Molecular Diversity of the Calcium Channel α₂δ Subunit

Abstract

Sequence database searches with the α₂δ subunit as probe led to the identification of two new genes encoding proteins with the essential properties of this calcium channel subunit. Primary structure comparisons revealed that the novel α₂δ-2 and α₂δ-3 subunits share 55.6 and 30.3% identity with the α₂δ-1 subunit, respectively. The number of putative glycosylation sites and cysteine residues, hydropathicity profiles, and electrophysiological character of the α₂δ-3 subunit indicates that these proteins are functional calcium channel subunits. Coexpression of α₂δ-3 with α_1C and cardiac β2a or α_1E and β3 subunits shifted the voltage dependence of channel activation and inactivation in a hyperpolarizing direction and accelerated the kinetics of current inactivation. The kinetics of current activation were altered only when α₂δ-1 or α₂δ-3 was expressed with α_1C. The effects of α₂δ-3 on α_1C but not α_1E are indistinguishable from the effects of α₂δ-1. Using Northern blot analysis, it was shown that α₂δ-3 is expressed exclusively in brain, whereas α₂δ-2 is found in several tissues. In situ hybridization of mouse brain sections showed mRNA expression of α₂δ-1 and α₂δ-3 in the hippocampus, cerebellum, and cortex, with α₂δ-1 strongly detected in the olfactory bulb and α₂δ-3 in the caudate putamen.

Voltage-gated calcium channels have been purified and cloned from various tissues such as skeletal muscle, heart, and brain. These channels are formed by heterooligomeric complexes consisting of various combinations of an α₁protein with auxiliary α₂δ, β, and γ subunits. To date, seven genes encoding α₁ subunits of the high voltage-activated (for review, see Hofmann et al., 1994; Strom et al., 1998; Bech-Hansen et al., 1998) and two genes of the low voltage-activated calcium channels have been identified (Perez-Reyes et al., 1998; Cribbs et al., 1998). This subunit accounts not only for the ion channel pore but also contains the voltage sensor and the determinants for binding of drugs and toxins. The current through the α₁ subunit is modulated by interactions with the β, α₂δ, and γ subunits. The molecular diversity of the β subunit is not only caused by the expression of four genes, but also by differential splicing. Until recently, only a single γ subunit in skeletal muscle had been described (Eberst et al., 1997), but a novel neuronal form has since been identified in brain (Letts et al., 1998).

Since the molecular cloning of the α₂δ subunit (Ellis et al., 1988), several splice variants have been detected, but no further α₂δ subunit genes have been identified. The different splice variants arise from various combinations of three alternatively spliced regions that result in five isoforms that are expressed in a tissue-specific manner (Angelotti and Hofmann, 1996). Structurally, the α₂δ subunit is a heavily glycosylated 175 kDa protein that is posttranslationally cleaved to yield a disulfide-linked α₂ and δ protein (DeJongh et al., 1990; Jay et al., 1991). The δ part anchors the α₂protein to the membrane via a single transmembrane segment (Gurnett et al., 1996). The membrane topology of the α₂δ subunit was further refined using anti-α₂ antibodies and C-terminal deletion mutants (Brickley et al., 1995; Wiser at al., 1996). Structural studies have shown that the extracellular α₂ domain provides the structural elements required for channel stimulation (Gurnett et al., 1996) and that the δ domain, which contains the only transmembrane segment of α₂δ complex, harbors the regions important for the shift in voltage-dependent activation, steady-state inactivation, and the modulation of the inactivation kinetics (Felix et al., 1997). The identification of new α₂δ subunits could present further possibilities for differential and specific regulation of calcium channels.

In studies designed to address the diversity and function of α₂δ, cloning, hybridization, and patch-clamp techniques were used to identify and characterize novel α₂δ subunits. Two new α₂δ genes were found by searching databases with the α₂δ cDNA sequence (Ellis et al., 1988). These genes encode proteins with essential features of α₂δ subunits, such as the high number of potential glycosylation sites and cysteine residues, the hydrophobicity plots, and electrophysiological characteristics. The novel α₂δ-2 gene was found to be predominately expressed in heart, pancreas, and skeletal muscle, with α₂δ-3 expressed only in brain.

MATERIALS AND METHODS

Isolation of RNA and cDNA library construction. Total RNA from mouse brain was isolated by the guanidinium thiocyanate method, and the poly(A) RNA was separated by oligo-dT cellulose chromatography (Poly(A) Quik mRNA Isolation kit; Stratagene, Heidelberg, Germany). Poly(A) RNA from mouse brain was reverse-transcribed, and double-strand cDNA was synthesized using the Superscript plasmid system (Life Technologies). The cDNA fragments were ligated with BstXI/EcoRI adaptors (Invitrogen, San Diego, CA) and size-fractionated in a low-melting agarose gel. Only gel slices containing fragments >2000 bp were excised and digested with Gelase (Biozym, Hessisch Oldendorf, Germany). Recovered cDNA was ligated into the BstXI site of the pcDNAII vector (Invitrogen) and transformed in the Escherichia coliXL1-blue mrf′ strain (Stratagene, Heidelberg, Germany). The cDNA library was screened with a random-primed labeled PCR probe (see below, nt 1501–1821). Sequencing of the clones was performed using the dideoxy chain termination method on both strands. A full-length clone was used for the preparation of the NotI-SphI fragment with the entire open reading frame in pcDNA3 (Invitrogen) yielding the expression plasmid pc3α₂δ-3.

PCR amplification. The expressed sequence tag (EST) with the accession number AA190607 was used to design the primers NKAD1, GGC ACA GAT GTC CCA GTT AAA GA and NKAD2, TGT ATA GTA GTA GTC ATT GGT CAT, with which the partial α₂δ-3 subunit cDNA was amplified. The PCR fragment was cloned in pUC18 and was sequenced on both strands.

Northern blot analysis. Human and mouse multiple tissue Northern Blots were obtained from Clontech (Heidelberg, Germany) and hybridized according to the manufacturer’s instructions. Random-primed labeled fragments (α₂δ-2, nt 2877–3249; α₂δ-3, nt 2893–3377) were used as probes; these regions share no significant homology with each other. A 3 hr prehybridization step was followed by an overnight hybridization with 5 × 10⁶ cpm/ml of probe at 42°C. The final stringency wash performed was with 0.1× SSC, 0.1% SDS at 42°C.

In situ hybridization. Intact brains of adult female mice were removed immediately after cervical dislocation and frozen in isopentane cooled to −40°C. Brain was sectioned into 16 μm slices in a cryostat at −20°C and thaw-mounted onto polylysine slides. The tissue was vacuum-dried, fixed in 4% paraformaldehyde in PBS, and washed in 0.5× SSC. The sections were acetylated in 0.25% acetanhydride in 0.1 m triethanolamine, pH 8.0, and washed in 2× SSC. The tissue was dehydrated through a series of ethanol solutions, from 50 to 100%. The slides were vacuum-dried and stored at −80°C until used.

Murine α₂δ-1 (nt 2760–3170), human α₂δ-2 (nt 2877–3249), and murine α₂δ-3 (nt 2744–3228) specific riboprobes were generated by in vitro transcription as described previously (Ludwig et al., 1997). Briefly, probe template DNA was amplified by PCR from murine α₂δ-1 and -2 clones using primers specific for the variable C-terminal region of α₂. BamHI andAsp718 restriction sites were added to the α₂δ-1 forward and reverse primers, respectively, to allow for sticky-end ligation. α₂δ-3 was ligated to the linearized vector by blunt end ligation. The integrity of each probe was verified by dideoxy termination sequencing. ³⁵S-UTP (DuPont NEN, Wilmington, DE)-labeled sense and antisense riboprobes were generated using a standard T3 and T7 polymerase in vitro transcription procedure (T3 and T7 from New England Biolabs, Beverly, MA). Unincorporated nucleotides were removed using Sephadex G50 columns (Pharmacia, Freiburg, Germany).

Messenger RNA in situ hybridization was performed on cryostat sections of mouse brain. A prehybridization step was performed at 45°C for 2 hr. Hybridization with 1 × 10⁷cpm/ml probe proceeded for 16 hr at 55°C. After a 20 μg/ml RNase A (Boehringer Mannheim, Mannheim, Germany) digestion, a high-stringency wash of 0.1× SSC, 1 mm EDTA, and 1 mmdithiothreitol was done for 2 hr at 60°C. The slides were dehydrated in ethanol and analyzed by autoradiography.

Transfection of HEK293 cells and electrophysiological recordings. The full-length cDNAs of all subunits, i.e., α_1C, α_1E, β2a, β3, α₂δa-1, and α₂δ-3 were cloned into the pcDNA 3 vector (Invitrogen). For more details, see Schuster et al. (1996). HEK 293 cells were transfected with various combinations of an α₁ subunit (α_1C, α_1E), a β subunit (β2a, β3), and an α₂δ subunit (α₂δ-1, α₂δ-3). This was achieved by lipofection with Lipofectamine (Life Technologies) at a DNA mass ratio of 1:1 for expression of two subunits or 1:1:1 for three subunits.

Electrophysiological recordings. Ionic currents from transfected cells were recorded in whole-cell configuration of the patch-clamp method. Ba²⁺ was used as the charge carrier. The extracellular solution contained (in mm):N-methyl-d-glucamine, 125; BaCl₂, 20; CsCl, 5; MgCl₂, 1; HEPES, 10; and glucose, 5, pH 7.4 (HCl). The intracellular solution contained (in mm): CsCl, 60; CaCl₂, 1; EGTA, 11; MgCl₂, 1; K₂ATP, 5; HEPES, 10; and aspartic acid, 50, pH 7.4 (CsOH). Currents were recorded using an EPC-9 patch-clamp amplifier and corresponding Pulse software from Heka Electronics (Lambrecht, Germany). Patch pipettes were pulled from borosilicate glass. The pipette input resistance was typically between 1.8 and 2.2 MΩ. The capacity of individual cells ranged between 25 and 90 pF, and series resistance ranged between 3.5 and 5.0 MΩ. Capacity transients were compensated using build-in procedure of the Heka system. Curve fitting and statistical analysis were performed using the Origin 5.0 software package (Microcal, Northampton, MA). The significance of observed differences was evaluated by nonpaired Student’s t test. Probability of 5% or less was considered to be significant.

RESULTS

Primary structures

A database search using the α₂δ subunit cDNA as probe revealed three additional sequences similar to this calcium channel subunit (Fig. 1), which were derived from two genes named 2 and 3. Two of these sequences are human full-length sequences of closely related isoforms of the α₂δ subunit (GenBank accession numbers: AF042792, isoform I; AF042793, isoform II; M. H. Wei, F. Latif, F. M. Duh, D. Andreazzoli-Angeloni, V. Kashuba, E. Zabarovsky, B. Johnson, and M. I. Lerman, unpublished observations). These sequences differ only at the N terminus of the α₂ protein, indicating that both isoforms are derived from the same gene 2 and are generated by differential splicing. The 5′-untranslated region upstream of the ATG codon from isoform I is in good agreement with the Kozak sequence for initiation of translation in eukaryotic cells, whereas the same region of isoform II shows only a limited homology with this sequence. Furthermore, only isoform I but not isoform II shows features of a potential signal sequence. Both observations suggest that only isoform I can form a functional calcium channel subunit. Isoform I, which we describe here as the α₂δ-2 subunit, (Wei, Latif, Duh, Andreazzoli-Angeloni, Kashuba, Zabarovsky, Johnson, and Lerman, unpublished observations) has 55.6% identity with the α₂δ-1 subunit (Ellis et al., 1988). Eighteen potentialN-glycosylation sites can be identified in the primary structure (Fig. 1), which is the same number of sites as in the α₂δ-1 subunit (Ellis et al., 1988).

Fig. 1.

a, Amino acid alignment of the α₂δ-1 (1), α₂δ-2 (2), and α₂δ-3 (3) subunits. The N-terminal region differing between the α₂δ-2 subunit isoform I and II isunderlined. Regions that are identical in all sequences are boxed, and conserved cysteine residues are additionally highlighted. The presumptive signal peptides for classes 1 and 3 are shown in italics. Potential N-glycosylation sites are printed inbold. The arrow indicates the cleavage site between the α₂ and δ proteins of the α₂δ-1 subunit. These sequence data are available from the EMBL database under accession numbers M21948 for the α₂δ-1 subunit, AF042792 for the α₂δ-2(I) subunit, and AF042793 for the α₂δ-2(II) subunit isoforms, respectively, and AJ010949for the α₂δ-3 subunit. b, Hydrophobicity profile of the α₂δ subunits computed according to Kyte and Doolittle (1982). The curve is the average of a residue-specific hydrophobicity index over a window of nine residues.

The third sequence identified by database searches was a partial human EST sequence AC: AA190607. This sequence was used to design primers for the amplification of cDNA from mouse tissues. The cDNA could be amplified only from mouse brain mRNA and not from other tissues (Fig.2). A cDNA library was constructed from mouse brain mRNA and screened with the PCR product as a probe. Several independent clones could be identified that encode the novel α₂δ subunit, which we name α₂δ-3 (Fig.1). A detailed restriction analysis of the different clones showed that there are no further isoforms in mouse brain. The sequence upstream of the start ATG is in agreement with that for the initiation of translation in eukaryotic cells (data not shown). The open reading frame consists of 3273 bp encoding a protein with 1091 amino acid residues. The α₂δ-3 subunit has 30.3% identity with the α₂δ-1 subunit and 31.2% with the α₂δ-2 subunit. The primary structure of α₂δ-3 contains nine potentialN-glycosylation sites (Fig. 1a).

Fig. 2.

Northern blot analysis of the α₂δ-2 and α₂δ-3 subunits. For each lane, ∼2 μg of poly(A) RNA was run on a denaturing formaldehyde-containing agarose gel, transferred to a nylon membrane, and fixed by UV irradiation. a, Human multiple tissue blot using a specific probe for the α₂δ-2 subunit.b, Mouse multiple tissue blot for the α₂δ-3 subunit. Arrows indicate predominant species of mRNA with sizes of 5.2 (α₂δ-2) and 4.3 (α₂δ-3) kb.

Hydrophobicity analysis of all three α₂δ subunits indicates a similar membrane topology, including a hydrophobic transmembrane segment at the C terminus of the δ domain (Fig.1b). A sequence comparison reveals that as many as 14 cysteine residues are conserved in all of the three α₂δ genes, further strengthening the postulate that these subunits are disulfide-linked proteins with similar higher order structures.

Tissue distribution

The expression of the novel α₂δ subunits was examined by Northern and in situ hybridization. The α₂δ-2 subunit is highly expressed in heart, pancreas, and skeletal muscle tissue (Fig. 2), which can be seen even after 6 hr of autoradiography of Northern blots. After 1 d of exposure, the α₂δ-2 subunit can also be detected in kidney, liver, placenta, and brain. This broad expression pattern was also observed for the α₂δ-1 subunit. An 8.0 kb transcript of the α₂δ-1 subunit was detected in brain, heart, aorta, skeletal muscle, and ileum (Ellis et al., 1988). In contrast to this ubiquitous expression pattern, the α₂δ-3 subunit is only found in brain (Fig. 2). In each case, the Northern analysis reveals one predominant species of mRNA with sizes of 5.2 (class 2) and 4.3 kb (class 3).

The mRNA expression of α₂δ-1 and α₂δ-3 calcium channel subunits in mouse brain was mapped by in situ hybridization. Both α₂δ forms were detected in several regions of the brain, with differential expression of the mRNAs in some structures. Strong expression of α₂δ-1 was seen in the pyramidal cell layer of Ammon’s horn in the hippocampus (CA1–3) and in the granular cell layer of the dentate gyrus (Fig. 3). The olfactory bulb also stained strongly, and expression in the mitral and glomerular cell regions was seen by dark-field microscopy of slides coated with photographic emulsion. Signals in the cerebellar cortex, and to a lesser extent in thalamic nuclei, were also observed (Fig. 3). Examination of emulsion-coated slides showed that the expression of α₂δ-1 in the cerebellum was restricted to the granular layer (data not shown).

Fig. 3.

Autoradiographs of α₂δ-1 and α₂δ-3 riboprobe hybridization to horizontal mouse brain sections. Central (a, c) and more basal (b, d) sections of the brain are shown. Expression of α₂δ-1 is seen in the (a) hippocampus (H), cerebral cortex (Co), cerebellum (Ce), and (b) olfactory bulb (Ob). α₂δ-3 mRNA was detected in the caudate putamen (CPu), hippocampus (H), entorhinal complex (E), cortex (Co), and thalamic nuclei (T) (c, d).

The α₂δ-3 mRNA was predominantly expressed in the caudate putamen, entorhinal complex, hippocampus, and cortex (Fig. 3). As with α₂δ-1, the pyramidal cell layer of the hippocampus (CA1–3) and granular cell layer of the dentate gyrus showed the highest degree of hybridization with the antisense probe (Fig. 3). Hybridization of both probes was specific as judged by the absence of signals when sense probes were applied (data not shown).In situ hybridization for the α₂δ-2 subunit indicated a ubiquitous expression in cardiac tissues (data not shown).

Functional characterization

Because the Northern blot hybridization indicated a high expression of the α₂δ-3 subunit in brain, we performed cotransfection studies with α₁ subunits that have been characterized in this tissue. Because in situ analysis did not show an exact colocalization of α₂δ-3 with known α₁ subunits, coexpression studies were done with an α₁ subunit of the dihydropyridine-sensitive class C and a dihydropyridine-insensitive class E calcium channel. This approach allows for the identification of interactions with both evolutionary distant calcium channel α₁ families.

Effects of the α₂δ-3 subunit on barium current through the α_1C-type calcium channel

To study the effects of α₂δ-3 subunit on channel gating, we coexpressed this subunit with the α_1C subunit alone or in combination with both α_1C and β2a subunits. Voltage protocols are described in detail in the legend to Figure4, and quantitative analyses are summarized in Table 1. When coexpressed with α_1C alone, the effects of α₂δ-3 on I_Ba were less noticeable than when β2a was coexpressed. α₂δ-3 did not affect the current density, time course of current inactivation, and voltage dependence of steady-state inactivation of α_1C expressed singly (data not shown). However, when expressed with α_1C only, α₂δ-3 shifted the activation curve by 4.4 mV in the hyperpolarizing direction and slightly accelerated the time of current activation during the depolarizing pulse (data not shown). The effects on the channel kinetics became more prominent in the presence of the β2a subunit (Fig. 4, Table 1). In this combination, α₂δ-3 significantly increased the current density, shifted the voltage dependence of current activation by 8.7 mV in a hyperpolarizing direction, accelerated the time course of current activation during the depolarizing pulse, accelerated the current inactivation at positive membrane potentials, shifted the steady-state inactivation curve in a hyperpolarizing direction, and significantly changed its slope.

Fig. 4.

The α₂δ subunit affects current through the α_1C-type calcium channel. The α₂δ-3 subunit was coexpressed with α_1C and β2a. For comparison, the current through cells coexpressing α₂δ-1 subunit was coanalyzed.a shows the voltage dependence of current activation measured as the amplitude of current activated by a 40-msec-long depolarizing pulse from a holding potential of −80 mV to voltages marked on the ordinate and normalized to the maximal amplitude. Each voltage dependence was fitted to the Boltzmann equation. Results of these fits are summarized in Table 1. The inset in thetop left of a shows voltage dependence of the kinetics of current activation. The ascending phase of the current time course was fitted to single exponential. The resulting time constants were averaged and plotted against corresponding membrane potentials. In both graphs, ○ represents the α_1Cβ2a channel, • the α_1Cβ2aα₂δ-1 channel, and ▪ the α_1Cβ2aα₂δ-3 channel. Theinset in the right of aillustrates a typical family of currents measured during a series of depolarizing pulses from the holding potential of −80 mV to membrane potentials ranging from −20 to +70 mV with a step of +10 mV. ○, α_1Cβ2a channel. The cell capacity was 83 pF, and the resulting maximal current density was approximately −13 pA/pF; ▪, α_1Cβ2aα₂δ-3. The cell capacity was 31 pF, and the corresponding maximal current density was approximately −41 pA/pF. b shows averaged steady-state inactivation curves measured from a holding potential of −80 mV. Current was inactivated by a 5-sec-long prepulse to the potentials marked on the ordinate. This was followed by a 5-msec-long return to the holding potential and a 40-msec-long test pulse to the maximum of the current–voltage relationship. Solid lines are fitted to the Boltzmann equation. The inset shows the time course of the current during a 5-sec-long depolarizing pulse to +20 mV with zero level indicated by a horizontal line. Currents shown are averaged time courses from 9 to 12 experiments scaled to the same amplitude. Individual measurements were fitted to the sum of two exponentials. Results of all fitting procedures are summarized in Table1. Symbols are as in a.

Table 1.

Comparison of the effects of α₂δ-1 and α₂δ-3 subunits on I_Ba through α_1C or α_1E-type calcium channels

	α1Cβ2a	α1Cβ2aα2δ-1	α1Cβ2aα2δ-3	α1Eβ3	α1Eβ3α2δ-1	α1Eβ3α2δ-3
IBa density (pA/pF)	13.8  ± 1.4 (16)	30.3  ± 6.6* (15)	28.6  ± 4.9* (21)	24.4  ± 5.4 (12)	98.1  ± 20.0*** (10)	95.4  ± 15.6*** (10)
V0.5 activ. (mV)	+5.2  ± 0.1 (16)	−5.0  ± 0.6*** (13)	−3.5  ± 0.5*** (11)	−13.0  ± 1.1 (10)	−14.5  ± 0.2 (9)	−20.1  ± 0.7***, aaa (11)
kactiv	5.4  ± 0.1	5.1  ± 0.4	5.1  ± 0.4	5.6  ± 0.9	3.7  ± 0.1	4.6  ± 0.6
V0.5 inact. (mV)	−12.5  ± 0.9 (9)	−19.6 ± .3*** (14)	−20.5  ± 0.4*** (8)	−64.4  ± 0.4 (9)	−69.7  ± 0.4*** (11)	−76.2  ± 0.5***, aaa (9)
kinact	11.1  ± 0.7	8.5  ± 0.2***	8.0  ± 0.3**	7.1  ± 0.4	8.4  ± 0.4*	9.5  ± 0.4***
A1 (%)	61.1  ± 3.9 (12)	73.7  ± 4.5* (11)	78.1  ± 2.6** (9)	93.0  ± 2.1 (8)	97.3  ± 0.5* (10)	98.5  ± 0.5** (10)
τ1 (msec)	273  ± 24	260  ± 19	156  ± 10***, aaa	31.9  ± 5.7	23.6  ± 0.9	24.1  ± 1.7
τ2 (sec)	1.16  ± 0.08	3.54  ± 0.46***	0.90  ± 0.10aaa

The data are the means ± SEM. The number of experiments performed is indicated in parentheses. The first row compares the average current density for each channel. The second row compares voltage dependence of current activation of different subunit combinations. The ascending part of the I–Vcurve, which is shown in panel a of Figures 4 and 5 was fitted to the Boltzmann function. V_{0.5 activ} and k_activ are the parameters of these fits. The fourth row compares voltage dependence of current inactivation. The steady-state inactivation curves shown in panel b of Figures 4 and 5 were fitted to the Boltzmann function. V_{0.5 inact} and k_inact are parameters of Boltzmann fits. The last three rows show the results of fitting the time course of inactivation ofI_Ba during a 5-sec-long (α_1C-derived channels, traces shown in Figure4b) or 300-msec-long (α_1E-derived channels, traces shown in Figure 5b) depolarizing pulse to +20 mV. Currents through α_1C-derived channels were fitted to a sum of two exponentials. A1 represents the relative amplitude of the current inactivating with the fast time constant τ1. α_1E-derived channels were found to inactivate with single time constant. For these channels, 100-A1 represents the relative amplitude of the noninactivating component of the current remaining at the end of the depolarizing pulse (see also Fig. 5b). The significance of the difference was tested using the unpaired Student’st test. Asterisks (*) indicate the statistical significance between α_1Cβ2a and α_1Cβ2aα₂δ-1 or α_1Cβ2aα₂δ-3 and between α_1Eβ3 and α_1Eβ3α₂δ-1 or α_1Eβ3α₂δ-3. Significant difference between α_1Cβ2aα₂δ-1 and α_1Cβ2aα₂δ-3 and between α_1Eβ3α₂δ-1 and α_1Eβ3α₂δ-3 is marked by (^a).

*p < 0.05; **p < 0.01; ***p < 0.001; ^aaap < 0.001.

We further compared the effects of α₂δ-3 with those of the previously described α₂δ-1 subunit (Fig. 4, Table1). The changes in gating-related channel characteristics elicited by both subunits were qualitatively similar, and in most cases the measurements (current density, voltage dependence of current activation and inactivation, and time course of current activation) were not significantly different. Both α₂δ-1 and -3 subunits accelerated the time course of current inactivation at a membrane potential of +20 mV by enhancing the proportion of the fast time constant, τ1. α₂δ-3 changed this constant from 273 ± 24 msec to 156 ± 10 msec. Surprisingly, α₂δ-1 significantly increased the value of the slow time constant τ2 from 1.16 ± 0.08 sec to 3.54 ± 0.46 sec, but this effect was apparently overwhelmed by the effect of the increased proportion of the current inactivating with fast time constant τ1, with the result that the overall time course of inactivation was accelerated (Fig. 4b, inset). The effects of both α₂δ-3 and α₂δ-1 subunits on whole-cell current parameters, which reflect the gating of the α_1C channel, are virtually identical and they require the presence of β (in this case β2a) to become prominent.

Effects of the α₂δ-3 subunit on barium current through α_1E-type calcium channel

Although both α_1C and α₂δ-1 subunits are fairly abundant in mammalian tissues, α_1E and α₂δ-3 are predominantly expressed in neuronal tissue. We therefore selected α_1E for studying the effects of α₂δ-3. For all experiments, β3 subunit was coexpressed with α_1E. This β subunit was suggested to modulate the current through the α_1E channel (Ludwig et al., 1997). As with the α_1C channel, both α₂δ-1 and α₂δ-3 affected most of the gating-related parameters except for the time constant of current activation during membrane depolarization (Fig.5, Table 1). In contrast to α_1Cβ2a channel, the effects of α₂δ-3 and α₂δ-1 on the voltage dependence of current activation (Fig. 5a) and inactivation (Fig. 5b) of α_1Eβ3 channel were significantly different. The α₂δ-1 subunit shifted both activation and steady-state inactivation curves in a hyperpolarizing direction, but the change in current activation was not statistically significant. In both curves, the shift evoked by α₂δ-3 was significantly larger (Table 1). In contrast, the effects of both α₂δs on the time course of current inactivation at a membrane potential of +20 mV were identical and restricted to diminution of the noninactivating part of the current (Table 1).

Fig. 5.

The α₂δ subunit affects current through the α_1E-type calcium channel. Unless otherwise indicated, the voltage protocols used were the same as those described in the legend to Figure 4. ○ represents the α_1Eβ3 channel; • the α_1Eβ3α₂δ-1 channel, and ▪ the α_1Eβ3α₂δ-3 channel. Boltzmann fits of voltage dependencies of current activation are summarized in Table 1. The inset in theright of a shows a typical family of currents measured during a series of depolarizing pulses from a holding potential of −100 mV to membrane potentials ranging from −30 mV to +60 mV with step of +10 mV. ○, α_1Eβ3 channel, with a cell capacity of 19 pF and a maximal current density of approximately −34 pA/pF; ▪, α_1Eβ3α₂δ-3 channel, with a capacity of 43 pF and a maximal current density of approximately −74 pA/pF. b shows the steady-state inactivation curve measured from a holding potential of −100 mV using a 5-sec- long conditioning pulse to membrane potentials marked on the ordinate.Solid lines represent Boltzmann fits. Theinsert illustrates the inactivation ofI_Ba during a 300-msec-long depolarizing pulse from a holding potential of −100 mV to +20 mV scaled to the same amplitude. Eight to 10 measurements were averaged for each channel type. The α_1Eβ3α₂δ-1 and α_1Eβ3α₂δ-3 current traces are indistinguishable from each other. Individual time courses of current inactivation were fitted to a single exponential with a small proportion of noninactivating current. Results of all fits are summarized in Table 1. Symbols are as ina.

DISCUSSION

We present here the first account of the existence of multiple α₂δ calcium channel subunits. The previously known α₂δ we have named α₂δ-1, and the new forms have been designated α₂δ-2 and α₂δ-3 based on their similarity to the original subunit. An amino acid alignment reveals that there is only a significant degree of homology in the core region of the α₂ protein, whereas the δ proteins show identities only with respect to the cysteine residues. Despite the low degree of homology at the primary structure level between these forms, other features such as the number of glycosylation sites, cysteine residues, and hydrophobicity profiles are very similar. For these reasons we conclude that all three α₂δ subunits are disulfide-linked proteins with similar higher order structures. The α₂δ-1 and -2 subunits are more ubiquitously expressed than α₂δ-3, which has only been identified in brain. Comparisons of the expression patterns of the 1 and 3 class α₂δ subunits with α₁ subunits in brain slices (Ludwig et al., 1997) do not indicate specific α₁-α₂δ combinations. We cannot exclude the possibility that α₂δ-3 also interacts with other ion channels or even other membrane proteins.

Functional coexpression of the α₂δ-1 subunit with various combinations of α₁ and β subunits results in an increase in the current densities or dihydropyridine (DHP)-binding sites (Singer et al., 1991; Welling et al., 1993;De Waard et al., 1995; Shistik et al., 1995; Bangalore et al., 1996;Gurnett et al., 1996; Felix et al., 1997; Parent et al., 1997;Jones et al., 1998), acceleration of current activation and inactivation (Singer et al., 1991; De Waard et al., 1995; Bangalore et al., 1996; Felix et al., 1997; Qin et al., 1998; Shirokov et al., 1998), and a shift of the current–voltage curve in a hyperpolarizing direction (Singer et al., 1991; Felix et al., 1997). Regardless of the α₁ subunit used in this study, the α₂δ-3 subunit was found to increase the current density, which is in agreement with the results of these groups. In addition to this effect, coexpression of α₂δ-3 caused a shift of the voltage dependence of channel activation and inactivation in a hyperpolarizing direction and an acceleration in the kinetics of current inactivation.

When the results shown in Figures 4 and 5 are compared, it can be seen that α₂δ-1 caused a smaller shift in the voltage dependence of activation and inactivation of α_1Eβ3 as compared with the α_1Cβ2a channel. The effects of α₂δ-3 on the electrophysiological properties of α_1C coexpressed with β2a was found to be similar to those of α₂δ-1. In contrast, coexpression of α₂δ-3 with α_1E and β3 produced more pronounced differences in the current characteristics associated with channel gating than the coexpression of α₂δ-1.

The mechanism whereby α₂δ modulates the conductances of α₁ is not clearly understood. The increase in density of current and DHP-binding sites can be explained by an improved targeting of expressed α₁ subunit to the cell membrane and maturation of the channel complex (Shistik et al., 1995), which leads to an increased amount of charge moved during channel activation (Bangalore et al., 1996; Qin et al., 1998). It was suggested that the increase in current requires the presence of an intact α₂protein, whereas the shift of voltage-dependent activation and steady state inactivation as well as the acceleration of the inactivation kinetics are caused by the transmembrane δ protein (Felix et al., 1997). However, based on the amino acid similarity of the three subunit forms, it seems more likely that α₂ harbors the relevant residues responsible for the observed effects on α₁ and that the δ domain functions only as an membrane anchor for α₂. This interpretation is further strengthened by the sequences of the δ proteins, which are not conserved.

These results, together with the brain-specific expression of α₂δ-3, suggest that the α₂δ-3 subunit may have a distinct physiological role in neuronal tissue. The α₂δ protein has been implicated as the in vivo target for the antiepileptic drug gabapentin (Gee et al., 1996), which apparently inhibits calcium currents in isolated rat brain neurons (Stefani et al., 1998). It was proposed that gabapentin binds preferentially to the α₂δ-1 subunit. This is supported by evidence that the partial N-terminal amino acid sequence of the gabapentin-binding protein obtained from porcine brain membranes is identical with an aminoterminal peptide of α₂δ-1. This sequence is not present in classes 2 or 3. Further support for this postulate is that our mRNA in situ analysis of α₂δ-1 in brain showed the same distribution as that of gabapentin-binding sites (Taylor et al., 1998), which differs from that of α₂δ-3. For these reasons it is more likely that gabapentin targets the α₂δ-1 subunit. However, another study using a different purification procedure showed that there may be an additional gabapentin-binding protein in brain, which was detected with a polyclonal α₂ antibody (Brown et al., 1998) and which could be another α₂δ form. Further investigations need to be undertaken to elucidate the binding of gabapentin to the α₂δ subunit.