In cardiac myocytes, LTCCs (L-type calcium channels) form a functional signalling complex with ryanodine receptors at the JM (junctional membrane). Although the specific localization of LTCCs to the JM is critical for excitation–contraction coupling, their targeting mechanism is unclear. Transient transfection of GFP (green fluorescent protein)–α1S or GFP–α1C, but not P/Q-type calcium channel α1A, in dysgenic (α1S-null) GLT myotubes results in correct targeting of these LTCCs to the JMs and restoration of action-potential-induced Ca2+ transients. To identify the sequences of α1C responsible for JM targeting, we generated a range of α1C–α1A chimaeras, deletion mutants and alanine substitution mutants and studied their targeting properties in GLT myotubes. The results revealed that amino acids L1681QAGLRTL1688 and P1693EIRRAIS1700, predicted to form two adjacent α-helices in the proximal C-terminus, are necessary for the JM targeting of α1C. The efficiency of restoration of action-potential-induced Ca2+ transients in GLT myotubes was significantly decreased by mutations in the targeting motif. JM targeting was not disrupted by the distal C-terminus of α1C which binds to the second α-helix. Therefore we have identified a new structural motif in the C-terminus of α1C that mediates the targeting of cardiac LTCCs to JMs independently of the interaction between proximal and distal C-termini of α1C.
Voltage-dependent LTCCs (L-type calcium channels) play important roles in the regulation of a variety of cellular functions of excitable cells, including EC (excitation–contraction) coupling, hormone secretion and transcriptional regulation [1,2]. LTCCs are composed of the pore-forming α1S (CaV1.1), ancillary β1, α2δ and γ subunits in skeletal muscle cells, and of α1C (CaV1.2), β2 and α2δ subunits in cardiac muscle . An α1 subunit comprises four homologous repeats (I–IV), each containing six transmembrane segments (S1–S6) that together form the Ca2+ channel pore, the voltage sensor, and activation and inactivation gates .
LTCCs in skeletal and cardiac muscle cells are specifically localized to the JM (junctional membrane) of the triad or peripheral couplings, where sarcolemmal membranes are closely apposed to the membranes of the terminal cisternae of the SR (sarcoplasmic reticulum) [5–7]. RyRs (ryanodine receptors) are concentrated in the terminal cisternae and form a functional complex with LTCCs. In skeletal muscle cells, the pore-forming α1S subunit directly interacts with RyR1 through its cytoplasmic II–III loop and induces VDCR (voltage-dependent Ca2+ release) from the SR in response to membrane depolarization. On the other hand, cardiac LTCCs induce EC coupling by CICR (Ca2+-induced Ca2+ release). Whether cardiac EC coupling requires physical interaction of LTCCs with RyRs is a matter of debate. However, a rapid increase in the local Ca2+ concentration by LTCC openings in the vicinity of RyRs is essential for efficient CICR in cardiac myocytes [8,9]. Thus, despite the difference in the Ca2+ release mechanisms, the precise co-localization of LTCCs with RyRs in the JM is essential for normal EC coupling in both skeletal and cardiac muscle cells.
Although identification of the JM-targeting mechanisms of α1C is critical for understanding how functional LTCC–RyR signalling complexes are formed and normal EC coupling is maintained in cardiac myocytes, the JM-targeting motif of α1C is still unknown. This is, in part, due to the fact that isolated cardiac myocytes in primary culture are not suitable for these analyses because of their poor viability, rapid change in cell shape and phenotype, and their resistance to liposome-based transfection. Therefore, in the past, the plasma-membrane-targeting signal of α1C was analysed in non-muscle cells, such as tsA201 cells, leading to the identification of a role for the C-terminus . Furthermore, in previous studies LTCCs were co-expressed with RyRs in Chinese-hamster ovary cells [11,12]. However, these cells did not exhibit the Ca2+ release from the ER (endoplasmic reticulum) in response to membrane depolarization [11,12], indicating that the functional Ca2+-releasing complex of LTCCs and RyRs could not be reconstituted. Thus these cells are not suitable for the analysis of the JM-targeting motif of α1C. To overcome these problems, we used GLT cells, a muscle cell line derived from α1S-deficient dysgenic (mdg) mice . In dysgenic myotubes, transiently expressed α1S [with an N-terminal GFP (green fluorescent protein) tag] is correctly localized to the JM with RyRs and restores skeletal LTCC currents, VDCR and EC coupling [14–16]. Moreover this muscle expression system has previously been used to identify a 55-amino-acid sequence in the C-terminus of α1S that is important for JM targeting of the skeletal muscle Ca2+ channel. Importantly, heterologously expressed cardiac α1C is also co-localized with RyRs in the JM of GLT myotubes and restores cardiac-type CICR and EC coupling [14,17]. Thus GLT cells express the machinery required for the JM targeting of α1C and therefore are a suitable cell system to analyse the mechanism underlying the JM targeting of the cardiac LTCC α1C subunit.
In the present study, we constructed multiple α1C-based α1C–α1A chimaeras and mutants, expressed them in GLT cells and analysed their subcellular localizations with immunofluorescence staining. The results identify a novel targeting motif consisting of two adjacent α-helices within amino acid residues 1681–1700 at the proximal C-terminus of α1C that are necessary for the JM targeting. The distal half of the JM-targeting motif overlaps with sequences to which the proteolytically cleaved distal C-terminus binds [18–20]. This interaction is essential for β-adrenergic stimulation of LTCCs . However, the JM targeting of α1C was not compromised by expression of the distal C-terminus. Therefore a new structural motif in the C-terminus of α1C identified in the present study mediates the specific targeting of the cardiac LTCCs to JMs of the EC coupling apparatus, without interfering with the interaction between the proximal and distal C-termini of α1C.
The cDNA encoding rabbit α1C was provided by Professor William Catterall (Department of Pharmacology, University of Washington, Seattle, WA, U.S.A.). The cDNAs encoding GFP–α1A and GFP–α1M were supplied by Professor Manfred Grabner (Innsbruck Medical University, Innsbruck, Austria). Most chimaera constructs were prepared with a ‘gene SOEing’ technique . All primer sequences used in the present study are available upon request from the authors. DNA polymerase PrimeSTAR HS (Takara Bio) was used for all PCRs. To generate GFP–α1C, the cDNA of rabbit α1C was amplified by PCR and then subcloned into HindIII/SalI sites of pEGFP-C1. An HA (haemagglutinin) epitope was inserted into the extracellular loop between S5 and S6 in domain II of α1C as described previously (HA–α1C) [23,24]. Chimaera constructs (GFP–α1CNTA, GFP–α1CI–IIA, GFP–α1CII–IIIM, GFP–α1CIII–IVA, GFP–α1CCTA, HA–α1C1507–1650A, HA–α1C1624–1665A, HA–α1C1666–1733A, HA–α1C1666–1676A, HA–α1C1666–1688A, HA–α1C1677–1688A, HA–α1C1689–1733A, HA–α1C1709–1733A, GFP–α1ACTC, GFP–α1AJMTC and CD8–α1CCT) were prepared using a SOEing technique. Two or three fragments of PCR products were fused and subcloned into a blunt-ended pZErO-2 vector (Invitrogen) and then back to the corresponding region of each vector (GFP–α1C, GFP–α1A or HA–α1C) using additional or native restriction sites. The deletion mutants (HA–α1CΔ1821, HA–α1CΔ1733, HA–α1CΔ1666 and HA–α1CΔ1624) were generated by PCR using antisense primers containing a stop codon and a SalI restriction site. The PCR products were subcloned into a blunt-ended pZErO-2 vector and again subcloned into the EcoRV/SalI site of HA–α1C. Alanine substitution mutants were generated with the QuikChange® Site-Directed Mutagenesis kit (Stratagene) according to the manufacturer's protocol.
The DCT (distal C-terminus) of α1C-(1821–2171) was amplified by PCR and then subcloned into p3×FLAG-CMV10. The cDNAs of CaM (calmodulin), sorcin and CD8a were isolated with RT (reverse transcription)–PCR from mouse heart and blood cDNA. The cDNAs of CaM and sorcin were subcloned into p3×FLAG-CMV10. The Ca2+-insensitive mutant CaM1234 (D21A/D57A/D94A/D130A) was prepared with the QuikChange® Site-Directed Mutagenesis kit according to the manufacturer's protocol. CD8a and CD8a–α1CCT were subcloned into pcDNA3.1. The nucleotide sequences of all of the constructs were verified with an ABI 3130 genetic analyser (Applied Biosystems).
Cell culture and transfection
Myotubes of the GLT cell line derived from myoblasts of dysgenic (mdg) mice were cultured as described previously . Briefly, GLT cells were plated on carbon-coated coverslips coated with 0.1% gelatin (day 0) in GM (growth medium): low-glucose DMEM (Dulbecco's modified Eagle's medium; Invitrogen) containing GlutaMAX (Invitrogen), 10% FBS (fetal bovine serum), 10% horse serum, 100 units/ml penicillin and 100 μg/ml streptomycin. To differentiate myoblasts into myotubes, 2 days after plating (day 2), the medium was changed to DM (differentiation medium): DMEM containing 2% horse serum (Invitrogen), GlutaMAX (Invitrogen), 100 units/ml penicillin and 100 μg/ml streptomycin (Invitrogen). Again 2 days later (day 4), cells were transfected with the expression vectors using FuGENE™ HD (Roche Diagnostics) according to the manufacturer's protocol (1 μg of plasmid DNA/35-mm dish). For co-expression of channels and other proteins [CaMWT (wild-type), CaM1234 and sorcin], equal amounts of plasmid DNAs (0.75 μg) were transfected. Plasmid vectors coding for channels (α1CΔ1821 or α1C) and the DCT of α1C were co-transfected with various molar ratios in the GLT myotube (DCT/channel vector molar ratios were 0.8, 1.6 and 3.2). Between days 7 and 10 myotubes were fixed and immunofluorescence labelling was performed.
Cells were fixed in 4% (w/v) paraformaldehyde in PBS at room temperature (24°C) for 10 min. After washing the samples with PBS, cells were blocked and permeabilized with PBS containing 0.2% Triton X-100 and 5% FBS for 1 h at room temperature and then incubated with primary antibodies at 4°C overnight. After washing with PBS, cells were incubated with fluorescent dye-conjugated secondary antibodies and Hoechst 33342 (Dojindo) at room temperature for 1 h. Cells were again washed with PBS, and coverslips were mounted with Fluoromount-G (Beckman Coulter). To label extracellular HA epitope or CD8, cells were incubated with the corresponding primary antibodies at 37°C for 60 min before permeabilization. The following primary and secondary antibodies were used: anti-GFP rabbit polyclonal antibody (Invitrogen; 1:1000 dilution), anti-RyR mouse monoclonal antibody (Affinity Bioreagents; clone 34C, 1:1000 dilution), anti-HA rat monoclonal antibody (Roche Diagnostics; clone 3F10, 1:200 dilution), anti-mouse CD8a rat monoclonal antibody (BioLegend; clone 53-6.7, 1:1000 dilution), anti-FLAG M2 mouse monoclonal antibody (Sigma; 1:1000 dilution), Alexa Fluor® 488-conjugated goat anti-rabbit IgG (Invitrogen), anti-myosin heavy chain mouse monoclonal antibody (R&D Systems; clone MF20, 1:100 dilution), Alexa Fluor® 488-conjugated goat anti-mouse IgG (Invitrogen) and Alexa Fluor® 555-conjugated goat anti-rat antibody (Invitrogen). Fluorescence images were acquired with an LSM 5 exciter laser-scanning microscope (Carl Zeiss). At least 200 transfected myotubes were analysed for each clone, and representative photos are shown in the Figures. To quantify the efficiency of targeting of each construct to JMs, the percentage of myotubes showing LTCC clustering relative to the total number of LTCC-expressing myotubes was assessed. For each construct, three to five coverslips (40–205 myotubes/coverslip) were tested to obtain means±S.E.M. A myotube was classified as ‘clustered’ if any region of the myotube displayed the characteristic clustered immunolabelling pattern.
To verify that under non-permeabilized conditions the anti-HA antibody recognized only membrane-targeted proteins, permeabilized and non-permeabilized myotubes were stained in parallel (Supplementary Figure S1 at http://www.BiochemJ.org/bj/448/bj4480221add.htm). The fluorescence signal in the ER/SR under the permeabilized condition was very intense compared with that in the plasma membrane under the non-permeabilized condition. No non-permeabilized myotubes exhibited such an intense ER/SR signal, indicating that under non-permeabilized conditions the anti-HA antibody detected exclusively membrane-targeted proteins. Under the permeabilized condition, a clustered distribution was observed in ~20% of myotubes transfected with HA–α1C and in 57% of myotubes transfected with GFP–α1C (Table 1). Apparently different high expression levels of the two constructs in the ER/SR system mask the label of membrane-expressed channels to different degrees.
|Construct||Percentage of cells with clusters (means±S.E.M.)||Total cell number|
|Construct||Percentage of cells with clusters (means±S.E.M.)||Total cell number|
Using a co-localization function of the ZEN program (Carl Zeiss), the co-localization co-efficient of HA-tagged channels with RyR signals was calculated as PHA+RyR/PHAtotal, where PHA+RyR and PHAtotal are the number of pixels containing both HA and RyR signals and those containing HA signals respectively. All of the images were acquired with the same microscope and camera setting, and individual myotubes were selected by a ROI (region of interest) tool for calculation.
GLT myotubes cultured on carbon-coated coverslips (113 mm2) were incubated with 5 μM Fluo 4/AM (acetoxymethyl ester; Dojindo) plus 0.01% Cremophore EL (Sigma) and 0.02% BSA (Sigma) in serum-free DMEM for 45 min at 37°C followed by de-esterification. Myotubes were superfused with modified Tyrode solution [136.5 mM NaCl, 5.4 mM KCl, 1.8 mM CaCl2, 0.53 mM MgCl2, 5.5 mM Hepes and 5.5 mM glucose (pH 7.4)] at room temperature and paced with 1-ms pulses of 50 V at 0.3 Hz across the 20-mm incubation chamber. Fluorescence images were acquired with an LSM 7 LIVE laser-scanning microscope with a 20×/0.8 plan apochromatic objective (Carl Zeiss). Fluo 4 was excited by 488 nm light, and emission light passed through a high-pass filter of 495 nm and was imaged with a CCD (charge-coupled device) camera. Each image was taken with 128 pixels×128 pixels every 2.8 ms. The time course of Ca2+ transients was obtained from the fluorescence change in individual myotubes selected by the ROI tool. To block Ca2+ influx through the plasma membrane, 0.5 mM Cd2+ and 0.1 mM La3+ were added. Application of 6 mM caffeine proved the functionality of SR Ca2+ release.
For quantification of the numbers of channel-expressing myotubes, cultures on coverslips were stained with anti-HA or anti-GFP, and anti-myosin heavy chain (differentiation marker) antibodies in permeabilized conditions (see the section on Immunofluorescence staining above). Low magnification (10× objective) fluorescence images were acquired at five randomly selected areas for each specimen, and the number of myotubes expressing HA or GFP, and myosin heavy chain were quantified.
Results are means±S.E.M. Statistical significance was evaluated with Student's unpaired t test. For multiple comparisons of data, ANOVA with Bonferroni's test was used. P<0.05 was considered statistically significant.
The C-terminal domain is necessary for JM targeting of the α1C subunit
Previous studies have revealed that transiently transfected α1S subunits of skeletal muscle LTCCs, but not α1A subunits of neuronal P/Q-type Ca2+ channels, were localized to JMs in GLT myotubes [14,15]. These studies further showed that heterologously expressed α1C subunits of cardiac LTCCs were also targeted to the JM and supported CICR in GLT myotubes [17,25]. Thus, by using the same muscle expression system, in the present study we wanted to identify the critical motif of α1C subunits for JM targeting. We made several chimaeras and transiently expressed them in GLT cells (Figure 1A). The N-terminus (α1CNTA), I–II loop (α1CI–IIA), III–IV loop (α1CIII–IVA) and C-terminus (α1CCTA) of α1C were substituted with corresponding regions of α1A. Since the II–III loop of α1A is >3 times longer than that of α1C, a chimaera with the II–III loop of house fly (Musca domestica) α1 (α1CII–IIIM) [14,26] was used instead to examine the role of the II–III loop chimaera in JM targeting. House fly α1 subunits were not targeted to JMs in GLT myotubes (results not shown). All chimaeras were fused with GFP at the N-terminus of α1C.
Localization of the JM-targeting motif in the C-terminus of α1C
The localization of the GFP–α1C and related chimaeras relative to the JM identified by the anti-RyR antibody was assessed by immunofluorescence labelling. GFP–α1C showed a clustered distribution and was co-localized with RyRs in the JM as reported previously  (Figure 1B). The quantitative results of the clustering assay of all of the channels tested in the present study are shown in Table 1. GFP–α1CNTA, GFP–α1CI–IIA, GFP–α1CII–IIIM and GFP–α1CIII–IVA exhibited a localization virtually identical with that of GFP–α1C, whereas GFP–α1CCTA was not clustered or co-localized with RyRs (Figure 1C). The distribution of GFP–α1CCTA was very similar to that of α1A, which was mainly localized in the ER/SR system and did not form clusters with RyRs (Figure 1B) [13,14]. These results indicated that the C-terminus of α1C is necessary for the JM targeting of α1C. Although the JM targeting rate of GFP–α1CIII–IVA was reduced to ~24% of that of GFP–α1C, a considerable number of cells still exhibited a clustered distribution of the chimaera (Table 1), indicating that the III–IV loop is not indispensable for the JM targeting of α1C.
The proximal C-terminus is necessary for JM targeting of α1C
To determine the sequences within the C-terminus critical for JM targeting, we next tested C-terminal truncation mutants of α1C and α1C–α1A chimaeras. It is of note that the C-terminus of α1C is also important for plasma membrane expression of α1C . To assess JM targeting of chimaera proteins that were properly trafficked to the plasma membrane, we used an α1C construct with an HA epitope inserted into the extracellular loop between segments S5 and S6 in domain II of α1C (HA–α1C) [23,24] for the construction of these mutants. Membrane-expressed channels were detected by anti-HA antibody labelling without prior membrane permeabilization. Successful expression of all of HA-tagged channels was confirmed by Western blotting (Supplementary Figure S2 at http://www.BiochemJ.org/bj/448/bj4480221add.htm).
We analysed JM targeting with the following deletion mutants (Figure 2A): an HA–α1CΔ1821 mutant lacking the distal part of the C-terminus, which is known to be subject to proteolytic cleavage in cardiac myocytes ; an HA–α1CΔ1733 mutant further lacking the region corresponding to the JM-targeting motif of α1S ; an HA–α1CΔ1666 mutant lacking the whole C-terminus distal to the IQ motif; and an HA–α1CΔ1624 mutant lacking the membrane-targeting motif identified previously . Immunofluorescence analysis of transfected myotubes revealed the clustered distributions of HA–α1C, HA–α1CΔ1821 and HA–α1CΔ1733 (>75% of cells, n≥200) (Figures 2B and 2C, and Table 1). In contrast, HA–α1CΔ1666 was diffusely distributed in the plasma membrane (Figure 2B), suggesting that amino acids between 1666 and 1733, but not those distal to 1733, are necessary for the JM targeting of α1C. The co-localization co-efficient of HA–α1CΔ1666 with RyR was significantly decreased compared with HA–α1C (Figure 2D). No fluorescence signal was observed in myotubes transfected with HA–α1CΔ1624 (Supplementary Figure S1).
Identification of the JM-targeting motif in the proximal part of the C-terminus of α1C
Next we analysed three α1C-based chimaeras in which partially overlapping sequences of the proximal C-terminus containing the critical domain between 1666 and 1733 were replaced with the corresponding sequences of α1A (Figure 2A). Substitution of amino acids 1507–1650 or 1624–1665 did not affect channel clustering, but chimaera 1666–1733 failed to cluster in JMs (Figures 2C and 2D). Thus amino acid residues 1666–1733, but not 1507–1665 or 1734–2171, in the C-terminus of α1C are necessary for proper JM targeting.
To further narrow down the JM-targeting motif, we prepared four partially overlapping α1C–α1A chimaeras within this critical region of the C-terminus (substitutions of 1666–1676, 1666–1688, 1689–1733 and 1709–1733) (Figure 2A). Of these, the two chimaeras with substitutions of the peripherally located sequences HA–α1C1666–1676A and HA–α1C1709–1733A were correctly clustered in myotubes (Figures 2C and 2D). In contrast both chimaeras with substitutions of the central sequences HA–α1C1666–1688A or HA–α1C1689–1733A failed to cluster in JMs. These results limit the sequences in the C-terminus of α1C necessary for JM targeting to the 31 amino acid residues between 1677 and 1708.
To determine critical domains within this sequence we performed systematic alanine scanning between 1677 and 1708. Within this sequence four consecutive amino acid residues at a time were sequentially substituted with alanine residues (Figure 3). The numbers of the cells with α1C clusters was dramatically decreased with the AAAA mutation of 1681–1684 (12%) and 1685–1688 (7%). A small, but still significant, decrease in clustering rates was also observed with AAAA mutations in 1689–1692 (78%), 1693–1696 (60%) and 1697–1700 (24%). In contrast, the clustering rates of AAAA mutants 1677–1680, 1701–1705 and 1705–1708 were not significantly different from those of WT α1C (Figure 3B). A significant decrease in the co-localization co-efficient was observed with the AAAA mutations in 1681–1684, 1685–1688 and 1697–1700, but not 1689–1692 and 1693–1696 (Figure 3C). Taken together these analyses demonstrate that the 20 amino acid residues 1681–1700 are necessary for the proper JM targeting of α1C and that, within this sequence, two motifs (L1681QAGLRTL1688 and P1693EIRRAIS1700) are of particular importance for the targeting mechanism.
Alanine substitution of the JM-targeting motif in the C-terminus of α1C
The C-terminus of α1C is not sufficient to confer the JM-targeting property of α1C to α1A or CD8
To assess whether the C-terminal-targeting motif of α1C can confer a JM-targeting property to neuronal Ca2+ channels or unrelated transmembrane proteins, we generated the following three chimaeras: the intracellular C-terminal domain of CD8 was replaced with the entire C-terminus of α1C (CD8–α1CCT); the JM-targeting motif of α1C was substituted for the corresponding sequences of GFP–α1A (GFP–α1AJMTC); and the entire C-terminus of GFP–α1A was swapped with that of α1C (GFP–α1ACTC). However, none of these three chimaeras was clustered in JMs (n>200) (Figure 4). These results indicate that, although the C-terminus of α1C is necessary for JM targeting, it is not sufficient.
The C-terminus of α1C is not sufficient for JM targeting
The effect of transient expression of proteins interacting with the motif necessary for JM targeting of α1C
In cardiac myocytes the DCT of α1C is truncated by proteolytic processing and the cleaved DCT re-associates with the PCRD (proximal C-terminal regulatory domain) (amino acid residues 1694–1700) of the truncated channel [18–20]. The interaction of DCT and PCRD auto-inhibits LTCCs, and PKA (protein kinase A) augments LTCC currents by removing this inhibition. This is a vital regulation of LTCCs in the flight-or-fight response . Interestingly the distal half of the motif necessary for the JM targeting of α1C overlaps with the PCRD. Thus we assessed the effects of co-expression of DCT (α1C1821–2171) on JM targeting of HA–α1CΔ1821 and HA–α1C to examine whether DCT inhibits JM targeting. However, co-expressed DCT did not reduce JM targeting of either one of these proteins (Figure 5), indicating that the JM targeting of α1C did not disrupt the responsiveness of LTCCs to PKA.
Effect of co-expression of the distal C-terminal domain on the JM-targeting motif of α1C
The motif (1681–1700) necessary for JM targeting of α1C is immediately distal to the IQ-motif known to be the Ca2+/CaM-interaction site [27,28]. Also the Ca2+-binding protein sorcin is known to bind to amino acid residues 1622–1748 in the C-terminus of α1C . Thus we examined the possibility of whether co-expression of CaM, its Ca2+-insensitive mutant (CaM1234) or sorcin interferes with the JM targeting of α1C. However, co-expression of none of these proteins significantly altered the clustering rates of HA–α1C (Supplementary Figure S3 at http://www.BiochemJ.org/bj/448/bj4480221add.htm).
Effect of mutations within the JM-targeting motif on EC coupling of α1C
Previous studies have shown that transient expression of α1C, but not α1A, restored cardiac-type CICR in GLT myotubes [14,25]. Thus we examined whether JM targeting contributed to the restoration of CICR by measuring action-potential-induced Ca2+ transients of myotubes transfected with HA–α1C, HA–α1C1685AAAA1688 or HA–α1C1666–1733A. Transient expression of HA–α1C restored action-potential-elicited Ca2+ transients (Figure 6A). Complete block of the Ca2+ transients by dantrolene (100 μM), an inhibitor of RyRs (results not shown), identified the SR as the major Ca2+ source. Cd2+/La3+ abolished the Ca2+ transients, whereas caffeine subsequently applied induced a large Ca2+ transient, indicating that the action-potential-induced SR Ca2+ release required LTCC currents. Thus HA–α1C restored CICR. Interestingly, transfection with two constructs which failed to show JM targeting, HA–α1C1685AAAA1688 and HA–α1C1666-1733A, also restored Ca2+ transients (Figure 6A), indicating that these mutations did not disturb channel activity. Figure 6(B) shows that these constructs, however, were significantly less efficient than HA–α1C in restoring the Ca2+ transients. Nevertheless, there was no significant difference between the mutant constructs and WT in the time-to-peak (HA–α1C, 0.389±0.037 s; HA–α1C1685AAAA1688, 0.389±0.036 s; and HA–α1C1666-1733A, 0.327±0.035 s) or the peak amplitudes (HA–α1C, 0.381±0.028; HA–α1C1685AAAA1688, 0.429±0.055; and HA–α1C1666–1733A, 0.445±0.061) of the restored Ca2+ transients. We also tested GFP–α1A- or GFP–α1ACTC-transfected myotubes. However, with these constructs no myotubes showed action-potential-elicited Ca2+ transients (Figure 6B). The number of channel-expressing myotubes in each dish was not significantly different among the groups (Figure 6C and Supplementary Figure S4 at http://www.BiochemJ.org/bj/448/bj4480221add.htm).
Action-potential-elicited Ca2+ transient in GLT myotubes expressing WT and mutant α1C
In the present study, we identified amino acids residues 1681–1700 in the proximal C-terminus of α1C as the motif necessary for the JM targeting of α1C. Systematic alanine-scanning in this sequence further showed that two separate motifs (L1681QAGLRTL1688 and P1693EIRRAIS1700) are of particular importance for the targeting mechanism. Using HEK (human embryonic kidney) tsA201 cells, Gao et al.  showed that deletion of amino acid residues 1623–1666 of the C-terminus completely disrupted the plasma membrane localization and clustering of α1C, and thus suggested that the sequence is a membrane-targeting motif. This is consistent with our finding that HA–α1CΔ1666, but not HA–α1CΔ1624, was expressed in the plasma membrane in GLT myotubes (Figure 2 and Supplementary Figure S1). However, Gao et al.  showed that the deletion of amino acids 1668–1733 of α1C harbouring the JM-targeting motif resulted in an only slightly decreased membrane-targeting rate (90% in WT and 72% in α1CΔ1668–1733) and did not disrupt the clustering of α1C . Thus HEK tsA201 cells share the common mechanism for membrane targeting of α1C with GLT cells, but their clustering mechanism does not resemble JM targeting in muscle cells. In contrast, GLT myotubes support JM targeting of α1S as well as α1C and are therefore suitable for analysing the JM-targeting mechanism of α1C . Using this system we, for the first time, identified the JM-targeting motif of α1C and demonstrated that it is distinct from the previously identified membrane-targeting motif.
Interestingly the JM-targeting motif of α1C is also different from that of α1S . The JM-targeting motif of α1S corresponds to amino acid residues 1735–1791 of α1C. Since we found that the HA–α1CΔ1733 mutant was normally clustered (Figure 2), amino acid residues 1735–1791 are not necessary for the JM targeting of α1C. On the other hand, the JM-targeting motif of α1C (amino acid residues 1681–1700) corresponds to residues 1555–1575 of α1S. In our previous study, substitution of this region of α1S with the corresponding part of α1A did not affect the JM-targeting rate of α1S . Thus, although both α1S and α1C are similarly targeted to JM in GLT myotubes, their JM-targeting mechanism appears to be different.
Expression of α1C restored action-potential-induced Ca2+ transients triggered by CICR. With two α1C channels possessing mutations in the JM-targeting motif, the probability of CICR restoration was significantly decreased compared with WT α1C channels (Figure 6B). However, the reduction in the number of responsive cells by these mutations (i.e. ~60%) was much lower than that in their respective JM-targeting rate [i.e. ~90% in HA–α1C1685AAAA1688 and 100% in HA–α1C1666–1733A, GFP–α1A and GFP–α1ACTC (Table 1)]. This indicates that some of the α1C- but not α1A-based channels are still functionally coupled with RyRs, even though JM targeting was below detectablility. Apparently, JM targeting is not an all-or-none process, and Ca2+ recording is somewhat more sensitive in detecting channels coupled to RyRs than immunofluorescence analysis. This is to be expected if individual rather than clustered channels are coupled to RyRs. Immunocytochemical studies showed that the clustering rates of the tested mutants were paralleled with the pixel-based LTCC–RyR co-localization index (Figures 2 and 3), suggesting a good correlation between the channel clustering and co-localization with RyRs. However, we could not assess the exact percentage of channel clusters co-localized with RyRs. Therefore we cannot exclude the possibility that a minor fraction of non-clustered channels still functionally couples to RyRs.
According to previous structural modelling of the C-terminus of α1C, proximal (L1681QAGLRTL1688) and distal (P1693EIRRAIS1700) parts of the JM-targeting motif form two adjacent α-helices connected by a short loop . Interestingly the distal helix of the motif overlaps with PCRD (amino acid residues 1694–1700) . The DCRD (DCT regulatory domain) of the cleaved DCT reassociates with the PCRD of the truncated channel to auto-inhibit LTCC currents [10,20]. This auto-inhibition plays a pivotal role in the response of LTCCs to β-adrenergic stimulation . Fuller et al.  also showed that in HEK cells Ba2+ currents of the truncated channel (Δ1800) were strongly inhibited by co-transfection of DCT at a molar ratio of DCT/channel higher than 0.75 . We co-expressed DCT at various molar ratios with α1C to examine whether the interaction of PCRD and DCRD might compete with the JM targeting of α1C. However, the co-expression did not affect the JM-targeting rates of either full-length α1C or truncated α1C (α1CΔ1821) under any of the conditions tested (Figure 5). Our alanine scanning indicated that the proximal helix of the JM-targeting motif is more important for JM targeting than the distal helix (Figure 3). On the other hand, DCRD interacts with Arg1696 and Arg1697 in the distal helix of the PCRD . Therefore putative protein–protein interactions involving the JM-targeting motif and the PCRD–DCRD interaction may be mediated by closely adjacent, but still distinct, amino acids in the proximal C-terminus of α1C subunit. These results indicate that the JM targeting of α1C and the responsiveness of LTCCs to PKA may not be mutually exclusive. This fact is important in light of the essential role of β-adrenergic augmentation of LTCC currents within the cardiac dyad and peripheral junctions in the flight-or-fight response of animals.
Several proteins bind directly to the C-terminus of α1C. In cardiac myocytes these include AKAP15 (A-kinase-anchoring protein 15) , PP2A (protein phosphatase 2A) [31,32], PP2B (protein phosphatase 2B, also known as calcineurin) , CaM [27,28,34–37] and sorcin [29,38,39]. Among these, AKAP15, PP2A and PP2B interact with the cleaved DCT. Because the deletion mutants lacking the DCT (α1CΔ1821 and α1CΔ1733) were normally targeted into JMs, these proteins are probably not involved in JM targeting of α1C. CaM and sorcin bind to the C-terminal segment close to the JM-targeting motif of α1C. However, co-expression of CaM, CaM1234 and sorcin with GFP–α1C did not alter the JM targeting of α1C (Supplementary Figure S3), indicating that excess CaM and sorcin does not affect the function of the JM-targeting motif of α1C. On the other hand, it has been reported that association of α1C with CaM is involved in membrane trafficking of α1C in hippocampal neurons and HEK-293 cells . Thus it is possible that endogenous CaM also participates in the JM targeting of α1C in myotubes. Future studies, such as knockdown of CaM and/or sorcin, will be needed to clarify the requirements of endogenous CaM and sorcin for the JM targeting of α1C.
The most straightforward interpretation of our results would be that the JM-targeting motif significantly contributes to a protein–protein interaction that supports the JM targeting of α1C. However, addition of the JM-targeting motif to CD8 or α1A was not sufficient for the targeting of these proteins to the JM (Figure 4). This may indicate that the motif identified in the present study is not the actual protein–protein-binding site. Alternatively, the insufficiency of the motif in JM targeting may indicate that parts of α1C other than the identified targeting motif contribute to JM targeting. For instance, mutation or deletion of this motif might disrupt the structure and/or folding of other parts of α1C meditating the JM targeting and thus secondarily impairing the targeting of α1C. However, the results of the present study on the C-terminal deletion mutants and chimaeras (Figure 2) exclude the possibility that parts of the C-terminus other than the JM-targeting motif are necessary for the targeting. Furthermore it is unlikely that the N-terminus, I–II, II–III or III–IV loop is required for JM targeting because their substitution with the corresponding sequences of α1A did not disrupt the JM targeting (Figure 1). Apparently JM targeting is a complex mechanism that necessitates the overall structure of the LTCC channel family. Indeed, a hemi-α1S channel composed of III–IV domains and the C-terminus, including the JM-targeting motif by itself, was not targeted to the JM, but JM targeting was restored when it was co-expressed with the corresponding I–II hemichannel .
To summarize, the results of the present study, for the first time, has identified the sequence motif in the proximal C-terminus of α1C that is necessary for the JM targeting of this LTCC in muscle cells. The mutations in the JM-targeting motif of α1C significantly impaired the efficiency of α1C-induced restoration of EC coupling in GLT myotubes. Because this motif is necessary, but not sufficient, for JM targeting of α1C, further studies will be necessary to identify additional structural elements that contribute to the targeting function. Moreover it is important to identify the putative binding partner of the motif to reveal the molecular mechanism underlying efficient EC coupling of cardiac myocytes and its derangement in heart failure.
A-kinase-anchoring protein 15
Ca2+-induced Ca2+ release
distal C-terminus regulatory domain
Dulbecco’s modified Eagle’s medium
fetal bovine serum
green fluorescent protein
human embryonic kidney
L-type calcium channel
proximal C-terminal regulatory domain
protein kinase A
protein phosphatase 2A
protein phosphatase 2B
region of interest
voltage-dependent Ca2+ release
Tsutomu Nakada, Bernhard Flucher and MitsuhikoYamada designed the research, analysed the data and wrote the paper. Tsutomu Nakada, Toshihide Kashihara, Xiaona Sheng, Toshihide Shibazaki, Miwa Horiuchi-Hirose, Simmon Gomi and Masamichi Hirose performed the experiments.
We are grateful to Professor William Catterall and Professor Manfred Grabner for providing the cDNA of α1 subunits. We are grateful to Dr Nagomi Kurebayashi (Juntendo University School of Medicine, Tokyo, Japan) for helpful advice on cytosolic Ca2+ measurements. We are grateful to Ms Reiko Sakai for her secretarial assistance.
This work was supported by the Ministry of Education, Culture, Sport, Science and Technology of Japan (MEXT) [Grant-in-Aid for Scientific Research, grant number 22790206], the Shinshu Foundation for Promotion of Medical Sciences [Medical Scientific Research grant], the Japan Research Promotion Society for Cardiovascular Diseases [a Sakakibara Memorial Research Grant] (all to T.N.), the Japan Society for the Promotion of Sciences [Grant-in-Aid for Scientific Research, grant number 21590275] (to M.Y.), and the Austrian Science Fund [grant number P23479-B19] to B.E.F.