing 1% Triton X-100. Approximately 10 g of GST, or GST-fusion proteins, bound to beads were used in each binding assay.
Ten micrograms of purified bovine brain G o ␣ (26) and G␤␥ 2 complex (27) were incubated at 4°C for 12 h with the beads that had been equilibrated with buffer solutions supplemented with 0.1% Lubrol PX and 0.6% sodium cholate, respectively. These buffers contained 20 mM Tris-HCl (pH 8), 0.1 mM EDTA, 0.5 mM dithiothreitol, 20 mM NaPO 4 , and a mixture of protease inhibitors (2.5 g/ml pepstatin, 2 g/ml phenylmethylsulfonyl fluoride, 0.02 mg/ml leupeptin, and 0.5 mM benzamidine). Then, the beads were washed with 40 volumes of the same buffer solutions and denatured by heating to 90°C for 3 min in 200 l of sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) loading buffer. Proteins (70 l) were separated by SDS-PAGE and transferred to Immobilon membrane (Millipore) for Western blotting and detection by the ECL system (Amersham Pharmacia Biotech) as described previously (28). The antibody, G o AB, raised against the C-terminal decapeptide of G o ␣ (FG o ) (29) was kindly provided by Dr. Tatsuya Haga, and the antibodies, K-20 and M-14, against G o ␣ and G␤ 1 , respectively, were purchased from Santa Cruz. The first antibody when used was diluted appropriately to reduce nonspecific reactivity, and the second antibody (biotinylated anti-rabbit Ig, Amersham Pharmacia Biotech) and the peroxidase-streptavidin (Vector) were both diluted at 1:2000. Each incubation was for 1 h at room temperature. The decapeptide FG o (Sawady, Japan) was synthesized based upon the amino acid sequence corresponding to the amino acid residues 345-354 of G o ␣ (29). The peptide antigens sc-387P (for K-20) and sc-261P (for M-14) were also purchased from Santa Cruz. The GST fusion proteins (GST-B3T, GST-B1T, GST-B3L1, and GST-B1L1) and GST proteins were tested more than three times for their ability to bind G o ␣ and G␤␥ 2 . There was no essential difference between G o AB and K-20 in detecting specific bindings of G o ␣ to the fusion proteins.

Effects of G␣ on the N-and P/Q-type Ca 2ϩ
Channels-In the previous study (8), N-type (␣ 1B ), P/Q-type (␣ 1A ), and L-type (␣ 1C ) Ca 2ϩ channels were shown to be functionally expressed in Xenopus oocytes, and ␣ 1B and ␣ 1A channels were negatively regulated by G i3 ␣ and G␤ 1 ␥ 2 . To determine further which G␣ isoforms regulate HVA Ca 2ϩ channels, either G␣ cRNA (for six different isoforms: G i1 ␣, G i2 ␣, G i3 ␣, G o1 ␣, G z ␣, and G s ␣) or G␤ 1 plus G␥ 2 cRNAs were injected into oocytes in combination with the Ca 2ϩ channel ␣ 1 (␣ 1B or ␣ 1A ), ␣ 2 and ␤ 1a subunits cRNAs and receptor (DOR or ␤ 2 AR) cRNA. When G s ␣ was expressed, ␤ 2 AR was co-expressed, and the receptor was stimulated by 1 M isoproterenol, an agonist of the ␤-adrenergic receptor.
Agonist-induced inhibitions of ␣ 1B channels were less pronounced in oocytes co-injected with G i1 ␣, G i2 ␣, or G␤ 1 ␥ 2 cRNA, as compared with control oocytes injected with Ca 2ϩ channel subunits (␣ 1B , ␣ 2 , and ␤ 1a ) and DOR (Fig. 1A, upper). Also, Leu-EK-induced inhibition was diminished in oocytes implanted with G z ␣. Alternatively, the response to Leu-EK of ␣ 1B channels was not affected significantly in oocytes co-expressed with G i3 ␣, G o1 ␣, or G s ␣. As shown in Fig. 1A (upper), PTX (200 ng/ml) blocked the agonist-induced inhibition of ␣ 1B channels with G i3 ␣ but not with G z ␣ nor G s ␣. These observations are consistent with the fact that G i3 ␣ is PTX-sensitive, and G z ␣ and G s ␣ are PTX-insensitive G-proteins (30). Because a PTXinsensitive component of the response was increased in oocytes co-expressed with the PTX-insensitive G z ␣ or G s ␣, it was presumed that a maximal inhibition of the ␣ 1B channel was already attained by endogenous oocyte G-proteins and that introduced G z ␣ or G s ␣ was capable of replacing endogenous G-proteins when exerting current inhibition. Therefore, in order to unmask the effects of exogenous G␣, the antisense oligonucleotide, AGO, directed against mRNAs encoding Xenopus G o ␣ (8), was injected prior to the electrophysiological studies. As expected, Leu-EK-induced inhibition of ␣ 1B channels in control oocytes (without exogenous G␣ subtypes) was reduced by 42.3 Ϯ 7.4% (n ϭ 25) in the presence of antisense oligonucleotide AGO. As a result, the hierarchy of exogenous G␣ subtypes in inhibiting ␣ 1B channels could be more clearly recognized ( Fig. 1A, lower) when compared with antisense-free control experiments (Fig. 1A, upper). Sense oligonucleotide, SGO, to Xenopus G o ␣ had no effects on the inhibition of ␣ 1B channels (n ϭ 6).
After antisense treatment, the agonist-induced inhibition of N-type ␣ 1B currents was further pronounced in oocytes injected with G i3 ␣, G o1 ␣, G z ␣, or G s ␣ cRNA (Fig. 1A, lower). Here, the action of G s ␣ would not be associated with adenylate cyclase, since the injection of 50 nl of 10 M cyclic AMP (n ϭ 3) or 10 units/l catalytic subunit of PKA (n ϭ 3) failed to influence ␣ 1B currents. Moreover, the pretreatment with 100 M H7, a inhibitor of cyclic nucleotide-dependent protein kinase and protein kinase C, did not affect the agonist-induced inhibition of ␣ 1B currents (n ϭ 6). However, in the case of L-type ␣ 1C channels, the catalytic subunit of PKA increased their current amplitude by 82.8 Ϯ 10.9% (n ϭ 6). In oocytes injected with cRNAs for Ca 2ϩ channel subunits (␣ 1B , ␣ 2 , and ␤ 1a ) and ␤ 2 AR, isoproterenol-induced inhibition of ␣ 1B channels was 17.5 Ϯ 1.2% (n ϭ 45).
To exclude further the possibility that such diffusible second messengers might be involved in the effects of G-proteins in potentiating the agonist-induced inhibition of N-type ␣ 1B channels, cell-attached patch recordings were performed (see "Experimental Procedures"). In oocytes implanted with N-type Ca 2ϩ channel subunits, DOR and G i3 ␣, multiple single channel currents through a membrane patch were suppressed by Leu-EK applied to the patch, but the application to the rest of the cell membrane was ineffective (n ϭ 4). The extent of inhibition was 48.0 Ϯ 8.0% (n ϭ 4), which was comparable to that observed for the whole cell currents of oocytes (Fig. 1A, lower). Thus, Leu-EK has to be applied directly to the recording membrane patch to induce current inhibition, indicating that the Leu-EK-induced inhibition of ␣ 1B channels associated with G i3 ␣ employs a membrane-delimited pathway as predicted in native neurons.
As shown in Fig. 1C, the extent to which Leu-EK inhibited ␣ 1B channels was dependent on the amount of G␣ cRNA injected. The rank order of efficiency among G␣ subtypes examined was G i3 ␣ Ͼ G o1 ␣ Ͼ G z ␣. The agonist-induced current inhibition was considerably reduced when a low concentration (15 ng/l) of G z ␣ cRNA was applied. By contrast, oocytes implanted with G i2 ␣ or G␤ 1 ␥ 2 cRNA showed no effect on the Leu-EK-induced current inhibition, whereas oocytes injected with G i1 ␣ showed an attenuating effect (Fig. 1A, lower).
Unlike N-type ␣ 1B channels (Fig. 1A, upper), P/Q-type ␣ 1A channels revealed more conspicuous intensifications of the agonist-induced inhibition by exogenous G␣, probably a result of weak masking effects of endogenous G-proteins (Fig. 1B, upper). The agonist-induced inhibition of currents was potentiated in those oocytes co-expressed with G i1 ␣, G i2 ␣, G i3 ␣, or G o1 ␣ but not with G z ␣ or G s ␣, regardless of antisense oligonucleotide (AGO) injection (Fig. 1B), although AGO did attenuate the Leu-EK-induced inhibition of ␣ 1A channels by 23.2 Ϯ 8.6% (n ϭ 6), as compared with its effect on ␣ 1B channels (42.3 Ϯ 7.4%, n ϭ 25). In addition, the agonist-induced inhibition of ␣ 1A channels was diminished in PTX-treated oocytes co-expressed with G i3 ␣ (Fig. 1B, upper). In oocytes co-expressed with the PTX-insensitive G s ␣, the blockade of agonist-induced inhibition of ␣ 1A channels by PTX was at the same level as in control oocytes absent of exogenous G␣. This finding is consistent with the observation that the agonist-induced inhibition of ␣ 1A currents was not potentiated in oocytes co-expressed with G s ␣. The inhibition of ␣ 1A channels was not potentiated in oocytes co-expressed with G␤ 1 ␥ 2 , similar to ␣ 1B channels.
By contrast, the agonists never evoked an inhibition of Ltype ␣ 1C currents in oocytes expressed with the Ca 2ϩ channel ␣ 1C , ␣ 2 , and ␤ 1a subunits and the receptor (DOR or ␤ 2 AR) in combination with either of the six different isoforms of G␣ (n ϭ 6 -30) or G␤ 1 ␥ 2 (n ϭ 6), even if the concentration of G i3 ␣ cRNA injected was increased to 150 ng/l (n ϭ 4).
To examine whether specifications of G␣-mediated inhibition of HVA Ca 2ϩ channels are reproducible in neuronal cells, we investigated the mechanism by which native HVA Ca 2ϩ channels are regulated by G␣ in NG108-15 neuroblastoma-glioma hybrid cells. NG108-15 cells express N-and L-type Ca 2ϩ channels (31,32) and DOR (24) and muscarinic ACh receptors (21) which inhibit the Ca 2ϩ channel activity. These native HVA Ca 2ϩ channels were sensitive to 10 M nifedipine (n ϭ 10) or 0.3 M -conotoxin GVIA (-CTx) (n ϭ 6), but not to 0.3 M -agatoxin IVA (n ϭ 3). Puff application of 1 mM ACh inhibited HVA Ca 2ϩ currents in the presence of nifedipine (n ϭ 5). On the other hand, ACh did not inhibit HVA Ca 2ϩ currents in the presence of both nifedipine and -CTx (n ϭ 4). These results indicate that ACh inhibits N-type Ca 2ϩ channel currents in NG108-15 cells. When exogenous G␣ isoforms were stably expressed in NG108-15 cells according to the procedures described previously (20), the ACh-induced inhibition of HVA Ca 2ϩ currents was potentiated in the clones transfected by the exogenous G i3 ␣ and G o1 ␣ but not potentiated by G i1 ␣ or G i2 ␣ (Fig. 1D). Thus, the specificities of G␣ in inhibiting native N-type Ca 2ϩ channels were similar to those observed in Xenopus oocytes. PTX (500 ng/ml, a supramaximal dose) impaired the potentiation by G i3 ␣ and G o1 ␣ of the ACh-induced inhibition of Ca 2ϩ channels (Fig. 1D).
These results indicate that the N-type ␣ 1B channel is negatively regulated by G i3 ␣, G o1 ␣, G z ␣, and G s ␣ and that the P/Q-type ␣ 1A channel is also negatively regulated by G i1 ␣, G i2 ␣, G i3 ␣, and G o1␣ . It is further suggested that the N-and P/Q-type Ca 2ϩ channels are regulated differentially by distinct G␣ subtypes. In order to unmask the effect of endogenous G␣, the antisense oligonucleotide, AGO, was routinely used in the following experiments using Xenopus oocytes (Figs. 2 and 3).
Effects of G␤␥ on the N-and P/Q-type Ca 2ϩ Channels-Leu-EK-induced inhibitions of N-type, ␣ 1B (Fig. 1A), and P/Q-type, ␣ 1A (Fig. 1B), Ca 2ϩ channels were not potentiated when G␤ 1 ␥ 2 was co-expressed. Alternatively, Ba 2ϩ currents recorded from . The receptor (DOR or ␤ 2 AR) and Ca 2ϩ channel ␣ 1 (␣ 1B or ␣ 1A ), ␣ 2 , and ␤ 1a subunits were co-expressed with G␣ or G␤ 1 ␥ 2 as indicated in Xenopus oocytes. In control oocytes, no exogenous G␣ nor G␤ 1 ␥ 2 was expressed. When G s ␣ was co-expressed, ␤ 2 AR, instead of DOR, was expressed, and 1 M isoproterenol, instead of 1 M Leu-EK, was used for stimulating the receptor. In practice, the membrane was held at Ϫ80 mV and depolarized by a 250-ms test pulse from Ϫ80 mV to ϩ10 mV. Since the peak current is inhibited prominently by Leu-EK (8), the amplitude of peak currents before and during exposure to Leu-EK or isoproterenol was used as a measure of the response to the agonist, and the change was expressed as a ratio of inhibition. Pretreatment (hatched bars) with 200 ng/ml PTX was carried out according to the procedures described under "Experimental Procedures." The number of oocytes examined are indicated in parentheses. * and ** indicate significant differences (p Ͻ 0.05 and p Ͻ 0.01, respectively, by analysis of variance with post hoc test) when compared with controls for the three types of experiments such as antisense(Ϫ)/ PTX(Ϫ), antisense(Ϫ)/PTX(ϩ) and antisense(ϩ). C, dose-dependent effects of G␣ such as G o1 ␣ (open bars), G i3 ␣ (filled bars), and G z ␣ (hatched bars) in potentiating Leu-EK-induced inhibition of ␣ 1B currents. Ca 2ϩ channel ␣ 1B , ␣ 2 and ␤ 1a subunits cRNAs and DOR cRNA were co-injected with G␣ cRNA at various concentrations indicated into Xenopus oocytes. The responses to 1 M Leu-EK were measured and expressed as ratios of them to those in control oocytes, in which only Ca 2ϩ channel subunits and DOR cRNAs were injected. The antisense oligonucleotide, AGO, was used. The number of oocytes examined are 8. The original responses for each before being normalized were 32.7 oocytes expressed with the Ca 2ϩ channel ␣ 1B , ␣ 2 , and ␤ 1a subunits, DOR and G␤ 1 ␥ 2 were facilitated by application of a large conditioning depolarization to ϩ80 mV, in the absence of receptor stimulation ( Fig. 2A, left). As mentioned before (8), this may indicate that the exogenous G␤␥ can inhibit the N-type Ca 2ϩ channel by itself, therefore without need for receptormediated activation of G-proteins. The prepulse facilitation was not prominent, but still significant, in the ␣ 1A channel ( Fig.  2A, right).
Changing the prepulse duration more clearly revealed the difference in facilitation between ␣ 1B and ␣ 1A channels in oocytes co-expressed with G␤ 1 ␥ 2 (Fig. 2B). A positive correlation between the extent of facilitation and prepulse duration was clearly observed in both ␣ 1B and ␣ 1A channels. However, the currents through ␣ 1A channels became suppressed as the duration of the prepulse was increased.
As shown in Fig. 2C (horizontal bar), Leu-EK-induced inhi-bition of ␣ 1A channel currents was markedly potentiated in oocytes co-expressed with G i1 ␣, G i3 ␣, or G o1 ␣, but this was not the case in oocytes co-expressed with G␤ 1 ␥ 2 . The prepulse procedure did not abolish the current inhibitions by G␣ isoforms (filled circles), whereas it was abolished in ␣ 1B channels (Fig. 3A) as shown in the previous paper (8). Thus, the difference between ␣ 1B and ␣ 1A channels in G-protein modulation appeared to be more prominent for G␣ than G␤ 1 ␥ 2 . Prepulse depolarization in the presence of G␤ 1 ␥ 2 facilitated both ␣ 1B and ␣ 1A channels, but facilitation of currents that were suppressed by G␣ was only observed for the ␣ 1B channel. These results suggest that G␣s are capable of distinguishing between HVA Ca 2ϩ channel types. Effects of Peptides Derived from Loop 1 and C terminus on the N-and P/Q-type Ca 2ϩ Channels-Based upon evidence uncovering the structural determinants of G i3 ␣ and G␤ 1 ␥ 2 interactions with ␣ 1B (8), an attempt was made to see whether FIG. 2. Differential responses of ␣ 1B and ␣ 1A channels to the depolarizing prepulse in Xenopus oocytes co-expressed with G␤ 1 ␥ 2 or G␣. A, N-type (␣ 1B , left), but not P/Q-type (␣ 1A , right), Ca 2ϩ channels in an oocyte implanted with DOR, Ca 2ϩ channel ␣ 1 (␣ 1B or ␣ 1A ), ␣ 2 and ␤ 1a subunits, and G␤ 1 ␥ 2 were prominently facilitated by a depolarizing prepulse (30 ms in duration) to ϩ80 mV in the absence of Leu-EK. A 200-ms test pulse was applied to ϩ10 mV from a holding potential of Ϫ100 mV. When preceded by the conditioning depolarization, the test pulse was applied 20 ms after cessation of the prepulse. B, effects of changing duration of the prepulse on ␣ 1B (filled circles) and ␣ 1A currents (open circles) in oocytes expressed with DOR, Ca 2ϩ channel ␣ 1 (␣ 1B or ␣ 1A ), ␣ 2 and ␤ 1a subunits, and G␤ 1 ␥ 2 . In practice, peak currents were measured before and after application of a prepulse to ϩ80 mV, and the ratios were expressed as a function of the duration of prepulse. The number of oocytes examined are 5. C, prepulse-resistant responses to Leu-EK in ␣ 1A channels. DOR and Ca 2ϩ channel ␣ 1A , ␣ 2 , and ␤ 1a subunits were co-expressed with G␣ or G␤ 1 ␥ 2 as indicated. In control oocytes (None), no exogenous G␣ nor G␤ 1 ␥ 2 was expressed. The responses of ␣ 1A currents to 1 M Leu-EK (horizontal bars), prepulse (open circles), and both (filled circles) were measured and expressed as ratios. The same experimental protocols were used as in Fig. 1 for Leu-EK and as in Fig. 2A for prepulse. The number of oocytes examined for each data are 4 -9. The antisense oligonucleotide, AGO, was used in A-C. ␣ 1B peptides derived from these interaction sites could quench the signaling accompanying exogenous expression of G-proteins in Xenopus oocytes. As shown in Fig. 3A (G␤ 1 ␥ 2 ; None, PL1; open circles), the prepulse facilitation observed in oocytes co-expressed with G␤ 1 ␥ 2 was almost abolished by the injection of PL1, an ␣ 1B -(or ␣ 1A )-derived peptide comprising an Nterminal cysteine and the amino acid residues 366 -384 of ␣ 1B (or the amino acid residues 370 -388 of ␣ 1A ) in loop 1 (see Fig.  4). However, this peptide did not suppress the potentiation of Leu-Ek-induced inhibition of ␣ 1B channels via G i3 ␣ (Fig. 3A, G i3 ␣, PL1, horizontal bar). By contrast, such potentiated inhibition with G i3 ␣ was reduced, as shown in Fig. 3A (G i3 ␣, PL1ϩPB3T1 and PL1ϩPB3T4, horizontal bars), by the injection of PL1 in combination with PB3T1 or PB3T4, an ␣ 1Bderived peptide comprising a cysteine and the amino acid residues 1934 -1943 (PB3T1) or 1931-1949 (PB3T4) in the C terminus (see Fig. 4). The combination of PL1 and PB3T4, a longer version of PB3T1, blocked more efficiently the potentiation of current inhibition than that of PL1 and PB3T1. However, injection of PL1 in combination with another ␣ 1B -derived C-terminal peptide, PB3T2 or PB3T3 comprising a cysteine and the amino acid residues 2016 -2025 or 1907-1925, respectively (see Fig. 4), failed to suppress this inhibitory potentiation via G i3 ␣ (Fig. 3A, G i3 ␣, PL1ϩPB3T2 and PL1ϩPB3T3, horizontal bars). Moreover, injection of PB3T4 alone did not affect either the agonist-induced inhibition of ␣ 1B channels via G i3 ␣ (G i3 ␣, PB3T4, horizontal bar) or the prepulse facilitation via G␤ 1 ␥ 2 (G␤ 1 ␥ 2 , PB3T4, open circle). On the other hand, both PL1 and PB3T4 diminished prepulse facilitation when DOR was stimulated by Leu-EK in oocytes co-expressed with G i3 ␣ (Gi3␣, PL1 and PB3T4, filled circles). These experiments using synthetic peptides revealed that both the loop 1 peptide (PL1) and the C-terminal peptide (PB3T4) were necessary to interrupt the interaction of the N-type ␣ 1B channel with G␣, whereas that the loop 1 peptide alone was capable of impairing the interaction with G␤␥.
G o ␣, but Not G␤␥ 2 , Binds to the C Termini of N-and P/Qtype Ca 2ϩ Channels-The results so far obtained from electrophysiological observations in this and the previous (8) studies suggest that the inhibition of N-(␣ 1B ) and P/Q-type (␣ 1A ) Ca 2ϩ channels by G-proteins involves direct interactions between the ␣ 1B loop 1 and G␤␥ and between the ␣ 1B /␣ 1A C termini and G␣. To examine these possibilities, the C terminus (corresponding to amino acid residues 1912-2339) and the loop 1 (corresponding to amino acid residues 361-483) of N-type Ca 2ϩ channels were expressed in E. coli as GST fusion proteins, GST-B3T and GST-B3L1, respectively. Similarly, the C terminus (corresponding to amino acid residues 1975-2424) and the loop 1 (corresponding to amino acid residues 365-489) of the P/Q-type channel were also fused with GST (GST-B1T and GST-B1L1), respectively. Then, as shown in Fig. 5, their ability to bind G␣ and G␤␥ was tested by immunoblot analysis using the antibodies G o Ab and K-20 against G o ␣ and the antibody M-14 against G␤ 1 (see "Experimental Procedures").
The antibody G o Ab reacted with a 39-kDa G o ␣ purified from brain (Fig. 5A, lanes 1 and 5) and with several polypeptides released from the GST-B3T-bound beads that had been incubated with the purified G o ␣ (Fig. 5A, lane 2). Among these immunoreactive polypeptides, a 39-kDa polypeptide obviously became devoid of its reactivity with G o Ab by co-incubation with the peptide antigen FG o for G o Ab (Fig. 5A, lane 4, arrowhead). This co-incubation also abolished the reactivity of the antibody with G o ␣ (Fig. 5A, lanes 3 and 7). These results indicate that the 39-kDa immunoreactive polypeptide was recognized specifically by the antibody against G o ␣ when released from the C terminus. By contrast, this antibody never detected a 39-kDa polypeptide released from GST-B3L1-bound beads (Fig. 5A,  lanes 6 and 8) or from GST-bound beads (n ϭ 3), the two kinds of beads that had been incubated with the purified G o ␣.
On the other hand, the antibody M-14 reacted with a 36-kDa FIG. 3. Effects of loop 1 and C-terminal peptides derived from ␣ 1B or ␣ 1A subunit on N-and P/Q-type Ca 2؉ channels. Synthetic peptides (10 M) derived from the loop 1 and C terminus of ␣ 1B or ␣ 1A subunit (see Fig. 4) were injected 15 h prior to electrophysiological measurements by various combinations indicated into Xenopus oocytes, which were implanted with Ca 2ϩ channel ␣ 1 , ␣ 2 , and ␤ 1a subunits, DOR, and/or G-protein subunit (G i3 ␣, G o1 ␣, or G␤ 1 ␥ 2 ). The amino acid sequence of PL1 is shared by the ␣ 1B and ␣ 1A subunits. The responses of wild-type ␣ 1B (A) and ␣ 1A (B) channels to 1 M Leu-EK (horizontal bars), prepulse (open circles), or both (filled circles) were measured in oocytes co-expressed with or without G i3 ␣, G o1 ␣, or G␤ 1 ␥ 2 , as indicated, and expressed as ratios. The pulse protocols were the same as those in G␤ purified from brain (Fig. 5B, lanes 1, 3, 5 and 7), as well as a 36-kDa polypeptide released from the GST-B3L1-bound beads (Fig. 5B, lane 4, arrowhead) that had been incubated with the purified G␤␥ 2 . Both reactivities of M-14 with the 36-kDa polypeptides were inhibited by preincubation of M-14 with the peptide antigen sc-261P for M-14 (n ϭ 3). No 36-kDa polypeptide, however, was detected by M-14 in polypeptides released from GST-B3T-bound beads despite the incubation with the purified G␤␥ 2 (Fig. 5B, lane 2, arrowhead). These results support the idea that G␣ and G␤␥ inhibit the N-type Ca 2ϩ channel by directly interacting with the C terminus and the loop 1 of the channel, respectively.
As also shown in Fig. 5A (lane 9), another antibody K-20 against G o ␣ detected the purified G o ␣ as well. K-20 mainly reacted with a 39-kDa polypeptide released from the GST-B1Tbound beads that had been incubated with the purified G o ␣ (Fig. 5A, lane 10). This reactivity of K-20 with the 39-kDa polypeptide was diminished by preincubation of the antibody with the peptide antigen sc-387P for K-20 (Fig. 5A, lane 12), as observed in the purified G o ␣ (lane 11). As shown in Fig. 5B, the antibody M-14 did not detect polypeptides released from the GST-B1T-bound beads (lane 6, arrowhead) but detected a 36-kDa polypeptide from the GST-B1L1-bound beads (lane 8, arrowhead), when these beads had been incubated with the purified G␤␥ 2 . The reactivity of M-14 with the 36-kDa polypeptide released from the GST-B1L1-bound beads was inhibited by preincubation of M-14 with the peptide antigen sc-261P (n ϭ 3). These results are consistent with the idea, obtained from the electrophysiological experiments, that the P/Q-type Ca 2ϩ channel, like N-type, is under the regulation of G␣ and G␤␥ directly interacting with the C terminus and the loop 1 of the channel, respectively.

DISCUSSION
In the present study, we found that distinct sets of G␣ co-expressed in oocytes mediated receptor agonist-induced inhibitions of N-type ␣ 1B and P/Q-type ␣ 1A channels; agonistinduced inhibition of the ␣ 1B channel was potentiated by coexpression of G i3 ␣, G o1 ␣, G z ␣, and G s ␣, whereas that of the ␣ 1A channel was intensified by co-expression of G i1 ␣, G i2 ␣, G i3 ␣, or G o1 ␣. Single channel recordings indicated that the molecular species G i3 ␣ is a PTX-sensitive G-protein in native tissues and FIG. 4. Schematic representation of the sites on ␣ 1B subunit for synthesizing the peptides. A, positions of the loop 1 and the C terminus, together with those of the deletion (L⌬1, L⌬2, L⌬3, and T⌬1) as described previously (8), are indicated by the number of the amino acid residues for ␣ 1B subunit (42) and ␣ 1A (BI-1 ␣ 1 ) subunit (14) in parentheses. The deletion sites are indicated by the crossing bars, and the cytoplasmic side below the horizontal lines. The peptides were synthesized on the basis of the amino acid sequences indicated by the open circles attached. The asterisk denotes the binding site for Ca 2ϩ channel ␤ subunit (43), and the filled circle denotes the phosphorylation sites for protein kinase C (9). B, primary sequences of the synthetic peptides used representing their alignments with the predicted protein sequences of the ␣ 1B and the ␣ 1A subunits. An extra cysteine (in parentheses) was added to each peptide on the N-terminal side. elicits channel inhibition through a membrane-delimited pathway (33,34). Alternatively, a depolarizing prepulse relieved current inhibition caused by the G␤ 1 ␥ 2 complex, with facilitation being more pronounced in ␣ 1B than in ␣ 1A channels. Finally, we defined the loop 1 of ␣ 1B and ␣ 1A as an interaction site for G␤␥ and the C termini of ␣ 1B and ␣ 1A for G␣, based on the direct binding of G o ␣ and G␤␥ 2 in vitro to channel segments, as well as the responses of wild-type channels to synthetic peptides. G o ␣, but not the G␤␥ 2 complex, purified from bovine brain bound in vitro to the C terminus of ␣ 1B and ␣ 1A channels, which was fused as a GST protein. Conversely, G␤␥ 2 bound in vitro to the loop 1 of ␣ 1B and ␣ 1A channels as described (9,10). The obtained results provide evidence that G␣ as well as G␤␥ directly interact with Ca 2ϩ channel ␣ 1 subunits to inhibit their activity.
Differential Regulation of ␣ 1B and ␣ 1A Channels by G␣ and G␤␥-G o ␣ (1, 2, 4, 6) and, more recently, G␤␥ (35,36) have been shown to be involved in inhibitory modulation of HVA Ca 2ϩ channels, including N-type Ca 2ϩ channels. Among the six different subtypes of G␣ examined, particular subtypes (G i3 ␣, G o1 ␣, G z ␣, and G s ␣) further intensified the agonist-induced inhibition of ␣ 1B channels, whereas G i1 ␣ and G i2 ␣ did not, despite the fact that they are able to couple to DOR (37,38). Therefore, these G␣ subtypes seem to be unable to inhibit the ␣ 1B channel. Moreover, it seems unlikely that the inhibitory action of G s ␣ is associated with adenylate cyclase, since neither intracellular application of cyclic AMP nor a catalytic subunit of PKA nor pretreatment with H7 altered the ␣ 1B channel activities. Our findings of G s ␣-induced inhibition of N-type Ca 2ϩ channels are consistent with evidence that inhibition of N-type Ca 2ϩ channels by vasoactive intestinal polypeptide is attenuated by cholera toxin and anti-G s ␣ antibodies (39). Similar preferences among G␣ subtypes in potentiating ACh-induced inhibition were observed in NG108-15 cells, in which N-type, but not P/Q-type, Ca 2ϩ channels were natively expressed. Inhibition of ␣ 1A channels by Leu-EK was potentiated when G i1 ␣, G i2 ␣, G i3 ␣, or G o1 ␣ was co-expressed. Thus, the ␣ 1B and ␣ 1A channels presumably carry interaction sites that are capable of selectively recognizing certain subtypes of G-protein ␣ subunits.
When G i1 ␣ cRNA or a low concentration of G z ␣ cRNA was co-injected, the Leu-EK-induced inhibition of ␣ 1B channels was considerably reduced. The potency of G z ␣ in inhibiting ␣ 1B channels was lower than those of G i3 ␣ and G o1 ␣. This attenuation of the channel inhibition by G i1 ␣ and G z ␣ may be due to a trap of the endogenous G␤␥ by the exogenous G␣, leading to occlusion of an inhibitory signal by G␤␥. Our previous studies have suggested the presence of blockade by exogenous G␣ in the metabotropic glutamate receptor-induced phosphoinositide hydrolysis (12).
When G␤ 1 ␥ 2 was co-expressed, the Leu-EK-induced inhibition was not potentiated in either ␣ 1B or ␣ 1A channels. In the case of ␣ 1B , however, a depolarizing prepulse to ϩ80 mV facilitated the currents in the absence of the receptor agonist, suggesting that the exogenous G␤␥ inhibits the ␣ 1B channel by itself (35,36). As shown in the previous paper (8), the prepulse did not affect ␣ 1B channels in oocytes co-expressed with G i3 ␣ unless DOR was stimulated by Leu-EK. Thus, it appears that the exogenous G␣ does not affect the ␣ 1B channel by itself and that the channel inhibition observed with the exogenous G␣ results from interaction of the channel with the exogenous G␣ and/or an endogenous G␤␥ released from the exogenous G␣.
In the case of the ␣ 1A channel, the prepulse facilitation was not prominent when G␤ 1 ␥ 2 was co-expressed. Furthermore, as the duration of prepulse was increased from 30 to 50 ms, the facilitation practically disappeared in ␣ 1A channel, whereas it remained unchanged in ␣ 1B channel, probably reflecting the difference in voltage-dependent channel inactivation (16). On the other hand, Leu-EK-induced inhibition of the ␣ 1A channel was markedly potentiated in oocytes co-expressed with G␣ subtypes, similar to the ␣ 1B channel. The prepulse failed to abolish this inhibition potentiated by G␣ subtypes in ␣ 1A but not in ␣ 1B channels (8).
All these findings indicate that the N-and P/Q-type Ca 2ϩ channels are regulated differentially by distinct G␣ subtypes and that the N-type is preferentially regulated by the G␤ 1 ␥ 2 subunit. In addition, it has been suggested that in N-and P/Q-type Ca 2ϩ channels, there is a structural domain associated with G-proteins, which has a voltage sensitivity distinguishable by prepulse (40).
The C termini of ␣ 1B and ␣ 1A Channels as an Interaction Site for G␣ and the Loop 1 of ␣ 1B Channel as a Regulatory Site for G␤␥-A synthetic loop 1 peptide (PL1) (see Fig. 4) blocked the prepulse facilitation of ␣ 1B channels via G␤ 1 ␥ 2 . This indicates that the loop 1 plays an essential role for the interaction of the ␣ 1B channel with G␤␥ (9, 10). On the other hand, PL1 did not influence the response to Leu-EK via G i3 ␣ in the ␣ 1B channels, indicating that there is an additional interaction site for Gprotein subunits outside of loop 1 (40). The same conclusions were drawn by using channel mutation and chimerization in the previous study (8).
The G i3 ␣-dependent potentiation in ␣ 1B channels was blocked by co-application of the peptide PL1 and an ␣ 1B Cterminal peptide (PB3T1 or PB3T4) but not blocked by the C-terminal peptide alone. These results indicate that both loop 1 and C-terminal segment of ␣ 1B play an essential role for the G i3 ␣-dependent potentiation and that the interaction site for G i3 ␣ seems to be mainly assigned to the ␣ 1B C terminus.
In the P/Q-type ␣ 1A channel, an ␣ 1A version of the C-terminal peptide PB3T4 (PPQT1), was able to diminish the potentiation of agonist-induced current inhibition via G o1 ␣ without the aid of the loop 1 peptide PL1. The results indicate, as in the case of ␣ 1B , that the C-terminal segment of ␣ 1A is essential for the interaction with G-protein subunits, whereas the loop 1 of ␣ 1A is not essential.
Direct Binding of G␣ with the C Terminus of ␣ 1B and ␣ 1A Channels, and G␤␥ with the Loop 1 of the Channels-Finally, we found that bacterial fusion proteins containing the C-terminal segment of the N-(␣ 1B ) and P/Q-type (␣ 1A ) Ca 2ϩ channels were capable of binding bovine brain purified G o ␣ but not G␤␥ 2 . In addition, GST fusion proteins containing the loop 1 of ␣ 1B and ␣ 1A were able to bind the G␤␥ 2 but not the G o ␣ as reported recently (9,10). It has been shown more recently that G␤␥ also binds to the C terminus of the ␣ 1E channel in vitro (41). This result suggests that each type of neuronal Ca 2ϩ channel is differentially regulated by each subunit of the G-protein complex as observed with the ␣ 1B and ␣ 1A channels, in which contribution of the ␣ 1A loop 1 to channel modulation by G␣ was smaller than that of the ␣ 1B loop 1. In addition, the C-terminal short fragments of ␣ 1B and ␣ 1A have been shown to be bound by G␤␥ (41). The discrepancy may imply the presence of an additional C-terminal domain that affects the interaction with G␣ and G␤␥. This speculation is consistent with the fact that the corresponding positions of the C-terminal peptides (PB3T4 and PPQT1) on ␣ 1B and ␣ 1A are different from those of the G␤␥bound fragments reported. Further studies using mutagenesis will be necessary to identify the specific amino acid residues on ␣ 1B and ␣ 1A determining the interactions with G␣ and/or G␤ 1 ␥ 2 , or determining the differences in modulatory properties between N-and P/Q-type Ca 2ϩ channels.
All of these findings from the present and the previous experiments (8) indicate that the C terminus of N-and P/Q-type Ca 2ϩ channels contain an interaction site for G␣ and that the loop 1 contains an interaction site for G␤␥. It is further indicated that the G␣ species G i3 ␣ and G o1 ␣ are shared by the N-and P/Q-type Ca 2ϩ channels, whereas G z ␣ and G s ␣ are rather specialized for inhibiting the N-type and G i1 ␣ and G i2 ␣ for selectively suppressing the P/Q-type. Since multiple Gprotein isoforms co-exist with more than two types of HVA Ca 2ϩ channels in a single neuronal cell, switching on/off the expression of a particular G␣ subtype would convert either one or both of the N-and P/Q-type channels to a sensitive or insensitive response to an inhibitory signal by the same transmitter/agonist stimulation. Thus, the regulation of the N-type and P/Q-type Ca 2ϩ channels by different G␣ and G␤␥ would allow a variability and a flexibility in synaptic efficacy by alteration of transmitter release.