Abstract
Chromaffin cells express N-type calcium channels identified on the basis of their high sensitivity to block by ω-conotoxin GVIA (ω-CgTx GVIA). In contrast to neuronal N-type calcium currents that inactivate during long depolarizations and that require negative holding potentials to remove inactivation, many chromaffin cells exhibit N-type calcium channel currents that show little inactivation during maintained depolarizations and that exhibit no decrease in channel availability at depolarized holding potentials. N-type calcium channels are thought to be produced by combination of the pore-forming α1B subunit and accessory β and α2/δ subunits. To examine the molecular composition of the non-inactivating N-type calcium channel, we cloned the α1B and accessory β (β1b, β1c, β2a, β2b, and β3a) subunits found in bovine chromaffin cells. Expression of the subunits in either Xenopusoocytes or human embryonic kidney 293 cells produced high-threshold calcium currents that were blocked by ω-CgTx GVIA. Coexpression of bovine α1B with β1b, β1c, β2b, or β3aproduced currents that were holding potential dependent. In contrast, coexpression of bovine α1B with β2aproduced holding potential-independent calcium currents that closely mimicked native non-inactivating currents, suggesting that non-inactivating N-type channels consist of bovine α1B, α2/δ, and β2a.
- N-type calcium channel
- α1B subunit
- β subunits
- non-inactivating calcium current
- chromaffin cells
- voltage-dependent calcium channel
Adrenal chromaffin cells express L-, N-, and P/Q-type calcium channels (Hans et al., 1990; Albillos et al., 1993; Artalejo et al., 1994). In many chromaffin cells, N-type calcium channels do not inactivate during long depolarizations, nor do they exhibit decreased availability at depolarized holding potentials (Artalejo et al., 1992). Similar non-inactivating N-type channels have been described in presynaptic nerve terminals (Stanley and Goping, 1991). Interestingly, a significant fraction of N-type channels in chromaffin cells exhibits robust inactivation, raising the question as to why inactivation varies so much from cell to cell.
Voltage-dependent calcium channels are composed of at least three different subunits designated α1, α2/δ, and β and possibly a γ subunit (Dunlap et al., 1995; Jones, 1998; Letts et al., 1998). The pore-forming α1 subunit alone encodes a voltage-dependent calcium channel with kinetic properties different from those of native channels (Lacerda et al., 1991; Varadi et al., 1991; Stea et al., 1993). When β and α2/δ subunits are coexpressed with a variety of cloned α1 subunits, current amplitude is greatly augmented, and the kinetics of activation and inactivation more closely resembles the kinetics of native channels (Birnbaumer et al., 1998). Ten different calcium channel α1subunits have been cloned [α1A–α1I and α1S (Jones, 1998)]. α1B subunits are required to make N-type calcium channels. Four β-subunit genes have been identified [β1–4 (Birnbaumer et al., 1998)], each of which has different effects on calcium channel inactivation. Inactivation rates of channels produced by α1Aand α1E are slowest with β2a, fastest with β3a, and intermediate with β1b and β4 (Sather et al., 1993; Olcese et al., 1994;Zhang et al., 1994; DeWaard and Campbell, 1995). Inactivation of α1B currents was greatest with β3a and least with β2a(Patil et al., 1998). Thus expression of different accessory subunits provides a potentially important regulatory mechanism for channel inactivation.
Certain α1-subunit regions appear to influence channel inactivation. When the IS6 and I–II loop of α1A or α1B were substituted for the analogous regions in α1E, the chimeric channels had inactivation rates much closer to those of α1A and α1B than to those of α1E (Zhang et al., 1994; Page et al., 1997). It was reported recently that the splice variant of α1A with a valine inserted in the I–II loop inactivated more slowly and less completely than did the variant lacking this valine (Bourinet et al., 1999).
Our study was performed to determine whether the inactivation properties of chromaffin cell N-type calcium channels result from a unique α1B gene or whether specific accessory subunit combinations were required to form these channels. We cloned the α1B subunit and five β subunits from chromaffin cells and expressed the clones in Xenopus oocytes and human embryonic kidney (HEK) 293 cells. Non-inactivating currents that mimicked those found in many chromaffin cells were observed when the bovine α1B clone was coexpressed with β2a and α2/δ. Channels made by coexpressing the α1B clone with either β1b, β1cβ2b, or β3a (and α2/δ) showed clear inactivation and resembled the inactivating N-type calcium currents observed in some chromaffin cells.
MATERIALS AND METHODS
Amplification of bovine chromaffin cell α1BcDNA. Fragments of bovine chromaffin cell α1B cDNA were amplified from total chromaffin cell RNA by reverse transcriptase-PCR (RT-PCR) using the Superscript Preamplification System (Life Technologies, Gaithersburg, MD) and degenerate primers designed from amino acid sequences conserved in the N and C terminals of cloned human, rabbit, rat, and mouse α1B subunits. A 563 bp cDNA corresponding to amino acids 65–252 of the N terminal of bovine α1B and a 683 bp cDNA corresponding to amino acids 1751–1969 of the C terminal of α1B were obtained.
Cloning of α1B from a bovine chromaffin cell library. The bovine α1B cDNAs obtained by PCR were used to screen 0.9 × 106plaques from a random- and polydT-primed bovine chromaffin cell library in λZapII (a gift from Dr. H. Pollard, National Institutes of Health). The plaques were transferred to Hybond N+ membranes and hybridized with 32P-labeled α1B cDNA using standard methods (Sambrook et al., 1989). Positive plaques were picked from the library plates and purified by two successive rounds of plating and hybridization. The α1B cDNA clones were excised in vivofrom λZapII using the ExAssist helper phage (Stratagene, La Jolla, CA). The ends of each cDNA insert were manually sequenced using the T7 Sequenase version 2.0 plasmid sequencing kit. The largest cDNAs from each screening were used to rescreen the library until clones corresponding to the entire bovine α1B cDNA were obtained. Ten screenings of the library yielded 44 independent α1B clones ranging in size from 150 to 3600 bp.
Construction of a full-length α1B cDNA. Eight clones that spanned the entire coding region of α1B were used to construct a full-length α1B cDNA in the mammalian expression vector pcDNA3.1(+) (Invitrogen, Carlsbad, CA). Fragments were cut from each clone using selected restriction endonucleases and subcloned into recipient plasmids using standard molecular biology techniques. The full-length bovine α1B cDNA was sequenced in its entirety by the University of Chicago Cancer Research Center Sequencing Facility using an ABI 377 automated fluorescent sequencer. Sequencing revealed a 2 bp deletion at the splice site for the +SFMG splice variant described by Lin et al. (1997). This deletion was repaired by subcloning the XbaI–Bsu36I fragment (bp 3245–3852) from a different library clone into the α1B construct in pcDNA3.1(+). The sequence of bovine α1B has been deposited in GenBank with the accession number AF173882.
Cloning of β1b, β1c,β2a, β2b, and β3a from a bovine chromaffin cell library. A 2690 bp fragment of human β1b was labeled with 32P and used to screen the bovine chromaffin cell library by standard hybridization techniques (Sambrook et al., 1989). Positive plaques were picked from the library plates and purified by two more rounds of plating beforein vivo excision of the phagemid pBSII containing each cDNA clone. This screening of the library yielded full-length clones for bovine β1b, β1c, and β3a but no β2a or β2b clones. Therefore degenerate PCR primers were designed to the β2 C-terminal amino acid sequences EAYWKAT and EWNRDVYI and used in an RT-PCR protocol to generate a 600 bp cDNA fragment of bovine β2. This fragment was labeled with 32P and used to screen the library as described above. This screening yielded clones containing all but the first 48 bp of the β2a-coding region and the first 36 bp of the coding region of β2b. The missing 5′ end of β2a was cloned by RT-PCR from chromaffin cell RNA and ligated to the longest library clone. The missing region of the bovine β2b was supplied by ligating the appropriate fragment of rabbit β2b to the bovine β2b. (The N-terminal amino acids 13–105 are identical between the bovine and rabbit clones. It appears likely that the first 12 amino acids may be identical as well.) All β-subunit clones were sequenced and subcloned into pcDNA3.1(+) for expression studies. The sequences of the five bovine β subunits have been deposited in GenBank with the accession numbersAF174415–AF174419.
Other cloned calcium channel subunits. Other cDNA clones used in this study were generous gifts from the following investigators: human α2/δ (GenBank accession number M21948) and β1b (M92303) from Dr. R. J. Miller (The University of Chicago), rat β2a (M80545) from Dr. E. Perez-Reyes (Loyola University), and rabbit β2b (X64298) from Dr. R. W. Tsien (Stanford University).
Xenopus oocyte expression and electrophysiology.Calcium channel proteins were expressed in Xenopus laevisoocytes after injection of in vitro-transcribed mRNA. Full-length cDNAs of each of the cloned subunits were used to generate mRNA from the T7 RNA polymerase promoter using an in vitrotranscription kit (mMessage Machine; Ambion, Austin, TX). Oocytes were harvested from mature Xenopus laevis female frogs and separated from follicle cells with 2 mg/ml collagenase type IA (Sigma, St. Louis, MO). Oocytes were injected with 25 ng of both bovine α1B and human α2/δ RNA and 10 ng of various β RNAs in a total of 50 nl of DEPC-treated H2O with a Drummond automatic microinjector. Three days after injection, oocytes were voltage-clamped at various potentials with two glass electrodes (filled with 3m KCl and having a resistance of 0.5–2 MΩ) using an Axoclamp 2A (Axon Instruments, Foster City, CA). Oocytes were initially superfused with 96 mm NaCl, 2 mm KCl, 1 mmMgCl2, and 5 mm HEPES, pH 7.6, and N-type calcium channel currents were measured in a Ba2+ recording solution designed to minimize the oocyte's endogenous Ca2+-dependent Cl− current [40 mmBa(OH)2, 25 mmtetraethylammonium (TEA)-OH, 25 mm NaOH, 2 mm CsOH, 5 mm HEPES, and 1 mm niflumic acid, pH adjusted to 7.5 with methanesulfonic acid]. Membrane currents were recorded on a computer using pCLAMP software (Axon Instruments). Leak currents were subtracted on-line with a P/4 protocol.
Cell culture and transfection. HEK 293 (American Type Culture Collection, Rockville, MD) cells were used for transient transfection studies and were grown in MEM and 10% calf serum supplemented with 2 mm glutamine, 100 U/ml penicillin, and 100 μg/ml streptomycin. Mammalian expression plasmids, containing the complete cDNAs for each of the calcium channel subunits, and the marker CD8 were purified on Qiagen (Hilden, Germany) columns and used to transfect HEK 293 cells. Cells were transiently transfected with cDNAs encoding the bovine α1B, human α2/δ, various β subunits, and CD8 as a reporter gene in a ratio of 15:12:5:3 using Lipofectin or Lipofectamine (both from Life Technologies). The human α2/δ and various β cDNAs were coexpressed with α1Bbecause they have been reported to increase the expression levels of calcium channels and to normalize current kinetics (Lacerda et al., 1991; Stea et al., 1993). Immediately after the transfection, cells were replated onto polylysine-coated coverslips. Calcium currents were recorded 2–3 d after transfection. Transfected cells were detected visually by binding of anti-CD8 beads (Dynal, Great Neck, NY).
Bovine chromaffin cells were prepared and cultured as described previously (Artalejo et al., 1992). Briefly, bovine adrenal glands were digested with collagenase, and cells were purified by density gradient centrifugation. Cells were resuspended in bovine chromaffin cell media and plated onto collagen-coated glass coverslips at a density of 0.15 × 106cells/cm2. Cells were maintained at 37°C in an atmosphere of 92.5% air and 7.5% CO2 and 90% humidity.
Electrophysiology on HEK 293 and chromaffin cells. Cells were voltage-clamped in the whole-cell configuration of the patch-clamp technique (Hamill et al., 1981) with a List model L/M-EPC7 patch clamp. Gigaohm seals were obtained in an extracellular Tyrode's solution (130 mm NaCl, 20 mm glucose, 10 mmHEPES, 1 mm MgCl2, 2 mmKCl, and 5 or 10 mm CaCl2, pH adjusted to 7.3 with NaOH). Ionic currents were measured in an extracellular TEA-Ba2+ solution (140 mm TEA-Cl, 10 mm glucose, 10 mm HEPES, and 5 or 10 mmBaCl2, pH adjusted to 7.3 with TEA-OH). For chromaffin cell recordings, 1 μm tetrodotoxin and 1 μm nisoldipine were added to the TEA-Ba2+ solution. Electrodes were pulled from glass capillary tubes (Drummond, Broomall, PA), coated with Sylgard (Dow Corning, Midland, MI), and fire-polished to a final resistance of ∼2.0 MΩ when filled with a CsCl-based internal solution (110 mm CsCl, 10 mm EGTA, 40 mm HEPES, 5 mm MgCl2, 0.3 mm GTP, and 2 mm ATP, pH adjusted to 7.3 with CsOH, or 110 mm CsCl, 10 mm EGTA, 20 mm HEPES, 4 mm MgCl2, 0.3 mm GTP, 6 mm ATP, and 14 mmcreatine phosphate, pH adjusted to 7.3 with CsOH). ω-Conotoxin GVIA (Alomone Labs, Jerusalem, Israel) was diluted in extracellular recording solution and added to the bath at a final concentration of 1 μm.
Ionic currents were activated by step depolarizations of 20–200 msec duration from various holding potentials. For chromaffin cell recording, the cells were continuously perfused with recording solution at a rate of 2–3 ml/min by gravity flow. Leak currents were generated by averaging hyperpolarizing steps. All data presented here are leak and capacitance subtracted. Series resistance was partially compensated (>60%) using the series resistance compensation circuit of the patch clamp.
Data analysis. For I–V curve calculations, the peak current from each cell was recorded, and groups were pooled to calculate the average and SEM. For steady-state inactivation curves, currents from each cell at each holding potential were normalized to the peak current at the most hyperpolarized holding potential. The normalized data were averaged across cells and fit to single Boltzmann functions.
RESULTS
We have described previously a biophysically distinct N-type calcium channel found in bovine chromaffin cells (Artalejo et al., 1992). This “atypical” N-type calcium channel does not exhibit voltage-dependent inactivation while retaining other defining characteristics of N-type channels such as ω-conotoxin GVIA (ω-CgTx GVIA) sensitivity. In addition, a large subset of chromaffin cells exhibits N-type calcium channel currents that show robust inactivation and thus resemble neuronal N-type channels. Examples of both varieties of chromaffin cell N-type calcium channels are shown in Figure1. The method used to isolate N-type calcium currents from whole-cell calcium currents is illustrated in Figure 1A. The whole-cell calcium currents in Figure1A labeled Total consist only of N- and P/Q-type currents because the specific L-type calcium channel blocker nisoldipine was included in the extracellular medium. Whole-cell currents were elicited by test depolarizations to +10 mV from holding potentials in the range of −90 to −50 mV. Note that currenttraces from five holding potentials are superimposed in each part of Figure 1A, indicating that the calcium currents in this cell are not sensitive to changes in the holding potential. Addition of 1 μm ω-CgTx GVIA to the extracellular medium blocked approximately two-thirds of the current at each of the holding potentials. Five superimposed traces illustrating the P/Q-type calcium currents remaining after ω-CgTx GVIA application are plotted in Figure 1A,middle (labeled ω-CTx GVIA insensitive). Subtraction of the ω-CgTx GVIA-insensitive current from the total current yields the ω-CgTx GVIA-sensitive N-type calcium current in isolation (Fig. 1A, labeled N-type). The currents from this cell were stable for >30 min and exhibited little rundown. Although many chromaffin cells exhibit calcium currents that are holding potential insensitive, other cells exhibit strong inactivation of currents. Figure 1B shows N-type current from a different cell, isolated in a similar manner, that showed a strong holding potential dependence; over one-half of the current was inactivated by a change in holding potential in the range of −90 to −60 mV. Figure 1C shows a plot of peak N-type current elicited at each holding potential, using the data from the non-inactivating cell shown in Figure 1A, right. Each current has been normalized to the current observed at the −90 mV holding potential. The data demonstrate that the N-type channels in this cell are not sensitive to changes in holding potential. Channel availability was not tested at very depolarized potentials (more than −40 mV) because these potentials activate Ca2+ currents in chromaffin cells. N-type Ca2+ current inactivation appears to involve a component of Ca2+-dependent inactivation (Cox and Dunlap, 1994), which would complicate our studies of voltage-dependent inactivation.
Some chromaffin cell N-type calcium channel currents observed in isolation are insensitive to changes in the holding potential. A, The strategy used to isolate N-type calcium channel currents in chromaffin cells is shown. Leak-subtracted calcium current traces were obtained by depolarizing a chromaffin cell to +10 mV for 24 msec from holding potentials in the range of −90 to −50 mV. Each holding potential was maintained for 60 sec before eliciting the test depolarization. Left,Total represents the total current obtained in the absence of any calcium channel antagonist. Middle,ω-CTx GVIA-insensitive shows calcium currents elicited after application of 1 μm ω-CgTx GVIA.Right, N-type shows currents after subtraction of ω-CgTx GVIA-insensitive currents from the total currents. For each condition there are five superimposed current records, one from each holding potential in the range of −90 to −50 mV (10 mV steps). B, Isolated N-type current from a different cell that exhibited strong holding potential dependence is shown. More than one-half of the current was inactivated when the holding potential was changed from −90 to −60 mV. C, Peak N-type current amplitude as a function of holding potential (HP) is shown. The peak current at eachHP was normalized by dividing by the current observed at an HP of −90 mV. Inset, The voltage-clamp protocol.
We have attempted to determine the molecular basis of the inactivating or non-inactivating N-type calcium channels found in bovine chromaffin cells by cloning the α1B subunit and a variety of accessory β subunits from a bovine chromaffin cell cDNA library and then characterizing the calcium currents resulting from the expression of the cloned subunits. The cDNA library was screened as described in Materials and Methods, and a full-length bovine α1B cDNA was assembled from eight overlapping library clones. The full-length clone was sequenced, and the deduced amino acid sequence for the bovine chromaffin cell α1B subunit was aligned with reported sequences for human α1B (Williams et al., 1992a), rabbit α1B (Fujita et al., 1993), rat α1B (Dubel et al., 1992), and mouse α1B (Coppola et al., 1994), and a consensus sequence was generated (Fig.2). The bovine α1B was most similar to the human α1B with 93% of the deduced amino acid residues being identical. Identities with the α1B subunits cloned from mouse, rat, and rabbit ranged from 89 to 91%. The N terminal and the four putative transmembrane domains of α1B were most highly conserved with 98–99% of the deduced amino acids being identical to the consensus. There was significantly more variation among species in the C terminal and the putative intracellular loop between transmembrane domains II and III with 89–93% of the C-terminal residues for each species being identical to the consensus sequence and 83–84% of the II–III loop residues agreeing with the consensus.
Alignment of the deduced amino acid sequence of bovine chromaffin cell α1B with those from human, rabbit, rat, and mouse α1B. The consensus sequence (Con) calculated from this alignment is shown on thetopline, and only differences from the consensus are shown for the other sequences. The putative transmembrane domains and the α1 interaction domain (AID) are underlined andlabeled in bold above the consensus sequence. The splice variants identified here and byLin et al. (1997) are indicated by *. The GenBank accession numbers and the sources of the sequences are as follows: Bov, bovine chromaffin cell (AF173882) (this study); Hum, human neuroblastoma (M94172) (Williams et al., 1992a); Mus,mouse neuroblastoma (U04999) (Coppola et al., 1994);Rab, rabbit brain (D14157) (Fujita et al., 1993); and Rat, rat brain (M92905) (Dubel et al., 1992). (Figure 2 continues.)
We also compared the bovine α1B sequence with the reported splice variants for cloned α1B. All of the three library clones containing the C terminal of bovine α1B were similar to the longer splice form described by Williams et al. (1992a) for human α1B. None of the four clones covering the II–III loop region of bovine α1B contained the 22 amino acid insert found in mouse α1B(Coppola et al., 1994), but interestingly a different splice variant was observed at this same splice site. Two of the four library clones and the full-length bovine α1B construct contained an arginine residue at this splice site (Fig. 2, R757), and the other two library clones did not. Three sites for alternative splicing of rat α1B have been described as follows: ±A(415), ±SFMG (1242–1245), and ±ET (1574–1575) (Lin et al., 1997). None of three library clones coded for A, and none of three library clones coded for SFMG, but we found one library clone that was +ET and one that was −ET. The full-length expressed bovine α1B was −A, +R, −SFMG, and −ET. The IS6 region, which has been suggested to be important in N-type channel inactivation (Zhang et al., 1994), contained a single amino acid substitution, leucine in bovine α1B rather than isoleucine.
Five accessory β subunits (β1b, β1c, β2a, β2b, and β3a) were cloned from the chromaffin cell cDNA library. The bovine β1b, β1c, and β3a subunits were recovered from the library as full-length clones and were sequenced in their entirety. The predicted amino acid sequences of these three β subunits were 97–98% identical with those of human clones (Powers et al., 1992; Williams et al., 1992b; Collin et al., 1994). The initial screen of the chromaffin cell cDNA library yielded several clones corresponding to the region common to β2a and β2b, but no clones contained the complete and unique N terminals of β2a or β2b. Therefore the N terminal of β2a was cloned from chromaffin cell RNA by RT-PCR and ligated to a library clone to produce full-length β2a. Overall the bovine β2a subunit was 94% identical to human β2a, but almost all of the differences were confined to the common C terminal shared by β2aand β2b. This region was only 82% identical to human β2a. The full-length β2b was obtained by ligating the incomplete β2b library clone with a cDNA fragment from rabbit β2b that coded for the first 12 missing amino acids. No human β2b clone has yet been reported.
The cloned α1B and β subunits were expressed both in HEK 293 cells and in Xenopus oocytes (Fig.3). Although the experiments could be done more efficiently in oocytes where virtually every injected cell exhibited large currents, HEK 293 cells were also used in certain experiments to exclude possible effects of the endogenous β3xo subunit in oocytes (Tareilus et al., 1997). HEK 293 cells transiently transfected with mammalian expression plasmids containing bovine α1B and human α2/δ and β1b subunits expressed a high-threshold calcium current (Fig. 3A) that was blocked by 1 μm ω-CgTx GVIA (Fig.3B). Sham-transfected HEK 293 cells did not have any observable calcium currents (data not shown). Oocytes were injected with in vitro-transcribed mRNAs for each subunit, and current recordings were done 2–3 d later. Oocytes expressing all three calcium channel subunits exhibited a large inward Ba2+ current that was not present in uninjected oocytes. Small outward currents (<50 nA) were occasionally seen in oocytes injected with only the α2/δ and α1B subunits (data not shown). Figure3C shows data from an oocyte injected with mRNA for bovine α1B and human α2/δ and β1b. The high-threshold currents observed were blocked by ω-CgTx GVIA. Thus in both oocytes and HEK 293 cells the bovine α1B subunit when coexpressed with a β and an α2/δ subunit appears to make N-type calcium channels.
Coexpression of bovine α1B, human β1b, and α2/δ in HEK 293 cells and Xenopus oocytes. A, Peak currents in HEK 293 cells plotted as a function of voltage. Transfected cells were depolarized for 200 msec to a variety of test potentials from an HP of −80 mV; peak current was measured and then plotted as a function of test potential. B, Peak currents plotted as a function of time. The HEK 293 cell was depolarized to +10 mV every 15 sec from an HP of −80 mV. ω-CgTx GVIA was added to the bath as indicated by the horizontalbar. Toxin application produced >95% inhibition of the calcium current.C, Peak currents in oocytes expressing α1B, β1b, and α2/δ plotted as a function of voltage. Oocytes were depolarized for 200 msec to a variety of test potentials from an HP of −120 mV. The experiment was repeated after the addition of 1 μm ω-CgTx GVIA.
We next investigated the inactivation properties of the cloned α1B subunit expressed with α2/δ and a variety of β subunits. Figure4A shows data from three different oocytes expressing bovine α1Band human α2/δ in addition to the β subunit indicated in the figure. The families of current records were obtained by depolarizing cells for 200 msec to +10 mV from a variety of holding potentials in the range of −90 to −30 mV. Each holding potential was maintained for 60 sec before the test depolarization. Expression of either β1b, β1c, or β3a produced currents that were strongly dependent on holding potential. Figure 4B plots current–voltage relationships for each of the cells shown in Figure4A. The cells were depolarized to a variety of test potentials from an HP of −100 mV; peak current was measured and then plotted as a function of test potential. Figure 4C graphs voltage-dependent inactivation curves made by averaging data from groups of cells expressing different β subunits. The curves were constructed by normalizing the currents; the currents observed at each holding potential were divided by the current recorded at −90 mV. Note that the currents generated by the channels containing any of the three β subunits used in this experiment produced strong inactivation.
Coexpression of bovine α1B with bovine β1b, β1c, or β3a and α2/δ inXenopus oocytes produced inactivating calcium currents.A, Families of current traces obtained by depolarizing oocytes for 200 msec to +10 mV from holding potentials in the range of −90 to −30 mV. Each holding potential was maintained for 60 sec before the test depolarization. B, Current–voltage relationship for each of the cells inA. The oocytes were depolarized for 200 msec to a variety of test potentials from an HP of −100 mV. C, Mean normalized peak current as a function of holding potential for groups of cells expressing the β subunit indicated. Thenumber of cells in each group is indicated inparentheses.
In contrast, expression of rat β2a with α1B produced currents that were virtually holding potential independent. Figure5A shows a family of current records obtained by depolarizing an oocyte for 200 msec to +10 mV from a variety of holding potentials in the range of −90 to −40 mV. Each holding potential was maintained for 60 sec before the test depolarization. Figure 5B plots the current–voltage relationship for this cell. The cell was depolarized to a variety of test potentials from an HP of −90 mV; peak current was measured and then plotted as a function of test potential. Figure 5Cplots the voltage-dependent inactivation curve obtained by averaging data from five oocytes expressing bovine α1Band rat β2a and illustrates the holding potential independence of N-type calcium channels constructed with β2a subunits. Similar results were obtained when the human α2/δ subunit was expressed along with α1B and rat β2a (data not shown).
Coexpression of bovine α1B with rat β2a in Xenopus oocytes produced non-inactivating calcium currents. A, Family of currenttraces obtained by depolarizing an oocyte for 200 msec to +10 mV from holding potentials in the range of −90 to −40 mV. Each holding potential was maintained for 60 sec before the test depolarization. B, Mean normalized peak current as a function of holding potential for eight cells expressing the bovine α1B and rat β2a subunits. C, Current–voltage relationship for the cell in A. The oocyte was depolarized for 200 msec to a variety of test potentials from an HP of −100 mV.
Bovine β2a differs significantly from rat β2a primarily in the C terminal where only 81% of the predicted amino acids are identical. Coexpression of bovine β2a along with α1B and α2/δ produced non-inactivating N-type calcium channel current (Fig.6A). In contrast, coexpression of bovine β2b along with α1B and α2/δ produced a robustly inactivating N-type calcium channel current (Fig.6B). These results suggest that the C terminal is not likely to contribute to the unique properties of β2a because similar results were obtained with both rat and bovine β2a. The N terminal of β2a is, however, a good candidate for mediating the non-inactivating calcium currents seen with β2a, because β2adiffers from β2b only in its short N terminal.
Coexpression of bovine α1B and human α2/δ with bovine β2a but not β2b produced a non-inactivating calcium channel current in HEK 293 cells. A, Family of currenttraces obtained by depolarizing an HEK 293 cell expressing bovine α1B and β2a along with human α2/δ for 140 msec to +10 mV from holding potentials in the range of −100 to −60 mV. Each holding potential was maintained for 60 sec before the test depolarization. B, Family of current traces obtained by depolarizing an HEK 293 cell expressing bovine α1B and β2balong with human α2/δ. The voltage-clamp protocol is described in A.
DISCUSSION
In an attempt to identify the molecular basis for the non-inactivating N-type calcium channels observed in bovine chromaffin cells (Fig. 1) (Artalejo et al., 1992), we cloned the α1B, β1b, β1c, β2a,β2b, and β3a calcium channel subunits from a chromaffin cell library and expressed these clones in both HEK 293 cells and Xenopus oocytes. In both systems expression of the cloned α1B and β2a along with human α2/δ yielded N-type calcium currents that resembled the non-inactivating N-type currents in bovine chromaffin cells. However, expression of the α1B and α2/δ with β1b, β1c, β2b, or β3a yielded calcium currents that showed distinct inactivation during prolonged depolarizations as well as decreased availability from depolarized holding potentials. These results suggest that the β2a subunit is a key determinant of the non-inactivating properties of the N-type calcium channel in chromaffin cells. Previous studies with cloned α1A, α1B, and α1E have shown that inactivation rates of these calcium channels depended on which β subunit was coexpressed; inactivation was slowest with β2, fastest with β3, and intermediate with β1 and β4 (Ellinor et al., 1993; Sather et al., 1993; Olcese et al., 1994; Zhang et al., 1994; DeWaard and Campbell, 1995; Parent et al., 1997; Patil et al., 1998). In addition coexpression of β2a has been found to shift the voltage-dependent inactivation curves obtained with α1E to more depolarized potentials and to result in incomplete steady-state inactivation (Sather et al., 1993;Olcese et al., 1994; Parent et al., 1997). However, steady-state voltage-dependent inactivation of α1A and α1E was never totally abolished even with the β2a subunit as was seen in this study that used bovine α1B and either the bovine or rat β2a. A recent study using single-cell RT-PCR linked β2a expression to a slowly inactivating Q-type current in neostriatal neurons, whereas β1b expression was linked to fast-inactivating Q-type currents in cortical neurons (Mermelstein et al., 1999).
Several previous studies have suggested that the N terminal of the β subunit is an important region regulating channel inactivation (Olcese et al., 1994; Cens et al., 1999). The different inactivation rates that we observed with bovine β2a and β2b (Fig. 6) support this hypothesis because β2a and β2b differ from each other only in their short N terminal. Palmitoylation of the two cysteines in the N terminal of β2a has been shown to be essential for certain modulatory effects of β2a on α1C and α1E (Chien et al., 1996; Qin et al., 1998). The C terminals of bovine β2a and β2b were markedly more different from this region of the human, rabbit, rat, and mouse β2aor β2b than these were from each other. To determine whether these C-terminal differences were functionally important, we also expressed the rat β2a (in which 19% of the amino acids in the C terminal differ from those in bovine β2a) with the bovine α1B and found that this combination of subunits also yielded non-inactivating N-type calcium channels. Thus it appears that the differences in the C terminal of β2aare not functionally important in terms of regulating calcium channel inactivation.
Despite the apparent importance of the β2asubunit in regulating the rate of inactivation of N-type calcium channels, a role for the α1B subunit itself cannot be excluded. Both the IS6 transmembrane domain and the adjacent I–II cytoplasmic loop have been implicated previously in the rate of inactivation exhibited by cloned α1A and α1B subunits (Zhang et al., 1994; Page et al., 1997). Bovine α1B differs from the consensus in only one position within the IS6 region and in two positions in the I–II loop region, but one of these is the absence of alanine 415. The presence or absence of alanine 415 is a known splice variant in α1B (Lin et al., 1997) that corresponds to the recently described splice variant ±valine 421 in α1A (Bourinet et al., 1999). The presence of valine 421 confers much slower inactivation on α1A. None of the three library clones corresponding to the alanine 415 region that were isolated from the bovine chromaffin cell library contained alanine 415. Another splice variant of α1B [+A, −SFMG, and +ET (Lin et al., 1997)] was originally suggested to have slower inactivation kinetics than +A, +SFMG, and −ET, but more recent work has suggested that the differences between these splice variants are primarily in the rates of activation and depend primarily on +ET (Lin et al., 1999). Our expressed α1B was −A, −SFMG, and −ET and thus did not correspond exactly to either of these forms. However, two of the four library clones covering the ±ET region did have +ET. Work is in progress to incorporate this splice variant into full-length α1B. In discussing the various splice variants of α1B, it should also be noted that we do not know the exact splice variant composition of the native α1B in chromaffin cells. The clones we used to construct the full-length α1B may not be the ones normally spliced together, a limitation that is true of all full-length α1 subunits that have been constructed from more than one cDNA library clone. RT-PCR of smaller portions of α1B could easily determine which of the individual known splice variants of α1B(e.g., ±A415 or ±ET1575) are expressed in chromaffin cells, but it still would not answer the question of which combinations of splice variants make up a single mRNA.
In addition to modifications to inactivation made by different β subunits or by changes to the α1 subunits described above, inactivation of N-type Ca2+ channels can be altered by interactions with synaptic vesicle docking and fusion machinery. Coexpression of syntaxin 1A with N-type channels in Xenopusoocytes dramatically decreased channel availability because of the stabilization of channel inactivation (Bezprozvanny et al., 1995). Interestingly, we observed no such changes in inactivation when we coexpressed syntaxin 1A with α1B, β2a, and α2/δ (our unpublished observations), which suggests that syntaxin 1A alters inactivation exclusively in N-type channels that exhibit inactivation.
N-type calcium channels, which are widely distributed throughout the nervous system, have been unambiguously linked to a variety of important physiological processes including synaptic transmission (Takahashi and Momiyama, 1993; Turner et al., 1993; Wheeler et al., 1994). In secretory cells, like chromaffin cells, activation of N-type calcium channels by themselves can trigger catecholamine secretion (Artalejo et al., 1994). It seems likely that non-inactivating N-type calcium channels would operate more effectively during periods of intense stimulation, which would lessen the availability of inactivating channels thereby diminishing Ca2+ influx. In addition to expressing inactivating or non-inactivating N-type calcium channels, chromaffin cells also expressed inactivating or non-inactivating P/Q-type calcium channels (our unpublished observation), which should also have dramatic effects on Ca2+ influx. Thus, the changes in channel inactivation properties because of alterations in β-subunit channel composition that are described in this study may have important functional consequences for a wide variety of Ca2+-dependent processes including catecholamine secretion.
Footnotes
This work was supported by National Institutes of Health (NIH) grants to A.P.F. and by an NIH training grant to J.H.H. We would like to thank Drs. H. Pollard (NIH) for the bovine chromaffin cell cDNA library, R. W. Tsien (Stanford University) for the rabbit β2bcDNA clone, E. Perez-Reyes (Loyola University) for the rat β2a, and R. J. Miller (The University of Chicago) for the human β1b and α2/δ cDNA clones.
Correspondence should be addressed to Dr. Anne L. Cahill, The Department of Neurobiology, Pharmacology, and Physiology, The University of Chicago, 947 East 58th Street, Chicago, Illinois 60637. E-mail: acahill{at}Drugs.bsd.uchicago.edu.
