Increased Expression of the Auxiliary β2-subunit of Ventricular L-type Ca2+ Channels Leads to Single-Channel Activity Characteristic of Heart Failure

Background Increased activity of single ventricular L-type Ca2+-channels (L-VDCC) is a hallmark in human heart failure. Recent findings suggest differential modulation by several auxiliary β-subunits as a possible explanation. Methods and Results By molecular and functional analyses of human and murine ventricles, we find that enhanced L-VDCC activity is accompanied by altered expression pattern of auxiliary L-VDCC β-subunit gene products. In HEK293-cells we show differential modulation of single L-VDCC activity by coexpression of several human cardiac β-subunits: Unlike β1 or β3 isoforms, β2a and β2b induce a high-activity channel behavior typical of failing myocytes. In accordance, β2-subunit mRNA and protein are up-regulated in failing human myocardium. In a model of heart failure we find that mice overexpressing the human cardiac CaV1.2 also reveal increased single-channel activity and sarcolemmal β2 expression when entering into the maladaptive stage of heart failure. Interestingly, these animals, when still young and non-failing (“Adaptive Phase”), reveal the opposite phenotype, viz : reduced single-channel activity accompanied by lowered β2 expression. Additional evidence for the cause-effect relationship between β2-subunit expression and single L-VDCC activity is provided by newly engineered, double-transgenic mice bearing both constitutive CaV1.2 and inducible β2 cardiac overexpression. Here in non-failing hearts induction of β2-subunit overexpression mimicked the increase of single L-VDCC activity observed in murine and human chronic heart failure. Conclusions Our study presents evidence of the pathobiochemical relevance of β2-subunits for the electrophysiological phenotype of cardiac L-VDCC and thus provides an explanation for the single L-VDCC gating observed in human and murine heart failure.


INTRODUCTION
Homeostasis of intracellular Ca 2+ concentration [Ca 2+ ] i is essential for cardiac function and integrity; its dysregulation is a hallmark of advanced heart failure [1,2]. Voltage-dependent L-type Ca 2+channels (L-VDCCs) are the source of trigger Ca 2+ entering cardiomyocytes [3]. Data derived from numerous studies support an involvement of L-VDCC in pathological changes of [Ca 2+ ] i in heart failure. Although still controversial, L-VDCC current density appears unchanged in failing cardiomyocytes [1,4,5]. Whole-cell currents are determined by a number of parameters, including number of channels, single-channel current amplitude and time spent in the open state. Therefore, altered number of active channels or activity of individual L-VDCC is not necessarily reflected by calcium current density. In fact, despite no change in whole-cell L-VDCC density (I Ca ), single-channel activity was significantly increased in ventricular myocytes from human endstage failing hearts [6]. Chen et al. [7] showed attenuated I Ca increase by (S)-BayK8644 in human failing myocardium whereas basal whole-cell currents were unchanged, indicating that singlechannel activity is enhanced while channel density is lowered. These findings confirm the idea of an ''electrophysiological heartfailure phenotype'' of single L-VDCCs. The biochemical nature of this change in phenotype has not been delineated, although phosphorylation [8,9] and dephosphorylation [10,11] have been implicated. Activities of kinases and phosphatases not only change channel function but interfere with neurohumoral modulation of the L-VDCC; e.g. b-adrenergic regulation is blunted in heart failure possibly due to hyperphosphorylation of L-VDCCs [6,7]. Using heterologous recombination we have shown that distinct subunit compositions of L-VDCC induce single-channel characteristics similar to the biophysical phenotype of ''hyperphosphorylated'' L-VDCC [12]. The latter suggests that changes in gene expression of L-VDCC subunits may form the basis of a heart-failure phenotype of L-VDCC. In mammalian hearts L-VDCCs are composed of an ion conducting pore (Ca V 1.2 or a 1C ), and two auxiliary subunits, an a 2 d and a b-subunit. Most investigators agree that b-subunit diversity is of physiological and pathophysiological importance [13][14][15][16][17][18]. In fact, some studies have revealed altered b-subunit patterns in human heart failure [19,20], suggesting that an altered b-subunit expression pattern is of functional relevance. Delineation of pathophysiological mechanisms in human heart is difficult because of wide inter-individual variance, including age, medication, state of disease etc. Human tissue also offers a limited choice of truly independent variables, such as time, disease stage and treatment options. Animal models offer control of any relevant factor to test pathophysiological concepts. We analyzed b-subunit gene expression in both human non-failing and failing hearts as well as in transgenic mice overexpressing the human Ca V 1.2 (a 1C ) subunit (tg Ca V 1.2). The latter was chosen because of phenotypical characteristics common with human heart failure, e.g. early blunting of b-adrenergic signaling, slow progression towards hypertrophy and calcium overload in failing myocytes [21,22]. Most importantly, in young (non-failing; ''Adaptive State'') tg Ca V 1.2 mice we previously found concordance of lowered b 2 -subunit expression and decreased activity of single L-VDCC [16]. In the present study we find an increase of single L-VDCC activity accompanied by enhanced expression of b 2 -subunits when these mice have entered the failing state (''Maladaptive State'' $9 months of age). By examination of a new, double-transgenic mouse bearing both constitutive Ca V 1.2 and inducible b 2 -subunit overexpression in the heart we show a relationship between subunit expression and channel function.

Gating parameters of single L-VDCC in failing human and transgenic myocardium
Single-channel activity of old ($9 months, and in heart failure) tg Ca V 1.2 was significantly increased compared to young (4 months, no hypertrophy or heart failure) tg Ca V 1.2 [16] (Table 1). Henceforth, these animals are referred to as ''young'' and ''old'' tg Ca V 1.2, respectively. Peak ensemble average current (I peak ) in old tg Ca V 1.2 mice was enhanced (256614 fA vs. 22367 fA, p,0.05) due to an increased fraction of active sweeps, mean open time, and mean open probability, and a decrease of mean closed time (t closed ). Of further interest, the changes of peak current, fraction of active sweeps and open probability mirror findings obtained from single L-VDCC measurements in human cardiomyocytes from non-failing or failing idiopathic dilated cardiomy-opathy (DCM) hearts, respectively [6] (Table 1).

L-VDCC subunit expression in non-failing and failing human hearts
Protein expression of Ca V 1.2, a 2 d, low molecular weight b 1 , and b 3 was similar in non-failing and failing human myocardium, but we found a significant up-regulation of b 2 (Figure 1a,b). There was no difference in gene expression of the Ca V 1.2, and the a 2 d at the protein level (mRNA data not shown). At least two b 1 -subunit isoforms (b 1a,c ), four b 2 -subunit isoforms (b 2a-d ), and two b 3subunit isoforms (b 3a,trunc ) are expressed at relevant levels in human myocardium [12]. b 1a (GenBank No NM_199247) and b 1c (GenBank_199248) are sequence-identical except for replacement of exon 7a by exon 7b in b 1c , consistent with previous work [23]. b 2a-d isoforms differ only with respect to the N-terminal region (D1 domain). Quantitation by real-time PCR revealed an increased expression of b 1c and all b 2 isoforms in heart failure, in line with the protein data ( Figure 1c).
L-VDCC subunit protein expression in old wild type and old tg Ca V 1.2 hearts b 1 , b 2 and b 3 isoforms are expressed at the protein level in old Ca V 1.2 mouse heart, although expression of b 1 -subunit isoforms was faint. Compared to old wild type mice ($9 months), the old tg Ca V 1.2 showed a significant up-regulation of b 3 -, and b 2 -subunits ( Figure 2), in striking contrast to the down-regulation of b 2subunits we previously observed in young tg Ca V 1.2 [16]. In the old tg Ca V 1.2 mouse myocytes, up-regulated b 2 -subunits and overexpressed Ca V 1.2 both localize to the surface sarcolemma and the T-tubules (Figure 3), suggesting the functional relevance of altered expression levels.
b-subunit-dependent modulation of Ca V 1.2 expressed in HEK293 The diversity of b-subunit expression patterns found in cardiomyocytes necessitated the functional characterization of each bsubunit isoform. Using HEK293 cells constitutively expressing the human Ca V 1.2 as a homologous recombination system we show that the gene as well as alternative splicing determines calcium channel gating, extending and elaborating on previous work [12]. This is highlighted by significant differences in peak current and Table 1. Single L-VDCC gating of young and old tg Ca V 1.2 resembles data obtained from human non-failing and failing ventricle . In a previous study [6] we found single-channel activity to be significantly increased in ventricular myocytes from human hearts failing due to idiopathic dilated cardiomyopathy compared to non-failing ventricles. In excellent agreement the present study reveals activity of single L-VDCC from $9 months old, i.e. failing murine hearts overexpressing the human Ca V 1.2 to be significantly increased compared to single-channel activity in 4 months old, i.e. non-failing young transgenics obtained in a previous study [16].   (Table 2), with b 2a and b 2b exerting the strongest effects. b 2c or b 2d as well as b 1a , b 1c , and b 3a induced a minor to moderate increase in single-channel activity with no significant effects detected for closed times. Taken together, the data support the view that the single-channel phenotype of failing cardiomyocytes is caused by channel complexes containing b 2a or b 2b .
Generation of an inducible, heart-specific b 2a -subunit overexpression mouse (tg ind b 2a ) Our functional analyses support the idea of pathophysiological relevance of b 2 -subunit up-regulation, but the parallel biophysical and biochemical changes in cardiomyocytes may still be coincidental. Rather than following the natural course of gene expression changes, transgene-controlled b 2 -subunit overexpression should prove its causative role in native tissue. A hybrid drosophila-bombyx ecdysone receptor (VgBmEcR) when coupled to an aMHC promoter should combine strictly drug-controlled, transgene-specific, and cardiac tissue-specific gene induction. We

DISCUSSION
The major, novel result of our study is the concomitant increase in b 2 -subunit expression and L-VDCC activity in three independent models: human dilated cardiomyopathy, old tg Ca V 1.2 mice spontaneously progressing into heart failure and young (''Adaptive State'') tg Ca V 1.2 mice with additional tissue-specific inducible overexpression of b 2 -subunits. We explain our results by the differential effects of the b-subunits, namely b 2a and b 2b as the most important modulators in recombinant assays. The three b-subunits (b 1-3 ) thus far known to be expressed in mammalian hearts have vastly different effects on current density [17,24], kinetics [12,24], and single-channel properties [12,14,17]. Beyond that, some b 2 -subunit isoforms have been implicated in mediating membrane targeting [3,25], cardiomyocyte apoptosis [26], cell death [17], adrenergic regulation [27,28] and modula-  . Protein expression of cardiac L-VDCC subunits in old wild-type and tg Ca V 1.2 mice (a) Specimens from old wild-type mice and tg Ca V 1.2 in heart failure were analyzed in immunoblots using specific polyclonal antibodies directed against the particular L-VDCC subunits. (b) Protein expression of L-VDCC subunits was always normalized to cardiac calsequestrin protein expression in the same sample. Quantitative analysis of subunit protein expression is depicted as ratio of 10 months old tg Ca V 1.2 vs. age-matched wild-type. b 1 protein bands were faint, and thus not analyzed quantitatively (number of WT/old tg Ca V 1.2 specimens was always identical for each subunit; n = 4). * p,0.05. doi:10.1371/journal.pone.0000292.g002 tion of CaM Kinase II [29,30]. It is of interest that the pattern of auxiliary subunits [3], including the prevalence of b-subunit isoforms varies among species, with b 2 -subunits predominating in small rodents like rats [17], and mice [31], and additional relevant expression of b 1 -and b 3 -subunits in humans [13,19,23].
In our study the homologous recombination of human L-VDCC subunits indicates the biophysical relevance of b 2a and b 2b isoforms for up-regulation of single L-VDCC activity. The increase of b 2 -subunit gene expression in human heart failure suggests that these L-VDCC subunits form the basis of the ''heartfailure phenotype'' of single L-VDCC found in human hearts [6]. Young tg Ca V 1.2 animals were chosen as a known example of dynamic adaptation of b 2 -subunit expression in heart. The functional relevance of this adaptation is illustrated by whole-cell current density, that is increased by only 50% [22,32], while Ca V 1.2 protein expression and density of single channels show a 3fold increment in these transgenic hearts [16,32]. This apparent discrepancy is explained by an up-regulation of b 1 -but substantial down-regulation of the b 2 -subunit expression [16]. This gene expression is characteristic for the ''Adaptive phase'' of the model [3,22] putatively limiting calcium overload at the younger ages. We now demonstrate that when this tg Ca v 1.2 mouse enters into the ''Maladaptive state'' with overt heart failure at $9 months of age, both single L-VDCC activity and b 2 -subunit expression increase, mimicking alterations of channel structure and biophysics in terminal human heart failure. Thus, the old tg Ca V 1.2 mouse may be regarded as a heart-failure model in which a primary calcium (current) overload can no longer be effectively counterbalanced by adaptive mechanisms, i.e. b-subunit expression. The transcriptional mechanisms underlying this bidirectional control of b 2 -subunit expression, however, remain to be elucidated in the context of changes in b 1 and b 3 subunit expression in human and old tg Ca V 1.2 mouse heart failure.
As a novel and first approach to induce an increased b 2 -subunit overexpression in intact animals, rather than in isolated cells [17,24,26] we generated a mouse model of cardiospecific inducible b 2a -subunit expression (tg ind b 2a ). Induction of b 2 -overexpression in this mouse model did not affect overall single L-VDCC gating significantly. As a more recent study indicates that an 1:1stoichiometry of pore-forming a 1 -and auxiliary b-subunit may be sufficient for modulation of channel gating [33], we assume that most calcium channel pores are saturated with native b-subunits in the induced tg ind b 2a . However, mean closed time was lower in induced tg ind b 2a suggesting that a portion of overexpressed b 2a exerts functional action similar to the recombinant channel data presented. To prove our concept that b 2 -subunit expression underlies the activity of single L-VDCC of the heart-failure phenotype we crossbred tg ind b 2a with tg Ca V 1.2 mice [21,22]. Induction of b 2a -subunit gene expression in the young double tg mice (tg Ca V 1.26tg ind b 2a ) led to a premature increase of single L-VDCC activity. This confirms our theory, derived from recombinant channel data, in the relevant tissue ex vivo.
Such deliberate overexpression of b 2 -subunit in vivo, when carried forward in a chronic manner, hopefully will pave the way for understanding the progression of heart failure if these alterations in single L-VDCC gating lead to decompensation at an earlier age of the animal. This knowledge will have direct implications because pharmacological agents which modulate L-VDCC function are in everyday clinical practice and have been shown to be beneficial in various clinical trials targeting different populations [34,35]. We wish to emphasize that, at this point in our studies, we show a relationship between electrophysiological parameters that is consistent with heart failure. In order to prove this, it is necessary to chronically imbue the young animals with a heart-specific increase in the b 2 subunit and follow their transition to heart failure at specific age points as they mature. These experiments are ongoing but will require considerable time.

Materials
Non-failing and failing human left ventricular specimens were obtained from explanted hearts not transplanted for technical reasons (n = 5), or from orthotopic heart transplantation recipients (n = 8). Heart failure patients were in NYHA class III-IV (17-63 y, 3 females) mean duration of symptomatic heart failure ranged from 9-60 months, peak oxygen exercise capacity was 13.3-15.5 ml kg 21 min 21 at time of listing. Heart transplant recipients were ambulatory at time of operation and received treatment with inhibitors of the angiotensin-system (100%), bblockers (75%), aldosterone antagonist (50%), diuretics (62.5%).

Animals
Mice with cardiac-specific heterozygous overexpression of the human Ca v 1.2 [21] were bred with non transgenic littermates. These animals were bred with the tg ind b 2a as described below. Non transgenic littermates served as WT controls in this study.

Western-blot analysis of Ca 2+ -channel subunits
Protein expression levels of the L-VDCC subunits were assayed by Western-blot analysis of human and mouse cardiac ventricular protein samples. Briefly, protein extracts were obtained by homogenizing frozen heart tissue in buffer (5% SDS, 50 mM TRIS-HCl, pH 7.4, 250 mM sucrose, 75 mM urea, and 10 mM DTT containing complete protease inhibitor cocktail tablet from Roche) using a Teflon homogenizer. The homogenate was denatured by incubation at 95uC (2 min) followed by centrifugation at 16,000g (5 min); supernatants (containing membrane fractions and cytosolic proteins) were collected for analysis. Protein was quantified using Bicinchoninic acid (BCA) Protein Assay (PierceH). For Ca V 1.2 and a 2 d-1 Western blots, 60 mg, and for b 2 and b 3 Western blots, 150 mg of total protein were separated on a 8% and 12% SDS-PAGE gel (BioRadH). Gels were transferred to nitrocellulose membranes (AmershamH) according to standard wet transfer procedure. L-VDCC subunits were detected using the following antibodies: anti-human Ca V 1.2 against the II-III loop (generous gift from Dr. Hannelore Haase, Max-Delbrück Center, Berlin, Germany; [37,38]); anti-b 2 (generous gift from Dr. Adolfo Cuadra (Dr. M. Hosey), Northwestern University, Chicago, USA; [39]); anti-b 3 and anti-a 2 d21 (Alomone; [40,41]), and anticalsequestrin (Santa Cruz;[42]). The anti-b 1 (Swant; cp. [16]) stains a band at ,57 kDa in the membrane of human skeletal muscle (data not shown) where b 1a is pre-dominant [43] suggesting that the antibody detects a b 1a in human and mouse myocardium. In the present study this antibody detected an additional band of ,65 kDa in murine myocardium and ,70 kDa in human heart. Though our present mRNA data indicate that there are two b 1isoforms in cardiac tissue we cannot exclude a cross-reaction of the antibody with b 3 -subunits since the second band is quite close to the band detected by the b 3 -specific antibody from Alomone (see above). Thus we decided to avoid any quantitation of this ''high molecular band'' detected in murine and human cardiac tissue, respectively.

Immunofluorescence analysis of Ca 2+ -channel subunits
Ventricular myocytes were freshly isolated from 10 month old tg Ca V 1.2 and controls as previously described, stored in Kraft-Bruehe solution and plated on laminin-coated with poly-L-lysine and 50 mg/ml mouse laminin (InvitrogenH) coverslips for 1 h at 37u C, 5% CO 2 . After incubation myocytes were washed with relaxation buffer (mM: 100 KCl, 5 EGTA, 5 MgCl 2 , 0.25 dithiothreitol (DTT) in PBS, pH 6.8). Myocytes were then fixed in pre-cooled (220uC) methanol/acetone (1:1) for 5-10 min at 4uC. To prevent non-specific binding, myocytes were blocked with 10% normal donkey serum (SigmaH) in PBS overnight (labeling buffer) [44]. Primary antibodies were diluted in labeling buffer and incubated with myocytes overnight at 4uC. Primary antibody dilutions for different subunits of the L-VDCC studied were: 1:200 for a 1C (AlomoneH) and 1:500 for b 2 . In the case of Wheat Germ Agglutinin (WGA) labeling, myocytes were incubated overnight at 4uC with Oregon Green 488-conjugated WGA (Molecular ProbesH) at a concentration of 1 mg/ml. WGA selectively binding to N-acetyl-d-glucosamine in glycoproteins was used to label the peripheral sarcolemma, the T-tubules and the intercalated disks. After overnight incubation, myocytes were washed with PBS and incubated with secondary antibody in PBS-0.1% BSA for 1 h at room temperature. Secondary antibody for the study of the L-VDCC subunits was tetramethylrhodamine-isothiocyanate (TRITC)-conjugated donkey anti rabbit antibody at 1:400 dilution (Jackson ImmunoResearchH, USA). For negative control experiments, myocytes were kept in labeling buffer overnight without primary antibody and only incubated with secondary antibody at the same concentration. After washing the cells with PBS, coverslips were mounted on slides using Gel/Mount aqueous mounting media (BiomedaH) and images were acquired on a Nikon PCM 2000 laser confocal scanning microscope as 0.5 mm ''optical sections'' of the stained cells, keeping gain and background values constant through the different samples.

Generation of transgenic mice with inducible cardiac overexpression of b 2a
Recent modifications in the Drosophila ecdysone receptor revealed better regulation of gene expression in mammalian cells, however, the dependence on steroidal ligand activation (i.e. ponasteron) with its potential additional effects on gene expression remains. The ecdysone receptor from Bombyx mori is activated by the non-steroidal ligand tebufenozide (effective drug in MIMIC, Dow AgroSciencesH, Munich, Germany) without known specific interaction in mammalian cells. This construct regulated bgalactosidase expression in HEK293 cells at concentrations of 1 mM tebufenozide as effectively as the Drosophila ecdysone receptor (data not shown). For our experiments we intraperitoneally injected 9 mg tebufenozide (i.e. twice the dose leading to maximum serum concentration of the drug) 48h before isolation of cardiac myocytes. The hybrid drosophila-bombyx ecdysone receptor (VgBmEcR) was constructed by fusion of the binding and transactivation domain of the modified drosophila system (pVgRXR, InvitrogenH) to the ligand binding domain of the bombyx ecdysone receptor (BmEcR in pBSII KS+BmB1 = ecdysone receptor type B1 of Bombyx mori, obtained from Fujiwara H., Tokyo, Japan) using the restriction enzyme BsrGI and NotI. The coding sequence of VgBmEcR was set under control of the promoter of aMHC for cardio-specific expression. The aMHC promoter was excised from pBlue-MHCb 1 ARSV40polyA (obtained from Stefan Engelhardt, Wuerzburg, Germany) using DraI and PvuII and inserted into pVgBmEcR opened with the same enzymes. The ecdysone-regulated plasmid pInd-b 2a was constructed by excision of the b-galactosidase coding sequence from pInd-LacZ (InvitrogenH) using HindIII and XbaI and insertion of the coding sequence for rat b 2a excised from pCR3-b 2a -6myc (obtained from A.J. Chien, Chicago, USA) with the same restriction enzymes. The linearized coding sequence of constructs were injected simultaneously into embryonic stem cells, and mice transgenic for the VgBmEcR and the b 2a -subunit were identified by PCR using construct-specific primers (not depicted), and by Southern blot using aMHC-VgBmEcR and b 2a specific probes labeled radioactively. Mouse DNA was obtained from mice 3 weeks post delivery and digested with EcoRI for proof of aMHC-VgBmEcR-, and HindIII/BamHI for Ind-b 2a -genomic integration. Probes specific for aMHC-VgBmEcR and Ind-b 2a were obtained by SacI and KpnI/HindIII restriction of the respective coding sequences. Probes were radio-labeled with a-32 P-CTP using the Klenow fragment. Animals positive for integrated coding sequences were identified by 3.7 kb hybridization signal for aMHC-VgBmEcR and a 2.4 kb signal for Ind-b 2a (Figure 4a).

Isolation of ventricular myocytes
Single ventricular myocytes were isolated from murine hearts by enzymatic dissociation using the method described earlier [47]. In brief, hearts were perfused with a collagenase solution (Worthington type I and II, 75 U l 21 ) in a Langendorff setup and subsequently cut into small chunks. Myocytes were harvested by pouring the suspension through cheesecloth.

Data analysis and statistics of single-channel recordings
Linear leak and capacity currents (averaged non-active sweeps) were digitally subtracted. Openings and closures were identified by the half-height criterion. The fraction of active sweeps within a patch (f active ), the open probability within active sweeps (P open ), and the peak value of single-channel ensemble average currents (I peak ) were determined as described [16]. Where necessary, these parameters were corrected for the number of channels in a patch, as described [6]. For comparisons unpaired Student's two-tailed ttest or Mann-Whitney test was used where appropriate. Throughout, a level of p,0.05 was considered significant. Values are given as mean6SEM.