HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) All Versions of this Article: 541/2/435    most recent 2002.018515v1 Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Colecraft, H. M. Articles by Yue, D. T. Search for Related Content PubMed PubMed Citation Articles by Colecraft, H. M. Articles by Yue, D. T.

Journal of Physiology (2002), 541.2, pp. 435-452
© Copyright 2002 The Physiological Society
DOI: 10.1113/jphysiol.2002.018515

Novel functional properties of Ca²⁺ channel subunits revealed by their expression in adult rat heart cells

Henry M. Colecraft *, Badr Alseikhan *, Shoji X. Takahashi *, Dipayan Chaudhuri *†, Scott Mittman ‡, Vasan Yegnasubramanian *, Rebecca S. Alvania †, David C. Johns §, Eduardo Marbán ¶ and David T. Yue *†

	ABSTRACT

Top Abstract Introduction Methods Results Discussion References

Recombinant adenoviruses were used to overexpress green fluorescent protein (GFP)-fused auxiliary Ca²⁺ channel subunits (₁-₄) in cultured adult rat heart cells, to explore new dimensions of subunit functions in vivo. Distinct -GFP subunits distributed differentially between the surface sarcolemma, transverse elements, and nucleus in single heart cells. All -GFP subunits increased the native cardiac whole-cell L-type Ca²⁺ channel current density, but produced distinctive effects on channel inactivation kinetics. The degree of enhancement of whole-cell current density was non-uniform between subunits, with a rank order of potency _{2a 4} > _1b > ₃. For each subunit, the increase in L-type current density was accompanied by a correlative increase in the maximal gating charge (Q_max) moved with depolarization. However, subunits produced characteristic effects on single L-type channel gating, resulting in divergent effects on channel open probability (P_o). Quantitative analysis and modelling of single-channel data provided a kinetic signature for each channel type. Spurred on by ambiguities regarding the molecular identity of the actual endogenous cardiac L-type channel subunit, we cloned a new rat ₂ splice variant, _2b, from heart using 5' rapid amplification of cDNA ends (RACE) PCR. By contrast with _2a, expression of _2b in heart cells yielded channels with a microscopic gating signature virtually identical to that of native unmodified channels. Our results provide novel insights into subunit functions that are unattainable in traditional heterologous expression studies, and also provide new perspectives on the molecular identity of the subunit component of cardiac L-type Ca²⁺ channels. Overall, the work establishes a powerful experimental paradigm to explore novel functions of ion channel subunits in their native environments.

(Received 7 February 2002; accepted 11 March 2002)
Corresponding author H. M. Colecraft: Department of Biomedical Engineering, Johns Hopkins University School of Medicine, Traylor Building, Room 710A, 720 Rutland Avenue, Baltimore, MD 21205, USA. Email: hcolecra{at}bme.jhu.edu

	INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

Ca²⁺ influx through cardiac L-type Ca²⁺ channels is critical for controlling both cardiac excitability and excitation- contraction (E-C) coupling. L-type channels are hetero-multimers comprised minimally of distinct _1C, , and ₂ subunits (Walker & De Waard, 1998; Catterall, 2000). The pore-forming _1C subunit contains the voltage sensor and is the major determinant of channel identity and pharmacology, but its expression and functional properties are importantly influenced by auxiliary subunits. In particular, Ca²⁺ channel subunits are powerful modulators of ₁ subunit function. Four subunit isoforms (termed ₁-₄, with multiple splice variants) encoded by distinct genes have been identified (Ruth et al. 1989; Pragnell et al. 1991; Perez-Reyes et al. 1992; Castellano et al. 1993a, b), and their functions extensively characterized in heterologous expression systems. The bulk of such experiments indicate that Ca²⁺ channel subunits increase membrane expression of ₁ subunits (Chien et al. 1995; Brice et al. 1997; Gao et al. 1999; Yamaguchi et al. 2000), normalize channel activation kinetics (Lacerda et al. 1991; Singer et al. 1991), cause hyperpolarizing shifts in the voltage dependence of channel activation (Perez-Reyes et al. 1992; Castellano et al. 1993a, b; Jones et al. 1998), and impart distinctive inactivation properties to the whole-cell current (Olcese et al. 1994; De Waard & Campbell, 1995; Wei et al. 2000).

Much less is known about the potential of auxiliary subunits to tune Ca²⁺ channel function in the context of native cells, although there are provocative hints that they may modulate cell signalling more profoundly than suggested by heterologous expression system studies. For example, distinct subunits reconstitute skeletal E-C coupling with different gains when expressed in -null skeletal myotubes (Beurg et al. 1999); and subunit identity is a major factor in determining the differential distribution of P/Q channels to either the apical or basolateral membrane in a polarized epithelial cell line (Brice & Dolphin, 1999). Recently, to explore new dimensions of subunit functions in heart, we utilized a plasmid-based adenoviral component system to transfect adult heart cells with exogenous subunits (Wei et al. 2000). Although the low transfection efficiencies achievable with this method limited the potential scope of this study, two important findings were presented. First, despite the hetero-multimeric nature of cardiac L-type channels, expression of either _2a or ₃ subunits alone was sufficient to enhance the L-type channel current density in adult heart cells. This finding is especially relevant because the expression levels of cardiac Ca²⁺ channel subunits are decreased in heart failure and atrial fibrillation (Schroder et al. 1998; Hullin et al. 1999; Grammer et al. 2001), and could possibly contribute to the decreased L-type Ca²⁺ current density observed in some studies of these pathophysiological conditions (Ming et al. 1994; Mukherjee et al. 1998; Van Wagoner et al. 1999; Boixel et al. 2001; He et al. 2001). Several important but unresolved questions relating to the regulation of native cardiac L-type current density by exogenous subunits remain. Do all subunits increase current density to the same extent? Is the increased current density mediated through an increase in the number of functional channels (N)? Alternatively, is it mediated mostly through an increase in channel open probability (P_o). Do different subunits affect these parameters in a similar manner? Resolution of these questions is crucial not only for assessing the rate-limiting role of subunits in expression of functional Ca²⁺ channel in native cells, but also for weighing the potential therapeutic benefits of manipulating expression levels of distinct subunits in vivo. A second critical finding from our previous study was that the putative cardiac subunit, _2a, reconstituted currents with markedly different inactivation kinetics compared with control. This suggested that the identity of the actual cardiac Ca²⁺ channel subunit remains unknown.

Based on these considerations, our aims here were to: (1) construct recombinant adenoviruses that permitted robust expression of distinct Ca²⁺ channel subunits in adult rat heart cells with a high efficiency, (2) compare the relative capabilities of different subunits to enhance the endogenous cardiac L-type Ca²⁺ channel current density, (3) determine the relative contributions of changes in N and P_o to the observed increase in L-type current density, and (4) clarify the molecular identity of the endogenous cardiac L-type Ca²⁺ channel subunit.

	METHODS

Top Abstract Introduction Methods Results Discussion References

Construction of shuttle plasmid vectors

The expression cassette of the vector pGFP-IRES (Johns et al. 1997) was cloned into the multiple cloning site of the adenovirus shuttle vector pAdLox (Hardy et al. 1997), creating the vector pAdCGI. The vector pAdCMV EGFP-N3 was constructed by transferring the Nhe1/Mun1 expression cassette of pEGFP-N3 (Clontech, Palo Alto, CA, USA) into pAdCGI. Coding sequences for Ca²⁺ channel subunits (_1b, Pragnell et al. 1991; _2a, Perez-Reyes et al. 1992; ₃, Castellano et al. 1993b; ₄, Castellano et al. 1993a) were amplified by PCR, and fused in-frame and upstream of EGFP in the adenovirus shuttle vector pAdCMV EGFP-N3. We chose to fuse EGFP to the C-termini of subunits because our previous data indicated that this configuration caused significantly less functional perturbations than fusions to the N-termini (Wei et al. 2000). To construct pAdCMV _1b-GFP, the entire coding sequence of _1b in pGW1, up to and including the penultimate codon, was PCR amplified using the following upstream and downstream primers, respectively,

AGAATTCGACCGGAGGATCCTCTCCATGGTCCAG and

AGGATCCGCGGATGTAGACGCCTTGTCC.

The resulting PCR fragment was ligated into pAdCMV EGFP-N3 by EcoRI and BamHI sites, yielding pAdCMV _1b-GFP. To construct pAdCMV _2a-GFP, the entire coding sequence of _2a in pGW1, up to and including the penultimate codon, was PCR amplified using the following upstream and downstream primers, respectively,

AAGATCTATGGACAGTGGCCGTGAC and

TCTGCAGATTGGCGGATGTATACATCCCT.

The PCR fragment was ligated into pEGFP-N3 by BglII and PstI sites, followed by excision and ligation of the _2a-GFP cassette into pAdCMV EGFP-N3 by NheI and XbaI sites, yielding pAdCMV _2a-GFP. To construct pAdCMV ₃-GFP, we PCR amplified approximately the final third of the coding sequence of ₃ in pGW, up to and including the penultimate codon, using the following upstream and downstream primers, respectively,

ACAGCTAGCCAAGACCTCACTG and

TGGATCCGTAGCTGTCCTTAGGCCAAG.

The PCR fragment was ligated into ₃ in pGW1 by NheI and BamHI sites, and the resulting ₃ expression cassette ligated into pAdCMV EGFP-N3 using HindIII and BamHI sites, to yield pAdCMV ₃-GFP. To construct pAdCMV ₄-GFP, we PCR amplified the entire coding sequence of ₄ in pMT2, up to and including the penultimate codon, using the following upstream and downstream primers, respectively,

AGCTAGCATGCGGCGCCTCTGAACCATGTCG and

AGGATCCAAGCCTATGTCGGGAGTCATGG.

The PCR fragment was ligated into pAdCMV EGFP-N3 by NheI and BamHI sites, yielding pAdCMV ₄-GFP. Pfu polymerase (Stratagene, La Jolla, CA, USA) was used in all PCR reactions to increase fidelity, and all amplified sequences were verified by sequencing.

Construction of adenoviral vectors

Adenoviral vectors were generated by Cre-lox recombination as previously described (Hardy et al. 1997; Johns et al. 1999). Briefly, 60 mm diameter dishes containing CRE 8 cells (Hardy et al. 1997) at ~20 % confluency were co-transfected with 3 µg pAdCMV -GFP shuttle vector (described above) and 3 µg 5 viral DNA (Hardy et al. 1997), using Fugene (Roche Molecular Biochemicals, Indianapolis, IN, USA). Following observation of cytopathic effects (CPE) usually after 7-10 days, cells were harvested and subjected to three freeze-thaw cycles, followed by centrifugation to remove cellular debris, and the resulting supernatant (2 ml) used to infect a 10 cm dish of 90 % confluent CRE8 cells. Following observation of CPEs after 2-3 days, cell supernatants were used to reinfect a new plate of CRE 8 cells. This procedure was repeated two or three more times, followed by viral expansion and purification as previously described (Johns et al. 1995; Wei et al. 2000). Viral particle numbers determined by absorbance at 260 nm were in the order of 10¹¹-10¹² particles ml^-1, with a ~30 : 1 particle to plaque-forming unit ratio.

Cell culture and transfection

Primary cultures of adult rat heart cells were prepared as previously described (Wei et al. 2000). Adult male Sprague-Dawley rats were killed with an overdose of sodium pentobarbitone in accordance with the guidelines of the Johns Hopkins University Animal Care and Use Committee. Hearts were excised and ventricular myocytes isolated by enzymatic digestion using a Langendorrf perfusion apparatus. Healthy, rod-shaped cardiomyocytes were cultured on laminin-coated coverslips at a density of ~50 000 cells ml^-1, in 35 mm tissue culture dishes. For biochemical studies, cells were plated in 60 mm dishes at a density of ~200 000 cells ml^-1. Cells were initially maintained in Medium 199 (Life Technologies) supplemented with 5 mM carnitine, 5 mM creatine, 5 mM taurine, 100 µg ml^-1 penicillin-streptomyocin and 10 % fetal bovine serum to promote attachment to the coverslips. After 2-3 h, the culture medium was switched to serum-free Medium 199, otherwise supplemented as described above. Cultures were equilibrated with 5 % CO₂ and 95 % air at 37 °C in a water-jacketed CO₂ incubator. Cells were transfected ~3 h after initial plating by adding 5-20 µl of the appropriate virus stock (~10¹¹-10¹² viral particles ml^-1) directly onto the cells, in a final volume of 1-2 ml. Virus-containing medium was aspirated after 4 h, or, in some cases, after overnight incubation, and replaced with fresh serum-free media. In all cases, robust transgene expression was explicitly verified by monitoring GFP fluorescence ~24 h after infection.

Low passage number human embryonic kidney (HEK 293) cell cultures (< 20 passages), maintained as previously described (Brody et al.1997), were transfected with 8 µg each of cDNAs encoding Ca²⁺ channel subunits (_1C, Wei et al. 1991; rat _2a, Perez-Reyes et al. 1992; rat ₂, Tomlinson et al. 1993).

Western blotting

Cultured heart cells were harvested 18-48 h after infection, as indicated, washed with ice-cold phosphate buffered saline (PBS), and resuspended in 0.3 ml lysis buffer containing 1 % Triton X-100 (Sigma, St Louis, MO, USA), 20 mM Tris-Cl (pH 7.4), and complete protease inhibitor cocktail (Roche Molecular Biochemicals). Cell lysis was completed with short bursts of mild sonication (Sonic Dismembrator Model 50; Fisher Scientific), and protein concentrations determined by the bicinchoninic acid assay system (Pierce Chemical Co., Rockford, IL, USA) using bovine serum albumin as a standard. Cell lysates (30 µg protein) were heated in SDS sample buffer and resolved on 10 % SDS-polyacrylamide gels (Laemmli, 1970). Molecular weights were determined using prestained markers (Kaleidoscope, Bio-Rad Labs, Hercules, CA, USA). Proteins were transferred to nitrocellulose membranes (Amersham Life Sciences), and detected using anti-GFP monoclonal antibody (Covance Research Products, Richmond, CA, USA) or anti-_GEN polyclonal antibody (Chien et al. 1996), followed by horseradish peroxidase-conjugated anti-mouse (Sigma) or anti-rabbit (Amersham) secondary antibody, respectively. Immunoreactive proteins were visualized by enhanced chemiluminescence (ECL, Amersham, UK).

5' RACE PCR

The whole heart was excised from an adult male Sprague-Dawley rat, and mRNA extracted using the MessageMaker mRNA Isolation System (Gibco Invitrogen) according to manufacturer's instructions. First strand cDNA synthesis and subsequent 5' RACE PCR were performed using the SMART RACE cDNA amplification kit (Clontech) according to manufacturer's instructions. The 3' primer for RACE PCR (5'-TCATTGGCGGATGTATACATCCCTGTTCC-3') was designed to hybridize to the last 29 nucleotides of the coding sequence of rat _2a (Perez-Reyes et al. 1992), including the stop codon, permitting the amplification of full-length ₂ splice variants from rat heart. Resulting PCR products were cloned and sequenced (4 clones), followed by transfer to a CMV mammalian expression vector for functional studies in HEK 293 cells.

Electrophysiology

All electrophysiological recordings were acquired at room temperature, using an Axopatch 200A patch-clamp amplifier (Axon Instruments, Union City, CA, USA). For whole-cell recordings in cultured heart cells, electrodes fashioned from borosilicate glass capillaries (World Precision Instruments, MTW 150-F4) were filled with an internal solution containing (mM): 150 caesium methanesulphonate (Cs-MeSO₃), 5 CsCl, 10 Hepes, 10 EGTA, 1 MgCl₂, 4 MgATP, supplemented with 0.5 µM ryanodine (pH 7.2, adjusted with CsOH). Typically, pipettes had resistances of 1-2 M, before series resistance compensation of 60-75 %. For formation of gigaohm seals and initial break-in to the whole-cell voltage clamp configuration, heart cells were perfused with normal Tyrode solution containing (mM): 138 NaCl, 4 KCl, 2 CaCl₂, 1 MgCl₂, 0.33 NaH₂PO₄, 10 Hepes, and 10 glucose (pH 7.4, adjusted with NaOH). Following successful break-in the perfusing medium was switched to an external recording solution containing (mM): 155 N-methyl-D-glucamine aspartate (NMG-Asp), 5 BaCl₂, 10 4-aminopyridine, 10 Hepes (pH 7.4, adjusted with NMG). To measure gating currents, ionic currents were blocked by adding 2 mM CdCl₂ and 0.1 mM LaCl₃ to the bath solution (Bean & Rios, 1989; Hadley & Lederer, 1989). Signals were filtered at 2 kHz (step currents) or 5 kHz (gating currents) (four-pole Bessel filter), and sampled at 25 kHz. Displayed gating current traces have been additionally processed with a Gaussian filter at 2 kHz. Data traces were acquired at a repetition interval of 30 or 45 s. Leaks and capacitive transients were subtracted by a P/8 protocol (ionic currents), or a P/-8 protocol (gating currents), from a -100 mV holding potential, as previously described (Hadley & Lederer, 1989). To facilitate better resolution of displayed gating currents, we subtracted smooth curves fitted to leak currents.

Single-channel currents were obtained in the cell-attached patch-clamp configuration as previously described (Colecraft et al. 2001). Cells were bathed in a membrane potential zeroing solution consisting of (mM): 132 potassium glutamate, 5 KCl, 5 NaCl, 3 MgCl₂, 2 EGTA, 10 glucose, and 10 Hepes (pH 7.4, adjusted with KOH). The pipette solution contained (mM): 90 BaCl₂, 20 tetraethylammonium methanesulphonate (TEA-MeSO₃), and 10 Hepes (pH 7.4, adjusted with TEA-OH). Patch pipettes fashioned from borosilicate glass and coated with Sylgard typically had resistances of 3-8 M when filled with the pipette solution. Signals were filtered at 2 kHz (-3 dB, four-pole Bessel), and sampled at 25 kHz. Voltage pulses were delivered at 4 s intervals, and reported voltages have been corrected for a liquid junction potential of 17 mV. We subtracted smooth functions fitted to leak and capacitive currents (P/8 protocol) from single-channel records in a semi-automated manner using custom software written in Matlab (Mathworks, Natick, MA, USA). Single-channel data were analysed using custom analysis software written in Matlab, as described (Colecraft et al. 2001). Records were additionally filtered to a 1.5 kHz band width using a digital Gaussian filter, and idealized using half-height criteria. Ensemble average currents, FL and P_OO were calculated from idealized records as previously described (Imredy & Yue, 1994).

Whole-cell recordings in HEK 293 cells were essentially as previously described (Colecraft et al. 2000). Internal and external solutions were as described above for heart-cell recordings, except that ryanodine was omitted from the internal solution. Pipette series resistances were typically < 1 M after 70-80 % compensation. Signals were filtered at 2 kHz and sampled at 25 kHz, and data traces were acquired at a repetition interval of 30 or 45 s, with leak and capacitive currents subtracted by a P/8 protocol.

Modelling

To represent the activation pathway in models simulating channel kinetics (Fig. 4 and Fig. 5), we adopted a structure similar to that proposed for voltage-gated K⁺ (Zagotta & Aldrich, 1990), Na⁺ (Armstrong & Bezanilla, 1977; Cha et al. 1999), and Ca²⁺ channels (Boland & Bean, 1993). Our model is a simple, eight-state Markov chain (Fig. 5C) that incorporates four voltage-dependent transitions between closed states, corresponding roughly to the movement of voltage sensors in each of four homologous domains. Actual opening of the channel involves a voltage-independent transition between the last closed state and the open state. Additionally, to incorporate inactivation in the simplest manner, we added two inactivated states that connect to both the last closed state and the open state, reminiscent of Na⁺ channel models (Armstrong & Bezanilla, 1977; Cha et al. 1999).

Simulations of first latencies (FL), open probabilities (P_o), conditional open probabilities (P_OO), and closed times were performed using custom-written scripts that made extensive use of the numerical matrix exponential function in Matlab (MathWorks). Simulations for these kinetic measures were performed simultaneously for each subunit by optimizing rate constants within the Markov model. We initially fitted simulated curves by eye to ensemble data from single-channel recordings. Then, we allowed the computer to optimize the rate constants using a Simplex algorithm, where we controlled the weight assigned to the sum of squared errors between the fit and data for each kinetic measure. The only fixed rate constant was the open-to-closed transition (), which was set to either the dominant rate constant (2.64 ms^-1) or weighted average (2.18 ms^-1) of biexponential fits to open times data. After computer simulation, we assessed the sensitivity of the model to ten-fold variations in each rate constant. The model was insensitive to large variations in the rate constants for the closed state 1 to closed state 2 (C₁-to-C₂) transition, so these were fixed to the values reported in Table 1 in subsequent simulations.

Single-channel data from the _1b-GFP subunit appeared to show bimodal behaviour, with each patch dominated by one or the other mode. Therefore, we separated patches into two sets according to the dominant mode, and modelled the average of each set independently. Similarly, comparisons to the single-channel data were made after averaging the independent fits. Best fits for each subunit are displayed as smooth curves plotted through data in Fig. 4 and Fig. 5, with corresponding rate constants detailed in Table 1.

Statistics

Pooled data are presented as means ± S.E.M., and P values were calculated using Student's two-tailed t test; P < 0.05 was considered significant.

	RESULTS

Top Abstract Introduction Methods Results Discussion References

Over-expression and subcellular targeting of Ca²⁺ channel -GFP subunits in adult heart cells

The use of recombinant adenoviral vectors enabled high-level expression of exogenous proteins in cultured adult rat heart cells, with efficiencies approaching 100 %. Fusion of Ca²⁺ channel subunits to GFP permitted evaluation of different aspects of subunit over-expression in heart cells (Fig. 1). First, Western blots on cultured heart cell lysates probed with GFP antibodies indicated the expression of full-length -GFP subunits with no evidence of protein degradation or cleavage (Fig. 1A). _2a-GFP was unique in being toxic to cardiac cells, with > 90 % cell death after 48 h (not shown). This limited harvest of _2a-GFP-expressing cells to periods (~18 h after infection) suboptimal for maximum protein expression, resulting in a relatively lightly stained protein band for this subunit (Fig. 1A). Second, the higher molecular weights of -GFP proteins were exploited to simultaneously monitor exogenous and endogenous subunit expression using a generic subunit antibody (_GEN) (Chien et al. 1996) (Fig. 1B). Uninfected heart cells displayed a single dominant subunit immunoreactive band of ~72 kDa, the level of which was unaffected by virally mediated over-expression of GFP, or -GFP subunits (Fig. 1B). The molecular weight of the endogenous cardiac subunit is consistent with it corresponding to the ₂ subunit, in agreement with previous studies (Pichler et al. 1997; Haase et al. 2000). At 48 h after infection, exogenous -GFP subunits were markedly over-expressed (> 6-fold) compared to their endogenous counterparts, as indicated by the relative intensities of ₃- and ₄-GFP bands to those of the endogenous subunit (Fig. 1B). Of note, contrary to the simple prediction from the primary amino acid sequence of a higher molecular weight for ₄ (58 kDa) than for ₃ (55 kDa), we observed that ₃-GFP migrated at a slightly higher molecular weight than ₄-GFP (Fig. 1A and B). Such discordance in the molecular weight of ₃ has previously been observed in HEK 293 cells (Chien et al. 1996), and may be due to as-yet-unidentified post-translational modifications of the ₃ subunit (Chien et al. 1998). Third, we used confocal microscopy to assess the subcellular localization of -GFP subunits (Fig. 1C). Images were acquired at < 24 h after infection to limit spurious targeting of subunits caused by marked protein over-expression. Control cells expressing GFP alone displayed fluorescence staining throughout the cell, with similar intensities in nuclear and sarcoplasmic compartments, as befits a molecule small enough (< 40 kDa) to passively diffuse across the nuclear membrane (Davis, 1995). Cells expressing _1b- or ₃-GFP displayed no gross difference in fluorescence pattern from cells expressing GFP alone. In striking contrast, cells expressing _2a- or ₄-GFP exhibited dramatically different and unique fluorescence localization patterns. The 'halo' appearance of _2a-GFP fluorescence bore testament to the predominant localization of this protein to the surface sarcolemma, with no representation in the sarcoplasmic or nuclear compartments. On the other hand, ₄-GFP was sharply segregated to either intense nuclear staining, or to regularly spaced transverse striations across the length of the cell (Fig. 1C). The unexpected intense nuclear localization of ₄-GFP (and to a lesser extent _1b- or ₃-GFP) cannot be explained by degradation or cleavage of the fusion protein yielding a smaller molecule capable of diffusing across the nuclear membrane (Fig. 1A). Since the molecular weights of these fusion proteins are clearly above the cut-off expected for passive diffusion, an active uptake process is likely to underlie concentration of -GFP proteins within the nucleus.

		View larger version [in this window] [in a new window]
Figure 1. Expression and subcellular targeting of -GFP subunits in adult rat heart cells A, Western blots of whole heart cell lysates probed with the GFP antibody. Cells were harvested 48 h post-infection with recombinant adenovirus, except for 2a-GFP-expressing cells, which were harvested after 18 h, because of toxicity. The higher molecular weight band in the GFP lane corresponds to GFP dimer and disappears with more intense boiling of the protein sample before electrophoresis (not shown). B, Western blots of whole-cell lysates probed with the GEN antibody, showing marked over-expression of 3- and 4-GFP compared with the endogenous subunit. From this perspective, some proteolysis of 3-GFP was evident, however, this did not result in a proteolytic fragment that was recognized by the GFP antibody ( see A). C, confocal images showing differential targeting of Ca2+ channel -GFP subunits in heart cells. Images were obtained in live cells < 24 h post infection, and represent a section through the centre of the cardiomyocytes. Images are representative of observations made in > 100 cells for each -GFP subunit.

Overall, these results demonstrate the successful application of recombinant adenoviruses to achieve over-expression of exogenous Ca²⁺ channel subunits in adult heart cells, and provide the novel observation that distinct subunits are differentially targeted in heart. We next turned our attention to the functional consequences of such subunit over-expression on endogenous L-type channels.

Exogenous subunits tune the function of native L-type Ca²⁺ channels

Electrophysiological recordings of whole-cell L-type Ca²⁺ channel currents permitted direct evaluation of the extent to which exogenous subunits tuned the functional properties of endogenous _1C subunits (Fig. 2). Exogenous subunit expression produced two visually identifiable effects on whole-cell Ba²⁺ currents in native heart cells. First, compared with control cells expressing GFP (Fig. 2Aa), each subunit enhanced the current amplitude as gauged by the relative sizes of average peak current densities evoked by 0 mV voltage steps (Fig. 2Aa-Ea). The increase in current density extended across all voltages as indicated by I-V relationships in cells expressing different -GFP proteins (Fig. 2Ab-Eb). Currents through GFP- and -GFP-expressing cells alike were abolished by 1 µM nimodipine, confirming that they are carried through L-type Ca²⁺ channels (not shown). Importantly, there were distinctions between subunits with regards to the degree of enhancement of current density, with a rank order of _{2 4} > _1b > ₃. These results confirm and extend our previous observations (Wei et al. 2000). By comparison with our previous work, a key distinctive of the present study was the use of recombinant adenoviruses to achieve high levels of expression of subunits, permitting evaluation of relative protein levels by Western blotting (Fig. 1B). This capability yielded the insight that the degree of subunit over-expression (Fig. 1B) outstrips the observed increases in current density (Fig. 2Da and Ea), suggesting that we have attained the maximum increase in L-type current density that can be obtained with each subunit.

		View larger version [in this window] [in a new window]
Figure 2. Tuning of L-type Ca2+ channel whole-cell current properties in ventricular myocytes over-expressing recombinant Ca2+ channel subunits Aa-Ea, current density waveforms averaged from multiple ventricular cells expressing GFP alone, or various -GFP constructs. Black traces plot the mean, and grey traces above and below the S.E.M. confidence range. Averaged traces were elicited by 0 mV step depolarizations from a holding potential of -90 mV. Compared with GFP cells, peak current density amplitudes were markedly elevated for channels expressing -GFP subunits, albeit to different extents. Ab-Eb, current density (J) vs. step voltage (V) relationships for GFP- and -GFP-expressing cells. Dashed traces in Bb-Eb reproduce the trace in Ab to facilitate direct visual comparison between the data from -GFP-expressing cells and control (GFP-expressing) cells. Traces show that the enhancement of native cardiac L-type channel current density extends across a wide range of voltages. Ac-Ec, plots of the fraction of current remaining after a 300 ms depolarization (r300) vs. step voltage (V). Dashed lines in Bc-Ec reproduce the trace in Ac to facilitate visual comparison between the data from GFP- and -GFP-expressing cells. Faster inactivation rates result in lower r300 values. Therefore, 2a- and 4-GFP expression in heart cells markedly slow L-type current inactivation rates over a broad range of voltages. *P < 0.05, n = 4-10 for each point.

Second, individual subunits uniquely altered the rate of current inactivation (Fig. 2Ac-Ec). Using the fraction of current remaining after a 300 ms depolarization (r₃₀₀) as an index of channel inactivation, cells expressing _2a-GFP (Fig. 2Cc) or ₄-GFP (Fig. 2Ec) displayed markedly reduced rates of inactivation compared with control. By contrast inactivation was not significantly altered in cells expressing _1b- or ₃-GFP (Fig. 2Bc and Dc, respectively). These results demonstrate that exogenously expressed -GFP subunits interact with endogenous _1C proteins, and explicitly confirm subunit identity as a major determinant of native cardiac L-type Ca²⁺ channel current inactivation rate, as previously found for recombinant Ca²⁺ channels in heterologous systems (Olcese et al. 1994; De Waard & Campbell, 1995; Patil et al. 1998).

Two critical questions directly ensue from these observations. First, by what mechanism does subunit over-expression result in increased L-type Ca²⁺ channel current density in heart cells? The leading candidate mechanisms for this effect are: an increase in the number of functional channels (N) expressed in the cardiac sarcolemma, an increase in single-channel open probability (P_o), or both. Second, what is the identity of the endogenous cardiac L-type Ca²⁺ channel subunit? While it is believed that ₂ is the predominant subunit isoform expressed in rat heart, our finding that expressing _2a in heart cells markedly slows L-type current inactivation kinetics suggests that the identity of the actual, endogenous Ca²⁺ channel subunit remains unknown. We focus on these two key questions in turn as outlined below.

Expression of exogenous subunits increases sarcolemmal Ca²⁺ channel gating current

One convenient measure of the number of voltage-gated channels expressed in the membrane is the maximum gating charge (Q_max) moved with depolarization (Jones et al. 1998). Accordingly, we measured gating currents in heart cells under conditions in which detection of charge movement associated with Ca²⁺ channel gating is selectively maximized. Ionic currents were blocked with solutions containing 2 mM Cd²⁺/0.1 mM La³⁺, and we operated from a holding potential of -50 mV to appreciably immobilize sodium channel gating charge (Bean & Rios, 1989; Hadley & Lederer, 1989). Exemplar Ca²⁺ channel gating currents elicited by voltage steps to -30, -10, and +10 mV in control cells are shown in Fig. 3Aa. The total charge moved during the test depolarization, Q_on, was obtained by integrating the on-gating current over the depolarization period, taking as baseline the average current over the last 3-5 ms of the test pulse (Jones et al. 1998). Q_on from multiple experiments was normalized to cell capacitance and plotted against voltage to obtain the Q_on-V relationship (Fig. 3Ab). As expected, this curve was sigmoidal and well described by a single Boltzmann relation of the form Q_on = Q_max/(1 + exp(-(V - V_1/2)/k)), where Q_max is the maximum amount of mobile charge, V_1/2 is the potential of half-maximal charge movement, and k is a slope factor. Q_max for control cells was 4.7 ± 0.9 fC pF^-1 (n = 7), in close agreement with previous estimates of 3.7 fC pF^-1 (Hadley & Lederer, 1989) and ~5 fC pF^-1 (Bean & Rios, 1989) in freshly isolated rat ventricular myocytes. In cells expressing -GFP proteins, the relatively larger sizes of exemplar gating currents (Fig. 3Ba-Ea) provided an initial indication of an increase in N. This was entirely confirmed in the population data where over-expression of all subunits resulted in an increase in the measured Q_max (Fig. 3Bb-Eb). The most robust increase was observed with ₄-GFP where Q_max = 11.4 ± 1.3 fC pF^-1 (n = 4, P < 0.01). These results suggest that the increase in macroscopic currents in heart cells seen with -GFP expression is at least mediated partly through an increase in N.

		View larger version [in this window] [in a new window]
Figure 3. Exogenous subunits increase L-type channel gating currents in heart cells Aa-Ea, top, voltage protocols, 20 ms step depolarizations from a holding potential of -50 mV were used to elicit gating currents. Ionic currents were blocked by including 2 mM Cd2+/0.1 mM La3+ in the bath solution. Lower traces, exemplar Ca2+ channel gating currents elicited by voltage steps to -30, -10 and +10 mV. Ab-Eb, plot of the integral of the normalized on-gating charge (Qon) vs. V. The Qon-V relationship for GFP (Ab) has been reproduced in Bb-Eb (dashed lines) to facilitate direct comparison with the Qon-V relationship for -GFP-expressing cells. Continuous curves through the data were derived from least-squares fits to a single Boltzmann function, with the following parameters (Qmax, V1/2 and k, respectively). GFP (Ab): 4.7 fC pF-1, -17.1 mV and 8.5; 1b-GFP (Bb): 6.9 fC pF-1, -19.6 mV and 7.2; 2a-GFP (Cb): 8.7 fC pF-1, -18.6 mV and 7.6; 3-GFP (Db): 7.1 fC pF-1, -17.3 mV and 7.5; 4-GFP (Eb) 11.4 fC pF-1, -15.3 mV and 6.5.

To determine whether changes in P_o contributed to the observed effects of subunits on macroscopic current density, we next turned to single-channel experiments.

Each subunit imparts unique single-channel gating properties to L-type Ca²⁺ channels

By contrast with their relatively uniform effects on Q_max, auxiliary subunits produced singular modifications on the kinetic profile of L-type Ca²⁺ channels, as initially hinted at by the different inactivation rates of macroscopic currents (Fig. 2). Single-channel recordings provided the most powerful means of resolving the distinctive effects of each subunit on L-type channel gating. Cell-attached patches containing one active L-type Ca²⁺ channel were repetitively depolarized to +40 mV from a -60 mV holding potential. Exemplar single-channel activity (Fig. 4Aa-Da) furnished a first hint that the distinctions in microscopic gating imparted by each subunit were in fact pervasive, extending well beyond differences of inactivation. Control channels (Fig. 4Aa) usually displayed a dense-burst gating pattern, and were characterized by a mixture of non-inactivating (Fig. 4Aa, traces 1 and 5), inactivating (traces 3 and 4), and null sweeps (trace 2). Channels incorporating _2a-GFP (Fig. 4Ca) also showed a dense-burst gating pattern. However, a key distinguishing feature of _2a-GFP channels was the almost exclusive prevalence of non-inactivating sweeps, with a dearth of null sweeps (Fig. 3Ca). This molecular signature fits nicely with the slowed inactivation rate of whole-cell currents conferred by this subunit (Fig. 2Ca and Cc). Channels partnered with _1b- or ₃-GFP subunits (Fig. 4Ba and Da, respectively) displayed a sparser gating pattern, while manifesting a mixture of inactivating and null sweeps. Despite numerous attempts, ₄-GFP single-channel currents remained elusive, a possible consequence of the subcellular localization of this subunit (Fig. 1C).

		View larger version [in this window] [in a new window]
Figure 4. Distinctive effects of -GFP subunits on single L-type Ca2+ channel gating Aa-Da, top, voltage protocols, single-channel currents were elicited by test depolarizations to +40 mV, from a holding potential of -60 mV. Lower traces, exemplar single-channel records from cells over-expressing GFP (Aa) or -GFP proteins as indicated (Ba-Da). Records shown are consecutive sweeps from patches containing a single active channel. Mean unitary current amplitudes were: GFP, 0.71 ± 0.03 pA, n = 4; 1b-GFP, 0.70 ± 0.02 pA, n = 4; 2a-GFP, 0.72 ± 0.01 pA, n = 4; 3-GFP, 0.71 ± 0.01 pA, n = 3. Ab-Db, ensemble currents (grey traces) averaged from the indicated number of patches were obtained from all sweeps, including nulls. Ac-Dc, first latency (FL) distributions (thick grey traces) obtained from all sweeps, and averaged from the same number of patches as indicated in Ab-Db, above. The FL histogram for GFP (Ac) is reproduced in Bc-Dc (dashed line) to permit direct comparison between the different infection conditions. Ad-Dd, conditional open probability (POO) distributions (thick grey traces) averaged from the indicated number of patches. The steady-state POO value for channels from GFP-expressing cells was 0.18, and this value is reproduced in Bd-Dd (dashed line) to facilitate direct comparison with values obtained in cells expressing -GFP subunits. Ab-Dd, continuous black lines through the data are fits generated from the kinetic model shown in Fig. 5C, with parameters given in Table 1.

Ensemble average currents (Fig. 4Ab-Db) provided further qualitative insight into the distinct effects of various subunits. Here, the currents have been divided by unitary current amplitude and averaged from multiple patches, providing a simple readout of channel P_o. Ensemble currents in control cells rapidly activated to a maximum P_o value of 0.12 (Fig. 4Ab, n = 4), and then declined as expected from the exemplar inactivating sweeps. Single-channel P_o was decreased in channels associated with _1b-GFP (P_o = 0.05, n = 4) or ₃-GFP (P_o = 0.08, n = 3), but enhanced in channels reconstituted with _2a-GFP (P_o = 0.31, n = 4). These results indicate that increased single-channel P_o contributes to the elevated macroscopic current observed with _2a-GFP, but not with _1b- or ₃-GFP. Moreover, these results help rationalize distinctions between subunits with respect to the degree of enhancement of macroscopic L-type channel currents (Fig. 2).

Beyond clarifying the molecular bases of differences in the whole-cell current amplitude, the distinctive effects of subunits on single-channel kinetics could potentially provide insight into the identity of the endogenous Ca²⁺ channel subunit. Reassuringly, during expression of recombinant subunits, single-channel activity nearly always adopted new characteristics different than in controls, as if the over-expressed subunits exchanged completely with pre-existing channels, and predominated within new complexes. Hence, quantitative analysis of single-channel gating behaviour probably provides in-depth insight into the kinetic mechanisms underlying subunit function, providing a 'kinetic fingerprint' of each channel type. Accordingly, we undertook further quantitative analysis of our single-channel data.

First-latency (FL) (Fig. 4Ac-Dc) and conditional open probability (P_OO) distributions (Fig. 4Ad-Dd) furnish a coarse overall synopsis of gating characteristics; together, FL and P_OO specify the total overall gating characteristics of a channel (Imredy & Yue, 1994; Colecraft et al. 2001). FL, defined as the probability that first opening occurs before time t in the test pulse, quantitatively portrays channel activation kinetics. P_OO, the conditional probability that a channel will be open with a delay t after first opening, reflects the aggregate gating behaviour of channels after first opening. Open and closed time distributions (Fig. 5A and B) supply additional, complementary representations of gating behaviour. All these descriptors of channel gating, along with the ensemble average P_o (Fig. 4D), constitute an extensive compendium of the gating behaviour imparted by each subunit. Even cursory examination of these descriptors rigorously confirmed key visual trends identified above (Fig. 4Aa-Da). The variable steady-state plateau levels of FL (Fig. 4Ac-Dc) verified the relative propensities for null sweeps; for example, the plateau level near unity for _2a-GFP channels established the virtual absence of null sweeps with this construct. The differing steady-state levels of P_OO (Fig. 4Ad-Dd) authenticated the sparser opening pattern of channels associated with _1b-GFP or ₃-GFP. Most of the changes in steady-state gating had to do with differences in the stability of closed conformations (Fig. 5B), rather than in the lifetime of open states (nearly identical across Fig. 5A).

		View larger version [in this window] [in a new window]
Figure 5. Open and closed time distributions, and quantitative modelling provide insights into distinctive subunit effects on L-type channel gating A, open time distributions (thick grey traces) of single L-type channels generated from cells expressing GFP (27, 321), 1b-GFP (8, 723), 2a-GFP (71, 773), or 3-GFP (12, 082) as indicated. Numbers in parentheses represent the total number of events used to generate the distributions. When reproduced onto the other plots the GFP distribution (dotted traces) was coincident with -GFP distributions, demonstrating that subunits did not affect channel open times. Continuous black curves through the data are single exponential fits utilizing the weighted rate constant from the open to closed transition ( = 2.18 ms-1) in the kinetic model shown in C. B, closed time distributions (thick grey traces) were generated from the following number of events: GFP (27, 031), 1b-GFP (8, 520), 2a-GFP (71, 556), and 3-GFP (11, 845). When reproduced onto the other plots, the GFP closed time distribution (dotted traces) was discordant with -GFP distributions, demonstrating that each subunit uniquely impacted L-type channel closed times. Continuous curves through the data are fits generated from the quantitative model in Fig. 5C, with the parameters shown in Table 1. C, top, kinetic scheme used to model subunit effects on L-type channel gating. Details of the model are presented in Methods. Bottom, fold changes in forward equilibria for activation and inactivation transitions for -GFP channels, when compared to control (GFP) channels. The forward equilibrium for each transition was normalized to the corresponding GFP value. Parameter values used to generate fits in Figs 4 and 5 are shown in Table 1.

Explicit fits of a voltage-gated channel mechanism to all the quantitative descriptors (see Methods), shown as continuous curves (Fig. 4 and Fig. 5) revealed the core bases for the distinctive gating effects of each subunit. Differences in the microscopic propensity for inactivation (Fig. 5C, /) clearly explained the variable inactivation rates of macroscopic data (Fig. 2). Moreover, the kinetic mechanism revealed a novel insight regarding the basis of null sweeps: such traces without activity could be completely accounted for by channels that were inactivated at the onset of test pulse depolarization (Table 1), and the slow rising phase of FL corresponded to channels gradually returning from inactivation during the test pulse. Previously, null sweeps were thought to reflect residence in an 'unavailable' state different from inactivation (Yue et al. 1990). Most importantly, apart from inactivation, the effects of subunits extended pervasively throughout the activation pathway, as shown by the fold changes in forward equilibria corresponding to each activation transition (Fig. 5C, bars). Interestingly, the recombinant subunits appeared to exert an alternating pattern of enhancement and inhibition of activation steps, each of which may coarsely represent voltage sensor movement in each of four homologous channel domains. If so, different subunits might induce distinctive gating signatures via different geometric footprints among the four channel domains. Previous studies that focused on interpretation of less discriminating, macroscopic gating properties (Neely et al. 1993), or limited single-channel information (Neely et al. 1995), led to the prevailing notion that subunits mainly exert their effects on steps near the open state. Our in-depth single-channel fingerprinting now provides compelling evidence that just the opposite scenario is true, and also underscores the profound and divergent effect of different subunits on L-type channel gating.

Identification of the endogenous cardiac Ca²⁺ channel subunit

The differences in single-channel gating signatures between _2a-GFP and control channels reinforced the notion that the endogenous Ca²⁺ channel subunit must be distinct from _2a. Yet, Northern analyses (Hullin et al. 1992; Perez-Reyes et al. 1992) and Western blot data (Ludwig et al. 1997; Pichler et al. 1997; Haase et al. 2000; Reimer et al. 2000) clearly indicate that ₂ is the dominant Ca²⁺ channel subunit isoform expressed in rat heart. How can these two divergent viewpoints be reconciled? One possible explanation is that heart cells express a ₂ variant that is only subtly different from _2a in terms of structure, and yet imparts markedly different kinetic properties to the native cardiac L-type Ca²⁺ channel. In this regard, the N-terminus has been found to be a prominent site of sequence variation among distinct ₂ subunit variants in different species (Hullin et al. 1992; Perez-Reyes et al. 1992; Rosenfeld et al. 1993; Massa et al. 1995). Therefore, we probed for the existence of rat cardiac-specific ₂ forms using 5'-RACE PCR. Using this method, we amplified the complete open reading frame of a new rat ₂ variant, _2b, which differs from _2a only in the N-terminus domain of the protein (see Fig. 7 for ₂ splice variant name designations). Figure 6A shows a multiple sequence alignment of the N-terminus domains of _2b from rat, rabbit (Hullin et al. 1992), human (authors' unpublished data), and guinea-pig (S. Ding, S. Kuroki, A. Kameyama, A. Yoshimura and M. Kameyama, 1998, direct GenBank submission, accession number AB016288).

		View larger version [in this window] [in a new window]
Figure 6. Identification and functional characterization of a novel rat heart 2 splice variant A, sequence comparison of the amino terminus (D1 domain) of newly cloned rat 2 variant 2b (GenBank Accession number AF423193) to the corresponding region in orthologues from other species. B, exemplar Ba2+ currents from recombinant L-type channels (1C 2) reconstituted in HEK 293 cells with either 2a (black trace) or 2b (grey trace) indicates greater inactivation with 2b. Currents were elicited with a 300 ms depolarization to +10 mV. C, averaged Ba2+ currents from heart cells infected with GFP virus (black trace, n = 5), or a recombinant virus encoding 2b and GFP in a bicistronic cassette (grey trace, n = 5). The GFP trace has been scaled up so as to match current amplitudes at the end of the test pulse, to permit explicit comparison of current waveforms. Currents were evoked with a 300 ms depolarization to +10 mV. Da-Df, single L-type channel gating properties in heart cells over-expressing 2b. Exemplar currents (Da) are consecutive traces from a patch containing one channel. Ensemble currents (Db, grey trace) were averaged from four single-channel patches, as were FL (Dc), Poo (Dd), open time histograms (De), and closed time histograms (Df). Continuous traces through the data are model fits using the same parameters as for GFP (Table 1), except that the percentage of cells initially in state C1 was 57 %.

		View larger version [in this window] [in a new window]
Figure 7. Molecular basis of Ca2+ channel 2 subunit diversity Top, generic modular domain structure proposed for Ca2+ channel subunits in which three variable domains (D1, D3 and D5) are interspersed by two conserved regions (D2 and D4). Bottom, analysis of human chromosome 10 genomic DNA sequence (contig NT-008682) clarifies the molecular basis of 2 subunit diversity. Five D1 variants arise from alternative splicing of six exons and have been designated 2a-2e. Three D3 variants arise from mutually exclusive splicing of exons 11, 12 and 13. 2a-e transcripts all contain exon 11.

To determine whether the subtle sequence differences between _2a and _2b could account for the flagrant discordance in channel inactivation kinetics between control and _2a-expressing heart cells (Fig. 2), we first turned to recombinant L-type channels (_1C2) reconstituted in HEK 293 cells. Ba²⁺ currents from recombinant channels reconstituted with _2b inactivated significantly during the test pulse, yielding waveforms that were more reminiscent of the cardiac phenotype, whereas _2a channels hardly inactivated during the 300 ms depolarization (Fig. 6B). To more directly determine whether _2b represented the dominant subunit species in heart cells, we over-expressed this subunit in cardiac myocytes by adenoviral-mediated gene transfer. Surprisingly, explicit overlay of averaged Ba²⁺ currents from _2b and control channels indicated a slight discordance in current inactivation kinetics (Fig. 6C). When the traces were scaled to match amplitudes at the end of the test pulse, it appeared as if about 90 % of the control current displayed a macroscopic inactivation rate similar to that of _2b-expressing channels, with the remnant exhibiting faster inactivation (Fig. 6C). Two possible interpretations for this result are that the actual cardiac subunit is distinct from _2b, or that heart cells express a mixture of two or more subunits, with _2b being the predominant form. To distinguish between these two possibilities we conducted single-channel experiments (Fig. 6Da-Df). Exemplar single-channel currents from heart cells over-expressing _2b (Fig. 6Da) appeared qualitatively similar to those recorded from GFP-expressing, control cells (Fig. 4Aa). Most importantly, there was quantitative agreement in microscopic gating kinetics between _2b and control channels (Fig. 6Db-Df), as indicated by the impressive coincidence between fits to control channel data (continuous curves are reproduced from fits to GFP data in Fig. 4 and Fig. 5), and the averaged quantitative descriptors of _2b channel gating. These results suggest _2b as a good candidate for the endogenous cardiac subunit.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

The important new findings in this study are that, (1) different subunits target to distinct subcellular localizations in heart cells; (2) subunits are rate limiting for expression of functional L-type Ca²⁺ channels in heart; (3) different subunits impart unique single-channel gating signatures to cardiac L-type channels; and (4) a ₂ splice variant, other than _2a, may predominate in heart. Overall, our results suggest novel functions for Ca²⁺ channel subunits in heart, with important implications for L-type Ca²⁺ channel operation in both normal and diseased myocardium. We discuss our findings in relation to previous work in terms of their biophysical, (patho)physiological, and possible therapeutic implications.

Differential targeting of subunits in heart cells

A well-known function of Ca²⁺ channel subunits is that they chaperone ₁ subunits to the plasma membrane, when the two proteins are co-expressed in heterologous expression systems (Chien et al. 1995; Brice et al. 1997; Gao et al. 1999; Yamaguchi et al. 2000). Here, we show that in the more complicated cyto-architecture of native heart cells, apparent distinctions in subunit localization point to a more sophisticated targeting capability of these proteins beyond a simple translocation of ₁ subunits to the membrane. In similar vein, the identity of the subunit associated with the _1A subunit was found to be a determinant of the relative distribution of the holochannel to either the apical or basolateral membrane in a polarized epithelial cell line (Brice & Dolphin, 1999). In heart, L-type Ca²⁺ channels are preferentially localized in T-tubules in close proximity and apposed to sarcoplasmic reticulum (SR) Ca²⁺ release channels at dyadic junctions (Carl et al. 1995; Sun et al. 1995; Gathercole et al. 2000; Scriven et al. 2000). The molecular determinants underlying this placement of cardiac L-type channels are presently unknown, despite this localization being critical to the high gain of Ca²⁺-induced Ca²⁺ release that underlies cardiac E-C coupling (Fabiato, 1983; Cannell et al. 1995; Wier & Balke, 1999). It is quite plausible that the endogenous Ca²⁺ channel subunit plays a role in targeting cardiac L-type channels to their strategic final destination. Our observations that distinct subunits are differentially targeted in heart provide a strong basis to investigate this intriguing notion.

_2a- and ₄-GFP especially provided strikingly distinctive subcellular localization when compared with control cells expressing GFP alone. The exclusive membrane localization of _2a in heart cells is reminiscent of its distribution when expressed alone in HEK 293 cells (Chien et al. 1995; Brice et al. 1997). Such membrane targeting of _2a is mediated through palmitoylation of two cysteine residues (Cys 3, 4) present at the N-terminus of the protein (Chien et al. 1996), and is responsible for the slowed whole-cell current inactivation kinetics observed with this subunit (Qin et al. 1998; Restituito et al. 2000). Thus, our observation that _2a targets exclusively to the surface sarcolemma in rat heart cells and slows L-type channel inactivation kinetics, indicates the functional operation of this post-translational modification pathway in heart. A particularly intriguing finding in this study was that ₄-GFP was strongly localized to the nucleus, in addition to its targeting to transverse elements throughout the cell. This nuclear localization was not due to cleavage of the fusion protein to a size that allowed passive diffusion into nucleus, because Western blots gave no indication of such protein degradation (Fig. 1A and B). It is tempting to speculate that this subunit may have an unappreciated role in the nucleus, such as in the regulation of gene expression. Future studies will investigate this possibility. Of relevance, ₄ subunit isoforms are found expressed in a temporally restricted manner in the developing heart (Haase et al. 2000).

Rate-limiting role of subunits for expression of cardiac L-type Ca²⁺ channel currents

Expression of all subunits increased whole-cell L-type Ca²⁺ channel current and Q_max, in conjunction with either a decrease (with _1b- and ₃-GFP) or increase (with _2a-GFP) in single-channel P_o. The most parsimonious interpretation of these data is that over-expression of subunits increases the number of functional channels (N) expressed in the sarcolemma. The strength of this suggestion is critically dependent on how faithfully Q_max mirrors N. Previous work suggests that the validity of utilizing Q_max for this purpose may be contingent on the specific experimental system. Co-expression of Ca²⁺ channel subunits with _1C or _1E (the putative R-type Ca²⁺ channel) in HEK 293 cells resulted in correlative increases in Q_max and ionic current (Josephson & Varadi, 1996; Kamp et al. 1996; Jones et al. 1998), suggesting Q_max to be a reliable measure of N in this system. By contrast, co-expression of _2a with _1C or _1E in Xenopus oocytes resulted in increased ionic currents with either no change or a decrease in Q_max (Neely et al. 1993; Olcese et al. 1996), despite evidence of markedly increased channel density in the surface membrane (Yamaguchi et al. 2000). Our results suggest that Q_max provides a good index for N in native heart cells, consistent with its recent use to demonstrate decreased expression of L-type Ca²⁺ channels in a canine heart failure model (He et al. 2001).

How does expression of exogenous subunits in heart cells result in increased Q_max? There are several distinct possibilities. The exogenous subunits may serve to chaperone pre-existing _1C subunits that are trapped in the Golgi. This scenario would suggest that _1C subunits are normally present in excess over endogenous subunits. Alternatively, the exogenous subunits may serve to stabilize _1C subunits present in the membrane. A third possibility is that there is an increased production of _1C in response to the elevated concentration of subunits in the heart cell. Distinguishing between these mechanisms will be crucial for in-depth understanding of how heart cells maintain a steady-state level of functional L-type Ca²⁺ channels in the sarcolemma. A key prediction from our conclusion that subunits are rate limiting for the expression of cardiac L-type Ca²⁺ channel currents is that over-expression of _1C would not appreciably enhance Ca²⁺ channel current density in heart cells. It will be interesting to determine if this prediction is met when the _1C subunit is over-expressed in adult heart cells. Notably, however, in a transgenic mouse model with cardiac-specific over-expression of the _1C subunit, there was no increase in ventricular Ca²⁺ channel current density when compared with control (Muth et al. 1999).

The proposed rate-limiting role of Ca²⁺ channel subunits lends extra significance to reports of down-regulation of subunit mRNA in pancreatic islets of diabetic rats (Iwashima et al. 2001), in allografts from diastolically failing hearts (Hullin et al. 1999), and in atrial fibrillation (Grammer et al. 2001). Such down-regulation of subunit expression may feature prominently in the decreased L-type Ca²⁺ current or channel density observed in some studies of heart failure (Ming et al. 1994; Ouadid et al. 1995; Santos et al. 1995; Mukherjee et al. 1998; He et al. 2001), and atrial fibrillation. In such cases, our studies suggest that somatic gene transfer of Ca²⁺ channel subunits may help rectify cardiac dysfunction. The ability to alter Ca²⁺ channel density by manipulating levels of the subunit alone greatly simplifies the potential utility of L-type channels as targets for gene therapy of cardiovascular diseases (Wei et al. 2000).

Distinctive subunit effects on single L-type Ca²⁺ channel gating

We provide here the first report that distinct subunits impart unique single-channel signatures to L-type Ca²⁺ channels. Previous whole-cell studies have hinted that distinct subunits may affect aspects of L-type channel gating apart from the well-known effects on channel inactivation kinetics (Jones et al. 1998). Our present results provide firm support for this notion. The existence of distinct single-channel gating modes in native L-type Ca²⁺ channels has been previously decribed: mode 0 (null sweeps), mode 0a (brief, infrequent openings), mode 1 (frequent, millisecond-long openings), and mode 2 (long-lasting openings) (Hess et al. 1984; Yue et al. 1990; McDonald et al. 1994). We observed the distinguishing hallmarks of all these modes in our recordings. Mode 2 openings occurred infrequently (present in < 2 % of sweeps) in agreement with previous reports (Hess et al. 1984; Yue et al. 1990; Wiechen et al. 1995). subunits had no impact on the frequency of mode 2 openings as evidenced both by visual inspection of traces, and the invariance of open time histograms (Fig. 5). This directly contradicts a recent preliminary report that expression of _2a in rabbit cardiomyocytes increases the frequency of mode 2 openings in L-type channels (Cuadra et al. 2001). Instead, distinct subunits had fundamentally different effects on fast gating transitions within the dominant gating mode (mode 1), as well as the propensity for sojourns into mode 0 or 0a. In particular, _2a virtually abolished mode 0 gating (Fig. 4).

The distinctive signatures of subunits on single L-type channel gating led to divergent effects on channel P_o, providing a molecular mechanism for the non-uniform increase in whole-cell current amplitude observed with different subunits (Fig. 2). This can be appreciated in the context of the following relationship,

I = NiP_o,

where I is the whole-cell current amplitude, N is the total number of channels, i is the unitary current amplitude, and P_o is the channel open probability. With _2a- and ₃-GFP expression in heart, the observed increases in whole-cell current amplitude (I) were 2.9- and 1.5-fold, respectively (Fig. 2). Single-channel experiments indicated no change in i with subunit expression (Fig. 4). Measured changes in N (Q_max,/Q_max,control, Fig. 3) were 1.85- and 1.51-fold for _2a- and ₃-GFP, respectively. Fold changes in P_o (P_OO,/P_OO,control; Fig. 4) were 1.72 and 0.94 for _2a- and ₃-GFP, respectively. Hence, for _2a- and ₃-GFP expression, the increases predicted by the product of N and P_o are 3.2- and 1.4-fold, respectively, in good quantitative agreement with the experimentally observed increase in I. For these calculations, we used the steady-state plateau level of P_OO rather than the peak of the ensemble average current, to represent P_o. This is because the faster repetition intervals used in single-channel recordings (4 s) may result in appreciable accumulation of channels in inactivated states, resulting in null sweeps which will lower P_o values reported by ensemble average currents. Under whole-cell recording conditions, the slower repetition interval (30-45 s) would limit the tendency of channels to accumulate in inactivated states. Hence, the steady-state P_OO value probably more accurately reflects channel P_o under whole-cell recording conditions. With _1b-GFP, this calculation fell short of the experimentally observed increase in whole-cell current amplitude. A possible explanation for the discrepancy stems from the bimodal single-channel behaviour we observed for _1b-GFP channels. Two out of the four patches recorded for this subunit were dominated by sparse mode 0a openings, which resulted in a low calculated P_OO value. Thus, the discordance between predicted and observed increases in whole-cell current may indicate that such low P_o channels are in a minority in the cardiac sarcolemma.

Quantitative modelling of the single-channel data provided novel mechanistic insights into subunit actions on _1C subunits. Most importantly, the influence of subunits on channel gating extended throughout the activation pathway, during the bulk of voltage sensor movement. This finding challenges current dogma that subunits modulate ₁ subunits primarily by affecting transitions near the open state, after most of the voltage sensor charge has moved. This conclusion was based primarily on experiments in which subunits produced hyperpolarizing shifts in steady-state activation (G-V) curves while having no effects on Q-V curves, when co-expressed with _1C (Neely et al. 1993). In simulations where we added voltage dependence to the quantitative model, predicted Q-V curves (not shown) were not very different between subunits, in overall agreement with our experimental observations (Fig. 3). This underscores the more discriminating power of single-channel data over whole-cell methods in providing mechanistic insights into channel gating.

Molecular basis of ₂ subunit diversity

A product of the ₂ gene is believed to be the predominant subunit present in heart cells. However, knowledge of the exact molecular identity of the predominant cardiac L-type channel subunit is complicated by the diversity of ₂ sequences present in public databases. The relation between the different ₂ forms is best understood in terms of a modular domain structure proposed for Ca²⁺ channel subunits (Fig. 7), in which regions of high variability (domains D1, D3 and D5) are interspersed by two conserved regions (domains D2 and D4) (De Waard et al. 1994; Birnbaumer et al. 1998). Presently, five ₂ D1 variants have been described among mammals. In order of discovery and the species they were originally described in, these encoded D1 peptides of length 16 (rat, Perez-Reyes et al. 1992), 17 (rabbit, Hullin et al. 1992), 43 (rabbit, Hullin et al. 1992), 71 (human, Rosenfeld et al. 1993), and 23 (mouse, Massa et al. 1995). To date, no more than three D1 variants have been described in any single species (Qin et al. 1998), making it difficult to determine whether this variation is the result of alternative splicing alone, or a combination of alternative splicing and interspecies differences. To resolve this issue, we made use of the recent publication of a draft human genome sequence to clarify the molecular basis of ₂ diversity. Analysis of human chromosome 10 genomic DNA sequence (contig NT- 008682) demonstrates the presence of regions encoding all five D1 variants, resulting from alternative splicing of six exons. Hence, these alternate D1 sequences represent genuine splice variants and have been designated _2a-_2e (Fig. 7), with the order of designation corresponding to the order of discovery in mammalian species cited above. Further, three human D3 variants have been observed (Rosenfeld et al. 1993), and analysis of NT-008682 reveals that the variants result from mutually exclusive splicing of exons 11 (134 nt), 12 (20 nt) and 13 (62 nt) (Fig. 7). _2a-_2e transcripts all contain exon 11, and lack exons 12 and 13. Hence, in theory, an additional ten ₂ variants could be derived via alternative splicing of the D1 and D3 exons. The insights from this analysis provide a more panoramic perspective from which to launch discussion of the molecular identity of the predominant cardiac subunit.

Molecular identity of the endogenous cardiac Ca²⁺ channel subunit

Using a generic subunit antibody, we detected the endogenous cardiac Ca²⁺ channel subunit as a single immunoreactive band of ~72 kDa (Fig. 1). The predominance of a single band, as well as its molecular weight are consistent with previous findings that ₂ is the major subunit present in heart (Hullin et al. 1992; Perez-Reyes et al. 1992; Ludwig et al. 1997; Pichler et al. 1997; Haase et al. 2000; Reimer et al. 2000). There is far less certainty regarding the representation of ₂ splice variants in heart. Molecular cloning studies have identified multiple ₂ splice variants in the hearts of several species: _2a in human (Yamaguchi et al. 2000); _2b in rabbit (Hullin et al. 1992), human (our unpublished data), and guinea-pig (Ding et al. direct GenBank submission); _2c in rabbit (Hullin et al. 1992); and _2d in human (Allen & Mikala, 1998). Until recently, _2a was the only ₂ splice variant described in rat. We have described here the cloning of a new rat ₂ splice variant, _2b, from heart. Recently, a rat homologue of _2d was also cloned from heart cDNA (Yamada et al. 2001). Does the multiplicity of ₂ splice variants cloned from the hearts of different species reflect a genuine heterogeneity of ₂ subunits in heart cells? One critical drawback of all these studies is the use of cDNA derived from whole heart, which contains non-cardiac cells such as smooth muscle and neuronal cells that are known to express multiple Ca²⁺ channel subunit isoforms (Ludwig et al. 1997; Pichler et al. 1997; Reimer et al. 2000). Such cellular heterogeneity may contribute to the diversity of Ca²⁺ channel subunit isoforms and ₂ splice variants reported in hearts from different species (Hullin et al. 1992; Collin et al. 1993; Allen & Mikala, 1998; Yamaguchi et al. 2000). Such concerns underscore the importance of obtaining functional data to complement cloning results.

Here, by monitoring changes in native cardiac Ca²⁺ channel kinetics upon over-expression of distinct subunits, we firmly exclude _2a as a major component of rat cardiac L-type channels. This result is in agreement with the inability to obtain amplification products corresponding to _2a in rabbit heart (Qin et al. 1998). By contrast, the close agreement in single L-type channel gating signatures between control and _2b-expressing cells support a major contribution of this ₂ splice variant to the endogenous cardiac L-type Ca²⁺ channel current. From our present data, we cannot unambiguously discount the potential contribution of other ₂ splice variants to ventricular L-type Ca²⁺ channels. We are undertaking in-depth comparison of the biophysical properties of the different ₂ splice variants, as well as attempting their individual knock-out in heart cells to resolve these remaining uncertainties.

	REFERENCES

Top Abstract Introduction Methods Results Discussion References

ALLEN, T. J. & M