Abstract
Voltage-gated calcium channel α1 subunits consist of four domains (I–IV), each with six transmembrane segments. A number of truncated isoforms have been identified to occur as a result of alternative splicing or mutation. We have examined the functional consequences for expression of full-length Cav2.2 (α1B) of its coexpression with truncated constructs of Cav2.2. Domains I-II or domains III-IV, when expressed individually, together with the accessory subunits β1b and α2δ-1, did not form functional channels. When they were coexpressed, low-density whole-cell currents and functional channels with properties similar to wild-type channels were observed. However, when domain I-II, domain III-IV, or domain I alone were coexpressed with full-length Cav2.2, they markedly suppressed its functional expression, although at the single channel level, when channels were recorded, there were no differences in their biophysical properties. Furthermore, when it was coexpressed with either domain I-II or domain I, the fluorescence of green fluorescent protein (GFP)–Cav2.2 and expression of Cav2.2 protein was almost abolished. Suppression does not involve sequestration of the Cavβ subunit, because loss of GFP–Cav2.2 expression also occurred in the absence of β subunit, and the effect of domain I-II or domain I could not be mimicked by the cytoplasmic I-II loop of Cav2.2. It requires transmembrane segments, because the isolated Cav2.2 N terminus did not have any effect. Our results indicate that the mechanism of suppression of Cav2.2 by truncated constructs containing domain I involves inhibition of channel synthesis, which may represent a role of endogenously expressed truncated Cav isoforms.
Voltage-gated calcium channels subserve a number of functions, including neurotransmitter release, regulation of gene transcription, and muscle contraction (Catterall, 2000). They are heteromeric complexes consisting minimally of three subunits, namely the pore-forming α1 subunit and the accessory β and α2-δ subunits. The α1 subunit is the structural and functional core of the channel and consists of four homologous domains (Dom I–IV), linked by intracellular loops and with intracellular N and C termini. Each domain contains six transmembrane-spanning segments (S1–S6). To date, 10 α1 subunits have been cloned and expressed (Birnbaumer et al., 1994; Perez-Reyes and Schneider, 1994; Catterall, 2000), termed α1A-α1I and α1S, now renamed Cav1–3 (Ertel et al., 2000).
Mutations in calcium channel α1 subunits can contribute to a number of pathological states, and some of these mutations involve the introduction of a premature stop codon. For example, in episodic ataxia type-2 (EA-2), a number of mutations in the Cav2.1 subunit predict truncated forms of this channel (Ophoff et al., 1996; Denier et al., 1999). Most identified mutations in EA-2 introduce stop codons at the end of domain II–S6, in domain III–S1, and in the S1 segments of domains III and IV. The truncation at S1 of domain III is of particular interest because a 95 kDa protein has been identified that normally copurifies with Cav2.1 (Scott et al., 1998). This protein appears to contain domains I, II, and part of the II–III loop of Cav2.1. Thus, it is very similar to the predicted truncation in EA-2. To date, no naturally occurring two-domain splice variants of Cav2.1 that would give rise to such a two-domain protein product have been found. Recently, novel splice variants of Cav1.2 have been identified, generated by alternative splicing in the II-III loop, which predict two truncated forms of Cav1.2, consisting of domains I and II (Wielowieyski et al., 2001). In addition, a splice variant of Cav2.2 has been identified in humans and rodents that would introduce a stop codon near the end of the II-III loop (Mittman, Agnew, 2000). The expression of this splice variant would give rise to a protein consisting of domains I and II of Cav2.2. Little is known about the tissue-specificity or developmental regulation of expression of such splice variants, except that an isoform consisting of the first two domains of Cav1.1 is the main transcript in newborn muscle, whereas the four-domain isoform is predominant in adult muscle (Malouf et al., 1992). Furthermore, it has recently been found that during development of a tunicate tadpole, a truncated calcium channel with homology to Cav1.1, consisting of domains III and IV with part of domain II was expressed from a maternal transcript (Okagaki et al., 2001).
In the present study, we have examined the expression, physiological function, and effects on channel protein levels of truncations of the N-type calcium channel Cav2.2. Our results indicate that constructs containing transmembrane domain I suppress the synthesis of full-length Cav2.2.
MATERIALS AND METHODS
Materials. The following cDNAs were used: rat β1b (Tomlinson et al., 1993), rabbit Cav2.2 (D14157), rat α2δ-1 (M86621), and green fluorescent protein (GFP) mut3b (Cormack et al., 1996).
Truncated Cav2.2 channel constructs. Constructs containing different domains of the rabbit Cav2.2 channel were made using the PCR with primers incorporating either start or stop codons and restriction enzyme sites. The following primers were used: N-term (forward): 5′-GCG ACT AGT ATG GTC CGC TTC GGG GAC −3′ and (reverse): 5′-GTA CTC GAG CTA AGG CCA CTC GGT GAT GCG-3′ (introduces a stop codon at the end of the N terminus); Dom I (reverse): 5′-TTA ACT AGT TTA CTG TGC CTT CAC CAT GCG-3′ (introduces a stop codon at the end of the I-II loop); Dom I–II (reverse): 5′-CTC GAC TAG TTA CAT GGT CAC AAT GTA GTG-3′ (introduces a stop codon at the end of the II-III loop); Dom III–IV (forward): 5′-TGG CCA CTA GT A TGG ACA ACC TTG CCA ATG-3′ (introduces a start codon at the beginning of the II-III loop).
In the GFP–Cav2.2, the stop codon of GFP was removed, and GFP was fused to the N terminus of Cav2.2 by PCR. The sequence for the forward primer was 5′-GAT GAA CTA TAC AAA ATG GTC CGC TTC GG-3′. The sequence in italics indicates the end of GFP (with the stop codon removed), and the underlined sequence indicates the beginning of Cav2.2. GFP was fused to the Dom I–II construct using the same primer. Enhanced yellow fluorescent protein (EYFP) (Clontech, Cowley, UK) was fused onto the N terminus of the Dom I construct using the primer, 5′-GAG CTG TAC AAG TCC GGA ATG GTC CGC TTC GGG-3′. The sequence in italics indicates the end of EYFP, and the underlined sequence indicates the beginning of Cav2.2. For the I–II loop construct, the following primers were used: 5′-GGAGAATTCGCTATGGAGCGCGAGAGAGTG-3′ (forward with an EcoRI site and a start codon) and 5′-CTGTGCTCTAGACATGCGCCGGATG-3′ (reverse with an incorporated XbaI site). The resulting fragment was digested with EcoRI and XbaI enzymes and ligated in frame with the myc- and His-tags of pcDNA3.1/myc-His(+)A (Invitrogen, Paisley, UK) vector. The validity of the construct was further confirmed by Western blot, probed with anti-His antibodies (Abs) (Santa Cruz Biotechnology, Santa Cruz, CA), after expression in COS-7 cells. The sequences of all constructs were verified by automated sequencing.
Yeast two-hybrid screening. Yeast two-hybrid studies were performed using the Matchmaker Gal4 system (Clontech). The N terminus of Cav2.2 was made by PCR using a forward primer against the vector and the reverse primer: 5′ GAT CTC GAG AGG CCA CTC GGT GAT GCG 3′. This gave a product with an NcoI site overlapping the 5′ ATG start codon and an XhoI site at the 3′ end. The digested PCR product was subcloned into theNcoI–XhoI sites of pACT2 and theNcoI-SalI sites of pAS2–1. The constructs containing the C terminus and the I-II loop were made using the following primers: C terminus: 5′ GTG ACC ATG GAC AAT TTT GAG TAC C 3′ and 5′ TAT CGA ATT CTA GCA CCG GCG GTC G 3′; I-II loop: 5′ GTA ACC ATG GCT AAG GAG CGC GAG AG 3′ and 5′ GTA GGA AAT CTG TGC CTT CAC CAT GC 3′. These PCR products, as well as the entire calcium channel β1b subunit, were subcloned into the NcoI–EcoRI sites of both pACT2 and pAS2–1. Competent yeast cells (Y190 strain) were cotransformed with both plasmids, and β-galactosidase colony-lift filter assays were performed according to the user manual (Clontech).
Cell culture and transfection. COS-7 cells were cultured as previously described (Campbell et al., 1995) and transfected using the Geneporter transfection reagent (Qbiogene, Harefield, UK). Cells were plated onto coverslips 2–3 hr before transfection. The cDNAs (all at 1 μg/ml) for Cav2.2 or truncated domain constructs (Dom I, I–II, III–IV, N terminus, or I–II loop), α2δ-1, β1b, and GFP (when used) were mixed in a ratio of 1.5 (or 3):2:1:0.2. When both Cav2.2 and truncated construct were both present, the ratios were 1.5:1.5:2:1:0.2. When particular subunits were not used, the volume was made up with water. The DNA mixture and Geneporter (6 μg and 30 μl, respectively) were each diluted in 500 μl of serum-free medium, mixed, and applied to the cells. After 3.5 hr, 1 ml of medium containing 20% serum was added to the cells, which were then incubated at 37°C for 3 d, followed by incubation at 27°C, where stated. Lactacystin (CN Biosciences, Beeston, UK) was stored at −20OC as a 3 mmstock solution in dimethylsulfoxide (DMSO) and, when used, was added to the transfected cells at 30 μm.
Immunocytochemistry and confocal microscopy. COS-7 cells were washed twice in Tris-buffered saline (TBS; 154 mm NaCl, 20 mm Tris, pH 7.4), then fixed in 4% paraformaldehyde in TBS as described (Brice et al., 1997). The cells were permeabilized in 0.02% Triton X-100 and incubated with blocking solution [20% (v/v) goat serum, 4% (w/v) bovine serum albumin (BSA), and 0.1% d,l-lysine in TBS]. In experiments using mouse monoclonal anti-GFP Ab (Clontech), or anti-myc Ab (9E10; Santa Cruz), they were used at 20 and 0.4 μg/ml, respectively, and the secondary Ab was 10 μg/ml goat anti-mouse IgG conjugated to Texas Red (Molecular Probes, Eugene, OR). In some experiments, cells were incubated for 20 min with Texas Red phalloidin (6.6 μm; Molecular Probes). The nuclear dye 4′,6-diamidino-2-phenylindole (DAPI; 300 nm; Molecular Probes) was also used to visualize the nucleus. Cells were then washed in TBS five times for 5 min each. Coverslips were mounted directly onto a microscope slide with Vectorshield (Vector Laboratories, Burlingame, CA), and the cells were examined on a laser-scanning confocal microscope (Leica TCS SP; Leica, Milton Keynes, UK). The optical sections were 0.2 μm, and all images were scanned sequentially to eliminate cross-talk. For the immunocytochemistry experiments, n = number of different transfections performed, with at least two coverslips of cells analyzed per transfection condition.
Western blotting. COS-7 cells were resuspended in hypotonic buffer (10 mm Tris, pH 7.4), containing protease inhibitors (complete EDTA-free; Roche Diagnostics, Lewes, UK) and 2 mm EDTA. Aliquots were taken for assay of total lysate protein (BCA; Perbio Science, Chester, UK), and the remainder of each sample was then solubilized in SDS-PAGE sample buffer containing 2% SDS. The samples were sonicated briefly (three times for 5 sec each on ice) and then centrifuged (10,000 × g, 15 min, 4°C) to remove any insoluble material. Samples (50 μg of total protein/lane) were separated by SDS-PAGE using 7.5% resolving gels and then transferred electrophoretically to polyvinylidene fluoride membranes. The membranes were blocked with 3% BSA for 5 hr at 55°C and then incubated overnight at 20°C with a 1:1000 dilution of either anti-GFP monoclonal Ab, or an anti-peptide Ab raised in rabbits against residues 846–861 within the II-III loop of rabbit brain Cav2.2 and purified by affinity chromatography using the immobilized synthetic peptide. Secondary Ab (a 1:1000 dilution of goat anti-mouse IgG or goat anti-rabbit IgG horseradish peroxidase conjugate, respectively) was added, and the membranes were incubated for 1 hr. After extensive washing, bound Abs were detected using enhanced chemiluminescence (Amersham Pharmacia Biotech, Little Chalfont, UK).
Whole-cell electrophysiology. Whole-cell patch-clamp recording was performed essentially as previously described (Meir et al., 2000), with 10 mmBa2+ as charge carrier. Only fluorescent cells expressing GFP were used for recording. The holding potential was −100 mV, and pulses were delivered every 10 sec. Currents were measured 10 msec after the onset of the test pulse, and the average over a 2 msec period was calculated and used for subsequent analysis. The current density–voltage (I–V) relationships were fitted with a modified Boltzmann equation as follows: where I is the current density (in picoamperes per picofarad), Gmax is the maximum conductance (in nanosiemans per picofarad),Vrev is the reversal potential,V50,act is the midpoint voltage for current activation, and k is the slope factor. Data are expressed as mean ± SEM of the number of replicates,n. Steady-state inactivation properties were measured by applying 10 sec pulse from −120 to +10 mV in 10 mV increments, followed by a 10 msec repolarization to −100 mV before the 40 msec test pulse to +20 mV. Steady-state inactivation data were fitted with a single Boltzmann equation of the form: where Imax is the maximal current, V50,inact is the half-maximal voltage for current inactivation,kinact is the slope factor, andA1 andA2 represent the proportion of inactivating and noninactivating current, respectively.
Single-channel electrophysiology. Recordings were performed on GFP-positive cells at 20–24°C. Recording pipettes were pulled from borosilicate tubes (World Precision Instruments, Sarasota, FL), coated with Sylgard (Sylgard 184, Dow Corning, Wiesbaden, Germany), and fire-polished. The bath solution, designed to zero the resting membrane potential (Meir and Dolphin, 1998) was composed of (in mm): 135 K-aspartate, 1 MgCl2, 5 EGTA, and 10 HEPES (titrated with KOH, pH 7.3), and the patch pipettes were filled with a solution of the following composition (in mm): 100 BaCl2, 10 TEA-Cl, and 10 HEPES, with 200 nm TTX, titrated with TEA-OH to pH 7.4. Both solutions were adjusted to an osmolarity of 320 mOsmol with sucrose. Data were sampled at 10 kHz and filtered on-line at 2 kHz. (Axopatch 200B and Digidata 1200; Axon Instruments, Foster City, CA). Voltages were not corrected for liquid junction potential (Neher, 1995), measured to be −15 mV in these solutions.
Leak subtraction was performed as described (Meir et al., 2000). Event detection was performed using the half-amplitude threshold method. Open time was determined by a single or double exponential fit to the open time distributions. Closed times were determined similarly using only patches with no overlapping openings. The latency to first opening was measured in 2 msec bins and analyzed as described (Meir et al., 2000). In brief, first latency histograms were accumulated and divided by the number of episodes, to represent the cumulative probability of a first latency event (PFL). If necessary these were corrected for the number of channels in the patch. We considered the number of detectable simultaneously overlapping openings as representing the number of channels active in the patch (Meir et al., 2000). To strengthen this assumption we included in the latency analysis only patches with up to three simultaneously overlapping openings.
RESULTS
Expression of GFP–Cav2.2 and truncated constructs
The functional expression of Cav2.2 in COS-7 cells was investigated using N terminal GFP-fusion proteins of Cav2.2 and several truncated forms. We first examined whether the GFP-tagged Cav2.2, expressed in combination with the accessory subunits β1b and α2δ-1, was able to reproduce the biophysical properties of wild-type Cav2.2. An example of a whole-cell recording ofIBa from COS-7 cells transfected with GFP–Cav2.2 cDNA is shown in Figure1a (top panel), together with the voltage protocol used. TheI–V relationship for the GFP-tagged Cav2.2 is shown in Figure 1b(filled circles). The GFP tag on the N terminus did not interfere with the functionality of the channel, because the current density at +20 mV was −55.1 ± 8.3 pA/pF for the untagged Cav2.2 channel (n = 12; data not shown), not significantly different from the GFP–Cav2.2 channel (−59.2 ± 17.9 pA/pF;n = 10). Similarly, there were no differences in other parameters of the I–V relationship (see legend to Fig.1).
GFP-Dom I–II, containing only the first two domains and the intracellular II-III loop of Cav2.2, when expressed with β1b and α2δ-1, did not elicit any detectable currents (Fig. 1b, open circles). The same was true for Dom III–IV, indicating that the hemichannels are unable to form functional channels alone. In contrast, coexpression of the two hemichannels (either with or without a GFP tag on Dom I–II) resulted in the reproducible expression of small whole-cell currents, with properties otherwise analogous to the native Cav2.2 (Fig.1a, middle and bottom panels). In cells expressing untagged Dom I–II and Dom III–IV, a +20 mV step elicited a current of −13.2 ± 4.2 pA/pF (n = 8). Coexpression of the GFP-Dom I–II with Dom III–IV resulted in the expression of currents with similar amplitude at +20 mV (−11.4 ± 3.8 pA/pF; n = 7). In addition, the steady-state inactivation properties of the reconstituted channel composed of Dom I–II and Dom III–IV did not differ from those of GFP–Cav2.2 (Fig. 1c).
Once the functional integrity of the GFP-tagged channel and hemichannels was proven, the expression of these fusion proteins was examined using confocal microscopy. Figure2a shows the localization of GFP–Cav2.2. This subunit was expressed throughout the cell (Fig. 2a, left panel) (n > 10). The cells were also stained with Texas Red phalloidin to visualize cortical actin, which delineates the plasma membrane (Fig. 2a, middle panel). The yellow color in the merged image (arrow) indicates that GFP–Cav2.2 and Texas Red phalloidin are colocalized at the plasma membrane (Fig. 2a, right panel), in accordance with the electrophysiological results.
Effect of coexpression of two domain constructs on Cav2.2 localization
Having determined the distribution of full-length GFP–Cav2.2, we next investigated the effect of Dom I–II or Dom III–IV on the expression of GFP–Cav2.2. As shown in Figure 2b, untagged Dom I–II strongly suppressed the expression of GFP–Cav2.2 (n = 6). This was evidenced by the complete absence of observable GFP-positive cells. This could be attributable to a low level of expression, below the detection capability of the imaging system. Thus, an anti-GFP Ab was used to amplify the signal of any expressed GFP–Cav2.2 within the cells. The anti-GFP Ab was able to detect GFP–Cav2.2 alone (Fig.2c) ( n = 4), but no staining was detectable when GFP–Cav2.2 was coexpressed with Dom I–II (Fig. 2d, center panel) ( n = 4). In this case, the presence of viable cells was established by using the nuclear stain DAPI (Fig. 2d, right panel). These results confirm that Dom I–II greatly reduces the expression of GFP–Cav2.2.
When GFP-Dom I–II was expressed alone, it was detectable at the plasma membrane and throughout the cell (Fig. 2e, right panel). The subcellular localization of GFP-Dom I–II was identical to that of GFP–Cav2.2 (n > 10). Interestingly, in the converse of the coexpression study described above, the expression of GFP-Dom I–II was not detectably reduced by the presence of untagged Cav2.2 (Fig. 2f).
The coexpression of GFP-Dom I–II with Dom III–IV did not alter the GFP expression or localization of GFP-Dom I–II (results not shown;n = 6). In contrast, coexpression of GFP–Cav2.2 with Dom III–IV altered the expression of GFP–Cav2.2 within individual cells, but did not entirely suppress it (Fig. 2g) (n = 6). In this case, GFP–Cav2.2 showed a perinuclear localization and was not readily detectable throughout the cytoplasm or at the plasma membrane by confocal microscopy (Fig. 2g).
Functional effects of coexpression of two-domain constructs with Cav2.2
Although immunofluorescence studies indicated that Dom I–II completely suppressed the expression of GFP–Cav2.2, it was plausible that small amounts of GFP–Cav2.2 were still expressed. This was confirmed by whole-cell recording, which showed that there was a marked reduction in IBa current density in cells expressing GFP–Cav2.2 together with either Dom I–II or Dom III–IV (Fig.3a), although theI–V parameters were unchanged (Fig. 3b).IBa at +20 mV was −59.2 ± 17.9 pA/pF for GFP–Cav2.2, and was reduced to −19.2 ± 3.6 pA/pF when GFP–Cav2.2 was coexpressed with Dom I–II (67% reduction; n = 26;p < 0.001) and −18.1 ± 6.1 pA/pF with Dom III–IV (69% reduction; n = 12; p < 0.001). The steady-state inactivation parameters for Cav2.2 were also unchanged by coexpression with Dom I–II (Fig. 3c). Because of the suppression effect, it was necessary to coexpress free GFP, to facilitate the identification of successfully transfected cells. This did not alter the amplitude of control Cav2.2 currents (data not shown). We also examined whether the effect of Dom I–II was a nonspecific result of coexpressing another transmembrane protein, but no reduction inIBa was observed when Cav2.2 was coexpressed with the α2A-adrenergic receptor under the same conditions (IBa = −80.1 ± 24.1 pA/pF at +20 mV; n = 6). We further assessed whether the decrease in Cav2.2 current amplitude when it was coexpressed with Dom I–II could be a result of an alteration of the ratio between the Cav2.2 and truncated construct cDNAs transfected. ThereforeIBa was examined in cells transfected with half of the normal amount of GFP–Cav2.2-pMT2 cDNA (1.5 μg/dish), with the same amount of accessory subunits. No reduction in current amplitude or alteration in I–V parameters were detected (data not shown), indicating that the Cav2.2 cDNA amount was saturating, consistent with the fact that COS-7 cells are SV40 transformed, and the vector pMT2 contains the SV40 origin of replication. We consistently observed that, within the range examined, reduction of the α1 subunit cDNA level decreased the number of cells transfected but not the current density.
To address the mechanism of suppression, we examined whether the suppressive effect of Dom I–II or Dom III–IV on Cav2.2 currents could be reduced by coexpression of both hemichannels together with Cav2.2, a result that might be predicted because our initial studies showed that Dom I–II and Dom III–IV were able to interact together to form functional channels. A protective effect was confirmed, because theIBa current density at +20 mV in cells expressing GFP–Cav2.2 together with both Dom I–II and Dom III–IV was −44.4 ± 11.9 pA/pF (n= 9), a nonsignificant 17% reduction (p = 0.54), compared with −53.5 ± 8.1 pA/pF−1 (n = 9) for control GFP–Cav2.2 currents (same controls as in Fig. 3e, because experiments performed in parallel).
We next examined whether the mechanism of suppression involved an accelerated degradation of Cav2.2, by examining the effect of the 26 S proteasome inhibitor lactacystin. When incubated with cells, at 30 μm, either for the entire period between transfection and visualization of GFP–Cav2.2, or for the final 16 hr, lactacystin did not increase GFP fluorescence of GFP–Cav2.2 coexpressed with Dom I–II (compared with controls receiving the same amount of solvent, n = 6, results not shown).
Is the suppression of expression of GFP–Cav2.2 caused by sequestration of β subunits by the truncated constructs?
The Cavβ subunits have been shown to act as chaperone proteins for the calcium channel α1 subunits, enhancing their translocation from the endoplasmic reticulum to the plasma membrane (Bichet et al., 2000). One possible explanation for the suppression of Cav2.2 currents is that Dom I–II acts to sequester free β1b via its I-II loop, and therefore limits the amount available for chaperoning GFP–Cav2.2 to the membrane. To test this hypothesis, cells were transfected with increased β1b cDNA, but this did not enhance Cav2.2 IBarecorded in the presence of Dom I–II (Fig. 3d) (IBa current density at +20 mV was −19.4 ± 10.7 pA/pF; n = 8). To examine further whether the effect of Dom I–II on GFP–Cav2.2 was attributable to scavenging of β1b by the I-II loop within Dom I–II, we also coexpressed GFP–Cav2.2 with a construct of the Cav2.2 I-II loop, which we have shown to be capable of binding β1b (Bell et al., 2001). No inhibitory effect was observed (Fig. 3e); theIBa current density at +20 mV in cells expressing GFP–Cav2.2 together with the free I-II loop was −50.3 ± 11.4 pA/pF (n = 11), a nonsignificant 5% reduction, compared with −53.5 ± 8.1 pA/pF (n = 9) for GFP–Cav2.2 currents recorded in parallel, from the same transfections. The steady-state inactivation parameters for GFP–Cav2.2 were also unchanged by coexpression with Dom I–II (Fig. 3f). In agreement with this, we saw no reduction of GFP–Cav2.2 fluorescence when it was coexpressed with the I-II loop, either in the presence or absence of coexpressed β1b (Fig. 3g) (data not shown; n = 2). Furthermore, omission of both coexpressed accessory subunits (β1b and α2δ-1) did not affect the ability of Dom I–II to suppress expression of GFP–Cav2.2, as determined by its fluorescence (Fig. 3h) (n = 6). Because there is very low functional expression of Cav2.2 channels in the absence of these accessory subunits (Meir et al., 2000), it was not possible to perform the corresponding electrophysiological experiments.
In yeast two-hybrid experiments, we observed no interaction of β1b with the intracellular N terminus or the C terminus of Cav2.2 (n = 3). In every experiment, an interaction of β1b with a I-II loop construct was obtained as a positive control (results not shown). This rules out β1b subunit scavenging as a mechanism of action of Dom III–IV, which contains the C terminus, but not the I-II loop.
Single channel properties of two domain constructs coexpressed together or with Cav2.2
These experiments were performed to determine whether the small whole-cell currents obtained either when Dom I–II was coexpressed with Dom III–IV or when Dom I–II was coexpressed with Cav2.2 were caused by altered properties of the channels formed. Single channels were recorded in the cell-attached mode of the patch clamp technique. The recordings were made from COS-7 cells transfected with GFP–Cav2.2 alone (Fig.4a), Dom I–II together with Dom III–IV (Fig. 4b), or GFP–Cav2.2 with Dom I–II (Fig. 4c). In all cases we could detect single channels with a similar mean conductance (Fig. 4d), mean open time (Fig. 4e), mean closed time (Fig.4f), and latency to first opening (Fig.4g; shown at +30 mV). These values are very similar to those obtained from wild-type Cav2.2 (not tagged with GFP) (Meir et al., 2000).
Effect of coexpression of Dom I or the cytoplasmic N terminus on Cav2.2 expression
We next investigated the minimal domain required for suppression of Cav2.2 expression. To this end, GFP–Cav2.2 was coexpressed with either Dom I or the cytoplasmic N terminus of Cav2.2. The Dom I construct consisted of the intracellular N terminus, domain I, and the intracellular I-II loop. Immunofluorescence studies using YFP-Dom I confirmed its expression throughout COS-7 cells (results not shown). In a similar manner to Dom I–II, untagged Dom I also appeared to abolish the expression of GFP–Cav2.2, as assessed by confocal microscopy (Fig. 5a) (n = 4). These results were confirmed by using the anti-GFP Ab, which did not reveal any GFP–Cav2.2 (Fig. 5b) (n = 4). Again, this suppression was not affected by the absence of accessory subunits (results not shown; n = 4). In contrast, the N terminus of Cav2.2 did not have any effect on the expression of GFP–Cav2.2 or on its subcellular localization (Fig. 5c) (n = 4). GFP–Cav2.2 was localized at the plasma membrane (colocalized with phalloidin) and throughout the cytoplasm. In confirmation of these results, IBarecorded from cells expressing GFP–Cav2.2 and Dom I was dramatically reduced, compared with controls (Fig.5d). IBa at +20 mV was −5.2 ± 1.9 pA/pF (88.4% reduction compared with control;n = 8). In contrast, the I–V parameters for cells expressing GFP–Cav2.2 together with the N terminus did not show any decrease in current amplitude or effect on activation (Fig. 5d). The steady-state inactivation parameters were also identical to those of GFP–Cav2.2 (Fig. 5e). This is in agreement with the lack of interaction observed between the β1b subunit and the N terminus of Cav2.2 in the yeast two-hybrid assay.
Effect of coexpression of truncated constructs on the Cav2.2 protein level
The GFP–Cav2.2 expressed alone was detectable by Western blotting using either an anti-GFP Ab or an Ab against the II-III loop of Cav2.2 (band at ∼250 kDa) (Fig. 6a,b, lane 1). A minor band at 100 kDa was also observed with both Abs, which might therefore represent an N terminal degradation product of Cav2.2. When Dom I–II was expressed alone it was detected by anti-Cav2.2, but not anti-GFP Abs (band at 120 kDa in Fig. 6b but not Fig. 6a, lane 2). However, when GFP–Cav2.2 was expressed together with Dom I–II, no band at 250 kDa was observed with either Ab (Fig. 6a,b, lane 3), although Dom I–II was detected by the anti-Cav2.2 Ab, to a similar level as when it was expressed alone (Fig. 6b, compare lanes 2 and 3). This is in agreement with the confocal imaging data (Fig. 2). No smaller molecular weight (MW) bands that might represent partially synthesized or degradation products of GFP–Cav2.2 were observed when it was cotransfected with Dom I–II, using either Ab (Fig. 6a,b,lane 3). Neither the 250 and 120 kDa bands nor the 100 kDa putative proteolytic product of GFP–Cav2.2 were present in nontransfected cells (Fig. 6a,b, lane 4). Similar results were obtained when GFP–Cav2.2 was expressed together with GFP-Dom I–II or Dom I (results not shown). In contrast, in the case of coexpression of GFP–Cav2.2 with Dom III–IV, there was little, if any, reduction in the amount of GFP–Cav2.2 (Fig. 6c, comparelanes 1 and 3), in agreement with the confocal imaging data (Fig. 2g).
DISCUSSION
It has been suggested that four-domain Na+ and Ca2+channels arose during evolution from two sequential gene duplications of a K+ channel (Plummer et al., 1997) and that expression of the two-domain isoforms still occurs in a developmentally regulated manner (Plummer et al., 1997). It is possible that the two-domain isoforms of Cav1.1, Cav1.2, and the Na+channel SCN8A may all serve a similar function (Plummer et al., 1997). It would be predicted from our results that this would be a dominant-negative function, to suppress expression of the full-length channel. In support of this, a three-domain construct of an ascidian calcium channel, thought to be expressed from maternal transcript, has recently been shown to suppress expression of the full-length ascidian calcium channel (Okagaki et al., 2001).
Dominant-negative suppression of K+channel tetramer function by transmembrane fragments has been studied previously for the Kv channel family (Tu et al., 1995). This suppression effect was found to involve multiple transmembrane peptides but not to affect synthesis (Tu et al., 1996). As discussed in that study, the formation of K+ channel tetramers will involve multiple interactions between domains, and disruption at any stage of biogenesis may be sufficient to cause suppression of functional expression (Tu et al., 1996). In contrast, another recent study showed the existence of a pathway involving arrest of synthesis and rapid degradation of mis-folded human ether-a-go-go-related gene (HERG) K+ channel tetramers induced by a point mutation (Kagan et al., 2000).
Here we have examined whether the mechanism of suppression of expression of full-length Cav2.2 by truncated constructs involves (1) interference with gating of the channel inserted in the plasma membrane, (2) interference with delivery to the plasma membrane, (3) increased protein degradation, or (4) synthesis arrest. Below we consider these possibilities in turn.
Is there an impairment of gating caused by association of channel fragments with full-length Cav2.2 in the plasma membrane? From the electrophysiological data, GFP-Dom I–II coexpressed with Dom III–IV resulted in small but reproducible whole-cell calcium channel currents and produced single channels whose properties, apart from frequency of observation, were indistinguishable from wild type. Therefore, at least a small proportion of the truncated constructs must be able to fold and assemble together correctly with the normal topology. Similar results have recently been obtained for coexpression of two domain constructs of Cav1.1 (Ahern et al., 2001).
Concerning the mechanism of suppression, a combination of the whole-cell and single channel analysis indicates that the inhibition of full-length Cav2.2 currents by Dom I–II is attributable to a reduction in the number of channels, because there is no alteration in any of their biophysical properties examined. Although fewer Cav2.2 channels reach the plasma membrane, their gating is not modified by an association with the truncated construct. The apparent discrepancy between the confocal imaging data, where almost no GFP fluorescence was seen when Dom I–II was coexpressed with GFP–Cav2.2, and the electrophysiological data, may be a function of the detection limit, which has been calculated to be ∼10,000 GFP molecules per tissue culture cell (Patterson et al., 1997). From our data, only ∼3000 channels would be required to give rise to the currents observed when Dom I–II is coexpressed with GFP–Cav2.2. It is also possible that the GFP tag is synthesized, but mis-folded, and therefore not fluorescent, but it was also not recognized by the GFP Ab.
Is there interference in the delivery of the channel to the plasma membrane? Cavβ subunits are involved both in trafficking α1 subunits and in modulating their biophysical properties (Chien et al., 1995; Brice et al., 1997; Bichet et al., 2000; Canti et al., 2001). It is conceivable that trafficking of Cav2.2 through the endoplasmic reticulum might be compromised by scavenging of Cavβ subunits by the truncated fragments. However, suppression was not prevented by expression of an increased amount of β subunit. Furthermore, the properties of the currents in the presence of the truncated constructs did not mimic those of Cav2.2 expressed without β subunits in COS-7 cells or Xenopus oocytes, where both the V50 for activation and steady-state inactivation were markedly depolarized (Canti et al., 2000; Meir et al., 2000; Stephens et al., 2000). Moreover, suppression of GFP–Cav2.2 protein expression remained evident in the absence of coexpressed accessory β subunits, indicating that it occurs before trafficking out of the endoplasmic reticulum [for which Cavβ is required (Bichet et al., 2000)]. From yeast two-hybrid experiments we found that, of the intracellular I-II loop, the N terminus and the C terminus of Cav2.2, only the I-II loop represents a high-affinity interaction site for β1b. However, coexpression of the I-II loop with Cav2.2 did not reduce Cav2.2 IBa, again indicating that scavenging of β subunits is not responsible for the suppressive effect of Dom I and Dom I–II. The lack of effect of the I-II loop, either to reduce expression or to affect theV50 for activation is presumably because the β subunit is present in excess, as also demonstrated in our recent study in Xenopus oocytes, where the maximum effect of β3 on expression occurred at ∼6 pg of β3 cDNA for 540 pg of Cav2.2 cDNA injected per oocyte (Canti et al., 2001).
Another potential mechanism of suppression would be prevention of correct folding of Cav2.2 by the truncated domains, so that endoplasmic reticulum retention signals are not masked, and the mis-folded channel is retained in the endoplasmic reticulum. Although this may be the mechanism of Cav2.2 suppression by the Dom III–IV construct, where loss of Cav2.2 protein was not observed (Figs. 2g, 6c), in the case of the truncated constructs containing domain I, instead of observing an accumulation of the GFP–Cav2.2 signal in the endoplasmic reticulum, we observed an almost complete loss of GFP–Cav2.2 fluorescence (Figs. 2b,d,5a,b).
This points to decreased synthesis or stability of the Cav2.2 protein. In experiments to distinguish between these possibilities, we found that inhibition of proteasome activity by lactacystin did not increase the amount of GFP–Cav2.2 observed in the presence of Dom I–II, suggesting that the mechanism does not involve enhanced proteolysis, in contrast to the finding with the HERG K+ channel mutant (Kagan et al., 2000). From these results, the most likely explanation for suppression by truncated constructs containing domain I is that synthesis of full-length Cav2.2 is arrested. Furthermore, the intracellular N terminus alone was ineffective, suggesting that suppression may occur by interaction of the nascent transmembrane segments of the first domain of Cav2.2 with Dom I of the truncated construct. Synthesis of polytopic proteins passes through a state where up to six nascent transmembrane α-helices span the endoplasmic reticulum membrane, but are not yet integrated in its lipid bilayer, associating via ionic rather than hydrophobic interactions (Borel and Simon, 1996). When the initial transmembrane α-helices of Cav2.2 are in this state, it may be that interference occurs with further synthesis of the full-length channel, because of interaction with the Dom I and Dom I–II proteins, where, in the absence of all four transmembrane domains for assembly, inappropriate residues would remain exposed. It is possible that this would effectively halt polysomal movement on each Cav2.2 mRNA. The lack of effect of the Dom III–IV construct to suppress synthesis of Cav2.2 could be attributed to the fact that synthesis and assembly of Cav2.2 is nearer completion before the interaction occurs with transmembrane segments of Dom III–IV, which then results in trapping in the endoplasmic reticulum. Furthermore, the fact that there is a reduction, rather than an increase, in suppression when Dom I–II and Dom III–IV are together coexpressed with Cav2.2, also points to the exposure of inappropriate residues on the singly expressed hemichannels as a mechanism of suppression.
In agreement with the hypothesis that synthesis of Cav2.2 is suppressed, we observed loss of full-length GFP–Cav2.2 protein, when it was coexpressed with either Dom I–II or Dom I. Furthermore, no smaller MW partially synthesized or degradation products of Cav2.2 were observed. This also points to synthesis arrest at an early stage, rather than enhanced degradation. However, although attenuation of translation is a well established aspect of the unfolded protein response that occurs when there is an accumulation of mis-folded proteins in the endoplasmic reticulum (Chevet et al., 2001), in the present case the synthesis inhibition was specific to Cav2.2, because coexpression of Cav2.2 did not appear to reduce the level of Dom I–II, and this may therefore be a novel mechanism. We are currently examining whether the Cav2.2 mRNA level is reduced, and if so, whether this is a primary event, or a consequence of synthesis inhibition.
In conclusion, it is likely that early in the process of synthesis, if Cav2.2 associates with domain I of the truncated constructs in the endoplasmic reticulum membrane, translation of the full-length Cav2.2 channel is largely prevented. This finding may generalize to all normally or pathologically occurring calcium channel splice variants that form such truncated proteins and represent a physiological mechanism for developmental or tissue-specific channel expression.
Footnotes
This work was supported by the Wellcome Trust and Medical Research Council (MRC). A.R. was an MRC PhD student. We thank Dr. E. Perez-Reyes for β1b cDNA, Dr. Y. Mori for Cav2.2 cDNA, Dr. H. Chin for α2δ-1 cDNA, and Dr. T Hughes for mut3-GFP cDNA. We thank Nuria Balaguero, Wendy Pratt, and Manuela Nieto-Rostro for technical assistance.
A.R. and F.B. contributed equally to this work.
Correspondence should be addressed to A. C. Dolphin, Department of Pharmacology, University College London, Gower Street, London WC1E6BT, UK. E-mail: a.dolphin{at}ucl.ac.uk.