Identification of Residues in the N Terminus of α1B Critical for Inhibition of the Voltage-Dependent Calcium Channel by Gβγ

To examine the role of the intracellular N terminus in the G-protein modulation of the neuronal voltage-dependent calcium channel (VDCC) α1B, we have pursued two routes of investigation. First, we made chimeric channels between α1B and α1C, the latter not being modulated by Gβγ subunits. VDCC α1 subunit constructs were coexpressed with accessory α2δ and β2a subunits inXenopus oocytes and mammalian (COS-7) cells. G-protein modulation of expressed α1 subunits was induced by activation of coexpressed dopamine (D2) receptors with quinpirole in oocytes, or by cotransfection of Gβ1γ2 subunits in COS-7 cells. For the chimeric channels, only those with the N terminus of α1B showed any G-protein modulation; further addition of the first transmembrane domain and I-II intracellular linker of α1B increased the degree of modulation. To determine the amino acids within the α1B N terminus, essential for G-protein modulation, we made mutations of this sequence and identified three amino acids (S48, R52, and R54) within an 11 amino acid sequence as being critical for G-protein modulation, with I49 being involved to a lesser extent. This sequence may comprise an essential part of a complex Gβγ-binding site or be involved in its subsequent action.

The inhibition of N-(␣1B) and P/Q-type (␣1A) calcium currents by receptors, usually acting through pertussis toxin-sensitive G-proteins, appears to be mediated by G␤␥ subunits (Herlitze et al., 1996;Ikeda, 1996). There has been some controversy concerning whether the ␣1E calcium channel is G-protein-modulated (Page et al., 1998). We have now established that, whereas an N-terminally truncated isoform of rat ␣1E is not subject to modulation, an isoform with a f ull-length N terminus is G-protein-modulated, either by coexpression of G␤␥ subunits or by activation of a G-protein-coupled receptor (Page et al., 1998), which would agree with results obtained previously for full-length human ␣1E .
A number of recent studies have established the importance of the intracellular loop that links transmembrane domains I and II, both in binding G␤␥ and in mediating its effects to produce inhibition of the channel (Herlitze et al., 1997;Z amponi et al., 1997). However, this result is controversial, and several studies have suggested either that the I-II loop plays no role in G-protein modulation of ␣1B (Z hang et al., 1996) or ␣1E , or that alone it cannot mediate the effects of the G␤␥ subunits (Page et al., , 1998Simen and Miller, 1998). Nevertheless it is not disputed that the I-II loops of ␣1A, B, and E comprise a major binding site or sites for G␤␥ and contain a QxxER amino acid consensus sequence common to many G␤␥-binding sites (De Waard et al., 1997;Herlitze et al., 1997;Zamponi et al., 1997;Dolphin et al., 1999). Secondly, a C-terminal G␤␥-binding site has recently been identified and proposed to be a region responsible for G-protein inhibition of human ␣1E . However, it is clear that there are also a number of other sites in the ␣1 subunit of G-protein-modulated calcium channels that are involved in expression of the inhibition by G␤␥. First, we have found that part of the intracellular N terminus of ␣1B and ␣1E is essential for their G-protein modulation (Page et al., 1998). Second, the transmembrane domain I has been found to have an important role (Zhang et al., 1996;Stephens et al., 1998b).
In the present study we have examined the critical nature of the intracellular N terminus of ␣1B, by making chimeric channels between ␣1B, which is strongly G-protein-modulated and ␣1C, which is not G-protein-modulated by this mechanism, and has a completely different N-terminal sequence. We have shown an absolute requirement for the ␣1B N terminus for observation of G-protein modulation in all the chimeric constructs. Second, we have made specific deletions and point mutations to identify the sequence in the N terminus of ␣1B that is responsible for conferring G-protein modulation.

Construction of N-terminal deletion and point mutations
The ␣1B N terminus was truncated at the 5Ј end by introducing a start codon before amino acid E7 to make ␣1B ⌬2-6, Y45 (␣1B ⌬2-44), and Q51 (␣1B ⌬2-50). The following primers were used; CGC AC T AGT ATG GAG C TG GGC GGC CGC TAT (⌬2-6), CAG AC T AGT ATG TAC AAA CAG TCG ATC GCG (⌬2-44), and CAG AC T AGT ATG CAG CGC GCG CGG ACC AT (⌬2-50). The ␣1B ⌬45-55 construct was made by using the primer GGC CAG CGG GTC C TC ATG GCG C TG TAC AAC to delete the 11 amino acids, YKQSIAQR ART. For all of the ␣1B point mutations, primers were designed so that single residues were mutated to alanines or so that a number of residues were mutated within the same primer. The following primers were used; R52A-R54A, TCG ATC GCG CAG GCC GCG GCG ACC ATG GCG C T; Y45A, CAG CGG GTC C TC GCC AAA CAG TCG ATC; K46A, CGG GTC C TC  TAC GCA CAG TCG ATC GCG; Q47A, GTC C TC TAC AAA GCG  TCG ATC GCG CAG; S48A, C TC TAC AAA CAG GCG ATC GCG  CAG C; I49A, TAC AAA CAG TCG GCC GCG CAG CGC GCG; Q51A, CAG TCG ATC GCG GCG CGC GCG CGG ACC; R52A, TCG ATC GCG CAG GCC GCG CGG ACC ATG; R54A, GCG CAG CGC GCG GCG ACC ATG GCG C TG; 45YKQSIA3 AAAAA, GCC GCA GCA GC T GCC GCG CAG CGC GCG CGG (forward) and GGC AGC TGC TGC GGC GAG GAC CCG C TG (reverse); and 45YKQ3 AAA, CGG GTC C TC GCC GCA GCG TCG ATC GCG CAG. The reverse primer used in each case was GTC GC T TC T GC T C TT C TT GG. For the PCR extension reactions, the forward primer used was AGC AC T AGT ATG GTC CGC TTC GGG GAC. The sequences of all constructs were verified.

Expression of constructs and electrophysiological recording
Xenopus ooc ytes. Adult female Xenopus laevis were killed by anesthetic overdose in a 0.25% solution of tricaine, decapitated, and pithed. Oocytes were removed and defolliculated by treatment with 2 mg /ml collagenase type Ia in a C a 2ϩ -free N D96 saline containing (in mM): NaC l, 96; KC l, 2; MgC l 2 , 1; and H EPES, 5, pH-adjusted to 7.4 with NaOH for 2 hr at 21°C. Plasmid cDNAs for the different ␣1 subunits, plus accessory ␤2a and ␣2␦ subunits and rat D2 receptors, were mixed in a ratio of 3:4:1:3 (except where stated), and ϳ10 nl was injected into the nuclei of stage V or V I oocytes. Injected oocytes were incubated at 18°C for 3-7 d in N D96 saline (as above plus 1.8 mM CaCl 2 ) supplemented with 100 g /ml penicillin, 100 I U/ml streptomycin (Life Technologies, Gaithersburg, MD), and 2.5 mM Na pyruvate. Whole-cell recordings from oocytes were made in the two-electrode voltage-clamp configuration with a chloridefree solution containing (in mM): Ba(OH) 2 , 5; TEA-OH, 80; NaOH, 25; C sOH, 2; and H EPES, 5 (pH 7.4 with methanesulfonic acid). In all experiments, oocytes were injected with 30 -40 nl of a 100 mM solution of K 3 -1,2-bis (aminophenoxy) ethane-N, N,NЈ,NЈ-tetra-acetic acid (BAP TA) in order to suppress endogenous C a 2ϩ -activated C l Ϫ currents. Electrodes contained 3 M KC l and had resistances of 0.3-2 M⍀. The holding potential (V H ) was Ϫ100 mV, and the test potential (V t ) used for time course studies was 0 mV. All illustrated traces are at this potential, and the current amplitude was always measured 20 msec after the start of the test pulse. Membrane currents were recorded every 15 sec, amplified, and low-pass filtered at 1 K Hz using a Geneclamp 500 amplifier and digitized through a Digidata 1200 interface (Axon Instruments, Foster C ity, CA). In all cases currents were leak subtracted on-line by a P/4 protocol. COS-7 cells. C ells were cultured and transfected, using the electroporation technique, essentially as described previously (C ampbell et al., 1995a). The ␣1, ␣2␦, ␤2a, and GFP cDNAs were used at 15, 5, 5, and 1 g, respectively. When used, G␤1 and G␥2 were included at 2.5 g each, or ␤-ARK was included at 5 g. Blank pMT2 vector was included where necessary to maintain the total cDNA at 31 g /transfection. C ells were replated using nonenzymatic cell dissociation medium (Sigma, St. L ouis, MO), and then maintained at 25°C for between 1 and 16 hr before electrophysiological recording. Maximum GFP fluorescence and voltagedependent calcium channel (V DCC) expression were observed between 2 and 4 d after transfection (Brice et al., 1997). C a 2ϩ channel currents were recorded using the whole-cell patch technique. Borosilicate glass 2-5 M⍀ electrodes were used. The internal (electrode) and external solutions were similar to those described previously (C ampbell et al., 1995b). The patch pipette solution contained in mM: C s aspartate, 140; EGTA, 5; MgC l 2 , 2; CaCl 2 , 0.1; K 2 ATP, 2; and H EPES, 10; pH 7.2, 310 mOsm with sucrose. GDP␤S (2 mM) was included where stated. The external solution contained in mM: tetraethylammonium (TEA) bromide, 160; KC l, 3; NaHC O 3 , 1.0; MgC l 2 , 1.0; H EPES, 10; glucose, 4; and BaC l 2 , 1 or 10, pH 7.4, 320 mOsm with sucrose. Whole-cell currents were elicited from V H of Ϫ100 mV and recorded using an Axopatch 1D amplifier. Data were filtered at 2 kHz and digitized at 5-10 kHz. The junction potential between external and internal solutions was 6 mV, the values given in the figures and text have not been corrected for this. Current records are shown following leak and residual capacitance current subtraction (P/4 or P/8 protocol) and series resistance compensation up to 85%. Current amplitudes were measured 50 msec after the start of depolarization.
All experiments were performed at room temperature (18 -20°C). Analysis was performed using pC lamp6 and Origin software. Data are expressed as mean Ϯ SEM. Statistical analysis was performed using paired or unpaired Student's t test, as appropriate.

RESULTS
In a previous study we made chimeras between the rat brain ␣1E (rbEII) clone, which is not G-protein-modulated, and the strongly modulated ␣1B. The results of this study showed that rbEII was not modulated because it was N terminally truncated, and a full-length rat ␣1E isoform showed clear G-protein modulation, although not to such a great extent as ␣1B. We further showed the importance of the first domain of ␣1B in increasing the extent of G-protein modulation of ␣1B/␣1E chimeras (Page et al., 1998;Stephens et al., 1998b), as has another recent study (Simen and Miller, 1998). In the present study, we wished to examine the distinct role of the N terminus of ␣1B in G-protein modulation. To do this we have taken two approaches. First, we have made chimeras between ␣1B and the ␣1C channel, which is not modulated by a G␤␥-mediated pathway under any conditions. Second, we have produced selective deletions and mutations of the ␣1B N-terminal sequence. With such constructs we can determine the domains necessary for the expression of G-protein modulation.

G-protein modulation of ␣1B/␣1C chimeras by activation of the dopamine D2 receptor
In this part of the study, all channels were expressed with the accessory subunits ␣2␦ and ␤2a (unless stated) in Xenopus oocytes, where they were coexpressed with the dopamine D2 receptor. A series of chimeras were made, in which the N terminus, first transmembrane domain, and I-II loop of ␣1B were systematically substituted for those in ␣1C, in different permutations. Figure 1 A shows the chimeras that were made, and the nomenclature employed, which uses capital letters for the transmembrane domains and small letters for the intracellular N-terminal and I-II loop. All chimeras contained the last three domains and C -terminal tail of ␣1C (denoted CCC), and all showed good expression levels with one exception (Table 1). However, because ␣1B, ␣1C, and the chimeras between them showed differences in their voltage dependence of activation (Table 1), we could not compare G-protein modulation at a single step potential (Fig. 2). Therefore, we have estimated the amount of G-protein modulation in two ways in Xenopus oocytes, first by determining the ability of the D2 agonist quinpirole to cause a depolarizing shift in the voltage dependence of activation, determined from current-voltage plots ( Fig. 1 B). Second, we have determined the percentage inhibition by quinpirole of the current activated at all potentials between Ϫ20 and ϩ30 mV (Fig. 2). In all cases, the modulation by quinpirole occurred within 30 -60 sec of its application and was fully reversible.
The modulation of ␣1B by activation of the dopamine D2 receptor with 100 nM quinpirole was voltage-dependent, as we have shown previously (Page et al., 1998;Stephens et al., 1998b). This is manifested by a depolarizing shift in the voltage for 50% activation of the current (V 50 ) ( Fig. 1 B), and also by a reduction in the percentage inhibition at increasing test potentials (Fig. 2 A). Maximum inhibition was usually seen at a test potential ϳ10 mV below the peak of the current-voltage relationship (47% at Ϫ10 mV for ␣1B; Fig. 2 A). The transfer of the entire N terminus, first transmembrane domain, and I-II loop sequence of ␣1B into ␣1C gave a chimera showing G-protein modulation that was smaller at all potentials than the ␣1B parent ( Figs. 1 B, 2B). The depolarizing shift in the V 50 for ␣1bBbCCC was less than for ␣1B ( Fig.  1 B), and the maximum modulation was 24% at Ϫ20 mV ( Fig.  2 B). With respect to both measurements, a similar degree of modulation by quinpirole was seen for ␣1bBcCCC (Figs. 1 B, 2C), providing strong evidence that the I-II linker from a modulatable channel such as ␣1B is not essential for exhibition of G-protein modulation. Modulation by quinpirole was also still present in the chimera ␣1bCbCCC (18% at Ϫ10 mV; Figs. 1 B, 2D). Furthermore, there was still a significant degree of modulation of the minimal chimera ␣1bCcCCC (13% at Ϫ10 mV; Figs. 1 B, 2E), again indicating that the I-II linker from a modulatable channel is not essential for the observation of G-protein modulation.
In contrast, none of the chimeras containing the N terminus of ␣1C instead of ␣1B showed any inhibition by quinpirole at any potential from Ϫ30 to ϩ40 mV under these conditions (inhibition by quinpirole at 0 mV: 0.66 Ϯ 1.0% for ␣1cBbCCC, -0.4 Ϯ 0.3% for ␣1cBcCCC, and -0.86 Ϯ 0.92% for ␣1cCbCCC; n values given in Table 1) (Fig. 2 F). There was also no quinpirole-induced depolarizing shift in the V 50 for activation ( Fig. 1 B). This was also the case for ␣1C (Ϫ0.25 Ϯ 0.21% inhibition by quinpirole at 0 mV; Figs. 1 B, 2F). Thus, the N terminus of ␣1B is essential and sufficient for the expression of any G-protein modulation, whereas the first transmembrane domain and I-II linker of ␣1B can be substituted by that of ␣1C, and significant, although reduced, G-protein modulation is still observed.
Antagonism by ␤2a of G-protein modulation of the ␣1B/␣1C chimeras It has previously been shown that the G-protein modulation of ␣1E currents is antagonized by ␤2a (Qin et al., 1998). To study the interaction between the presence of overexpressed VDCC ␤2a and the extent of G-protein modulation, we also examined the degree of G-protein modulation by dopamine D2 receptor activation in the absence of exogenously coexpressed VDCC ␤ subunit in Xenopus oocytes. Nevertheless, it should be stressed that Xenopus oocytes contain an endogenous ␤3-like subunit, and when this was depleted with an antisense construct, no functional currents were seen (Tareilus et al., 1997). The G-protein modulation of ␣1B and the chimera ␣1bBbCCC was found to be significantly greater in the absence of coexpressed ␤2a than in its presence ( Fig. 2 A, B). In contrast, the extent of quinpiroleinduced modulation of the ␣1bBcCCC and ␣1bCbCCC chimeras was not significantly increased in the absence of exogenous ␤2a (Fig. 2C,D). Furthermore, the absence of ␤2a did not uncover G-protein modulation in any of the chimeras lacking the N terminus of ␣1B that were not modulated in the presence of ␤2a (results not shown). We were unable to examine ␣1bCcCCC currents in the absence of ␤2a, because no expression was observed (n ϭ 3 experiments). These results suggest that the presence of the I-II linker and first transmembrane domain of ␣1B, although not being essential for G-protein modulation, are together required for the reduction of G-protein modulation in the presence of the exogenously expressed VDCC ␤2a subunit, seen under these conditions.
Coexpression of ␤ subunits with ␣1 subunits in Xenopus oocytes and other systems results in a hyperpolarizing shift in current activation (for review, see Walker and De Waard, 1998), and it is of interest that this is greatest for ␣1B, ␣1C and those chimeras in which all the transmembrane domains are identical (␣1bCbCCC and ␣1cCbCCC; Table 1). However, despite the reduced ␤2a-induced hyperpolarizing shift in the activation of the ␣1bBbCCC chimera, compared to ␣1B, there was still a clear ␤2a-induced reduction in the amount of G-protein inhibition at all potentials (Fig. 2 B), indicating that ␤2a was influencing this channel.

G-protein modulation of ␣1B/␣1C chimeras by coexpression of G␤␥ subunits
The role of G␤␥ in mediating the inhibition observed was confirmed by coexpression of the chimeric ␣1 channels with ␣2␦, ␤2a, and G␤1␥2 in COS-7 cells. A prepulse protocol was used (Fig. 3, Figure 1. G-protein modulation of chimeras between ␣1B and ␣1C. A, Chimeras made between ␣1B (white) and ␣1C (black), together with the nomenclature used. B, Chimeras and parental constructs were expressed in Xenopus oocytes together with ␣2␦ and the dopamine D2 receptor. The V 50 in the absence and presence of the D2 agonist quinpirole (100 nM) was determined from current-voltage relationships performed before and during its application, as described in the legend to Table 1, and the ⌬V 50 was calculated (mean Ϯ SEM). The number of experiments is given for each histogram bar. The statistical significance of ⌬V 50 was determined by paired t test; **p Ͻ 0.01. left panels), giving steps to potentials between Ϫ40 and ϩ40 mV, before (P1) and 10 msec after (P2) a large depolarizing step to ϩ120 mV (Page et al., 1998). The prepulse reverses G␤␥mediated modulation, and hence P2 acts as an internal control. The G␤␥-mediated modulation was determined from the hyperpolarizing shift in the V 50 of the current-voltage relationship in P2 compared to that in P1 (Fig. 3, right panels). For ␣1B, this shift was almost Ϫ10 mV (Figs. 3A, 4), and it was not significantly smaller for the chimeras ␣1bBbCCC and ␣1bCbCCC (Figs. 3B, 4). It was reduced, but still significantly different from ␣1C for the ␣1bBcCCC and ␣1bCcCCC chimeras (Figs. 3C, 4). Of the other chimeras, all of which had the N terminus of ␣1C, none showed any greater shift in V 50 than ␣1C itself (Figs. 3D, 4). In control experiments recorded in the absence of coexpressed G␤1␥2 and in the presence of intracellular GDP␤S, the shift in V 50 caused by a depolarizing prepulse was approximately Ϫ1.8 mV for ␣1C (n ϭ 10), very similar to the value for ␣1C coexpressed with G␤1␥2 (Ϫ2.1 mV; Fig. 4). A similar level of control facilitation was observed for ␣1B (n ϭ 10) (Fig. 4). Similar control results were obtained when the ␤-ARK1 G␤␥-binding domain was coexpressed, to act as a sink for endogenous G␤␥ and prevent tonic modulation (Stephens et al., 1998a,b) (results not shown). This control prepulse potentiation is therefore likely to be caused by a mechanism other than G-protein modulation (Dolphin, 1996). The main discrepancy between the results examining direct G␤␥ modulation and those examining receptor-mediated modulation involve the ␣1bCbCCC chimera, which is strongly modulated by overexpression of G␤1␥2 (Figs. 3B, 4), and more weakly modulated by receptor-mediated inhibition (Figs. 1 B, 2D). The reason for this may relate to differences in G␤␥ subtype or concentration between the two systems.

Isolation of the amino acid residues of the N terminus essential for G-protein modulation
We have made a number of deletions to determine the amino acid sequences that are essential for G-protein modulation. From our previous study (Page et al., 1998), we found that the truncated ⌬1-55 ␣1B construct was not G-protein modulated, in agreement with the N-terminally truncated ␣1E (rbEII) isoform, that is also not G-protein-modulated. In the present series of experiments, the effect of quinpirole (100 nM) was determined during steps to 0 mV, because none of the constructs showed major shifts in voltage dependence of current activation, compared to ␣1B. Inhibition of wild-type ␣1B was 35.3 Ϯ 2.2% at 0 mV under these conditions (n ϭ 8). We made a number of truncations: ␣1B⌬2-6 and ⌬2-44, in line with regions of homology between the N-terminal sequences of all the G-protein modulated ␣1 subunits (Fig. 5A). These two constructs were as strongly G-proteinmodulated as ␣1B itself [respectively, 35.8 Ϯ 2.5% (n ϭ 5), and 36.1 Ϯ 5.3% (n ϭ 6) inhibition by quinpirole; Fig. 5 B, C]. This identifies the 11 amino acid sequence of ␣1B 45-55 (YKQSIAQRART) (Fig. 5A) as being required for the G-protein modulation of ␣1B. To confirm this finding, deletion of only this sequence created a construct, ␣1B ⌬45-55, in which G-protein modulation was completely abolished [Ϫ0.7 Ϯ 2.1% inhibition by quinpirole (n ϭ 5); Fig. 5 B, C].
Point mutations to alanine (A) were then carried out to identify the specific amino acids in this 11 amino acid sequence that are essential for G-protein modulation. Mutation of both arginines to alanine (R52A, R54A) produced a construct that showed no G-protein modulation (Ϫ2.5 Ϯ 2.5% inhibition by quinpirole; n ϭ 8). Point mutations of the individual amino acids in the QRART sequence (Q51A, R52A, and R54A) subsequently identified both arginines as being critical for G-protein modulation, because either mutation produced a construct that showed almost complete loss of inhibition by quinpirole (Fig. 5 B, C).
The N-terminal part of this 11 amino acid sequence also contains residues that are critical for G-protein modulation. When YKQSI was mutated to AAAAA (Fig. 5A), the channel was not G-protein-modulated [0.3 Ϯ 2.1% inhibition by quinpirole (n ϭ 4); Fig. 5C]. To confirm the importance of the amino acids 45-50 (YKQSIA), an intermediate deletion ␣1B ⌬N2-50 was made, to give a construct starting with methionine followed by Q51. This was also found not to be G-protein-modulated (Fig.  5C). Subsequent point mutations were made of the individual amino acids in the YKQSIA sequence to A (Y45A, K46A, Q47A, S48A, and I49A). This identified only the serine and, to a lesser extent, isoleucine in the sequence as being involved in G-protein modulation. These mutations resulted in reduced quinpiroleinduced inhibition of I Ba to 4.5 Ϯ 1.0% (n ϭ 6) for S48A and 17.4 Ϯ 2.1% (n ϭ 11) for I49A, respectively (Fig. 5C). Although the individual point mutants Y45A, K46A, and Q47A were all strongly G-protein-modulated by quinpirole, the modulation of the construct containing the triple mutation YKQ3 AAA was reduced (18.8 Ϯ 3.9% inhibition by quinpirole; n ϭ 5; Fig. 5C), indicating an influence of these amino acids.

Modulation of N-terminal mutants of ␣1B by G␤␥
We have confirmed that the identified amino acids are similarly involved in direct G␤␥-induced modulation of ␣1B by performing The parameters determined for the different ␣1 constructs (cotransfected with ␣2␦, and either with or without ␤2a) were measured as described in Materials and Methods and in the legend to Figure  1. Individual current density-voltage relationships were fitted with a Boltzmann equation where G max is the maximum conductance; V rev is the reversal potential; k is the slope factor, and V 50 is the voltage for 50% current activation. No systematic variation was seen in k, which was between 4.5 and 8 for all constructs, or in V rev , which was between ϩ45 and ϩ51 mV for all constructs. For simplicity, these values are not given. The current amplitude is given at the peak of the I-V relationship, which is between 0 and ϩ10 mV. experiments with coexpressed G␤1␥2 in C OS-7 cells. Examples of results obtained are shown in Figure 6. For the Q47A mutation, G-protein modulation was still observed, with slowly activating currents in P1 and a clear hyperpolarizing shift in the V 50 for current activation resulting from the depolarizing prepulse ( Fig. 6 A). In contrast, for the R52A mutation, no G-protein modulation was observed (Fig. 6 B). The mean results for all the constructs are given in Figure 7, expressed as P2/ P1 facilitation ratio at Ϫ10 mV (Fig. 7A). The V 50 for the I Ba current-voltage relationship was also plotted, because this shows a depolarizing shift in G-protein-modulated channels, compared to the control ␣1B expressed in the absence of G␤␥ (Fig. 7B). These two sets of measurements are strongly correlated (r ϭ 0.76, data not shown), and the depolarizing shift in activation V 50 is also highly correlated to the percentage inhibition by quinpirole observed for the same constructs in the Xenopus oocyte experiments (Fig. 7C), suggesting that direct G␤␥ modulation and quinpirole-induced modulation of these constructs are using the same mechanism. The I49A mutation stands out in both these systems as pro-ducing a reduction, but not a complete inhibition of modulation (Fig. 7C).

Basis for the reduction in receptor-mediated modulation of the N-terminal point mutation I49A
G-protein modulation of calcium channels is strongly voltagedependent, in that more inhibition is observed at low than at high depolarizations (Bean, 1989). To examine the basis for the reduced modulation of the partially modulated mutant (␣1B I49A) compared to ␣1B, we first examined, in Xenopus oocytes, the voltage dependence of the removal of inhibition by quinpirole, during a depolarizing prepulse (see Fig. 8 A for voltage protocol). There was no significant effect of the I49A mutation on the voltage dependence of the prepulse-induced facilitation in the presence of quinpirole (Fig. 8 B), or on the time course of removal of quinpirole-induced inhibition during a 100 mV depolarizing prepulse. Single exponential fits gave values for removal of inhibition (possibly representing dissociation of G␤␥ at this depolarized potential) of ϳ20 msec for both constructs (Fig. 8C). ; and E, ␣1bCcCCC. E xperiments were performed both in the presence (white bars) and the absence (black bars) of overexpressed ␤2a, except for ␣1bCcCCC, where no expression was seen in the absence of ␤2a. The numbers of experiments (with, without ␤2a) are 8, 6 ( A); 12, 6 ( B); 7, 6 (C); 10, 6 ( D); and 9 (E). The statistical significance of the differences at each potential between inhibition in the presence and absence of ␤2a is indicated by *p Ͻ 0.05; **p Ͻ 0.01. E xample currents in the presence of ␤2a are given as insets to parts A-C for ␣1B and for all the chimeras shown. They were expressed as described in the legend to Figure 1. These traces were evoked by a pulse from Ϫ100 to 0 mV, and therefore do not show the maximum inhibition. Traces are shown before (con) and during quinpirole (100 nM) application (quin). F, E xample traces showing the lack of effect of quinpirole on ␣1C, ␣1cBbCCC, ␣1cBcCCC, and ␣1cCbCCC, all expressed with ␤2a. The calibration bars are all 50 msec and 500 nA, unless otherwise stated.
The only clear difference between I49A ␣1B and wild-type ␣1B was in the more rapid time course of reinstatement of G-protein modulation after its removal by a 100 msec prepulse to ϩ100 mV (Fig. 8 D). This could be fit to a single exponential with reinhibition of 187 msec for ␣1B and 85 msec for the I49A mutant (Fig. 8 D,  inset).
If we consider G-protein modulation as a simple bimolecular reaction, as has been done previously (Z hang et al., 1996;Stephens et al., 1998a): where C is one of the closed states of the calcium channel ␣1B subunit, k 1 is the association rate constant, and k Ϫ1 is the dissociation rate constant for G␤␥. At equilibrium, from the law of mass action: at the holding potential, 1/ reinhibition ϭk 1 [G␤␥]ϩk Ϫ1 (1) and steady state inhibition ϭ k 1 ͓G␤␥͔/͑k 1 ͓G␤␥͔ ϩ k Ϫ1 ͒ From our previous study (Stephens et al., 1998a), we estimated the G␤␥ concentration to reach ϳ300 nM, when G␤␥ was overexpressed. Taking an approximate value of 100 nM for the concentration of G␤␥ resulting from quinpirole-induced receptor activation in the present study [a value of 130 nM can be calculated making the assumptions described in Stephens et al. (1998a)], we can obtain estimates by substitution of steady-state inhibition and reinhibition values into Equations 1 and 2, of k 1 and k Ϫ1 . For ␣1B, k 1 is 25.2 M Ϫ1 sec Ϫ1 , and k Ϫ1 is 2.8 sec Ϫ1 , whereas for the I49A mutant k 1 is 21.0 M Ϫ1 sec Ϫ1 , and k Ϫ1 is 9.6 sec Ϫ1 . Clearly, the major difference is an apparent 3.4-fold increase in the off-rate for G␤␥ in the I49A mutant. However, assuming that reassociation of G␤␥ is very slow at ϩ100 mV, the dissociation of G␤␥ during the prepulse to ϩ100 mV, found from Figure 8C, is 53.2 sec Ϫ1 for ␣1B and 46.7 sec Ϫ1 for ␣1B I49A, indicating that the apparent off-rate is more rapid for both constructs at this depolarized potential [as previously observed in Stephens et al. (1998a)], and the difference between the parental ␣1B and the I49A mutant is lost. Figure 3. Examples of direct modulation by G␤1␥2 of the chimeras between ␣1B and ␣1C. The ␣1 subunits shown were coexpressed with ␣2␦, ␤2a, G␤1, and G␥2 in C OS-7 cells. Lef t panel, Traces obtained before and after a depolarizing prepulse (ϩ120 mV, 100 msec). The prepulse protocol is above the top trace. Right panel, Current-voltage relationships (steps from Ϫ40 to ϩ50 mV in 10 mV intervals, from a holding potential of Ϫ100 mV), measured 50 msec after the start of the step, for the currents in P1 (open circle) and P2 ( filled circle). The current-voltage relationships were fitted (solid lines) with a modified Boltzmann equation as given in the legend to Table 1. The mean depolarizing shifts in V 50 resulting from the depolarizing prepulse are given in Figure 4. A, Currents resulting from ␣1B expression (currents shown resulting from steps Ϫ40 to 0 mV, and recorded in 1 mM Ba 2ϩ ). B, Currents resulting from ␣1bCbCCC expression (steps Ϫ40 to ϩ20 mV shown, recorded in 10 mM Ba 2ϩ ). C, Currents resulting from ␣1bCcCCC expression (steps Ϫ40 to ϩ20 mV shown, recorded in 10 mM Ba 2ϩ ). D, Currents resulting from ␣1C expression (steps Ϫ40 to Ϫ10 mV shown, recorded in 1 mM Ba 2ϩ ). In this example the depolarizing prepulse was not preceded by a 10 msec step to the holding potential, but this had no effect on the subsequent results. Histogram giving the mean Ϯ SEM of the hyperpolarizing shifts in V 50 after a depolarizing prepulse for the same chimeras as in Figure 1. *p Ͻ 0.05; **p Ͻ 0.001 compared to ␣1C/G␤1␥2. All ␣1B currents were recorded with 1 mM Ba 2ϩ and all chimeras with 10 mM Ba 2ϩ as charge carrier. It was checked for parental ␣1B that the use of 1 or 10 mM Ba 2ϩ did not affect the ⌬V 50 caused by a depolarizing prepulse. For the bars marked control, the parental constructs were expressed without G␤␥ subunits, in the presence of GDP␤S (1 mM), and a small prepulse-induced hyperpolarizing shift in V 50 was observed for ␣1B and ␣1C. A similar control shift was also observed for all the chimeras tested [for example for ␣1bCbCCC the control ⌬V 50 was Ϫ2.7 Ϯ 0.8 mV (n ϭ 8)]. This shift was not significantly different from that for ␣1C coexpressed with G␤1␥2. The number of experiments performed is given at the base of each bar.

DISCUSSION
The molecular determinants for the inhibition of neuronal VDCC ␣1 subunits by G␤␥ have been the subject of several studies. However, there remains no consensus of opinion concerning the functional importance of biochemically identified G␤␥-binding sites on the I-II loop and C terminus (De Waard et al., 1997;Page et al., 1997;Qin et al., 1997;Zamponi et al., 1997) (for review, see Dolphin, 1998). Furthermore, there has been little agreement on the extent of modulation of the E-type VDCCs (Bourinet et al., 1996;Toth et al., 1996;Yassin et al., 1996;Mehrke et al., 1997;Page et al., 1997;Qin et al., 1997). However, following our recent study (Page et al., 1998), it now seems clear that all ␣1E orthologues are G-protein-modulated when the long N terminus is present.

Requirement for the N terminus of ␣1B for G-protein modulation
The present study was performed to further our understanding of the involvement of the N terminus of the VDCC ␣1B in G-protein modulation, first identified by Page et al. (1998). We therefore made a series of chimeras between ␣1B, which is strongly G-protein-modulated, and ␣1C, which is not modulated by G␤␥, in the systems studied. Our conclusions are that the N terminus of ␣1B is absolutely essential for its G-protein modulation. No modulation was observed of any channel that contained the N terminus from ␣1C. The sequences of the intracellular N termini of ␣1B and ␣1C show little homology, and it is thus clear that the N terminus of ␣1B plays a role in G-protein modulation that cannot be substituted by that of ␣1C.

Role of the I-II linker and first transmembrane domain of ␣1B in G-protein modulation
In contrast to the results concerning the N terminus, the I-II linker of ␣1B was not completely essential; significant G-protein modulation was observed in the chimeras ␣1bBcCCC and ␣1bCcCCC, although the extent of modulation was less than for the control ␣1B. These results are of mechanistic interest because of the inability of the ␣1C I-II linker to bind G␤␥ (De Waard et al., 1997;Qin et al., 1997;Dolphin et al., 1999), presumably because of the lack of the QxxER-binding motif. It is possible that G␤␥ binding to the I-II linker of ␣1B increases its concentration close to its site of action, but is not directly involved in its functional effects. Both the I-II linker and the first transmembrane domain of ␣1B are, however, essential for the observation of a reduction of G-protein modulation by overexpression of exogenous ␤2a subunit in the Xenopus oocyte system. Whereas ␣1B itself and ␣1bBbCCC showed significantly greater modulation by quinpirole in the absence of coexpressed VDCC ␤2a subunit, the ␣1bBcCCC and ␣1bCbCCC chimeras exhibited a similar degree Figure 5. The effect of various deletions and point mutations of the N terminus of ␣1B on inhibition of I Ba by the D2 agonist quinpirole. The sequence of the N terminus of ␣1B, with the 11 amino acid sequence identified as being involved in G-protein modulation in bold, and the points at which deletions were made shown by arrows beneath the sequence. E xample traces, showing the effect of quinpirole (100 nM) on I Ba in the ␣1B ⌬2-44 mutant (lef t), the ␣1B ⌬45-55 mutant (center lef t), the ␣1B I49A mutant (center right), and the ␣1B R54A mutant (right). Traces (100 msec duration) were obtained at a test potential of 0 mV, from a holding potential of Ϫ100 mV. Con, Control traces; quin, after perf usion of quinpirole. Histogram of the percentage inhibition by 100 nM quinpirole (mean Ϯ SEM) of I Ba in the various deletion and point mutants of the N terminus of ␣1B. The currents were activated at 0 mV, and the degree of inhibition was determined from the currents activated every 15 sec. The number of experiments for each condition is given in parentheses, and the significance of the differences compared to the inhibition of ␣1B are given by *p Ͻ 0.005. of inhibition by quinpirole in the presence and absence of coexpressed ␤2a. The mechanism of this partial antagonism by ␤2a remains unclear, but is not completely shared by other ␤ subunits such as ␤1b (C. C anti and A. C. Dolphin, unpublished results).
The first transmembrane domain of ␣1B clearly has a role in G-protein modulation, as suggested previously (Z hang et al., 1996;Stephens et al., 1998b). We have found that, although it can be substituted by that of ␣1C, the ␣1bCbCCC chimera is less modulated than ␣1bBbCCC by quinpirole in the Xenopus oocyte system. It is possible that the first transmembrane domain mediates the effects of G␤␥ subunits to slow current activation, via interference with the f unction of its voltage sensor. Evidence suggests that only one G␤␥ subunit binds per ␣1 subunit, in a voltage-dependent manner (Stephens et al., 1998a;Z amponi and Snutch, 1998). We previously estimated the off-rate (k Ϫ1 ) of G␤␥ subunits to be ϳ1.3 sec Ϫ1 at Ϫ100 mV and 50 sec Ϫ1 at ϩ120 mV (Stephens et al., 1998a). Thus, the binding of G␤␥ is probably of higher affinity to the channel with the voltage sensors in their resting state. The action of G␤␥ subunits is to delay channel opening and to produce a depolarizing shift in the voltage dependence of activation (Patil et al., 1996). Presumably, this is achieved either by slowing the movement of the voltage sensors (and the IS4 sensor in particular), in response to a change in transmembrane voltage, or reducing the efficiency of coupling of the voltage sensor to channel opening (Jones et al., 1997).
Some of our chimera results and conclusions do not agree with those of a previous work (Furukawa et al., 1998), which also made a chimera with the I-II linker of ␣1B in ␣1C and showed it to be G-protein-modulated, thus indicating that the I-II linker alone could mediate G-protein modulation. However, their chimera, Figure 7. Mean effect of ␣1B deletions and mutations on G␤␥ modulation in C OS-7 cells. A, The P2/ P1 ratio was determined in COS-7 cells overexpressing G␤1␥2, from current amplitudes during steps to Ϫ10 mV before and after a depolarizing prepulse (ϩ120 mV, 100 msec), for the same N-terminal deletions and point mutations shown in Figure 5. Comparison is made with ␣1B in the absence of G␤␥, recorded with 1 mM GDP␤S in the patch pipette (1). The value [(P2/ P1) Ϫ 1] is plotted, which will be 0 if there is no facilitation. B, The activation V 50 was determined for the same constructs coexpressed with G␤1␥2, and compared to the value for ␣1B in the absence of G␤␥, recorded with 1 mM GDP␤S in the patch pipette ( 1). The dashed lines are 1 SEM more positive than the mean value for ␣1B ( 1), and 1 SEM more negative than the mean value for ␣1B/G␤␥ ( 2). C, Correlation between ⌬ activation V 50 (the data given in B, after subtraction of the V 50 for ␣1B) on the y-axis, and the data from Figure 5C (percentage inhibition of I Ba by 100 nM quinpirole), on the x-axis. The numbers identif ying the constructs refer to the bars in A and B. Regression analysis (dotted line) gives a coefficient, r of 0.92 ( p Ͻ 0.001). The data divide into a group of modulated and a group of nonmodulated constructs, as identified, except for constructs 14 (I49A) and 9 (YKQ3 AAA). together with other chimeras described in their paper, involved substitution of more than just the I-II linker of ␣1B. In the chimera in question, a region of ␣1B from part of IS5 through to IIS2 was substituted into ␣1C, with in addition several amino acid substitutions and deletions, and the results are thus not directly comparable. Furthermore, in their study the reciprocal chimera, made up of ␣1B with a region including the I-II linker of ␣1C, was also G-protein-modulated.
Our finding is that substitution into ␣1C of the region from the N terminus to the end of the I-II linker of ␣1B (␣1bBbCCC) does not produce a channel that is as strongly modulated as ␣1B in the Xenopus oocyte assay, although in the G␤␥ overexpression assay, there was no significant difference between ␣1bBbCCC and ␣1B. Thus, it is likely that other regions in the rest of ␣1B may also contribute to the extent of G-protein modulation of ␣1B, possibly including the C terminus Hamid et al., 1999), which may form part of a complex G␤␥-binding pocket.

Amino acids in the N terminus of ␣1B that are critical for G-protein modulation
From our mutational study of the N terminus, we identified the sequence between amino acids 45 and 55 (YKQSIAQRART) as being essential for G-protein modulation, because a deletion to amino acid 55 produced a construct that showed no G-protein modulation (Page et al., 1998), whereas a channel truncated to amino acid 44 was fully modulated, and a deletion of these 11 amino acids (45-55) resulted in a nonmodulated construct. Subsequently, we have identified three amino acids within this sequence, S48, R52, and R54, that when mutated to alanine, markedly reduce G-protein modulation of ␣1B, and a fourth amino acid (I49) that also shows an involvement. The substitution of just two amino acids (R52A, R54A) completely abolished G-protein modulation, whereas constructs containing the individual mutations still showed a small degree of modulation (4 and 9% inhi- Figure 8. Voltage dependence of inhibition, rate of loss of inhibition, and reinhibition rate for ␣1B and ␣1B I49A in Xenopus oocytes. A, Voltage protocol, showing variation of the prepulse voltage ( V ), the prepulse duration (⌬t dep ), and the interpulse interval (⌬t inter ) between the prepulse and the test pulse. The prepulse potential was 100 mV and 50 msec duration, and the interpulse interval was 20 msec, unless these parameters were varied. B, Effect of increasing the 50 msec prepulse voltage ( V ) on prepulse facilitation in the presence of quinpirole. Facilitation was measured as (I Ba in P2) Ϫ (I Ba in P1) and normalized to the maximum facilitation observed (normalized ⌬I). ␣1B (open circles), ␣1B I49A ( filled circles). The inset histogram gives the V 50 values (mean Ϯ SEM, determined by fitting Boltzmann f unctions to the data from the number of individual experiments given above each bar) for ␣1B (white bar) and ␣1B I49A (black bar). C, Effect of increasing the duration of the 100 mV prepulse (⌬t dep ) on prepulse facilitation in the presence of quinpirole. Facilitation was measured as described in B. ␣1B (open circles), ␣1B I49A ( filled circles). The inset histogram gives the dissociation values (mean Ϯ SEM, determined by fitting a single exponential to the data from the number of experiments given above each bar) for ␣1B (white bar) and ␣1B I49A (black bar). D, Effect of increasing the interval between the 100 mV, 50 msec prepulse and the subsequent test pulse P2 on the facilitation in the presence of quinpirole. Facilitation was measured as described in B: ␣1B (open circles), ␣1B I49A ( filled circles). The inset histogram gives the reinhibition values (mean Ϯ SEM, determined by fitting a single exponential to the data from the number of experiments given above each bar) for ␣1B (white bar) and ␣1B I49A (black bar). bition by quinpirole, respectively). The R52A and R54A constructs also individually showed some slowing of current activation when coexpressed with G␤␥, whereas the double mutant did not (Fig. 6 B; results not shown). The R AR motif is reminiscent of the R AK motif found in one of the G␤␥-binding sites on GIRK4 (Krapivinsky et al., 1998).
The I49A mutation stands out as producing a reduction in G-protein modulation in both systems (Fig. 6C). It is of interest that the 11 amino acid motif we have identified is identical in rat ␣1E and ␣1A, except for I49, whose equivalent is lysine in ␣1E, and methionine in ␣1A. Furthermore, both ␣1A and ␣1E long show less G-protein modulation than ␣1B in a number of systems (Zamponi et al., 1997;Page et al., 1998), possibly involving this amino acid substitution. In the present study we have observed that the reinhibition after a depolarizing prepulse is more than twice as fast for ␣1B I49A (85 msec) than for ␣1B (187 msec; Fig.  8 D). However, in our previous study we observed that the reinhibition for both ␣1B and ␣1E long was ϳ95 msec (Page et al., 1998). We are currently re-examining the comparison between ␣1B and ␣1E long under the present conditions (5 mM Ba 2ϩ , BAPTAinjected oocytes), to investigate whether our previous lack of observation of any difference in reinhibition between ␣1B and ␣1E long was caused by an influence of niflumic acid, which we have subsequently found to affect G-protein modulation of ␣1B currents.
It is possible that the N terminus forms a G␤␥ or VDCC ␤-binding site, or it may be involved in the downstream effects of G␤␥ binding. We have observed that inactivation is increased in a number of the ␣1B mutants, suggesting an impairment of interaction with ␤2a (G. J. Stephens and A. C. Dolphin, unpublished results). However, one can consider that these mutations in the N terminus of ␣1B may alter the binding affinity for G␤␥ (see Results for I49A). From this analysis the major difference is an apparent 3.4-fold increase in the off-rate for G␤␥ in the I49A mutant. Thus, YKQSIAQRART may form part of a G␤␥-binding site, with I49 playing a modulatory role in binding affinity, or it may be involved in the interaction between G␤␥ and VDCC ␤ subunits.