Current modulation and membrane targeting of the calcium channel α_1C subunit are independent functions of the β subunit

Abstract

1. The β subunits of voltage-sensitive calcium channels facilitate the incorporation of channels into the plasma membrane and modulate calcium currents. In order to determine whether these two effects of the β subunit are interdependent or independent of each other we studied plasma membrane incorporation of the channel subunits with green fluorescent protein and immunofluorescence labelling, and current modulation with whole-cell and single-channel patch-clamp recordings in transiently transfected human embryonic kidney tsA201 cells.

2. Coexpression of rabbit cardiac muscle α_1C with rabbit skeletal muscle β_1a, rabbit heart/brain β_2a or rat brain β₃ subunits resulted in the colocalization of α_1C with β and in a marked translocation of the channel complexes into the plasma membrane. In parallel, the whole-cell current density and single-channel open probability were increased. Furthermore, the β_2a isoform specifically altered the voltage dependence of current activation and the inactivation kinetics.

3. A single amino acid substitution in the β subunit interaction domain of α_1C (α_1CY467S) disrupted the colocalization and plasma membrane targeting of both subunits without affecting the β subunit-induced modulation of whole-cell currents and single-channel properties.

4. These results show that the modulation of calcium currents by β subunits can be explained by β subunit-induced changes of single-channel properties, but the formation of stable α_1C-β complexes and their increased incorporation into the plasma membrane appear not to be necessary for functional modulation.

Voltage-sensitive calcium channels are multimeric protein complexes formed by the α₁ subunit and the auxiliary subunits α₂δ, β and γ (Leung et al. 1987; Takahashi et al. 1987; Vaghy et al. 1987). The α₁ subunit by itself shows the characteristic properties of a voltage-gated ion channel, i.e. voltage sensing, ion permeation and drug binding. The physiological roles of the auxiliary subunits are currently the subject of intensive investigations. The β subunit modulates calcium currents by increasing the current density and by changing the current kinetics when coexpressed in heterologous expression systems (Lacerda et al. 1991; Varadi et al. 1991; Lory et al. 1992). Furthermore, it has been suggested that the β subunit is involved in the targeting of the α₁ subunit to the plasma membrane (Chien et al. 1995; Gregg et al. 1996). Both the α₁ and β subunits exist in multiple tissue-specific isoforms, which differ from one another in their primary structure and in their functional properties (Birnbaumer et al. 1994; Isom et al. 1994). The skeletal muscle α_1S isoform, for example, shows slow activation and inactivation kinetics compared with the other α₁ isoforms. Also, whereas α_1C expressed in heterologous expression systems exhibits currents even in the absence of auxiliary subunits (Perez-Garcia et al. 1995), expression of α_1S in heterologous systems rarely gives rise to measurable calcium currents (Johnson et al. 1997). The β subunit isoforms differ in their current modulation (Hullin et al. 1992; Sather et al. 1993; Parent et al. 1997) and, when expressed alone, in their subcellular distribution (Chien et al. 1995, 1996; Brice et al. 1997). For example, β_2a drastically reduced the speed of inactivation when coexpressed with the neuronal α_1E subunit in oocytes (Parent et al. 1997), whereas other β subunit isoforms showed only minor effects on current inactivation. Analogously, β_2a differs from most other β isoforms in that it was localized in the plasma membrane when expressed without an α₁ subunit in a heterologous expression system (Chien et al. 1995), whereas β_1a, β₃ and β₄ showed a cytoplasmic localization (Brice et al. 1997; Neuhuber et al. 1998b).

A conserved β subunit binding motif has been identified in the cytoplasmic loop between repeats I and II of α_1S, α_1A, α_1B and α_1C (Pragnell et al. 1994). Point mutations within this binding motif perturbed α₁-β binding and affected calcium current properties when the neuronal α_1A isoform was coexpressed with β_1b in oocytes. A point mutation (Y366S) within the β subunit binding motif in the I-II linker of the skeletal muscle α_1S resulted in the expected loss of α_1S-β_1a binding, but the probability that tsA201 human embryonic kidney cells cotransfected with α_1SY366S and β_1a exhibited calcium currents was still increased by the β_1a subunit (Neuhuber et al. 1998b). This indicates that stable binding of β_1a to the known motif in the I-II linker of α_1S is not necessary for the β subunit to increase the frequency of current expression. Thus, association of β with this interaction domain in the cytoplasmic I-II linker plays an important role in β subunit-dependent modulation of calcium currents, but other mechanisms for α₁-β interaction may exist. De Waard et al. (1994, 1996) identified a conserved 30 amino acid domain in the β subunit that is complementary to the binding site in the I-II linker on α_1A and is involved in the interaction with this subunit. Mutations within this domain of β perturbed binding to α_1A and affected modulation of calcium current properties. However, this domain in the β subunit cannot account for all observed modulatory effects of α_1A-β interactions, since certain truncated β subunits were only affected in their modulation of inactivation kinetics, not in current stimulation. Therefore, a region of the β subunit other than that interacting with the I-II linker of α₁ may be involved in the modulation of α_1A (De Waard et al. 1994). Moreover, using chimeras of different β isoforms Olcese et al. (1994) and Qin et al. (1996) have found that regulation of activation and inactivation of α_1E channels are two separable functions of the β subunit, suggesting the existence of two separate interaction domains on each of the subunits. Indeed, Tareilus et al. (1997) identified a second β subunit binding domain within the last 277 amino acids of the C-terminus of α_1E, and Walker et al. (1998) identified a low affinity binding site in the carboxy-terminal region of α_1A that accounts for β₄-induced modulation of current inactivation.

Despite this progress in understanding the function of the calcium channel β subunit the mechanism of current modulation by β remains largely unresolved. For example, it is still controversial whether increased insertion of channels into the plasma membrane and modulation of current properties are two independent functions of the β subunit or whether the latter is a direct result of the former. Also, we do not know whether α₁ and β form a stable complex in which β serves as a necessary cofactor or mediator of modulatory signals, or whether association and dissociation of the β subunit in itself is the modulatory mechanism. To address these questions we studied the interactions of three different β subunit isoforms with α_1C and an α_1C mutant with a single amino acid substitution in the β subunit interaction domain of the I-II linker (α_1CY467S) using a combination of structural and functional techniques. This approach allowed us to distinguish β isoform-specific effects from common effects of α_1C-β interactions and to demonstrate that increased membrane incorporation and modulation of channel properties are two independent effects of β subunits. Single-channel analysis showed that changes in the open probability are sufficiently large to explain the increase in whole-cell current density observed upon coexpression of the β subunit. Further, the comparison of β subunit effects on wild-type and mutant α_1C shows that this increase in current density occurs even without the formation of stable α_1C-β complexes or their increased incorporation into the plasma membrane.

METHODS

Cell lines

tsA201 cells, a HEK cell subclone stably transfected with the SV40 large T-antigen, were plated and allowed to proliferate in F12 medium (Gibco BRL, Vienna, Austria) containing 10% fetal bovine serum. Cells were grown to 80% confluency before passaging. For structural analysis (green fluorescent protein (GFP) and immunocytochemistry), cells were plated at dilutions of about 1:10 onto poly-L-lysine-coated 13 mm round coverslips and transfected on the following day (see below) when cells reached 30% confluency. For patch-clamp analysis cells were plated on 35 mm culture dishes at a dilution of 1:10. After transfection at about 50% confluency, cells were replated onto poly-L-lysine-coated 25 mm round coverslips at dilutions between 1:5 and 1:10 to get isolated cells for patch-clamp recordings on the following day.

Transfection

Transfections were carried out with a liposomal transfection reagent (DOTAP, Boehringer Mannheim) according to the manufacturer's instructions. The total amount of DNA used per 35 mm culture dish was 10 μg, 5-8 μg of which was specific DNA (expression plasmids encoding α_1C constructs, β subunit and GFP), with the rest made up with inert DNA (pUC18; Norrander et al. 1985). In co-transfection experiments two or more expression plasmids were combined at equimolar concentrations. This resulted in coexpression of β with any α_1C construct in approximately 70% of β subunit-transfected cells. The liposome-DNA mixture was diluted in 1.5 ml F12 medium and then added to the culture. On the following day the cells were processed for immunocytochemistry or replated for patch-clamp analysis.

Expression plasmids

Details of the expression plasmids are given in Table 1. The coding sequence of rabbit cardiac muscle α_1C DNA (Mikami et al. 1989) with an alternative exon in IVS3 (Koch et al. 1990) was inserted into the expression plasmid pcDNA3 (Invitrogen, San Diego, CA, USA) via HindIII and NotIdigestion. The tyrosine to serine substitution in position 467 in α_1C (α_1CY467S) was introduced by site-directed mutagenesis of α_1C-pcDNA3 using the splicing by overlap extension technique. The following mutagenic sense primer was used:

Table 1.

Expression plasmids

Name	Vector	Insert	Reference
α1C	pcDNA3	Rabbit cardiac muscle α1	Mikami et al. 1989; Koch et al. 1990
GFP-α1C	pcDNA3	GFP-α1C fusion protein	Grabner et al. 1998
α1CY467S	pcDNA3	Cardiac muscle α1C Y to S substitution	Present study
β1a	pcDNA3	Rabbit skeletal muscle β	Ruth et al. 1989
β1a-GFP	pcDNA3	β1a-GFP fusion protein	Neuhuber et al. 1998b
β2a	pCMV6	Rabbit heart/brain β	Perez-Reyes et al. 1992
β3	pCMV6	Rat brain β	Castellano et al. 1993
GFP(S65T)	pRK5	Green fluorescent protein	Heim et al. 1995
pUC18	pUC18	—	Norrander et al. 1985

To facilitate the identification of positive clones, a PvuII site (nucleotide 1376) was eliminated in the mutagenic primers by silent mutation. The mutated α_1C fragment was inserted into α_1C-pcDNA3 after digestion with BamHI (nucleotide 1263) and EcoRI (nucleotide 2213). The mutation was verified by sequence analysis. In addition, a GFP-α_1C fusion protein (Grabner et al. 1998) was used for double fluorescence labelling. The subcellular distribution and the electrophysiological properties of GFP-α_1C and α_1C were identical.

GFP and immunofluorescence labelling

Paraformaldehyde-fixed cultures were immuno-stained as previously described (Flucher et al. 1993). For double labelling with GFP, Texas Red-conjugated antibodies (Jackson Immuno Research, West Grove, PA, USA) were used to exclude bleed-through between the red and the green channels. Working dilutions and the sources of primary antibodies are listed in Table 2. Samples were evaluated on a Zeiss Axiovert microscope with epifluorescence and phase-contrast optics and documented on 35 mm high speed black and white film. Controls, such as the omission of primary antibodies and incubation with inappropriate antibodies, were routinely performed. Localization of α_1C and GFP-α_1C, or β_1a and β_1a-GFP gave the same results with respect to distribution patterns in all examined conditions.

Table 2.

Antibodies

Specificity	Code	Type	Dilution	Reference
DHP-receptor, α1C	CNC	Affinity purified, rabbit	1:1500	Safayhi et al. 1997
DHP-receptor, β	βcom	Affinity purified, rabbit	1:1200	Pichler et al. 1997; Neuhuber et al. 1998b.

Patch-clamp recording

Whole-cell and single-channel recordings were performed as described by Hamill et al. (1981). Cultures grown on 25 mm round coverslips were mounted in a recording chamber and viewed with a × 16 phase-contrast multi-immersion lens on a Zeiss Axiovert microscope. Fluorescent cells (indicating successful transfection with β_1a-GFP or GFP) were selected for recording. An Axopatch 200A patch-clamp amplifier controlled by the software pCLAMP 6.0 (Axon Instruments) was used for all recordings.

Whole-cell recordings

The bath solution contained (mM): 40 BaCl₂, 100 TEA-Cl and 10 Hepes (adjusted to pH 7.4 using TEA-OH). Patch pipettes, pulled from borosilicate glass and fire polished, were filled with (mM): 130 caesium aspartate, 10 Hepes, 2 Mg-ATP, 2 Cs-EGTA and 0.5 MgCl₂ (adjusted to pH 7.4 with CsOH). Resistances of the patch pipettes were between 4 and 7 MΩ. Capacitative currents were compensated using built-in analog circuits (series resistance error was corrected for 80%), and the value for the whole-cell capacity was determined from this adjustment of the amplifier and used to calculate the current density. Leak resistance in the cell-attached mode was normally larger than 8 GΩ. Current data were low-pass Bessel filtered at 2 kHz and sampled at 1 kHz with an IBM compatible PC. In the electrophysiological experiments α_1C was used in place of GFP-α_1C, which was primarily used in the structural analysis. In control experiments we did not find any difference between calcium currents from cells transfected with GFP-α_1C or α_1C. I-V curves, obtained by plotting the peak current density, i, against the test potential, V, were fitted by a modified Boltzmann function:

where g is the specific conductance (in pA (pF mV)⁻¹), V_rev is the reversial potential (in mV), V₅₀ is the potential of half-maximal activation (in mV), and k is the slope factor (in mV) which determines the steepness of the voltage dependence of activation.

Single-channel recordings

The bath solution contained (mM): 110 potassium aspartate, 20 KCl, 2 MgCl₂, 20 Hepes and 2 EGTA (adjusted to pH 7.4 with KOH). Patch pipettes, pulled from borosilicate glass, fire polished and coated with Sigmacote (Sigma), were filled with (mM): 80 BaCl₂, 30 TEA-Cl, 15 Hepes and 1 EDTA (adjusted to pH 7.4 with TEA-OH) and had resistances between 5 and 15 MΩ. If not otherwise specified in the text the following procedures for data acquisition and analysis were used. Test pulses of 200 ms duration from -70 to 0 mV were applied at 3 s intervals. Data were low-pass Bessel filtered at 2 kHz, sampled at 10 kHz, and analysed with Fetchan 6.0 (Axon Instruments) after digital Gaussian filtering at 1.5 kHz. The event detection threshold was 0.5 pA. With this event detection threshold events arising from endogenous low voltage-activated small conductance calcium channels, which we sometimes observed in non-transfected tsA201 cells, were excluded from the analysis. This also applies for the low threshold small conductance channels described by Meir & Dolphin (1998) in COS7 cells transfected with α₁ subunits. Furthermore, events with a duration of 0.2 ms or shorter were excluded from the analysis. The open probability for a patch, i.e. NP_o where N is the number of channels in the patch and P_o is the single-channel open probability, was calculated by dividing the sum of the duration of all events, which was obtained from the event list provided by Fetchan 6.0, by the total observation time. This simplified method could be applied since, due to the rare occurrence and short duration of channel openings under our experimental conditions, higher conductance levels were almost never observed. In order to estimate the number of channels in each patch, N, from these NP_o values we used two different approaches: (1) a mathematical procedure, which is based on theoretical considerations about the distribution of NP_o values for a set of patches with different unknown N values, and (2) an experimental approach, in which the highest conductance level in each patch observed after the addition of the calcium channel agonist (±)-Bay K 8644 to the bath solution was taken as an estimate for the number of channels in the patch (for details, see Results).

RESULTS

Differential localization of individually expressed α_1C and β subunits

tsA201 cells were transiently transfected with GFP-α_1C, β_1a-GFP, β_2a or β₃ and the subunit localization was determined with either GFP fluorescence or immunofluorescence (Fig. 1). In 79% of 543 examined cells (Table 3) α_1C was localized in a dense network of a tubular membrane compartment that extended throughout the entire cytoplasm. The tubular/reticular network was very dense in the perinuclear region but could be resolved in thin regions of the cells (Fig. 1a). Occasionally the nuclear envelope was also labelled. Based on these structural characteristics the α_1C-containing compartment was identified as the endoplasmic reticulum. In 15% of the cells α_1C was also localized in clusters in the plasma membrane (not shown). There were no differences in the distribution patterns between the wild-type α_1C and the mutant α_1CY467S. The distribution patterns of the β subunits differed from that of α_1C and were different among themselves. When expressed alone, β_1a and β₃ were both distributed diffusely throughout the cytoplasm of tsA201 cells (Table 4; Fig. 1d and o). In contrast, β_2a showed a continuous plasma membrane stain and the cytoplasm was essentially free of immunolabel (Table 4; Fig. 1i).

Figure 1. Changes in subcellular distribution patterns of calcium channel β_1a, β_2a and β₃ subunits coexpressed with α_1C or α_1CY467S.

tsA201 cells were transfected with α_1C (wild-type or mutant) and β isoforms alone or in different subunit combinations and the subunits were localized with immunofluorescence or GFP labelling. Left column: when expressed alone α_1C (a) is localized in a tubular network - presumably the endoplasmic reticulum - that is very dense in the perinuclear region (saturated fluorescence) but can be resolved in the cell periphery (arrows); β_1a(d) and β₃ (o) are diffusely localized in the cytoplasm; and β_2a is localized in the plasma membrane (i; arrowhead). Centre column: when wild-type α_1C and β_1a are coexpressed, the two colocalize in clusters in the plasma membrane (b and e; examples indicated by arrows); α_1C and β_2a(g and j) are colocalized throughout the plasma membrane, diffusely (arrowheads) and in clusters (arrows); and α_1C and β₃ are either colocalized in clusters (l and p; arrows) or β₃ remains diffusely distributed in the cytoplasm (m and q). The α_1C+β pairs were double labelled by using GFP-α_1C (b, g, l and m) and anti-β_com (e, j, p and q). Right column: when coexpressed with the mutant α_1CY467S, translocation and colocalization fail and all subunits remain in the same compartments as when expressed individually; α_1CY467S in the endoplasmic reticulum (c, h and n; arrows); β_1a(f) and β₃ (r) in the cytoplasm; and β_2a in the plasma membrane (k; arrowhead). N, nucleus; scale bar, 20 μm.

Table 3.

Subcellular distribution of the calcium channel α_1C subunits expressed alone or with β subunit isoforms

	α1C		α1CY467S
	ER	PM	ER	PM
Without β	79	15	87	7
β1a	14	83	87	7
β2a	32	64	93	0
β3	49	47	88	6

Numbers give percentage of cells in each category (n = 500–800 cells for each condition). Balance to 100% represents cells in which localization could not be unambiguously identified as endoplasmic reticulum (ER) or plasma membrane (PM).

Table 4.

Subcellular distribution of the calcium channel β subunits expressed alone or with α_1C

	Without α1C		α1C			α1CY467S
	Cytoplasm	PM	Cytoplasm	ER*	PM	Cytoplasm	PM
β1a	98	0	12	19	69	96	0
β2a	0	94	0	0	99	0	98
β3	99	0	44	28	28	98	0

Numbers give percentage of cells in each category (n = 400–900 cells for each condition). Balance to 100% represents cells in which localization could not be unambiguously identified as cytoplasm, ER or PM.

^*ER distribution of β subunits was only observed with α_1C and is therefore not included in the other conditions.

Colocalization of coexpressed α_1C and β subunits in the plasma membrane

α_1C was coexpressed with β_1a, β_2a or β₃ in tsA201 cells and the subcellular distribution of both subunits was determined with GFP fluorescence or immunofluorescence. Upon coexpression with any of the β subunits the localization of α_1C changed significantly from that observed when α_1C was expressed alone. In all cases the number of cells with an endoplasmic reticulum localization of α_1C was strongly reduced, whereas the number of cells with a plasma membrane localization was strongly increased compared with cells in which α_1C was expressed alone (Table 3). In parallel, β_1a and β₃ were no longer found in the cytoplasm but were localized in clusters in the plasma membrane (Fig. 1e and p) and to some extent in the endoplasmic reticulum (Table 4). In both locations α_1C and the β subunits were colocalized (Fig. 1b, e, l and p). The degree to which α_1C and β were translocated to the plasma membrane was considerably higher for β_1a than for β₃. Fewer than half of the cells coexpressing α_1C and β₃ showed a colocalization of the two subunits in plasma membrane clusters, and in the rest of the cells α_1C and β₃ remained in the endoplasmic reticulum and the cytoplasm, respectively (Fig. 1l, m, p and q; Tables 3 and 4). In contrast to the cytoplasmic β_1a and β₃ subunits, the plasma membrane-associated β_2a subunit showed little change in its distribution pattern when coexpressed with α_1C. β_2a remained evenly distributed throughout the plasma membrane, where it now was colocalized with α_1C (Fig. 1g and j). In addition to the even plasma membrane distribution, the two subunits were frequently colocalized in plasma membrane clusters. The observed changes in the subcellular distribution of the α_1C and β subunits upon coexpression indicate that all three examined β subunit isoforms bind to α_1C and that the α_1C-β complexes become inserted into the plasma membrane.

Translocation and colocalization of α_1C and β subunits fails when the β subunit binding domain in the cytoplasmic I-II linker of α_1C is mutated

To investigate the mechanism of the direct α_1C-β interactions that underlie the observed changes in the subcellular distribution of the channel subunits, we replaced tyrosine in position 467 of the β subunit interaction domain of α_1C with serine. The corresponding point mutations in α_1A and α_1S have been shown to disrupt β_1b and β_1a binding, respectively (Pragnell et al. 1994; Neuhuber et al. 1998b). Indeed, when α_1CY467S was coexpressed with any one of the β subunit isoforms, α_1CY467S and all β subunits remained localized in the same subcellular compartment as when expressed alone (Fig. 1). The subunits were not colocalized with one another in the plasma membrane or anywhere else in the cells and no increased expression of α_1CY467S clusters occurred. This shows that the translocation of the subunits to the plasma membrane that was observed with GFP fluorescence and immunofluorescence when any one of the β subunits was coexpressed with wild-type α_1C depends on an intact β subunit interaction domain in the cytoplasmic I-II linker of α_1C.

β subunits modulate current activation in wild-type and mutant α_1C channels

To investigate whether the observed structural interactions between α_1C and the β subunits are reflected in a modulation of calcium current properties, we performed whole-cell patch-clamp recordings. tsA201 cells were transfected with α_1C or α_1CY467S with or without one of the β subunit isoforms, plus a plasmid encoding GFP as an expression marker. In the case of β_1a, the fusion protein β_1a-GFP was used instead of a combination of the two separate plasmids. The fraction of GFP-expressing cells in which high voltage-activated calcium currents could be recorded was 50% for cells transfected with α_1C alone and 42% for cells transfected with α_1CY467S alone. The percentage of cells exhibiting high voltage-activated calcium currents was about equal or higher when α_1C or α_1CY467S was coexpressed with a β subunit.

To measure the voltage dependence of activation, 400 ms voltage steps were applied from a holding potential of -80 mV to various test potentials between -40 and +80 mV (Fig. 2). I-V curves were obtained from these measurements by plotting the peak current density against the test potential (Fig. 3A). Coexpression with any one of the β subunits increased the peak current density severalfold. When α_1C or α_1CY467S was expressed alone the largest peak current was observed at a test potential of +40 mV. Upon coexpression with β_2a the test potential at which the largest current was observed shifted to +30 mV. The Y467S point mutation had no effect on the intrinsic current properties of α_1C or on the characteristic modulation by the β subunit isoforms. The increase in current density observed with all β subunit isoforms and the shift in the peak of the I-V curve observed with β_2a were the same with α_1C and α_1CY467S. I-V curves were further analysed by fitting them with a modified Boltzmann function (see Methods; Fig. 3B). This analysis revealed that all β subunits increased the specific conductance, g, and that the strongest shift occurred with β_1a. The potential of half-maximal activation, V₅₀, and the slope factor, k, were both decreased by coexpression of the β subunits, with β_2a showing the strongest effect (with P < 0.001, t test). The value of k was significantly decreased with P < 0.05 by all β subunits. The reversal potential, V_rev, was not significantly altered by coexpression of the β subunits. Interestingly, none of these parameters showed a significant difference between α_1C and α_1CY467S in any of the examined subunit combinations, suggesting that despite the deficiency in complex formation α_1CY467S was still sensitive to modulation by β subunits.

Figure 2. Comparison of the time course of current traces for α_1C and α_1CY467S expressed without or with β_1a, β_2a or β₃ in tsA201 cells.

The membrane potential was stepped for 400 ms from a holding potential of -80 mV to 0, +20 or +40 mV (upper, middle and lower traces, respectively). The magnitude of currents varied greatly between individual cells (see also error bars in Fig. 3A and B). Current inactivation was slowed down on coexpression of α_1C or α_1CY467S with β_2a. There was no apparent difference between currents recorded from α_1C- or α_1CY467S-transfected cells.

Figure 3. Voltage dependence of current activation for wild-type and mutant α_1C expressed with and without β subunits.

A, I-V curves obtained by plotting the peak current density from voltage step measurements as shown in Fig. 2 against the test potential. Coexpression of α_1C (•) and α_1CY467S (○) with any one of the β subunits increased the current amplitude severalfold (means and s.e.m.). B, I-V curves for each individual recording were fitted by a modified Boltzmann function (see Methods) and the obtained parameters were subsequently averaged. □, α_1C; , α_1CY467S; error bars are s.e.m., and the number of cells recorded is shown. The specific conductance, g, was increased on β subunit coexpression whereas V₅₀ and k were decreased by β subunit coexpression. The decrease in k on coexpression with each of the β subunits compared with α_1C or α_1CY467S expression alone was significant with P < 0.05. The reversal potential, V_rev, was not significantly altered by the β subunits.

Current kinetics are modulated by the β subunits

As a measure of the activation kinetics the time from the onset of the voltage step to 70% of the total rise in current amplitude, τ_70%, was determined (Fig. 4A). Whereas β₃ had no effect, β_1a decreased the speed of activation and β_2a slightly accelerated activation. The activation kinetics of α_1CY467S were always somewhat faster than those of wild-type α_1C, regardless of whether they were expressed alone or together with a β subunit. This suggests that tyrosine in position 467 of α_1C is involved in determining the channel activation kinetics. The acceleration of activation kinetics by the mutation was statistically significant (P < 0.05) for coexpression with β_1a and β_2a, but not for β₃ and expression of α_1C alone.

Figure 4. Calcium current kinetics in tsA201 cells transfected with α_1C or α_1CY467S with and without β subunits.

□, α_1C; , α_1CY467S. A, activation kinetics of calcium currents are expressed as the time from the onset of a voltage step to +40 mV (holding potential, -80 mV) to 70% of the total rise in current amplitude, τ_70%. β_1a slowed down current activation, β_2a accelerated it and β₃ did not affect activation kinetics. In all cases α_1CY467S was slightly faster than the wild-type α_1C. B, inactivation kinetics were determined by fitting a single exponential function with a time constant τ_inact to the decay phase of the current during a voltage step to +40 mV from a holding potential of -80 mV. For α_1C and α_1CY467S coexpressed with β_2a, test pulse duration was 4 s; for all other conditions test pulse duration was 400 ms. Only β_2a severely slowed down current inactivation, by a factor of ≈10.

The most dramatic modulatory effect of a β subunit appeared in the inactivation kinetics (Fig. 4B). Current inactivation was determined by fitting a single exponential function to the decaying phase of the current (for cells cotransfected with β_2a, 4 s-long test pulses were used instead of 400 ms pulses). Figure 4B shows the inactivation time constants for test pulses to +40 mV. Whereas coexpression with β_1a or β₃ had only small effects (P > 0.04), β_2a slowed current inactivation by a factor of approximately 10 (P < 0.0001). The inactivation kinetics of the mutant α_1CY467S were indistinguishable from those of wild-type α_1C.

Steady-state inactivation is not significantly modulated by the β subunits

Steady-state inactivation was determined by measuring the current amplitude at a test potential of +40 mV (holding potential, -100 mV) after 10 s prepulses of different potentials (2 ms interpulse interval). Figure 5A shows the normalized current amplitudes as a function of the prepulse potential. The data were fitted by a Boltzmann function, the parameters of which are shown in Fig. 5B. The analysis did not indicate any modulatory effects of the β subunits on steady-state inactivation, or any significant effects of the point mutation in α_1C. Only the potential of half-maximal inactivation, V₅₀, was different, being about 7 mV higher for α_1CY467S than for α_1C, when the α subunits were not coexpressed with a β subunit, but this effect was not significant (P > 0.05). Although coexpression with the β subunits resulted in small changes of V₅₀, statistically there was no significant difference from the V₅₀ values for α_1C and α_1CY467S expressed alone. The slope factor, k, was the same for all subunit combinations tested. Thus, steady-state inactivation of α_1C was not subject to modulation by any one of the examined β subunit isoforms.

Figure 5. Steady-state inactivation in tsA201 cells transfected with α_1C or α_1CY467S with and without β subunits.

A, normalized current amplitude during a voltage step from -100 to +40 mV after 10 s prepulses to various potentials with an interpulse interval of 2 ms (•, α_1C; ○, α_1CY467S; means and s.e.m. of individual recordings are shown). There were only minor differences in steady-state inactivation between the different subunit combinations. B, steady-state inactivation curves for each individual recording were fitted by a Boltzmann function: i = 1/(1 + exp((V - V₅₀)/k)), where i is the test pulse current amplitude, V is the prepulse potential, V₅₀ is the prepulse potential of half-maximal inactivation and k is the slope factor, and the obtained parameters were subsequently averaged (□, α_1C; , α_1CY467S; error bars are s.e.m.). The apparent difference in the slopes of the averaged curves shown in the upper left figure in A is not reflected in the slope factor, k, shown in the right panel of B, due to the large variability of V₅₀ for α_1C expression (see error bar in left panel of B). This analysis does not indicate statistically significant differences in steady-state inactivation parameters between cells transfected with different subunit combinations.

Single-channel open probability of α_1C is increased by β_1a

The data presented above suggest that the increase in whole-cell current density observed when α_1C or α_1CY467S was coexpressed with a β subunit was not necessarily linked to an increase in plasma membrane localization of the α subunit, because the latter was not observed with α_1CY467S. Thus, the possibility that the β subunit-induced increase in current density is due to changes of single-channel properties, such as a higher open probability or single-channel conductance, needed to be addressed directly. Therefore we performed single-channel recordings of α_1C or α_1CY467S alone, and α_1C or α_1CY467S coexpressed with β_1a (Figs 6 and 7), the β isoform that induced the largest current increase in whole-cell recordings. The amplitude of single-channel opening events, i, was about the same in all examined subunit combinations, showing that the single-channel conductance was not modulated by β_1a (i = 0.90± 0.05 pA for α_1C, 1.00 ± 0.04 pA for α_1C+β_1a, 1.00 ± 0.16 pA for α_1CY467S, and 0.80 ± 0.03 pA for α_1CY467S +β_1a (means ±s.e.m.)). In contrast, the frequency and the duration of channel openings increased significantly when α_1C or α_1CY467S was coexpressed with β_1a (Fig. 6). The open-state dwell time histograms (Fig. 6) show that coexpression of β_1a increased the mean open time 2-fold. The observation that the mutation itself did not change single-channel properties, together with the results from the whole-cell experiments, shows that the Y467S substitution does not mimic the effects of β subunit coexpression. The fraction of null sweeps was highly variable from cell to cell but on average no significant differences between α_1C+β_1a (51%) and α_1CY467S +β_1a (55%) were observed.

Figure 6. Single-channel recordings from tsA201 cells transfected with α_1C (A), α_1C+β_1a (B), α_1CY467S (C) and α_1CY467S +β_1a (D).

Voltage pulses of 200 ms duration from a holding potential of -70 mV to a test potential of 0 mV were applied every 3 s. Coexpression of α_1C or α_1CY467S with the β_1a subunit resulted in an increased frequency of channel openings. The right panel shows open-state dwell time histograms. The mean open times, τ, obtained by fitting single exponential functions to the histograms, were increased by a factor of about 2 by β_1a coexpression.

Figure 7. Determination of the number of channels in a patch, N, and the single-channel open probability, P_o.

Open probabilities for individual patches, NP_o, were calculated from recordings from tsA201 cells transfected with α_1C (○), α_1C+β_1a (□), or α_1CY467S +β_1a (▿) and plotted against the number of channels in each patch, N. N values were determined by two different approaches. (i) Open symbols in A represent recordings in which at the end of an experiment 5 μm (±)-Bay K 8644 was added to the bath solution to visualize simultaneous openings of multiple channels in the patch. C shows an example of the short and long openings before and after addition of (±)-Bay K 8644, respectively, in a cell transfected with α_1CY467S and β_1a (upper trace, test pulse duration 200 ms at 0 mV; lower trace, test pulse duration 400 ms at +10 mV). The all-points amplitude histogram in D corresponding to the lower trace in C shows the number of conductance levels. (ii) Filled symbols in A represent recordings for which N values were estimated by finding the best fit of all NP_o values in one experimental group to a linear regression crossing the abscissa at zero. Multiple data points with the same or similar NP_o marked with 1 (▪, ▾, ▾) and 2 (•, •). The slope corresponds to the mean single-channel open probability. B, the P_o values determined experimentally with (±)-Bay K 8644 (□) closely match the values from the fitting procedure (▪; error bars give standard deviation). Coexpression of β_1a resulted in a 7- to 8-fold increase of P_o.

In order to determine the single-channel open probability, P_o, it was necessary to determine the number of channels, N, in each patch. The usual approach is to estimate N from the maximum number of conductance levels observed. This is reasonable only when P_o is sufficiently high and the mean open time is sufficiently long to observe simultaneous openings of multiple channels. Due to the rare occurrence and short mean open time of events under the conditions used here, simultaneous openings of multiple channels are extremely unlikely and were almost never observed. Therefore we applied two different approaches to estimate N for each patch. In the first approach, we added the calcium channel agonist (±)-Bay K 8644 to the bath solution to a final concentration of 5 μm after we had recorded several hundred sweeps without (±)-Bay K 8644 for the analysis of NP_o. With (±)-Bay K 8644 in the bath solution and depolarizations to +10 mV, channel openings lasting up to several hundred milliseconds occurred, allowing the observation of simultaneous openings of multiple channels in the patch (Fig. 7C). An all-points amplitude histogram of these recordings gives the number of conductance levels and thus the number of channels in the patch (Fig. 7D). With the value determined for N it was then possible to calculate P_o (open symbols in Fig. 7A and open bars in Fig. 7B). The second approach to estimate the unknown N for each NP_o value was a procedure based on theoretical considerations. This method does not require the application of the calcium channel agonist, allowing us to include recordings in the analysis that had not been tested for the number of channels with (±)-Bay K 8644, e.g. when the patch was not stable for long enough. The fitting procedure is based on the supposition that under the same experimental conditions the differences between the NP_o values from different patches can only be due to different numbers of channels in the patches. Thus, the NP_o values should always be an integer multiple (N) of the smallest possible value P_o. In other words, when plotting NP_o against N all data points should lie on a straight line crossing the abscissa at zero. According to this supposition, we estimated the unknown value N by assigning those values of N to each patch that resulted in the best linear regression through all data points and zero (Fig. 7A). The values for the single-channel open probability estimated with this method gave the same results as those obtained by the experimental approach (cf. open and filled bars in Fig. 7B). This indicates that the fitting method is a useful approach to estimate the number of channels in a patch, and thus to determine the single-channel open probability, when simultaneous openings of events are too rare to be detected or when the mean open time cannot be increased sufficiently by application of a channel agonist.

Open probabilities determined for each condition according to these procedures were approximately 7- to 8-fold higher in cells coexpressing α_1C or α_1CY467S with β_1a than for cells expressing the α subunit alone. These single-channel data show that the β_1a-induced 4- to 5-fold increase in the specific conductance, g, seen in whole-cell recordings (Fig. 3B) can be fully explained by the increase in single-channel open probability without the need for an increased number of functional channels in the plasma membrane.

DISCUSSION

In the present study we used heterologous expression of α_1C with three different β subunit isoforms in tsA201 cells to describe isoform-specific effects of the β subunit on the subcellular distribution of the calcium channel subunits and on the modulation of current properties. The role of the β subunit interaction domain in the cytoplasmic I-II linker of α_1C was analysed by comparing the properties of wild-type α_1C with those of an α_1C construct in which tyrosine in position 467 was replaced with serine. Immunofluorescence and GFP fluorescence were used to determine the subcellular distribution patterns of the α_1C and β subunits, and current properties were studied with whole-cell and single-channel patch-clamp recordings. The combination of structural and functional analyses revealed β isoform-specific effects on the targeting of both channel subunits and the modulation of calcium currents, but also showed that the two effects of the β subunit are not causally linked.

Isoform-specific effects of β_1a, β_2a and β₃ on targeting of α_1C and on current properties

Coexpression of α_1C with one of the β subunit isoforms resulted in changes of the subcellular distribution of both subunits and in changes in the current properties. Both structural and functional effects of the β subunit on α_1C showed differences depending on the β isoform. When expressed alone, only β_2a was localized in the plasma membrane, whereas β_1a and β₃ were diffusely distributed in the cytoplasm. This is consistent with previous studies (Chien et al. 1995; Brice et al. 1997; Neuhuber et al. 1998b) and can be explained by the intrinsic ability of β_2a to associate with the plasma membrane via a palmitoylation site in its N-terminus. Without a β subunit, α_1C accumulated in the endoplasmic reticulum of tsA201 cells. In only a small fraction of transfected cells could clusters of α_1C be visualized in the plasma membrane, even though calcium currents revealed the presence of functional channels in the majority of transfected cells. Apparently, without coexpression of the β subunit the density of functionally expressed channels in the plasma membrane is too low to be detected with immunocytochemistry or GFP labelling. A rough estimate of the density of functional α_1C subunits in the plasma membrane based on the single-channel properties and the whole-cell data yielded a channel density of 21 μm⁻² (channel density =I_peakC/iP_o= (0.72 × 0.01)/(0.9 × 0.00039) where I_peak (at 0 mV) is in pA pF⁻¹, i is in pA, and the specific membrane capacity (C) is in pF μm⁻², assuming a value of C of 1 μF cm⁻²= 0.01 pF μm⁻²). This value is far below the density of approximately 2200 channels μm⁻² found in skeletal muscle triads where calcium channels can be reliably detected with immunocytochemistry.

When α_1C was coexpressed with any one of the tested β isoforms, α_1C became detectable with immunocytochemistry and GFP labelling in the plasma membrane, where it was now colocalized with the β subunit. This was accompanied by an increase in current density for all subunit combinations, indicating that α_1C and β formed complexes that facilitated the incorporation of channels into the plasma membrane. However, the results of the single-channel recordings show that the current increase observed with the β subunits can be explained by changes of the single-channel properties, i.e. an increase in open channel probability. Thus the number of functional channels in the plasma membrane did not change during coexpression of α_1C with β. Neely et al. (1993) reached the same conclusion for experiments in which β subunit coexpression led to an increase in whole-cell currents without a parallel increase in gating charges. However, other studies suggested that the increase in whole-cell currents was entirely or in part due to an increase in the number of channels in the plasma membrane (Wakamori et al. 1993; Kamp et al. 1996; Josephson & Varadi, 1996). Our observation that the massive incorporation of α_1C-β complexes into the plasma membrane was not accompanied by an increased number of functional channels suggests that some additional factor is limiting the expression of functional channels. If such a factor is saturated to different degrees in different experimental systems, it could explain why some investigators observed a β subunit-induced increase in channel density while others did not.

The ability of the different β isoforms to facilitate incorporation of α_1C-β complexes into the plasma membrane differed qualitatively and quantitatively. Coexpression of the plasma membrane-associated β_2a isoform resulted in an even distribution of α_1C throughout the plasma membrane, suggesting that the intrinsic distribution of β_2a also determined the distribution of α_1C. In addition, clusters of both subunits in the plasma membrane could be observed. The cytoplasmic β_1a and β₃ subunits were not homogeneously expressed throughout the plasma membrane when coexpressed with the α₁ subunit, but instead were only found colocalized with α_1C in distinct clusters. This indicates that their own translocation from the cytoplasm to the plasma membrane is dependent on association with α_1C and that an intrinsic predisposition of α_1C to aggregate in clusters may determine the distribution of the α_1C-β complexes in the plasma membrane. Our observation that only about half of the cells coexpressing β₃ and α_1C showed α_1C-β complexes in the plasma membrane, and in the other cells α_1C and β₃ remained individually localized in the endoplasmic reticulum and the cytoplasm, respectively, suggests that β₃ binds to α_1C with a somewhat lower affinity than β_1a and β_2a. This is consistent with data from De Waard et al. (1995) who showed that β₃ bound to α_1A with lower affinity than β_1b, β_2a or β₄.

The isoform-specific modes of membrane targeting of the calcium channel α_1C and β subunits were paralleled by the isoform-specific modulation of whole-cell calcium currents. Whereas all three β isoforms caused an increase of current density due to an increased specific conductance, a shift in the voltage dependence of activation was most pronounced with β_2a, and only β_2a caused a dramatic decrease in the rate of inactivation. A specific effect of β_2a on current inactivation has previously been shown for α_1A and α_1E when expressed in oocytes (Sather et al. 1993; Qin et al. 1996; Parent et al. 1997), indicating that the distinct effects of β_2a shown in the present study are not the result of combining the two native subunit partners of cardiac muscles; instead this property must reside on β_2a itself. The observation that β_2a differs from β_1a and β₃ in its targeting properties as well as in its modulatory ability suggests that these properties may be related. This interpretation is supported by a recent finding by Qin et al. (1998) showing that removal of the palmitoylation site in β_2a reversed several β_2a-specific modulatory effects on α_1C and α_1E. Thus, the subcellular organization of isolated calcium channel complexes has a profound influence on the functional characteristics of the channel.

Dissociation of effects of the β subunit on targeting of α_1C and on calcium currents

The substitution of a single residue (Y467S) in the β subunit interaction domain of α_1C disrupted the formation and incorporation of α_1C-β complexes into the plasma membrane but not the modulatory effects of β on calcium currents. Pragnell et al. (1994) identified a motif of nine conserved amino acids that is critical for binding the β subunit, in the cytoplasmic loop between repeats I and II of the α₁ subunit. Mutations within this motif of α_1A perturbed β subunit binding (Pragnell et al. 1994). Here we have shown that α_1CY467S and a coexpressed β subunit did not colocalize in the plasma membrane but remained localized in distinct subcellular compartments. α_1CY467S was concentrated in the endoplasmic reticulum, β_1a and β₃ were diffusely distributed in the cytoplasm, and β_2a was evenly distributed in the plasma membrane - all as when wild-type α_1C and the β subunits were expressed individually. Therefore we conclude that the stable association of α_1CY467S and β failed and that this had a dramatic effect on the incorporation of all subunits except β_2a into the plasma membrane. Previously we have shown that the stable association of the skeletal α_1S isoform and β_1a can occur in the endoplasmic reticulum upon coexpression, as seen by translocation of β_1a from the cytoplasm to the endoplasmic reticulum, and that this translocation fails when the β subunit interaction domain in α_1S is mutated at the corresponding tyrosine residue (Y366S) (Neuhuber et al. 1998b). Here we observed in some cells the colocalization of the cytoplasmic β subunits with α_1C (but not with α_1CY467S) in the endoplasmic reticulum, indicating that in these cells the two subunits formed a complex but were not efficiently transported to the plasma membrane. Therefore, if the mutation in the cardiac α_1CY467S subunit had not perturbed binding, we would expect to see the colocalization of α_1CY467S and β in the endoplasmic reticulum, even if the export of α_1CY467S from the endoplasmic reticulum had failed for a reason other than the lack of α_1C-β complex formation; but instead α_1CY467S and the β subunits remained in distinct compartments. Moreover, α_1CY467S expressed alone generated calcium currents of the same magnitude as wild-type α_1C, indicating that α_1CY467S is a fully functional channel. Thus, the lack of colocalization of α_1CY467S and the β subunits indicates that none of the three examined β isoforms form stable complexes with the mutated α_1CY467S subunit.

Nevertheless, the increase in current density, the modulation of current kinetics, and the increase in single-channel open probability upon coexpression of β were equally observed with wild-type α_1C and the mutant α_1CY467S. This shows first that the increased plasma membrane localization of α_1C, observed when α_1C was coexpressed with β, is not a prerequisite for the increased current density, and second that the stable association of α₁ with a β subunit is not required for their functional interaction. Apparently, the majority of visible channels in the plasma membrane are not functional and unfortunately it cannot be unambiguously verified that the population of functional channels does not form complexes with the β subunit as well. However, if the large visible fraction of α_1CY467S channels fails to associate with β subunits one can expect that the functional fraction will equally fail to do so and, consequently, their functional modulation by β subunits most probably does not depend on the stable association of α_1CY467S with a β subunit. Our finding that upregulation of current density by β is due to a mechanism other than the increased channel density, such as an increased channel open probability, is in agreement with several previous studies in which β subunit-induced effects on single-channel properties have been reported (Neely et al. 1993; Wakamori et al. 1993; Shistik et al. 1995). In the present study we have shown an increase in dwell time and open probability of α_1C and α_1CY467S upon coexpression with β_1a. Whereas the analysis of the number of functional channels per patch gave no indication of a strong increase on coexpression with β, single-channel open probability increased about 7- to 8-fold - more than enough to explain the increase in whole-cell current density. Consequently, the simultaneously occurring increase of detectable channels in the plasma membrane must be a parallel but not causally linked effect of the β subunit; only in this way can one fail without affecting the other. In a recent study in Xenopus oocytes, Yamaguchi et al. (1998) showed similar separate effects of β₃ on the functional modulation of α_1C and on increased membrane incorporation. They achieved a temporal dissociation of the two effects by injecting a β₃ fusion protein into α_1C-expressing oocytes and were able to block membrane incorporation without affecting β subunit-induced modulation of current kinetics and voltage dependence. Together, the study by Yamaguchi et al. (1998) and the present study provide strong evidence for dual and independent roles of the accessory β subunit in targeting and modulation of the α₁ subunit.

To what extent these functions contribute to the normal incorporation and function of the channels in the excitation-contraction coupling apparatus in muscle is not clear. Expression of the skeletal α_1S isoform with the analogous mutation in the β subunit interaction domain in skeletal myotubes of the dysgenic mouse also resulted in the disruption of α_1S-β_1a complexes, although without perturbing the specific targeting of α_1S into the triads and the reconstitution of normal function (Neuhuber et al. 1998a). This is consistent with the interpretation of the results presented here that β subunits can exert their functions without forming stable α₁-β complexes. However, because in the muscle expression system calcium currents and excitation-contraction coupling could not be analysed in the absence of the endogenous β subunit, functional modulation of calcium currents by β could not be directly tested with α_1S. Our present result that specific modulatory effects of three different β subunits can all be observed with a mutated α₁ subunit (α_1CY467S), which fails to colocalize with the β isoforms, provides strong evidence that the formation of α_1C-β complexes is not required for functional interactions between the two calcium channel subunits.

Dual mode of α_1C-β interactions

We observed two effects of the β subunit on α_1C: an increased incorporation of α_1C into the plasma membrane which is dependent on the formation of stable α_1C-β complexes, and the modulation of current properties which is independent of the formation of stable α_1C-β complexes. The β subunit interaction domain in the cytoplasmic loop between repeats I and II of α_1C is clearly responsible for the stable association of α_1C and β, and for the increased incorporation of the complex into the plasma membrane. In contrast, it may or may not be the site of functional α_1C-β interactions. Either the point mutation lowered the affinity of this β subunit interaction domain on α_1C or a second, low affinity interaction domain exists apart from that in the I-II linker. In either case high concentrations of free β subunit in the cytoplasm or in the plasma membrane could interact with such a low affinity binding site and thus cause functional modulation without forming a stable complex. Distinct domains for the functional modulation of α₁ have been described on the β subunit (Olcese et al. 1994; Cens et al. 1998). Furthermore, a second low affinity binding domain in α_1E that mediates modulation of inactivation kinetics has been described (Walker et al. 1998). However, these results differ from the present data in that other functional effects of the β subunit on α_1E, like the increase in current amplitude, were mediated by the I-II linker, whereas in our study the loss of binding to the I-II linker of α_1C did not perturb any modulatory effects of β on current properties. Finally, it has been suggested that more than one β subunit can interact simultaneously with different binding sites on the α₁ subunit. Our results are consistent with such a model in that structural and functional effects of β subunit coexpression appeared to be independent. However, one can also envisage a mechanism for α_1C-β interactions by which binding of a β subunit to the β subunit interaction domain in the I-II linker brings β into a position to interact with the functionally important residues on α_1C, but that this binding is not necessarily required.

In all, the results of this study show that the β subunits of calcium channels serve a dual function, firstly in the formation of a complex with α_1C and its incorporation into the plasma membrane and secondly in the modulation of single-channel calcium current properties, but it also shows that these two functions of the β subunit are independent of each other.