The importance of occupancy rather than affinity of Ca_Vβ subunits for the calcium channel I–II linker in relation to calcium channel function

Abstract

The Ca_Vβ subunits of voltage-gated calcium channels regulate the trafficking and biophysical properties of these channels. We have taken advantage of mutations in the tyrosine residue within the alpha interaction domain (AID) in the I–II linker of Ca_V2.2 which reduce, but do not abolish, the binding of β1b to the AID of Ca_V2.2. We have found that the mutation Y388S decreased the affinity of Ca_Vβ1b binding to the Ca_V2.2 I–II linker from 14 to 329 nm. However, the Y388S mutation had no effect on current density and cell surface expression of Ca_V2.2/α2δ-2/β1b channels expressed in human embryonic kidney tsA-201 cells, when equivalent proportions of cDNA were used. Furthermore, despite the 24-fold reduced affinity of Ca_Vβ1b for the Y388S I–II linker of Ca_V2.2, all the key features of modulation as well as trafficking by Ca_Vβ subunits remained intact. This is in contrast to the much more marked effect of the W391A mutation, which abolished interaction with the Ca_V2.2 I–II linker, and very markedly affected the trafficking of the channels. However, using the Xenopus oocyte expression system, where expression levels can be accurately titrated, when Ca_Vβ1b cDNA was diluted 50-fold, all evidence of interaction with Ca_V2.2 Y388S was lost, although wild-type Ca_V2.2 was still normally modulated by the reduced concentration of β1b. These results indicate that high affinity interaction with the α1 subunit is not necessary for any of the modulatory effects of Ca_Vβ subunits, but occupancy of the interaction site is important, and this will occur, despite the reduced affinity, if the Ca_Vβ subunit is present in sufficient excess.

Voltage-gated calcium (Ca_V) channels are key components of all excitable cells. Three families of voltage-gated calcium channels have been identified, Ca_V1–3. The Ca_V1 and Ca_V2 classes are both high-voltage activated (HVA). These channels are heteromultimers composed of the pore-forming α1 subunit, associated with auxiliary Ca_Vβ and α₂δ subunits (Catterall, 2000). Ca_Vβ subunits are crucial for normal HVA channel function (for review see Dolphin, 2003), since they are required for the expression of functional channels at the plasma membrane (Bichet et al. 2000), and modulate their biophysical properties (for review see Dolphin, 2003). The Ca_V2 family calcium channels are inhibited by Gβγ dimers (Herlitze et al. 1996; Ikeda, 1996) which is the main mechanism of presynaptic inhibition by G-protein-coupled receptors (GPCRs). Ca_Vβ subunits promote the voltage dependence of modulation of Ca_V2.2 calcium channels by Gβγ dimers, although the mechanism involved remains unclear (Meir et al. 2000; Canti et al. 2001; Leroy et al. 2005).

We have previously investigated the role of Ca_Vβ subunits in the plasma membrane expression and modulation of Ca_V2.2 calcium channels, by mutating the tryptophan (W391 in Ca_V2.2) that is conserved in the AID sequence of all Ca_V1 and 2 channels. This sequence is QQIERELNGYLEWIFKAE in Ca_V2.2. This tryptophan has been shown both by the original study that identified the AID motif and by the recent structural studies to be key to the interaction between Ca_Vβ subunits and the AID (Pragnell et al. 1994; Chen et al. 2004; Opatowsky et al. 2004; Van Petegem et al. 2004). Our results showed that the W391A mutation reduced the binding of β1b to the I–II linker by at least 1000-fold, prevented the enhancement of functional expression of Ca_V2.2 by Ca_Vβ1b, and also prevented modulation of the biophysical properties of Ca_V2.2 by this subunit. In addition, although the G-protein modulation of Ca_V2.2 W391A was present, it was not voltage dependent (Leroy et al. 2005).

We have now investigated the role of the AID tyrosine Y388 in the role of Ca_Vβ subunits and in G protein modulation, since the crystal structure showed the W and Y to form a hairpin arrangement with their aromatic rings stacked together (Chen et al. 2004; Opatowsky et al. 2004; Van Petegem et al. 2004). The Y residue has previously been described as essential for Ca_Vβ binding to the AID and for functional expression (Pragnell et al. 1994; Witcher et al. 1995). However, a subsequent study disputed the importance of this residue in β subunit-induced modulation of Ca_V1.2 currents (Gerster et al. 1999), and its role remains open to question. Since in our previous study, measurement of Ca_Vβ binding to the I–II linker by surface plasmon resonance correlated extremely well with the maximum conductance (G_max) values for Ca_V2.2 currents and with cell surface biotinylation for the W391A mutation (Leroy et al. 2005), we performed similar studies following mutation of Y388. Our results allow us to conclude that there is no requirement for high affinity binding of Ca_Vβ to the AID, since this is reduced 24-fold by the mutation Y388S. However, occupancy of the site is a key factor, since reducing the concentration of β1b by 50-fold relative to Ca_V2.2 Y388S removed all influence of β1b on this channel, whereas the wild-type Ca_V2.2 was still modulated at this concentration of β1b.

Methods

Materials

The cDNAs used in this study were Ca_V2.2 (D14157), Ca_Vβ1b (X61394), α2δ-2 (Barclay et al. 2001) and D₂ dopamine receptor (X17458). The green fluorescent protein (GFP-mut3b, U73901) was used to identify transfected tsA-201 cells. All cDNAs were subcloned in pMT₂.

Construction, expression and purification of proteins

Ca_V2.2 Y388S was generated using standard molecular biological methods. The Y388S, Y388F and W391A mutations were introduced into Ca_V2.2 I–II loop in pGEX2T (Bell et al. 2001) (GST Ca_V2.2 I–II loop) by site-directed mutagenesis using standard molecular biological methods. The resulting mutated I–II linkers and the wild-type linker were subcloned into pETM6T1 (gift from Dr A. Okorokov, UCL), which encodes an N-terminal NusA-tag and a His-tag, using BamH I and EcoR I, generating NusA Ca_v2.2 I–II loop fusion proteins.

NusA Ca_v2.2 I–II loop fusion proteins were expressed in BL21(DE3) codon plus (RIL) E. coli in 1 litre cultures of LB medium containing 30 μg ml⁻¹ kanamycin, 34 μg ml⁻¹ chloramphenicol and 1% (w/v) glucose. Cultures were grown to optical density 0.6 at 560 nm and induced at 20°C for 4 h using 0.1 mm isopropyl-β-d-thiogalactopyranoside. For protein purification, cells were lysed by sonication in buffer A (20 mm Hepes pH 8.0, 150 mm NaCl, 1 mm DTT, 20 mm imidazole) containing protease inhibitors (Complete EDTA-free, Roche Diagnostics, Lewes, UK). The lysate was cleared by centrifugation at 20 000 g for 20 min and the resulting pellet was washed once with buffer A. The pellet was resuspended in buffer A containing 1.5% CHAPS and incubated for 1 h at 4°C. After centrifugation at 20 000 g for 20 min the supernatant was diluted 1 : 1 with buffer A and applied to 1 ml Ni–NTA resin (Qiagen, Crawley, West Sussex, UK) equilibrated with buffer B (buffer A containing 0.5% CHAPS). The column was washed with 25 volumes buffer B before proteins were eluted with 4 volumes buffer B containing 350 mm imidazole. Eluted proteins were analysed by SDS-PAGE followed by Coomassie blue staining. C-terminally His-tagged Ca_vβ1b (Ca_vβ1b-H6C) was expressed and purified as described by Bell et al. (2001).

Surface plasmon resonance

Assays were performed using a BIAcore 2000 (Biacore AB, Uppsala, Sweden) at 25°C using running buffer (20 mm Hepes, pH 7.5, 150 mm NaCl, 0.1% CHAPS). NusA fusion proteins and NusA only were immobilized directly onto the surface of a CM5 sensor chip. Using a 1 : 1 mixture of 400 mm 1-ethyl-3-[3-dimethyl-aminopropyl]carbodiimide hydrochloride and 100 mmN-hydroxysuccinimide to activate the chip surface, 2000 reference units (RU) of NusA I–II loop fusions and the molar equivalent of NusA were immobilized. Ca_vβ1b-H6C was dialysed against running buffer and diluted to the required concentrations in running buffer. These samples were applied for 5 min at a flow rate of 50 μl min⁻¹ over all flow cells and each injection was followed by a 5 min dissociation phase. The sensor chip surface was regenerated between injections by the application of 30 μl of 20 mm glycine pH 2.2 at 10 μl min⁻¹. Sensorgrams were processed using the BIAevaluation 3.0 software (Biacore AB). Sensorgrams recorded from the flow cells containing NusA Ca_v2.2 I–II loop, either wild-type, Y388S, or Y388F were corrected for passive refractive index changes and for non-specific interactions by subtraction of the corresponding sensorgram recorded from the flow cell containing NusA only. Sensorgrams were analysed using Biacore kinetic analysis software using a model of 1 : 1 interaction. In addition, the maximum responses for the Ca_v2.2 I–II linker and both mutants after 250 s of sample injection were plotted against Ca_vβ concentration. The resulting curves were analysed by fitting a rectangular hyperbola, using Origin 7 (Microcal Software, Northhampton, MA, USA), and the affinity constant K_D was estimated. The dissociation phase of the sensorgrams was also fitted to a single exponential, to determine the dissociation rate, k_off.

Cell culture and heterologous expression

The tsA-201 cells were cultured in a medium consisting of Dulbecco's modified Eagle's medium, 10% fetal bovine serum, and 1% non-essential amino acids. The cDNAs (all at 1 μg μl⁻¹) for Ca_Vα1 subunits, Ca_Vβ, α₂δ-2, D₂ dopamine receptor and GFP (when used as a reporter of transfected cells) were mixed in a ratio of 3 : 2 : 2 : 2 : 0.4. The cells were transfected using Fugene6 (Roche Diagnostics, Lewes, UK; DNA/Fugene6 ratio of 2 μg in 3 μl).

Cell surface biotinylation and Western blotting

Cell surface biotinylation experiments were carried out as described in Leroy et al. (2005). For Western blotting, samples (10 μg) from tsA-201 whole-cell lysates from biotinylation experiments (see above) were separated by SDS-PAGE on 4–12% Tris-glycine gels and then transferred to polyvinylidene fluoride membranes. Immunodetection was performed with antibodies to the Ca_v2.2 II–III linker as previously described (Raghib et al. 2001).

Whole-cell patch clamp in tsA-201 cells

The tsA-201 cells were replated at low density on 35 mm tissue culture dishes on the day of recording. Whole-cell patch-clamp recordings were performed at room temperature (22–24°C). Only fluorescent cells expressing GFP were used for recording. The single cells were voltage clamped using an Axopatch 200B patch-clamp amplifier (Axon Instruments, Inc., Union City, CA, USA). The electrode potential was adjusted to give zero current between pipette and external solution before the cells were attached. The cell capacitance varied from 10 to 40 pF. Patch pipettes were filled with a solution containing (mm): 140 caesium aspartate, 5 EGTA, 2 MgCl₂, 0.1 CaCl₂, 2 K₂ATP, 10 Hepes, titrated to pH 7.2 with CsOH (310 mosmol l⁻¹), with a resistance of 2–3 MΩ. The external solution contained (mm) 150 tetraethylammonium (TEA) bromide, 3 KCl, 1.0 NaHCO₃, 1.0 MgCl₂, 10 Hepes, 4 glucose, 10 BaCl₂, pH adjusted to 7.4 with Tris-base (320 mosmol l⁻¹). The pipette and cell capacitance, as well as the series resistance, were compensated by 80%. Leak and residual capacitance current were subtracted using a P/4 protocol. Data were filtered at 2 kHz and digitized at 5–10 kHz. The holding potential was −100 mV, and pulses were delivered every 10 s.

Two-electrode voltage clamp in Xenopus oocytes

Xenopus laevis females were killed by non-recovery anaesthesia in accordance with Schedule 1 to the Animals (Scientific Procedures) Act, 1986. Oocytes were then surgically removed and defolliculated by treatment with 1.5 mg ml⁻¹ collagenase type Ia in a Ca²⁺-free ND96 saline containing (mm): 96 NaCl, 2 KCl, 1 MgCl₂, 5 Hepes, pH adjusted to 7.4 with NaOH, for 2 h at 18°C. Ten nanolitres of cDNAs (1 ng nl⁻¹) were injected into the nuclei of stage V and VI oocytes. Injected oocytes were incubated at 18°C for 3–7 days in ND96 saline (as above plus 1 mm CaCl₂) supplemented with 100 μg ml⁻¹ penicillin and 100 i.u. ml⁻¹ streptomycin (Life Technologies, Gaithersburg, MD, USA). Further details have been previously described (Canti et al. 2001).

Whole-cell recordings from oocytes were performed in the two-electrode voltage-clamp configuration under continuous superfusion of a chloride-free solution containing (mm): 10 Ba(OH)₂, 50 NaOH, 10 TEA-OH, 2 CsOH, 5 Hepes (pH 7.4 with methanesulphonic acid). The oocytes were injected with 30–40 nl of a 100 mm solution of K₃-1,2-bis(aminophenoxy)ethane-N,N,N′,N′-tetra-acetic acid (BAPTA) to suppress endogenous Ca²⁺-activated Cl⁻ currents. Electrodes contained 3 m KCl and had resistances of 0.3–1.5 MΩ. The holding potential was −100 mV in all cases. The currents were amplified and low-pass filtered at 1 kHz by means of a Geneclamp 500 amplifier (Axon Instruments, Burlingame, CA, USA), digitized through a Digidata 1200 interface (Axon Instruments). All experiments were performed at 18–20°C.

Electrophysiological data analysis

Current amplitude was measured 10 ms after the onset of the test pulse, and the average over a 2 ms period was calculated and used for subsequent analysis. The current density–voltage (I–V) relationships were fitted with a modified Boltzmann equation as follows: I = G_max(V − V_rev)/(1 + exp(−(V − V_50,act)/k)) where I is the current density (in pA pF⁻¹), G_max is the maximum conductance (in nS pF⁻¹), V_rev is the reversal potential, V_50,act is the midpoint voltage for current activation, and k is the slope factor. Steady-state inactivation properties were measured by applying 5 s pulses from −120 to +20 mV in 10 mV increments, followed by a 11 ms repolarization to −100 mV before the 100 ms test pulse to +20 mV. Steady-state inactivation and activation data were fitted with a single Boltzmann equation of the form: I/I_max = (A₁ − A₂)/[1 + exp((V − V_50,inact)/k)] + A₂, where I_max is the maximal current, V_50,inact is the half-maximal voltage for current inactivation. For the steady-state inactivation, A₁ and A₂ represent the total and non-inactivating current, respectively. Inactivation kinetics of the currents were estimated by fitting the decaying part of the current traces with the equation: I(t) = C + Aexp(−(t − t₀)/τ_inact) where t₀ is zero time, C the fraction of non-inactivating current, A the relative amplitude of the exponential, and τ_inact, its time constant. Activation kinetics were estimated by fitting the activation phase of the current with either a single or a double exponential. Analysis was performed using pCLAMP6 and Origin 7. Data are expressed as mean ± s.e.m. of the number of replicates n. Error bars indicate standard errors of multiple determinations. Statistical significance was analysed using Student's paired or unpaired t test (as appropriate).

Results

Mutation of Y388S in the I–II linker of Ca_V2.2 reduces its affinity for the Ca_Vβ1b subunit

The amino acid Y388 in Ca_V2.2 is conserved in the AID sequence of all HVA calcium channels (Dolphin, 2003; Richards et al. 2004) and has been previously described to be crucial for the binding of the Ca_Vβ ancillary subunits to HVA calcium channels (Pragnell et al. 1994; Witcher et al. 1995; De Waard et al. 1996). The recent structural analysis of the interaction of Ca_Vβ subunits with the Ca_V1.2 I–II linker showed that the aromatic ring of the Y side chain is stacked with the side chain of the W residue and deeply embedded in the AID binding groove in Ca_Vβ (Chen et al. 2004; Opatowsky et al. 2004; Van Petegem et al. 2004).

We first examined by surface plasmon resonance analysis whether mutation of Y388 to either F or S in the AID of Ca_V2.2 affected the binding of Ca_Vβ1b to the I–II linker of Ca_V2.2. In these experiments, NusA fusion proteins corresponding to the entire I–II linker, including the AID of Ca_V2.2, Ca_V2.2 Y388S, Ca_V2.2 Y388F or NusA alone as control (Fig. 1A), were immobilized chemically onto individual flow cells of a CM5 dextran sensor chip. Ca_Vβ subunit solutions (20–1000 nm) were perfused over all flow cells. No concentration-dependent binding of the Ca_Vβ subunits to the control NusA fusion protein was detected (data not shown). Ca_Vβ1b exhibited specific binding to the full-length I–II linker of Ca_V2.2 (Fig. 1B). Significant binding of Ca_Vβ1b was also observed to both the Y388F and Y388S mutant I–II linkers (Fig. 1C and D). The dissociation constant (K_D) for Ca_Vβ1b binding to the I–II linker of Ca_V2.2 was calculated to be 13.7 nm for β1b binding to the wild-type I–II linker, and 78 and 329 nm for the Y388F and Y388S mutant I–II linkers, respectively, representing a 5.7-fold and a 24-fold reduction compared to the wild-type I–II linker (Fig. 1E). In contrast, negligible binding of the Ca_Vβ1b subunit to the Ca_V2.2 W391A I–II linker was detected, and thus the K_D values could not be determined (Fig. 1F), as we have previously shown for a GST fusion protein with a I–II linker construct truncated immediately after the AID sequence (Leroy et al. 2005). These results refine, rather than contradict, the findings of previous studies which indicated that mutation of Y to S in the AID sequence of other Ca_V channels abrogated Ca_Vβ subunit binding (Pragnell et al. 1994; Witcher et al. 1995; De Waard et al. 1996; Berrou et al. 2002), since all previous studies have used non-quantitative overlay or pull-down assays, where low affinity interactions may easily be missed.

Figure 1. The Y388F and Y388S mutant I–II linkers of Ca_V2.2 bind to Ca_Vβ1b with reduced affinity.

A, Coomassie blue-stained SDS-PAGE gel showing the purity of the wild-type and mutant NusA fusion proteins and the Ca_Vβ1b subunit used in surface plasmon resonance analysis. NusA only (lane 1), Ca_V2.2 I–II loop (lane 2), Ca_V2.2 Y388F (lane 3), Ca_V2.2 Y388S (lane 4), Ca_V2.2 W391A (lane 5) and Ca_Vβ1b (lane 6). Molecular weight markers are indicated. B–D, representative Biacore sensorgrams showing interactions between Ca_Vβ1b (20–1000 nm as indicated) with NusA I–II loop fusion proteins from Ca_V2.2 (B), Ca_V2.2 Y388F (C) and Ca_V2.2 Y388S (D). K_D values determined from the Biacore analysis software, assuming a 1 : 1 interaction, were 10.5, 64 and 200 nm, respectively, very similar to those determined from panel E. E, plot showing relationship between response at 250 s after β1b injection for the NusA Ca_V2.2 wild-type I–II linker (▪), Ca_V2.2 Y388F I–II linker (•) and Ca_V2.2 Y388S I–II linker (▴) and Ca_Vβ1b concentration. Curves were fitted using a rectangular hyperbola function, and the K_D values determined to be 13.7, 78.0 and 329 nm, respectively. F, representative Biacore sensorgrams showing negligible interaction between NusA Ca_V2.2 W391A I–II linker and Ca_Vβ1b at concentrations up to 1000 nm.

Single exponential fits were made to the dissociation phases of the sensorgrams, and the dissociation rate constants (k_off) of 20 nm Ca_Vβ1b from the I–II linkers of Ca_V2.2, Ca_V2.2 Y388F and Ca_V2.2 Y388S were calculated to be 8.1 × 10⁻³ s⁻¹, 16 × 10⁻³ s⁻¹ and 39 × 10⁻³ s⁻¹, respectively. As expected, there was little variation in the dissociation rates for each mutant across the range of Ca_Vβ1b concentrations used in these experiments. Our findings highlight the contributions made by the hydroxyl group and phenyl group of this Y residue to the high affinity interaction between AID and Ca_Vβ.

The effect of Y388S mutation on the functional expression of Ca_V2.2 calcium channels at the plasma membrane

Since the mutation of Y388S reduced the affinity for Ca_Vβ1b to a greater extent than the Y388F mutation, we used the Y388S mutation in full length Ca_V2.2 for further functional studies. We coexpressed full length Ca_V2.2 Y388S or wild-type Ca_V2.2, together with accessory Ca_Vβ1b and α₂δ-2 subunits and compared the biophysical properties of the wild-type and mutated channels. Surprisingly, and unlike the W391 mutant of Ca_V2.2 that we studied recently (Leroy et al. 2005), there was no significant difference in G_max determined from the current–voltage (I–V) relationships between wild-type Ca_V2.2 and Ca_V2.2 Y388S (Fig. 2A). The G_max of Ca_V2.2 Y388S was 97.5 ± 16% (n = 17) of that found for the Ca_V2.2/β1b/α₂δ-2 combination when coexpressed with β1b. Thus, the auxiliary β1b subunit was still able to significantly increase the G_max of Ca_V2.2 Y388S when compared with either the wild-type Ca_V2.2 expressed alone (Fig. 2A) or with the Ca_V2.2 Y388S/α₂δ-2 (n = 4, data not shown). In the transfection system used, the over-expression of Ca_Vβ may mean that the AID site is constantly occupied despite the 24-fold lower affinity of the Y388S mutant AID. Thus the high affinity interaction of Ca_Vβ, previously suggested to be essential for the trafficking to the plasma membrane may not be necessary, but occupancy of the site may be the most important factor.

Figure 2. Role of Ca_Vβ subunits in plasma membrane expression of Ca_V2.2.

A: upper panel, superposition of representative current traces for the subunit combinations indicated, recorded during 100 ms, 10 mV steps from −30 mV to +60 mV, from a holding potential of −100 mV. Same scale bars for the left and centre panels; lower panel, I–V relationships (mean ± s.e.m.) for Ca_V2.2/α₂δ-2 coexpressed with Ca_Vβ1b (▪, n = 29) or without a Ca_Vβ subunit (▴, n = 21), compared to the I–V relationship for Ca_V2.2 Y388S/α₂δ-2/Ca_Vβ1b (•, n = 17). The mean data are fitted with a modified Boltzmann function (see Methods). B, cell surface expression of either Ca_V2.2 or Ca_V2.2 Y388S expressed with α₂δ-2 and Ca_Vβ1b. Left panel, total expression of Ca_V2.2 in whole-cell lysate (WCL) is shown by Western blot in the upper row, and biotinylated Ca_V2.2, in the streptavidin agarose precipitate, is shown in the lower row. Cells were transfected with empty vector (lane 1), Ca_V2.2/α₂δ-2/β1b (lane 2) or Ca_V2.2 Y388S/α₂δ-2/β1b (lane 3). Right panel, bar chart showing quantification of the mean amount of Ca_V2.2 Y388S expressed at the plasma membrane when coexpressed with β1b (grey bar), given as a percentage of the amount of Ca_V2.2 expressed under the same conditions (white bar). Data are mean ± s.e.m. of 3 independent experiments. The cell surface expression of Ca_V2.2 was increased slightly by the Y388S mutation in all three experiments.

We then used a cell surface biotinylation assay to assess biochemically whether there was an alteration in the amount of channel protein present at the surface of the tsA-201 cells transfected with Ca_V2.2 Y388S and a Ca_Vβ subunit, compared with the wild-type Ca_V2.2/Ca_Vβ combination. The Ca_V2.2 Y388S mutation had no effect upon the total expression compared to wild-type Ca_V2.2, whereas the amount of biotinylated Ca_V2.2 Y388S channels at the plasma membrane was non-significantly elevated by 38 ± 19% (n = 3; Fig. 2B). Together these results show that the Y388S mutation in Ca_V2.2 has no detrimental effect upon the trafficking and functional expression of Ca_V2.2 channels. This implies that the Ca_Vβ1b can still associate with, and effectively exert its trafficking effects on, the Y388S mutant channel even though the affinity of the Y388S AID/Ca_Vβ interaction is reduced over 20-fold. This is in agreement with previous studies which have shown little or no effect of a Y to S mutation in AID of Ca_V1.1 or Ca_V2.3 channels on functional expression (Neuhuber et al. 1998; Berrou et al. 2002). However, it is in contrast to the earlier studies that identified the key amino acids responsible for Ca_Vβ subunit binding to the AID and showed that Y was one of the main amino acids whose mutation markedly reduced functional expression (Pragnell et al. 1994; De Waard et al. 1996).

Biophysical properties of Ca_V2.2 Y388S expressed with Ca_Vβ1b are similar to those of Ca_V2.2/β1b

As well as having effects on trafficking, Ca_Vβ1b is known to promote the activation of HVA calcium channels at more negative potentials (for review see Dolphin, 2003). In a previous study, we have estimated that the effects of Ca_Vβ subunits on the voltage-dependent properties of the channels occurred with an approximately 7-fold lower affinity than the effects on trafficking (Canti et al. 2001). It is therefore possible that Ca_V2.2 Y388S would be trafficked to the plasma membrane by Ca_Vβ1b, despite its reduced affinity, but would lose any interaction following trafficking. However, from the I–V relationships, the Ca_V2.2 Y388S had a similar V₅₀ for activation (+5.5 ± 1.5 mV; n = 17) to wild-type Ca_V2.2 in the presence of β1b (+6.0 ± 0.6 mV; n = 29), which were both significantly less positive compared with V₅₀ for activation for the wild-type Ca_V2.2 expressed alone (+14.4 ± 0.4 mV; n = 21, P < 0.05) (Fig. 2A). This indicates that the reduction in affinity conferred by the Y388S mutation was insufficient to alter the effect of Ca_Vβ1b on the voltage dependence of activation of the channel, and this mutant channel behaved like the wild-type Ca_V2.2, expressed with a Ca_Vβ1b subunit.

Another important feature of Ca_Vβ1, β3 and β4 subunits is that they hyperpolarize the steady-state inactivation curves of Ca_V2.2, as well as other HVA calcium channels (Bogdanov et al. 2000). The potential for half-inactivation (V_50,inact) was −45.4 ± 0.4 mV (n = 25) for Ca_V2.2 expressed with Ca_Vβ1b (Fig. 3), representing a −15 mV shift compared to Ca_V2.2 expressed in the absence of a β subunit. Again, we found no significant effect on the steady-state inactivation for Ca_V2.2 Y388S/β1b, as this was superimposed on that of Ca_V2.2 expressed with Ca_Vβ1b (Fig. 3A and B).

Figure 3. Inactivation properties of Ca_V2.2 and Ca_V2.2 Y388S coexpressed with Ca_Vβ1b.

A, voltage protocol and representative current traces to illustrate steady-state inactivation protocols. Inward Ba²⁺ currents were recorded after conditioning pulses of 5 s duration, applied from a holding potential of −100 mV in 10 mV steps between −120 and +10 mV, followed by a 50 ms test pulse to +20 mV. Same scale bars for the left and centre panels. B, voltage dependence of steady-state inactivation for Ca_V2.2/α₂δ-2 coexpressed with Ca_Vβ1b (▪), without any Ca_Vβ subunits (▴) or Ca_V2.2 Y388S/α₂δ-2 expressed with Ca_Vβ1b (•). The normalized data, obtained from recordings such as those shown in the upper panel, are plotted against the conditioning pulse (n = 10–25). The mean data are fitted with a Boltzmann function, whose V_50,inact values are given in the text. C, currents were recorded at +20 mV for 800 ms, and normalized to the peak current before averaging. Left panel, mean normalized current traces for Ca_V2.2 Y388S/α₂δ-2/β1b and wild-type Ca_V2.2/α₂δ-2/β1b combination. Right panel, mean τ_inact data for wild-type Ca_V2.2/α₂δ-2/β1b (filled bar, n = 14) and Ca_V2.2 Y388S/α₂δ-2/β1b (open bar, n = 14).

Ca_Vβ subunits are known to modulate not only the voltage dependence of the inactivation of calcium channels, but also the kinetics of current decay (for review see Dolphin, 2003). The time constant of inactivation (τ_inact) of the Ca_V2.2 Y388S/β1b current recorded during 800 ms pulses at +20 mV (550 ± 161 ms; n = 14) was the same as that of the wild-type Ca_V2.2/β1b combination (557 ± 88 ms; n = 14) (Fig. 3C). Altogether these results indicate that Ca_V2.2 Y388S is normally regulated by Ca_Vβ1b in the plasma membrane.

Ca_V2.2 Y388S channels show normal voltage-dependent modulation by G-protein activation

To investigate the importance of the Y388 residue for the modulation by G-proteins of Ca_V2.2 calcium channels, the wild-type or mutant Ca_V2.2, Ca_Vβ1b, α₂δ-2 combination was coexpressed with a D₂ dopamine receptor, and the receptor was activated with 100 nm quinpirole, which is a maximal concentration of this agonist. Figure 4A shows representative currents obtained before and during application of quinpirole, both before (P1) and immediately after (P2) a 100 ms prepulse to +100 mV. As previously observed (Leroy et al. 2005), the currents measured at +10 mV were inhibited by quinpirole by 64.2 ± 4.0% for the wild-type channel (n = 29; Fig. 4B). The P2/P1 ratio obtained from traces such as those represented in Fig. 4A reflects the voltage-dependent removal of G-protein inhibition. A value of 2.2 ± 0.1 (n = 29) was obtained for P2/P1 at +10 mV for the wild-type channel expressed with Ca_Vβ1b (Fig. 4C).

Figure 4. G-protein modulation of Ca_V2.2 and Ca_V2.2 Y388S currents.

A, the pulse protocol used consisted of a 100 ms test pulse (P1) from −30 mV to +60 mV applied from a holding potential of −100 mV. After 800 ms repolarization to −100 mV, a 100 ms prepulse to +100 mV was applied. The cell was repolarized for 20 ms to −100 mV and a second pulse (P2) identical to the first one was applied. Typical current traces obtained with this protocol are represented for Ca_V2.2 and Ca_V2.2 Y388S coexpressed with α₂δ-2 and Ca_Vβ1b, and for Ca_V2.2 expressed without a Ca_Vβ subunit. The D₂ dopamine receptor is coexpressed and the upper current traces (depicted by the open symbols) are in the presence of the agonist quinpirole (100 nm). B, I–V curves, obtained from currents evoked by P1, for the calcium channel combinations shown, obtained before (filled symbols) and during (open symbols) application of 100 nm quinpirole. I–V curves for Ca_V2.2/α₂δ-2 coexpressed with Ca_Vβ1b (squares, n = 29) or without β1b (triangles, n = 21) and for Ca_V2.2 Y388S/α₂δ-2 coexpressed with Ca_Vβ1b (circles, n = 17) are represented. I–V curves are fitted with modified Boltzmann functions whose V_50,act values are given in the Results. C, voltage-dependent facilitation was calculated by dividing the peak current value obtained in P2 by that obtained in P1 at the potentials of −10, 0, +10 +20 and +30 mV, for Ca_V2.2/α₂δ-2 with (black bars, n = 29) or without Ca_Vβ1b (white bars, n = 19) or for Ca_V2.2 Y388S/α₂δ-2 with Ca_Vβ1b (grey bars, n = 17) after application of quinpirole.

For the Ca_V2.2 Y388S/β1b currents, inhibition by quinpirole was 63.0 ± 8.5% at +10 mV (n = 17; Fig. 4A and B), and it showed similar voltage dependence to the wild-type currents, the P2/P1 ratio being 2.3 ± 0.2 at +10 mV (n = 17), very similar to that for Ca_V2.2/β1b (P < 0.01, Fig. 4C).

We have shown previously that lowering the concentration of expressed Ca_Vβ subunits leads to a slower rate of facilitation of the G-protein-modulated current, with two components of facilitation being present at intermediate Ca_Vβ concentrations (Canti et al. 2001). Thus a reduction in affinity of Ca_Vβ for the Ca_V2.2 Y388S channel might be manifested by a reduction in facilitation rate. We therefore determined the time constants of facilitation (τ_facil) by varying the prepulse duration during quinpirole application, and found that the τ_facil was very similar for the wild-type Ca_V2.2 (10.2 ± 0.7 ms; n = 11) and Ca_V2.2 Y388S (11.2 ± 0.8 ms; n = 12) (Fig. 5).

Figure 5. Facilitation rate of G-protein-modulated channels.

Left panels, the pulse protocol used to estimate the facilitation rate was similar to the one described in the legend to Fig. 4, but the duration of the prepulse was increased from 0 to 200 ms. The P2/P1 facilitation ratios are given for each prepulse, for Ca_V2.2 (▪, n = 11) and Ca_V2.2 Y388S (•, n = 12) coexpressed with Ca_Vβ1b. Data are fitted with a single exponential, whose time constant (τ_facil) is given in the text. Right panels, typical current traces obtained for Ca_V2.2/α₂δ-2 and for Ca_V2.2 Y388S/α₂δ-2, both coexpressed with Ca_Vβ1b and the D₂ dopamine receptor, are represented.

The interaction between Ca_V2.2 Y388S and Ca_Vβ1b is lost when the concentration of β1b is reduced

From the foregoing, it is clear that a 24-fold reduction in affinity of Ca_Vβ1b for the Ca_V2.2 AID containing the Y388S mutation is insufficient to have any effect on the ability of β1b to modulate the channel, by all the parameters we have studied, although we know from the W391A mutation that binding to the AID region is essential for these effects of β1b to occur (Leroy et al. 2005). We also know from our previous study in Xenopus oocytes that the amount of β1b expressed when Ca_V2.2 and β1b cDNAs are injected in an equivalent ratio is at least 30-fold in excess of that required to hyperpolarize the voltage dependence of steady-state inactivation of the entire channel population (Canti et al. 2001). We therefore examined the properties of wild-type Ca_V2.2 and Ca_V2.2 Y388S coexpressed with β1b either at a normal cDNA ratio or using 50-fold diluted β1b cDNA in Xenopus oocytes. For wild-type Ca_V2.2, the effects of both concentrations of β1b were identical, both in terms of peak current amplitude at +10 mV (Fig. 6A), and in terms of hyperpolarization of the steady-state inactivation (Fig. 6B). For the steady-state inactivation, the V_50,inact was −51.0 ± 1.1 mV for Ca_V2.2/β1b injected in a standard ratio (n = 10) and −46.8 ± 1.3 mV for Ca_V2.2/β1b using 50-fold-diluted β1b cDNA (n = 7). This result is in agreement with our previous findings (Canti et al. 2001). It can be attributed to the fact that Ca_Vβ subunits, being low molecular weight cytoplasmic proteins, are transcribed and translated more rapidly and are therefore likely to be present in the cytoplasm at much higher concentrations than the concentration of functional transmembrane Ca_V2.2 channels at the plasma membrane. However, for Ca_V2.2 Y388S there was a clear difference between the effects of the two concentrations of Ca_Vβ1b, in that the currents in the presence of the 50-fold diluted β1b were significantly reduced by 74% (P < 0.01) compared to those in the presence of the standard concentration of β1b (Fig. 6A), and the steady-state inactivation became more positive, to the same extent as in the absence of any β subunit (Fig. 6B). The V_50,inact was −44.9 ± 2.0 mV for Ca_V2.2 Y388S/β1b injected in the standard ratio (n = 7) and −16.5 ± 0.45 mV for Ca_V2.2 Y388S/β1b injected in the diluted ratio (n = 10, P < 0.001 compared to the standard ratio). This is very similar to the V_50,inact obtained for steady-state inactivation curves of Ca_V2.2/α₂δ-2 currents, which was −18.0 ± 1.5 mV (n = 10; Fig. 6B). Thus a reduction in the concentration of expressed Ca_Vβ subunits reveals the functional effect of the lower affinity of the Ca_V2.2 Y388S I–II linker for Ca_Vβ subunits.

Figure 6. The effect of a 50-fold dilution of β1b on the expression and steady-state inactivation properties of Ca_V2.2 and Ca_V2.2 Y388S in Xenopus oocytes.

A, peak current levels at +10 mV of Ca_V2.2/α₂δ-2 coexpressed with Ca_Vβ1b in the standard ratio (black bar, n = 10) or with 1 : 50 diluted β1b cDNA (grey bar, n = 7), or Ca_V2.2 Y388S/α₂δ-2 coexpressed with Ca_Vβ1b in the standard ratio (white bar, n = 7) or with 1 : 50 diluted β1b cDNA (hatched bar, n = 10). **P < 0.01, statistically significant compared to the standard ratio of Ca_V2.2 Y388S/Ca_Vβ1b, using Student's two-tailed t test. B, voltage dependence of steady-state inactivation of Ca_V2.2/α₂δ-2 coexpressed with Ca_Vβ1b in the standard ratio (▪, n = 10) or with 1 : 50 diluted β1b cDNA (□, n = 7), or Ca_V2.2 Y388S/α₂δ-2 coexpressed with Ca_Vβ1b in the standard ratio (•, n = 7) or with 1 : 50 diluted β1b cDNA (○, n = 10), and compared to data obtained without any β subunit coexpressed (, n = 10). The normalized data are plotted against the conditioning potential. The mean data are fitted with a Boltzmann function, whose V_50,inact values are given in the text.

Discussion

Ca_Vβ subunits are membrane-associated guanylate kinase proteins characterized by a guanylate kinase-like (GK) domain that binds to the AID motif in the I–II loop of HVA Ca_Vα1 subunits (Fig. 1A) and a Src homology 3 domain (SH3) (Hanlon et al. 1999). The 18 amino acid AID motif contains a conserved W that is crucial for binding Ca_Vβ (Pragnell et al. 1994; Berrou et al. 2002; Leroy et al. 2005), and also a conserved Y three residues proximal to the W. Recent structural data from three groups has provided detailed information about the interaction between the AID–Ca_Vβ complex and confirmed that both W and Y are deeply embedded in the binding groove within the GK of Ca_Vβ (Van Petegem et al. 2004; Opatowsky et al. 2004; Chen et al. 2004). The importance of the Y in Ca_Vβ binding and functional effects is controversial. It was first found that mutation of this Y to S in the AID of Ca_V2.1 completely abolished binding to β3, and almost completely abolished binding to β2a, whereas the mutation of Y to F appeared to be slightly less effective (Witcher et al. 1995). This Y residue was also originally described as being essential for functional expression (Pragnell et al. 1994; Witcher et al. 1995). However, Berrou et al. (2002) found that although no direct binding of the ³⁵S-labelled Ca_vβ3 subunit to a GST Ca_v2.3 AID containing the Y to S mutation could be detected, Ca_v2.3 Y383S was still in part modulated by the Ca_vβ3 subunit when coexpressed in Xenopus oocytes. Berrou et al. (2005) subsequently found that both conserved and non-conserved mutations in Y383 of Ca_V2.3 had little effect on Ca_Vβ modulation of whole-cell currents in Xenopus oocytes but the same mutations in an AID peptide almost abolished ³⁵S-labelled Ca_vβ3 binding, using a non-quantitative assay. In addition, Neuhuber et al. (1998) found that whilst Y366S mutation in Ca_V1.1 appeared to prevent co-localization of Ca_V1.1 with Ca_Vβ1a in transfected tsA-201 cells, expression of Ca_V1.1 currents was not affected. Similar results were found by the same group for the same Y to S mutation in Ca_V1.2, which prevented co-localization of Ca_V1.2 with β subunits at the plasma membrane, as determined by immunocytochemistry, but did not affect calcium current expression (Gerster et al. 1999). This group concluded that increased incorporation of channels into the plasma membrane is not required for the Ca_Vβ-mediated enhancement of functional calcium currents.

Our evidence that β subunits increase the number of Ca_V2.2 channels incorporated into the plasma membrane, as determined by cell surface biotinylation (Leroy et al. 2005), and that this is reduced by the W391A mutation (Leroy et al. 2005) but not by the Y388S mutation (as shown here using the same transfection conditions), are in agreement with the effect of these two mutations on Ca_V2.2 current density, although the effect on cell surface incorporation was always less than the overall influence on current density. Our results strongly support the view that both the Ca_Vβ-mediated increase in channel number in the plasma membrane, as well as the undisputed effect of Ca_Vβ subunits on channel properties (Neely et al. 1993; Gerster et al. 1999; Meir et al. 2000), both normally contribute to the increase in whole-cell current that is observed. It is likely that previous immunocytochemical results, using intracellular epitopes that require cell permeabilization (Neuhuber et al. 1998; Gerster et al. 1999), do not allow the distinction between sub- plasma membrane channels, and those that are actually in the membrane, whereas cell surface biotinylation is a more accurate reflection of proteins that are incorporated into the membrane.

Low affinity interactions of different Ca_Vβ subunits with the N- and C-termini of various calcium channels have also been reported (for reviews see Walker & De Waard, 1998; Dolphin, 2003), although in a yeast two-hybrid screen we did not observe any interaction of Ca_Vβ1b with the N- or C-terminus of Ca_V2.2, under conditions where the interaction of Ca_Vβ1b with the I–II linker was robust (Raghib et al. 2001). Furthermore, it is unlikely that such interactions could be responsible for the effects of Ca_Vβ subunits in the absence of an anchor to the AID region of the I–II linker, since all the effects of β1b were abrogated by the W391A mutation (Leroy et al. 2005). However, palmitoylated β2a was still able to modulate the biophysical properties of Ca_V2.2 W391A, indicating that the plasma membrane anchoring afforded by its palmitoylation can substitute for high affinity interaction with the I–II linker (Leroy et al. 2005). Thus it seems likely that several other regions of the calcium channel α1 subunits are involved in mediating the effects of Ca_Vβ subunits.

Lack of evidence for the binding of Ca_Vβ subunits to additional regions on the I–II linker, apart from the AID

In this study we obtained a similar affinity of Ca_Vβ1b for the full-length Ca_V2.2 I–II linker to that which we found previously for a I–II linker construct truncated immediately after the AID (13.8 nm; Leroy et al. 2005). If there were an additional binding site, for example for the β subunit SH3 domain, to a site on the I–II linker distal to the AID, as suggested previously (Maltez et al. 2005), the combination of two binding sites would lead to the measurement of a higher overall affinity of Ca_Vβ for the full-length I–II linker. Our results, combined with the complete lack of binding of β1b to the full-length Ca_V2.2 W391A I–II linker, do not provide evidence that there is an additional binding site for other domains of β1b on the distal I–II linker of Ca_V2.2, in contrast to the previous conclusion (Maltez et al. 2005).

The Y388S mutation in the AID of Ca_V2.2 appears not to influence the functional effects of β1b, despite producing a 24-fold reduction in affinity for β1b binding to the AID

One of the main effects of Ca_Vβ subunits on HVA calcium channels is to increase current density. Our studies have shown that there are fewer channels present at the cell surface when no Ca_Vβ subunits were coexpressed or when mutated Ca_V2.2 W391A channels were cotransfected with a Ca_Vβ (Leroy et al. 2005). It has been suggested that a Ca_Vβ bound to the I–II linker may mask an endoplasmic reticulum retention signal present in the I–II linker of HVA calcium channels and favour the trafficking of the channel to the cell surface (Bichet et al. 2000). Our previous data suggested that the endogenous Ca_Vβ3 that we have identified in tsA-201 cells was responsible for trafficking some wild-type Ca_V2.2 to the plasma membrane in the absence of a coexpressed β subunit, and that the markedly reduced affinity of the W391A mutated channel for Ca_Vβ subunits abolished interaction with the endogenous Ca_Vβ3 subunits, and thus prevented any trafficking to the plasma membrane (Leroy et al. 2005). Our results therefore provided very strong evidence that the binding of a Ca_Vβ subunit to the channel is an essential requirement for the functional expression of Ca_V2.2 at the plasma membrane.

In contrast, the markedly reduced affinity of the Y388S AID (from 13.7 to 329 nm) for β1b does not translate into a reduced expression of the channels at the plasma membrane, or any effect on the voltage dependence of activation or inactivation or voltage dependence of G-protein modulation. We have determined previously, from experiments in which varying concentrations of β subunits were expressed together with a constant amount of Ca_V2.2 in Xenopus oocytes, that there appeared to be two different affinities of β subunits for trafficking the channels (approximately 17 nm for β3) and for hyperpolarizing the steady-state inactivation (about 7-fold lower affinity, approximately 120 nm). However, in Xenopus oocytes the concentration of Ca_Vβ subunits obtained following the standard conditions of heterologous expression used in this study was estimated to be far in excess of this, at 2–3 μm (Canti et al. 2001). If similar amounts are expressed in the mammalian expression system then it is not surprising that little effect was observed of a 24-fold reduction in the affinity of β1b for the AID. Occupancy would remain very high because of the excess of free Ca_Vβ subunits.

Support is given to this conclusion by our experiments in Xenopus oocytes in which dilution of β1b by 50-fold abolished the influence of this Ca_Vβ subunit on the steady-state inactivation of Ca_V2.2 Y388S but had no effect on that of wild-type Ca_V2.2.

These experiments demonstrate the limitation of coexpression studies in that the concentration of the expressed proteins may be very different, particularly when coexpressing membrane proteins with cytoplasmic proteins, despite the use of similar cDNA concentrations, and in this way they may not mimic the ratios of subunits present in vivo. However, the stoichiometry and occupancy of the AID site on endogenous calcium channels by endogenous Ca_Vβ subunits remains an open question to be addressed in the future.

The Y388S mutation in the AID of Ca_V2.2 has no effect on G-protein modulation of Ca_V2.2 channels

It has been proposed that G-protein modulation of Ca_V2 channels involves competition between Gβγ and Ca_Vβ subunits, with displacement of β subunits being a key step (Bourinet et al. 1996; Sandoz et al. 2004). Our results do not support this view as, despite the 24-fold reduction in affinity of Ca_V2.2 Y388S for β1b, there was no enhancement of G-protein modulation. If there were simple competition between Gβγ and Ca_Vβ, then a 24-fold reduction in affinity of the I–II linker for Ca_Vβ1b should result in an increased occupancy by Gβγ at the peak of the response to the agonist quinpirole. The present result concurs with our previous results that did not provide evidence for simple competition between Ca_Vβ and Gβγ (Meir et al. 2000).

All parameters of G-protein modulation were identical, including the rate of facilitation, which has been interpreted as resulting from the dissociation of Gβγ (Ikeda, 1996; Zamponi & Snutch, 1998). In our previous study we found that the facilitation rate during a +100 mV prepulse was a sensitive measure of changes in Ca_Vβ subunit concentration (Canti et al. 2001). It was 20-fold slower in the absence than in the presence of coexpressed Ca_Vβ subunits, and could be resolved into different proportions of fast and slow components in the presence of intermediate concentrations of Ca_Vβ subunits (Canti et al. 2001). Our interpretation of these two components was that the fast component represented Gβγ dissociation from channels to which Ca_Vβ was already bound, and the slow rate represented increased Ca_Vβ binding at +100 mV, followed by Gβγ dissociation, since the slow rate depended on Ca_Vβ subunit concentration (Canti et al. 2001). In agreement with our previous results, this suggests that Ca_Vβ subunit displacement by Gβγ is not involved in G-protein modulation (Meir et al. 2000; Canti et al. 2000; Leroy et al. 2005), but in contrast the presence of bound β subunits is essential to promote the loss of Gβγ at positive potentials (prepulse facilitation) (Meir et al. 2000; Leroy et al. 2005).