Manipulating L-type calcium channels in cardiomyocytes using split-intein protein transsplicing

Significance

Manipulating expression of proteins directly in isolated heart cells is an important method to study cardiac biology and disease. However, this approach is challenging for large genes (>6 kb) because these exceed the packaging capacity of commonly used viral vectors used to deliver genes into heart cells. Here, we used a split-intein–mediated protein transsplicing strategy to reconstitute the large pore-forming subunit of L-type calcium channels from two separate fragments in heart cells. This approach provides a general method to express different large proteins in heart cells and will help advance new insights into basic mechanisms underlying cardiac function and disease.

Abstract

Manipulating expression of large genes (>6 kb) in adult cardiomyocytes is challenging because these cells are only efficiently transduced by viral vectors with a 4–7 kb packaging capacity. This limitation impedes understanding structure–function mechanisms of important proteins in heart. L-type calcium channels (LTCCs) regulate diverse facets of cardiac physiology including excitation–contraction coupling, excitability, and gene expression. Many important questions about how LTCCs mediate such multidimensional signaling are best resolved by manipulating expression of the 6.6 kb pore-forming α_1C-subunit in adult cardiomyocytes. Here, we use split-intein–mediated protein transsplicing to reconstitute LTCC α_1C-subunit from two distinct halves, overcoming the difficulty of expressing full-length α_1C in cardiomyocytes. Split-intein–tagged α_1C fragments encoding dihydropyridine-resistant channels were incorporated into adenovirus and reconstituted in cardiomyocytes. Similar to endogenous LTCCs, recombinant channels targeted to dyads, triggered Ca²⁺ transients, associated with caveolin-3, and supported β-adrenergic regulation of excitation–contraction coupling. This approach lowers a longstanding technical hurdle to manipulating large proteins in cardiomyocytes.

Adult ventricular cardiomyocytes have a unique cytoarchitecture and intracellular milieu (including transverse tubules, dyadic junctions, and ryanodine receptors) that is not replicated in many other cell types (1). Consequently, many fundamental questions regarding structure–function mechanisms of cardiac signaling proteins can only be pursued in the context of native cardiomyocytes. Knock-in mouse models are unarguably the gold standard for structure–function studies of signaling proteins in heart. However, the high cost (∼$50,000) and length of time (up to 2 y) required to generate a single knock-in mouse makes it impractical to routinely use this model system for investigative studies of structure–function mechanisms in heart cells. Directly manipulating expression of target proteins in adult cardiomyocytes is an important tool for structure–function studies in heart (2). However, this approach is challenging because (i) adult cardiomyocytes are refractory to transfection using conventional methods and (ii) they can only be maintained in culture for short periods (3–4 d) before they dedifferentiate (2). Fortunately, adult cardiomyocytes are efficiently transduced by viral vectors. A major technical hurdle is that commonly used viral vectors have a packaging capacity of 4−7 kb (2, 3).

Studies of the cardiac L-type calcium channel (LTCC) illustrate these challenges. In heart, LTCCs mediate excitation–contraction (EC) coupling, control excitability, and regulate gene expression (1, 4). In ventricular myocytes, the majority of LTCCs are targeted to transverse tubules where they are closely apposed to intracellular Ca²⁺ release channels, ryanodine receptors (RyR2), at dyadic junctions (4, 5) (Fig. 1A). This spatial arrangement is critical for the high gain of Ca²⁺-induced Ca²⁺ release (CICR) that underlies cardiac EC coupling (6). A subset of cardiac LTCCs are found located in caveolae (7), and are hypothesized to selectively signal through local effectors to the nucleus (8, 9) (Fig. 1A). Cardiac LTCC currents are increased several-fold by protein kinase A (PKA), which is activated via β-adrenergic signaling (10). The molecular mechanisms and determinants underlying LTCC targeting to dyads and caveolae and modulation by PKA are either unknown or ambiguous. Reconstituting LTCCs in heterologous cells to address such questions has proven inadequate because these cells are deficient in essential structural elements such as transverse (t) tubules (Fig. 1B), and also lack a permissive environment for PKA modulation (11, 12). Moreover, because resolving these structure–function questions will likely require many different constructs, it is impractical to use knock-in mice for such investigative purposes. Hence, directly manipulating LTCCs in isolated adult ventricular myocytes is critical for addressing these questions. Cardiac LTCCs are multisubunit proteins containing a pore-forming α_1C assembled with auxiliary β and α₂δ subunits, and calmodulin (13) (Fig. 1C). The 6.6 kb size of α_1C is at the packaging capacity of commonly used viral vectors, making it difficult to generate viral vectors incorporating full-length α_1C. This deficiency has been a major roadblock to progress in determining structure–function mechanisms of cardiac LTCCs.

Fig. 1.

Model systems for structure–function studies of cardiac LTCCs. (A, Upper) Adult rat ventricular cardiomyocyte. (A, Bottom) Differential LTCC targeting to distinct subcellular microdomains in cardiomyocytes. (B) Model systems for LTCC structure–function studies. (Scale bar, 20 μm.) (C) Topology of LTCC pore-forming α_1C and auxiliary proteins (β, α2δ, and calmodulin).

To overcome the obstacle of routinely expressing LTCC α_1C and other large proteins in adult cardiomyocytes, we developed an approach that relies on split-intein–mediated protein transsplicing. Inteins are naturally occurring protein-splicing elements found in archaeal, eubacterial, and eukaryotic genes (14). When attached to two different polypeptides (termed “exteins”), transacting split inteins can rapidly associate to form an active intein that uses a self-catalytic reaction to splice the two exteins together with a peptide bond while excising itself out of the resulting protein sequence (15, 16). We used the split DnaE intein from the cyanobacterium Nostoc punctiforme (17, 18) to reconstitute full-length α_1C in situ from two separate halves. The split-intein–tagged α_1C fragments readily incorporated into adenovirus and reconstituted dihydropyridine (DHP)-resistant channels in cardiomyocytes. Similar to endogenous LTCCs, intein-spliced α_1C subunits targeted to dyads, triggered Ca²⁺ transients, associated with caveolin-3, and supported PKA regulation of EC coupling.

Results

Design and Functional Properties of Split-Intein–Spliced α_1C Reconstituted in HEK293 Cells.

LTCC α_1C subunit is a 2,171-residue protein containing four homologous domains each with six transmembrane segments and cytoplasmic N and C termini (19) (Fig. 1C). The homologous domains are joined by long intracellular loops. We split α_1C cDNA at the II–III loop and cloned the split-intein fragments of N. punctiforme DnaE intein into the C and N termini of the left ([I–II]_N-intein) and right (_C-intein[III–IV]) halves, respectively (Fig. 2A and Fig. S1). To visualize and distinguish reconstituted channels from endogenous LTCCs, we attached CFP and YFP to the N and C termini of [I–II]_N-intein and _C-intein[III–IV], respectively; engineered an extracellular epitope tag (bungarotoxin-binding site, BBS) (20, 21) into domain II S5–S6 loop of _CFP[I–II]_N-intein; and introduced two mutations (T1066Y/Q1070M) into _C-intein[III–IV]_YFP that render LTCCs relatively insensitive to DHP blockers (DHP–) (22, 23) (Fig. 2A). When coexpressed in cells, we predicted _CFP[I–II]_N-intein and _C-intein[III–IV]_YFP would interact to form a spliced α_1C intermediate in which the split N. punctiforme DnaE inteins associate to form an active intein. The active intein would then use a self-catalyzed reaction to splice the two channel halves together while excising itself out (Fig. 2A). Typically, three intein residues—cysteine, phenylalanine, and asparagine—are left behind in the spliced product. Taking advantage of an endogenous asparagine residue at the split site, we designed constructs so that the splicing reaction leaves behind only two foreign residues, cysteine and phenylalanine (Fig. S1).

Fig. 2.

Reconstituting LTCCs by split-intein–mediated protein transsplicing. (A) Strategy for reconstituting full-length α_1C using split-intein–mediated protein splicing. The α_1C subunit is split into two halves tagged with fluorophores and flanking N. punctiforme DnaE split inteins, creating _CFP[I–II]_N-intein and _C-intein[III–IV]_YFP, respectively. A 13-residue BBS tag is introduced to permit selective labeling of surface channels with quantum dot (QD₆₅₅). (B and C) Anti-GFP Western blot showing successful reconstitution of covalently linked full-length α_1C from intein-flanked α_1C moieties in HEK293 cells. Lane 1, full-length α_1C[BBS]–YFP (open circle) + β_2a-CFP (open square); lane 2, _CFP[I–II]_N-intein (inverted triangle); lane 3, _C-intein[III–IV]_YFP (open diamond); lane 4, _CFP[I–II]_N-intein + _C-intein[III–IV]_YFP (open triangle); lane 5, _CFP[I–II]_N-intein + _C-intein[III–IV^TQ/YM]_YFP (open triangle). Untagged β_2a is coexpressed in lanes 2–5. (C) Titration of _CFP[I–II]_N-intein to _C-intein[III–IV]_YFP transfection ratio (plus CFP-β_2a). (D) Confocal images showing CFP, YFP, and QD fluorescence in cells coexpressing _CFP[I–II]_N-intein + _C-intein[III–IV]_YFP ± β_2a. (Scale bar, 10 μm.) (E) Flow cytometry dot plot of YFP and QD fluorescence. (F) Quantification of QD₆₅₅ signal. *P = 0.0000427 compared with –β using two-tailed unpaired t test. n = 5, 50,000 cells per experiment.

We first used Western blotting to determine whether coexpression of split-intein–tagged α_1C moieties in HEK293 cells resulted in a reconstituted full-length protein as conceptualized in Fig. 2A. In cells expressing either _CFP[I–II]_N-intein or _C-intein[III–IV]_YFP alone, respectively, anti-GFP (which recognizes both CFP and YFP) Western blots revealed bands corresponding to the two half proteins (Fig. 2B, lanes 2 and 3). In cells coexpressing _CFP[I–II]_N-intein and _C-intein[III–IV]_YFP, we detected a higher molecular weight band at the expected size of the spliced WT α_1C channel product (Fig. 2B, lane 4). We similarly detected the spliced DHP-resistant α_1C channel when _CFP[I–II]_N-intein was coexpressed with _C-intein[III–IV^TQ/YM]_YFP (Fig. 2B, lane 5). The transsplicing reaction was highly efficient, as evidenced by the complete consumption of the moiety with limiting expression (_C-intein[III–IV]_YFP) in the reaction (Fig. 2B, lanes 4 and 5). The high efficiency of the transsplicing reaction is likely aided by the fact that the I–II and III–IV domains of Ca_V channels have a natural affinity for each other and interact when coexpressed (24–26). By titrating the relative amount of _CFP[I–II]_N-intein to _C-intein[III–IV]_YFP cDNA transfected, it was possible to approach a balanced level of expression where both fragments were essentially used up in the splicing reaction (Fig. 2C).

We next determined whether intein-spliced α_1C had similar functional properties to full-length (WT) α_1C subunit expressed in HEK293 cells. First, WT α_1C requires coexpression of Ca_Vβ for effective targeting to the cell surface (27, 28) (Fig. 2F). To selectively label surface channels, nonpermeabilized HEK293 cells coexpressing _CFP[I–II]_N-intein + _C-intein[III–IV]_YFP ± β_2a were sequentially labeled with biotinylated α-bungarotoxin (BTX) and streptavidin-conjugated fluorescent quantum dots (QD₆₅₅). Channel surface expression was analyzed by confocal imaging (Fig. 2D) and flow cytometry (Fig. 2E). Quantification of fluorescence intensity signals indicated Ca_Vβ increased surface expression of intein-spliced α_1C eightfold, similar to the magnitude of response seen with WT α_1C (Fig. 2F).

We next compared functional characteristics of currents reconstituted with intein-spliced α_1C subunits to those obtained with their full-length WT counterparts. LTCCs reconstituted with WT α_1C + β_2a generated robust whole-cell currents (Fig. 3A) that activated at a threshold potential of –30 mV and peaked at 0 mV (Fig. 3B). Moreover, WT channels were strongly and rapidly inhibited by 1 μM nifedipine, with complete loss of current observed for 5 μM and 10 μM nifedipine (Fig. 3 A–C). Intein-spliced WT α_1C channels reconstituted with _CFP[I–II]_N-intein + _C-intein[III–IV]_YFP + β_2a displayed virtually identical whole-cell currents and similar sensitivity to nifedipine as WT α_1C + β_2a (Fig. 3 D–F and Table S1). By contrast, intein-spliced DHP-resistant α_1C channels displayed currents that were relatively insensitive to nifedipine (Fig. 3 G–I). Expressing either _CFP[I–II]_N-intein or _C-intein[III–IV]_YFP separately did not yield either ionic or gating currents in HEK293 cells (Fig. S2). This contrasts with previous reports that expressing the I–II domain of Ca_V1.1 (α_1S) in skeletal myotubes results in measurable gating charge (24, 25). The difference could be due to either the distinct Ca_V1 isoforms or cell types involved.

Fig. 3.

Electrophysiological properties and DHP sensitivity of WT and split-intein–reconstituted LTCCs in HEK293 cells. (A) Exemplar Ba²⁺ currents from cells expressing full-length WT α_1C + β_2a in the absence (black trace) and presence of different nifedipine concentrations (1 μM, red; 5 μM, blue; 10 μM, cyan). (B) Population current-density versus voltage (J−V) relationship for cells expressing full-length WT α_1C + β_2a channels in the absence (black symbols) or presence (cyan symbols) of 10 μM nifedipine. n = 7 for each point. (C) Diary plots of nifedipine inhibition of full-length WT α_1C + β_2a channels. (D–F) Data for intein-spliced WT α_1C channels are in the same format as A–C. n = 9. (G–I) Data for intein-spliced DHP-resistant α_1C channels are in the same format as A–C. n = 10.

Expression and Trafficking of Intein-Spliced α_1C in Adult Cardiomyocytes.

Having validated the split-intein strategy in HEK293 cells, we next used this method to express LTCCs in adult cardiomyocytes. We chose adenoviral vectors because they readily infect adult cardiac myocytes and have a relatively fast onset of protein expression (<24 h). This is important because adult cardiac myocytes begin to dedifferentiate after 3–4 d in culture (2). We were unable to generate adenoviruses incorporating full-length α_1C, a common finding for many investigators in the field, that is attributable to the large insert size of α_1C and regulatory elements (∼7.2 kb) being at the packaging capacity limit of adenoviral vectors. By contrast, both intein-tagged α_1C moieties were readily packaged into adenoviral vectors (Fig. 4A). We infected adult rat cultured cardiomyocytes with _CFP[I–II]_N-intein and either _C-intein[III–IV]_YFP or _C-intein[III–IV^TQ/YM]_YFP and examined cells for protein expression 48 h after infection. The cardiomyocyte infection was highly efficient, with >90% of cells exhibiting robust CFP/YFP fluorescence, confirming expression of the intein-flanked α_1C moieties (Fig. 4B). Western blotting confirmed that the split-intein–tagged α_1C moieties were spliced to generate full-length protein in adult cardiomyocytes. Whole-cell lysates probed with anti-α_1C antibody detected both endogenous and intein-spliced α_1C (Fig. 4B). Intein-spliced α_1C was readily distinguishable from endogenous α_1C due to the presence of fluorescent protein tags that endowed them with a higher molecular weight. Intein-spliced α_1C expression level was titratable, with increasing protein expression observed at higher adenoviral doses (Fig. 4B).

Fig. 4.

Expression and subcellular targeting of intein-spliced α_1C in cardiac myocytes. (A) Schematic showing incorporation of split-intein–tagged α_1C moieties into adenovirus and infection of cardiomyocytes. (B, Left) Confocal images showing high efficiency expression of CFP- and YFP-tagged split-intein α_1C moieties in adult cardiomyocytes. (Scale bar, 50 μm.) (B, Right) Western blot (anti-α_1C antibody) detection of endogenous (#) and intein-spliced (&) α_1C subunits. (C) Confocal images of CFP, YFP, and QD₆₅₅ fluorescence in a cardiomyocyte expressing intein-spliced α_1C. Control shows absence of QD₆₅₅ staining in an uninfected cell. (Scale bar, 20 μm.) (D, Upper) Immunostaining of endogenous α_1C (anti-α_1C, green) and RyR (anti-RyR, red) in an uninfected cardiomyocyte. (D, Lower) Immunostaining of intein-spliced α_1C (anti-GFP, green) and endogenous RYR (anti-RyR, red). (Scale bar, 20 μm.) (E) Colocalization analyses.

To determine whether intein-spliced α_1C expressed in cardiomyocytes trafficked to the cell surface, we sequentially exposed nonpermeabilized infected cardiomyocytes to biotinylated BTX and streptavidin-conjugated QD₆₅₅. Cardiomyocytes coexpressing _CFP[I–II]_N-intein and _C-intein[III–IV]_YFP displayed robust red QD₆₅₅ fluorescence staining of the surface sarcolemma (Fig. 4C and Fig. S3). As a negative control, uninfected myocytes showed negligible QD fluorescence (Fig. 4C and Fig. S3). Similarly, cardiomyocytes expressing only _CFP[I–II]_N-intein displayed minimal QD₆₅₅ surface staining, indicating this moiety alone does not traffic to the cell surface (Fig. S3).

In adult ventricular myocytes, the close proximity of LTCCs and RyR2s at dyadic junctions can be demonstrated using immunofluorescence analyses showing a high degree of colocalization between endogenous α_1C and RyR2 (Fig. 4D), in agreement with previous reports (5). To determine whether intein-spliced α_1C was properly targeted to dyads we performed coimmunofluorescence experiments using anti-GFP and anti-RyR antibodies (Fig. 4D). Much of the anti-GFP staining was observed in regularly spaced transverse patterns that occurred with a periodicity of 1.8 μm, consistent with the staining pattern of t tubules in cardiomyocytes (Fig. 4D). Most importantly, quantitative analyses indicated intein-spliced α_1C colocalized with RyR2s to the same extent as endogenous channels, as demonstrated by similar Pearson’s correlation coefficient values (Fig. 4E). Similar values were obtained by quantifying this colocalization independently using Li’s intensity correlation coefficient (29) (Fig. S4).

Beyond localization at dyads, a subset of LTCCs in ventricular myocytes associate with caveolin-3 and target to caveolae (7, 30). Consistent with this, endogenous LTCCs exhibit significant colocalization with caveolin-3 in adult cardiomyocytes (Fig. S4). It is unknown whether α_1C subunits that are targeted to dyads are molecularly distinct from those associated with caveolin-3. One possible mechanism for differential LTCC targeting in cardiomyocytes could be alternative splicing of α_1C. We discovered that intein-spliced α_1C also colocalizes with endogenous caveolin-3 in cardiomyocytes (Fig. S4), indicating molecularly identical α_1C subunits can target to both dyads and caveolae.

Overall, these results show that intein-tagged α_1C halves expressed in cardiomyocytes are efficiently spliced together to generate an intact α_1C that traffics to dyads and associates with caveolin-3 similar to the endogenous protein. Because no Ca_Vβ was coexpressed in these experiments, the results suggest intein-spliced α_1C associates with endogenous Ca_Vβ to traffic to the cell surface.

Functional Characterization of Intein-Spliced α_1C Subunits in Cardiomyocytes.

We used whole-cell patch clamp to examine the functionality of intein-spliced α_1C subunits expressed in cardiomyocytes. Surprisingly, cardiomyocytes coexpressing _CFP[I–II]_N-intein and either _C-intein[III–IV]_YFP or _C-intein[III–IV^TQ/YM]_YFP displayed only a marginal increase in Ba²⁺ current density compared with uninfected control myocytes (Fig. 5 A–C). One possible explanation for this result is that endogenous Ca_Vβ subunits are rate-limiting for LTCC functional expression in heart. This interpretation would be consistent with prior observations that overexpressing Ca_Vβs alone in heart causes a several-fold increase in LTCC current density (11, 31). An alternate possibility was that the intein-spliced α_1C subunits were not functional despite making it to the cell surface (Fig. 4). To distinguish between the two possibilities we recorded currents in cardiomyocytes expressing either intein-spliced WT or DHP-resistant α_1C in the presence of 10 μM nifedipine (Fig. 5D). Under these conditions, cardiomyocytes expressing intein-spliced DHP– α_1C subunits expressed currents that were on average threefold larger than those expressing spliced WT α_1C, demonstrating that the expressed channels are functional (Fig. 5 D and E). We obtained a similarly large DHP-resistant current in cardiomyocytes coinfected with _CFP[I–II]_N-intein and _C-intein[III–IV^TQ/YM] expressed in a bicistronic adenovirus vector (Fig. S5). The bicistronic strategy can be used to overcome potential concerns that directly fusing fluorescent proteins to the α_1C fragments could disrupt aspects of their functions in some unanticipated way. Because DHP block of Ca_V1.2 is known to be voltage- and state-dependent, we examined whether we could achieve greater nifedipine separation of WT and DHP-resistant α_1C by using a –50 mV holding potential and 2 mM Ca²⁺ as charge carrier (Fig. S6). Under these conditions, WT Ca_V1.2 currents elicited by a ramp protocol were deeply inhibited by 10 μM nifedipine, whereas myocytes expressing intein-spliced DHP-resistant α_1C displayed a sevenfold larger remnant current (Fig. S6).

Fig. 5.

Electrophysiological characterization of intein-spliced WT and DHP-resistant α_1C in cardiac myocytes. Population J−V relationships in (A) uninfected cardiomyocytes (control), (B) myocytes expressing intein-spliced WT α_1C, or (C) intein-spliced DHP– α_1C. (D and E) Exemplar current traces and population J–V curves from cardiomyocytes expressing either intein-spliced WT α_1C (black trace and symbols, n = 9) or DHP– α_1C (red trace and symbols, n = 8) channels in the presence of 10 μM nifedipine. *P = 0.018 (–10 mV), *P = 0.0324 (0 mV), *P = 0.048 (10 mV), and *P = 0.04 (20 mV) using two-tailed unpaired t test.

To determine the functional competence of intein-spliced α_1C to trigger CICR at dyads, we measured rhod-2–reported Ca²⁺ transients using line scan confocal microscopy (Fig. 6). Control uninfected cells responded to 1 Hz field stimulation with robust basal Ca²⁺ transients that were inhibited by 1 μM nifedipine (Fig. 6 A and D). In myocytes expressing either intein-spliced WT (Fig. 6B) or DHP– (Fig. 6C) α_1C, basal Ca²⁺ transients were marginally larger than observed in uninfected controls, consistent with a modest increase in the number of surface LTCCs (Fig. 6D). Ca²⁺ transients in cardiomyocytes expressing intein-spliced WT α_1C were inhibited by 1 μM nifedipine (90% inhibition), with most cells unresponsive to field stimulation (Fig. 6 B and D). By contrast, cardiomyocytes expressing spliced DHP-resistant α_1C were only modestly inhibited by 1 μM nifedipine (20% inhibition) and virtually all cells remained responsive to field stimulation. Exposure of cardiomyocytes expressing spliced WT α_1C to 5 μM nifedipine completely eliminated Ca²⁺ transients in response to field stimulation in eight out of eight cells (Fig. 6E), whereas six out of nine cells expressing spliced DHP-resistant α_1C remained responsive at this higher DHP concentration (Fig. 6F).

Fig. 6.

Participation of intein-spliced WT and DHP– α_1C in CICR in cardiomyocytes. (A, Upper) The rhod-2-reported Ca²⁺ transients in an uninfected (UT) cardiomyocyte paced with 1 Hz field stimulation. (A, Lower) Ca²⁺ transients from UT myocyte in the presence of 1 μM nifedipine. (B and C) Ca²⁺ transients from cardiomyocytes expressing intein-spliced WT (α_1C) and DHP– α_1C, respectively. Same format as A. (D) Impact of 1 μM nifedipine on Ca²⁺ transient amplitudes. * P = 0.0014 (UT), * P = 0.0003 (α_1C), two-tailed unpaired t test. #, significantly different from control and intein-spliced WT α_1C +nifedipine using one-way ANOVA [F(2, 12) = 47.48, P = 2E-6] and Bonferroni pairwise comparisons, n = 5 cells. (E) CICR is abolished by 5 μM nifedipine in cardiomyocytes expressing intein-spliced WT α_1C. (F) Persistence of CICR in cardiomyocytes expressing intein-spliced DHP– α_1C in 5 μM nifedipine. (G) A total of 1 μM isoproterenol potentiates Ca²⁺ transients triggered by intein-spliced DHP– α_1C, *P = 0.018.

With the ability to completely isolate intein-spliced DHP-resistant α_1C-stimulated Ca²⁺ transients with 5 μM nifedipine, we examined whether this channel was permissive for sympathetic up-regulation of EC coupling under these experimental conditions. Exposure of cardiomyocytes expressing spliced DHP-resistant α_1C to 5 μM nifedipine + 1 μM isoproterenol resulted in Ca²⁺ transients with an elevated peak compared with those obtained with 5 μM nifedipine alone (Fig. 6G). The sympathetic elevation of Ca²⁺ transients under these conditions is likely due to a combination of PKA-dependent phosphorylation of Ca_V1.2, RYR2, and phospholamban (32).

Discussion

We have developed a unique method to readily manipulate tagged α_1C expression in isolated adult cardiomyocytes. The unique technology offers clear advantages over using heterologous expression systems to study the functional properties of cardiac LTCCs. Heterologous cells lack the unique cytoarchitecture of adult cardiomyocytes. Consequently, questions such as the mechanisms underlying differential LTCC targeting to dyads and caveolae, or their putative role in pathological cardiac hypertrophy, cannot be studied in noncardiomyocytes. It has also proven difficult to determine the mechanism of PKA-induced enhancement of LTCCs from heterologous expression experiments. Based on studies in HEK293 cells, it was suggested that phosphorylation of Ser1928 in α_1C and Ser478/Ser479 in β_2a was critical for PKA regulation of LTCCs (33, 34). However, the putative role of these phosphorylation sites in PKA regulation of LTCC was ruled out using α_1C[Ser1928Ala] mutant channels and truncated-β₂ knock-in mice (35, 36), as well as experiments in isolated adult cardiomyocytes (11, 37). More recent data using HEK293 point to an essential role for phosphorylation of Ser1700 and Thr1704 in PKA regulation of LTCCs (38). The importance of these residues has not been directly confirmed in cardiac myocytes. Overall, these discrepancies emphasize the importance of addressing questions pertaining to LTCC trafficking and functional regulation in the context of actual heart cells. A caveat is that the isoproterenol-induced increase in peak evoked Ca²⁺ release we measured was relatively modest and likely also includes contributions by PKA regulation of RYR and phospholamban. In part, the relatively modest response may be attributable to the use of rat cardiomyocytes in which PKA modulation of Ca_V1.2 is much less robust than in other species such as guinea pig (11). We are currently working to establish conditions for strong PKA modulation of spliced DHP-insensitive α_1C in guinea pig cardiomyocytes.

Knock-in mouse models are the gold standard for defining structure–function mechanisms of proteins in heart. The approach of manipulating α_1C expression with adenoviral vectors in isolated adult cardiomyocytes offers several advantages over knock-in mouse models. First, knock-in mice are costly and take up to 2 y to generate, making their routine use for in-depth structure–function studies requiring many different constructs impractical. By comparison, split-intein adenoviruses are more cost and time effective. Second, certain questions cannot be pursued in knock-in mice due to lethality of the knocked in gene. For example, knock-in mice featuring C-terminal truncations of α_1C unexpectedly display a marked decrease in L-type calcium current and die shortly after birth due to heart failure (39, 40). This precludes the use of knock-in mice to pursue relevant questions regarding the role of α_1C C terminus in LTCC trafficking and functional regulation in adult cardiomyocytes. Third, the adenoviral infection approach permits questions to be easily addressed in different species. This is important because the functional properties of murine cardiomyocytes differ in several critical ways from those of human, such as the action potential duration being considerably shorter in mice (41). There have been scattered reports of successful expression of LTCC α_1C subunits in adult cardiomyocytes (37, 42, 43). Previously, by additionally deleting the fiber gene from the standard adenoviral vector backbone (E1 and E3 deleted) and generating a complementing stable cell line, we were able to generate adenovirus containing full-length α_1C (37, 42). However, this system proved unstable and further attempts to generate new adenoviruses were unsuccessful. There has been one report using lentivirus to express α_1C subunits in adult rat cardiomyocytes (44). We generated lentiviruses expressing GFP-tagged α_1C truncated in the C terminus at residue 1905, which transduced HEK293 cells efficiently, expressing α_1C[1905]–GFP within 48 h. However, in our hands, the transduction efficiency of adult cardiomyocytes was low, and required several days for transgene expression, a problematic feature given the short half-life of rod-shaped cardiomyocytes in culture. Recently, a biolistic transfection method was adapted to express α_1C in adult rat ventricular myocytes (41). Compared with our approach, the biolistic method suffers from high toxicity (∼10% viable cells) and relatively low transfection efficiency (∼30% of surviving cells express transgene) (45).

Our unique application of split-intein technology engineers functional expression of a large protein with high efficiency in adult heart cells. Beyond LTCCs there are many other large cardiac proteins (e.g., voltage-gated sodium channel, ankyrin, spectrin, myosin, ryanodine receptor, and titin) whose structure–function properties, trafficking, and functional modulation in heart are of major interest, but currently only accessible using knock-in mice. The split-intein approach can be extended to explore the functional properties of many other large proteins in cardiomyocytes. Overall, the split-intein method expands the toolkit available, which includes high-capacity HSV-1 amplicon vectors (46), for manipulating expression of large genes in primary cardiac myocytes.

Materials and Methods

Generation of Plasmid and Adenoviral Vectors.

GenBank IDs of cDNAs used are provided in Table S2. Detailed cloning strategies to generate expression plasmids are provided in SI Materials and Methods. PCR primer sequences used for cloning are provided in Table S3. Replication-deficient adenoviruses were generated using the Clontech Adeno-X System 3. Viral expansion and purification was carried out as described (31).

Cell Culture and Transfection/Infection.

HEK293 cells cultured on 60 mm dishes were transfected with the appropriate cDNA (6 μg each) and T-antigen (1 μg) using calcium phosphate precipitation. Adult rat cardiomyocytes were prepared as described (31, 47) and cultured in medium supplemented with 0.5 μM cytochalasin D to preserve cell morphology (48). Cells were infected with 5–15 μL of viral stock in a final volume of 1–2 mL.

Quantum Dot Detection of Cell Surface BBS-Tagged LTCCs.

Surface LTCCs in HEK293 cells or cardiomyocytes were labeled with quantum-dots as described (21), and assayed by flow cytometry using a BD LSR II Cell Analyzer (BD Biosciences). Flow cytometry data were analyzed using FloJo Software (21).

Intracellular Calcium Transient Measurements.

Primary myocytes in culture were loaded with the calcium-sensing dye rhod2–AM. Evoked intracellular calcium transients were measured as described in SI Materials and Methods.

Image Processing and Analysis.

Confocal images were analyzed using ImageJ software and background subtracted. Colocalization of CFP/YFP, α_1C/RyR, and α_1C/caveolin-3 was carried out using the ImageJ JaCoP plugin. Colocalization in two channels was quantified using Pearson’s and Li’s colocalization coefficient algorithms in JaCoP (29).

Electrophysiology.

Whole-cell recordings were conducted on HEK293 cells 48 h posttransfection using an EPC-8 patch clamp amplifier (HEKA Electronics) controlled by PULSE software as previously described (21, 28). Whole-cell recordings of rat ventricular myocytes were performed as described previously (31).

Analysis and Statistics.

Data were plotted and statistical analyses performed in Origin software using built-in functions. Significant differences between means (P < 0.05, P < 0.01) were determined using Student t test for comparison between two groups, or one-way ANOVA followed by Bonferroni post hoc analyses for comparisons involving more than two groups. Data are represented as means ± SEM.