mitoBK_Ca is encoded by the Kcnma1 gene, and a splicing sequence defines its mitochondrial location

Abstract

The large-conductance Ca²⁺- and voltage-activated K⁺ channel (BK_Ca, MaxiK), which is encoded by the Kcnma1 gene, is generally expressed at the plasma membrane of excitable and nonexcitable cells. However, in adult cardiomyocytes, a BK_Ca-like channel activity has been reported in the mitochondria but not at the plasma membrane. The putative opening of this channel with the BK_Ca agonist, NS1619, protects the heart from ischemic insult. However, the molecular origin of mitochondrial BK_Ca (mitoBK_Ca) is unknown because its linkage to Kcnma1 has been questioned on biochemical and molecular grounds. Here, we unequivocally demonstrate that the molecular correlate of mitoBK_Ca is the Kcnma1 gene, which produces a protein that migrates at ∼140 kDa and arranges in clusters of ∼50 nm in purified mitochondria. Physiological experiments further support the origin of mitoBK_Ca as a Kcnma1 product because NS1619-mediated cardioprotection was absent in Kcnma1 knockout mice. Finally, BK_Ca transcript analysis and expression in adult cardiomyocytes led to the discovery of a 50-aa C-terminal splice insert as essential for the mitochondrial targeting of mitoBK_Ca.

The large-conductance Ca²⁺- and voltage-activated K⁺ channel (BK_Ca) is ubiquitously expressed and especially abundant at the plasma membrane of smooth muscle and neuronal cells, where they control contractility, resting membrane potential, neurotransmitter release, and action potential repolarization. In the adult heart, BK_Ca expression at the plasma membrane of myocytes has not been reported; however, electrophysiological, biochemical, and pharmacological evidence support its presence in mitochondria (1, 2). The presumed activation of mitochondrial BK_Ca (mitoBK_Ca) with NS1619 protects the heart from ischemia and reperfusion (1), which highlights the significant role of this channel in cardiac function.

Cardiac mitoBK_Ca shares functional characteristics, such as large conductance and inhibition by paxilline and charybdotoxin, with its plasma membrane counterpart. This resemblance, together with the reported Ca²⁺/voltage-dependency of mitoBK_Ca from human glioma (3), indicates that the molecular structure of this protein is closely related to the plasmalemmal BK_Ca. The plasmalemmal BK_Ca is minimally assembled by four pore-forming α-subunits encoded by a single gene, Kcnma1, which can undergo significant splice variation. The minimal protein, which is encoded by 27 constitutive exons, has a molecular mass of ∼125–130 kDa, depending on the start site (4, 5). However, proteins that migrate at ∼55 kDa (1, 6)—but also at ∼125 kDa (2)—have been proposed to correspond to the α-subunit of cardiac BK_Ca. In addition, the cloning of cardiomyocyte BK_Ca cDNAs revealed the standard 27 constitutive exons of Kcnma1, but their expression in H9c2 cardiac cells produced a BK_Ca protein that was unable to reach the mitochondria (7). Taken together, these studies questioned whether mitoBK_Ca was related to Kcnma1.

Here, we investigated the molecular identity of mitoBK_Ca, tested whether splice variation of Kcnma1 gene underlies BK_Ca targeting to cardiac mitochondria, and addressed the role of Kcnma1 gene expression in NS1619-mediated cardioprotection.

Results

Fingerprinting mitoBK_Ca.

We first evaluated the size of cardiac mitoBK_Ca using immunoblotting and mass spectrometry. For this purpose, we used Percoll-purified mitochondria from the left ventricle (8) (Fig. 1A) or from cardiomyocytes and specific antibodies raised against BK_Ca C terminus that generated no signals in BK_Ca knockout (Kcnma1^−/−) samples. Immunoblots using lysates of purified mitochondria from mouse ventricular myocytes showed two specific bands, one at ~140 kDa and a smaller one near 100 kDa; the latter is likely to be a proteolytic fragment. These bands were present in the WT but not in the knockout mitochondrial samples (Fig. 1B, arrows) (n = 3). For comparison, whole-brain lysates were also used; a band of ∼125 kDa was observed only in the WT samples (Fig. 1C, arrow). The differences in BK_Ca separation characteristics in ventriculocyte mitochondria and brain samples are consistent with the use of different start sites and/or the inclusion of splicing-derived sequences in the immunodetected BK_Ca proteins. As shown later in Fig. 4, analyses of ventriculocyte transcripts predict proteins containing large 50- and 58-aa inserts. Proteins starting from Met 1 or Met 3 (4) and containing both inserts would have a molecular mass of 145 kDa or 139 kDa, respectively, in close agreement with the larger-size band in mitochondria. The size of 125 kDa in the brain is similar to that found in smooth muscle (9) and matches the mass of a predominant protein starting at Met 3 without large inserts.

Fig. 1.

Identifying cardiac BK_Ca in purified mitochondria. (A) Electron micrograph of purified mitochondria. Dark spots correspond to mitochondrial granules. (B) Immunoblot for BK_Ca in mitochondrial lysates from WT and knockout (^−/−) ventriculocytes. (C) Immunoblot of whole-brain lysate. Protein = 50 µg/lane. Arrows, specific bands. (D) Example of spectra identifying BK_Ca from mitoplasts. (E) Identified peptides (underlined) span the C terminus (gray area).

Fig. 4.

Molecular determination and quantification of cardiac BK_Ca. (A) Scheme of the detected 27 constitutive exons and translated alternatively-spliced exons (STREX, SV27, and DEC). Boxes, exons; line, introns. Numbers correspond to nucleotide numbers. White boxes, 5′UTR and alternative 3′UTRs. Black arrows, forward primers; gray arrows, reverse primers. Dashed gray arrows, 3′UTR primers used for cDNA synthesis (Table S2). (B) Example of real-time PCR fluorescence vs. cycle number plots for known amounts of DEC insert (closed circles) and in ventriculocytes’ cDNA (open circles). Dashed line, level of threshold cycles (Cq). (C) Corresponding dose–response curve together with experimental value (open circle). (Inset) Corresponding melting curves. (D) Agarose gel of PCR end-products for β-actin, total BK_Ca, and DEC, STREX, and SV27 exonic sequences. For explanation, see text. (E) Percentage expression of BK_Ca-SV27, BK_Ca-STREX, and BK_Ca-DEC normalized to total BK_Ca in left ventricle and in left ventriculocytes. *P < 0.05 vs. total BK_Ca.

For mass spectrometry analysis, Percoll-purified mitochondria from rat left ventricle were used to prepare inner mitochondrial membranes (mitoplasts). mitoBK_Ca was immunoprecipitated from mitoplast lysates using anti-BK_Ca antibodies that generated no signals in purified mitochondria of Kcnma1^−/− (Fig. 2M). Immunopurified samples from 10 rat hearts were combined for each experiment (n = 5). LC/MS/MS spectra, such as the one shown in Fig. 1D, confirmed the presence of BK_Ca encoded by the Kcnma1 gene in the mitochondrial bands excised at ∼100–150 kDa (three of five experiments). Notably, we did not obtain BK_Ca sequences in the gel plugs that corresponded to molecular masses of <100 kDa or >150 kDa. In contrast, in gel plugs at ∼100–150 kDa, we identified 11 unique peptides throughout the C terminus of BK_Ca (Fig. 1E and Table S1). The coding sequences of the identified peptides were localized in constitutive exons 9, 12, 14, 17, 19, 21, 25, and 27, and spanned exons 21 and 22. In addition, we found a peptide of a 27-aa splice insert, which we refer to as SV27. The presence of peptides covering 778 residues, or 95% of the constitutive BK_Ca C terminus, ruled out the possibility that the cardiac mitoBK_Ca α-subunit consists of an ∼55-kDa protein (1). Supporting this view, LC/MS/MS analysis of mitoBK_Ca immunoprecipitated from lysates of mouse brain mitoplasts (from Percoll-purified mitochondria) yielded 21 unique peptides covering 1,113 residues, or 92% of the entire BK_Ca sequence. The peptides included one within the first intracellular loop (between S0 and S1) and two overlapping parts of the S1 and S4 transmembrane domains (Fig. S1 and Table S1). Taken together, these results support the presence of classic BK_Ca in mitochondria and revealed mitoBK_Ca as a protein of ∼140 kDa encoded by the Kcnma1 gene.

Fig. 2.

Localization of native BK_Ca in mitochondria. (A and B) Ventriculocytes loaded with MitoTracker, fixed, permeabilized and labeled with anti-BK_Ca Ab. (C) Overlay of A and B. (D and E) Ventriculocytes labeled with WGA under nonpermeabilized conditions, and subsequently permeabilized and labeled with the anti-BK_Ca Ab. (F) Overlay of D and E. (G–I) Purified mitochondria from left ventricle triple-labeled with anti-BK_Ca pAb (G), MitoTracker (Mito) (H), and anti-VDAC1 Ab (I). (J–L) Overlays of G and H, G and I, and H and I, respectively. (M–O) Kcnma1^−/− mitochondria triple-labeled for BK_Ca (M), MitoTracker (N), and VDAC1 (O).

Direct Visualization of mitoBK_Ca.

We then wondered whether BK_Ca could be visualized within the mitochondria of adult ventriculocytes. Notably, BK_Ca immunolabeling indicated a longitudinal distribution pattern (Fig. 2 A and D) that largely coincided with MitoTracker labeling (Fig. 2 B and C) but not with surface membrane labeling by wheat germ agglutinin (WGA) (Fig. 2 D–F). Colocalization analysis using the protein proximity index (PPI) algorithm (10) indicated that the WGA to BK_Ca PPI was low (0.15 ± 0.01; n = 32 cells from five independent experiments). In contrast, the MitoTracker to BK_Ca PPI was considerable (0.52 ± 0.02; n = 56 cells from 10 independent experiments), which is consistent with the view that BK_Ca resides in mitochondria.

To increase the sensitivity of our method, we directly visualized BK_Ca in purified mitochondria. Regular confocal and superresolution (stimulation emission depletion, STED) fluorescence microscopies were used. For regular confocal microscopy, mitochondria were triple-labeled for BK_Ca, with MitoTracker and for the voltage-dependent anion channel 1 (VDAC1) (Fig. 2 G–I). Supporting the presence of BK_Ca in mitochondria, BK_Ca signals frequently coincided with those of MitoTracker (Fig. 2J) and VDAC1 (Fig. 2K) in the same way as VDAC1 coincided with MitoTracker labeling (Fig. 2L). Furthermore, mitochondria from Kcnma1^−/− mice produced no BK_Ca-specific signals (Fig. 2M) in MitoTracker-loaded mitochondria, in which MitoTracker and VDAC1 colabeling was preserved (Fig. 2 N and O) (n = 3). In fact, in the BK_Ca knockout samples, the PPI for mitoBK_Ca to MitoTracker and vice versa were identical and within background levels (0.14 ± 0.01, n = 12 experiments, three fields per experiment). In contrast, WT mitochondria displayed a PPI for mitoBK_Ca to MitoTracker of 0.84 ± 0.03 and for MitoTracker to mitoBK_Ca of 0.66 ± 0.04 (n = 8 experiments, three fields per experiment), which were comparable to the PPI values for VDAC1 to MitoTracker and for VDAC1 to mitoBK_Ca (0.63 ± 0.03 and 0.56 ± 0.03, n = 5 experiments, three fields per experiment). The results imply that the majority of mitoBK_Ca colocalizes with MitoTracker-loaded mitochondria but not all MitoTracker-labeled mitochondria express BK_Ca proteins.

Increasing resolution with STED microscopy allowed us to map mitoBK_Ca with unprecedented detail. STED transformed the indistinct regular confocal image (Fig. 3 A and D) into an image in which discrete mitoBK_Ca clusters were observed (Fig. 3 B and E). Overlaid images of the same field contrast the difference in resolution (Fig. 3 C and F). Random line scans (Fig. 3E) were used to construct line scan plots (Fig. 3G) to measure the full-width half-maximum of mitoBK_Ca clusters. A frequency histogram analyzing 500 clusters was fitted with a Gaussian function; the fit indicates that the majority of mitoBK_Ca arranges in clusters of 50 ± 1.4 nm (Fig. 3G, inset) (n = 3 independent preparations). Overall, the microscopy results established beyond doubt the presence of BK_Ca in cardiac mitochondria, and Kcnma1^−/− experiments verified its Kcnma1 origin.

Fig. 3.

Nanovisualization of cardiac mitoBK_Ca. Confocal (A, green) and STED images (B, red) of same Percoll-purified mitochondria from mouse left ventricle labeled for BK_Ca and using Atto-647N secondary antibody conjugate. (C) Overlay of A and B. (D–F) Amplification of squared region in A–C, respectively. (G) STED intensity profile of the line scan in E (white line). (Inset) Cluster size histogram fitted with a Gaussian function (red line). Images were acquired at 3.5 nm/pixel and pseudocolored for presentation.

BK_Ca Exonic Composition in Ventriculocytes.

The molecular composition of mitoBK_Ca was addressed by first purifying mRNAs from isolated mouse ventriculocytes (avoiding contamination from other cardiac cell types) and performing a systematic exon scan of Kcnma1’s 41 predicted exons (National Center for Biotechnology Information Evidence Viewer, Gene ID 16531). Because the Kcnma1 gene has two predicted 3′UTRs (Fig. 4A), we used oligo-dT primers and primers from both 3′UTRs to reverse-transcribe mRNAs. All 41 predicted exons were scanned with specific primers designed to amplify exonic regions 200–400 bp in size (Table S2). This approach detected the standard 27 constitutive exons that encode the insertless BK_Ca beginning from the first possible start codon of Met 1 (NP_034740.2; n = 3) (4). Furthermore, we found three spliced exons in the C terminus encoding: a 57-aa insert corresponding to the stress-axis regulated exon (STREX) (11), SV27, and a 50-aa C-terminal sequence named after the three last amino acids, DEC (n = 3) (Fig. 4A and Fig. S2).

To determine the relative contribution of the three spliced exons to cardiac BK_Ca mRNAs, we performed quantitative real-time PCR using cDNAs prepared from DNase pretreated total RNA of the left ventricle and isolated ventriculocytes. Controls included no reverse transcriptase (−RT), PCR primers spanning introns (Table S3), and H₂O instead of cDNA. Standard curves using linearized BK_Ca plasmids were used to extrapolate the cDNA concentration in the cardiac samples. Fig. 4 B–D illustrates, as an example, the analysis for BK_Ca-DEC in ventriculocyte samples. Fig. 4B presents the fluorescence vs. PCR cycle curves for different concentrations of the BK_Ca-DEC standard (closed circles) together with the curve for a cardiac cDNA sample (open circles). Fig. 4C is the corresponding threshold cycle number (Cq, read at ∼100 relative fluorescence units) vs. the BK_Ca-DEC plasmid concentration curve (closed circles); also shown is the experimental point (open circle, arrow). The corresponding melting curves at the end of PCR cycles (Fig. 4C, Inset) featured a single peak, which is indicative of a single PCR product. The latter was confirmed by examining the PCR end-products using agarose gel electrophoresis as shown in Fig. 4D, which also shows results for β-actin, total BK_Ca, and the other two splice sequences (+RT refers to the PCR product in ventriculocyte cDNA samples, −RT and H₂O are negative controls as described above, and plasmid refers to products from constructs BK_Ca, BK_Ca-DEC, BK_Ca-STREX, and BK_Ca-SV27). Absolute quantities of total BK_Ca were nearly four orders-of-magnitude greater in the left ventricle than in isolated ventriculocytes (1.5 ± 0.3, n = 4 vs. 5.4 × 10⁻⁴ ± 0.0002 pg cDNA/µg RNA, n = 6) pointing to the large abundance of BK_Ca transcripts in other cardiac cells. STREX and DEC were only 10- to 25-times more abundant in the left ventricle than in the ventriculocytes (STREX, 4.0 × 10⁻³ ± 0.001 vs. 3.8 × 10⁻⁴ ± 0.0001 pg cDNA/µg RNA, n = 4 each; DEC, 1.1 × 10⁻² ± 0.003, n = 4 vs. 4.8 × 10⁻⁴ ± 0.0001, n = 6, pg cDNA/µg RNA). The abundance of SV27 was less than the detection limit in samples from isolated ventriculocytes (n = 6) but present in the left ventricle (4.7 × 10⁻² ± 0.003 pg cDNA/µg RNA, n = 4). For comparison, Fig. 4E shows the percent expression with respect to total BK_Ca for all three species. It is clear that in the left ventricle, the STREX, SV27, or DEC transcript levels were minimal with respect to total BK_Ca (0.3 ± 0.03%, 3.8 ± 1.06% and 0.7 ± 0.1%, n = 4, respectively). In contrast, in ventriculocytes the DEC sequences were as abundant as total BK_Ca (91 ± 10%, n = 6), followed by STREX (65 ± 18%, n = 4) and SV27, which was negligible (n = 6). These results imply that ventriculocytes express BK_Ca transcripts containing primarily DEC splice sequences, whereas other cardiac cells (e.g., from vessels, fibroblasts, or neurons) express abundant insertless BK_Ca. In both cases, BK_Ca originates from the Kcnma1 gene.

DEC Splicing and mitoBK_Ca Targeting.

Because constitutive (insertless) BK_Ca fails to target BK_Ca to mitochondria in cardiac H9c2 cells (7), the finding of STREX, SV27, and DEC splice sequences in cardiomyocytes mRNAs prompted us to investigate whether a splicing mechanism could be responsible for targeting BK_Ca to mitochondria. We hypothesized that the most-abundant splice insert, DEC, should have a preponderant role in BK_Ca targeting; in accordance with this conjecture, the least-abundant insert, SV27, would play an insignificant role.

To test our hypothesis, adenovirus constructs containing insertless BK_Ca, and isoforms BK_Ca-SV27 and BK_Ca-DEC (with splice sequences in their native positions) were transduced into rat adult ventriculocytes for immunochemical analysis. All constructs contained a c-Myc epitope at the N terminus. Because the plasma membrane BK_Ca has an extracellular N terminus (12) (Fig. 5, schemes), the N-terminal c-Myc tag allowed us to examine nonpermeabilized and permeabilized cells to discern whether newly synthesized BK_Ca isoforms were targeted to the plasma membrane or found intracellularly, perhaps within mitochondria. To visualize mitochondria, cardiomyocytes were preloaded with MitoTracker. Using this strategy under nonpermeabilized conditions, we found that insertless BK_Ca was readily targeted to the plasma membrane (Fig. 5A) and that its signals were not linked to MitoTracker labeling (Fig. 5 B–D) (n = 3, 12 of 12 cells). In permeabilized cells, the peripheral expression pattern remained the same, and there was only a low intracellular signal that did not coincide with MitoTracker labeling (Fig. 5 E–H) (n = 9, 27 of 27 cells). Similarly, the BK_Ca-SV27 isoform displayed a pronounced plasmalemmal localization detected under nonpermeabilized conditions (Fig. 5 I–L) (n = 4, 10 of 10 cells) that was the same in permeabilized conditions and without overlap with MitoTracker labeling (Fig. 5 M–P) (n = 4, 15 of 15 cells).

Fig. 5.

Insertless BK_Ca and BK_Ca-SV27 do not colocalize with mitochondria in transduced adult ventriculocytes. (A) Insertless BK_Ca was observed at the plasma membrane of a nonpermeabilized cell. (B) Same cell showing clear MitoTracker signals. (C) Overlay of A and B. (D, H, L, and P) Amplified images (squares in C, G, K, and O, respectively). (E) In permeabilized cells insertless BK_Ca was predominantly observed at the cell periphery. (F) MitoTracker signals of same cell. (G) Overlay of E and F. (I) BK_Ca-SV27 labeling was also observed at the plasma membrane under nonpermeabilized conditions. (J) MitoTracker labeling of the same cardiomyocyte. (K) Overlay of I and J. (M) BK_Ca-SV27 at the cell periphery in a permeabilized cell. (N) MitoTracker signal of the cell in M. (O) Overlay of M and N. Cartoons show orientation of constructs and tag at the N terminus. PM, plasma membrane.

Finally, when cells were transduced with BK_Ca-DEC, nonpermeabilized cells displayed no BK_Ca-DEC signal at the surface membrane (Fig. 6A), which demonstrated that the DEC insert prevented expression of BK_Ca at the plasma membrane; as a positive control, MitoTracker labeling was clearly observed (Fig. 6 B–D) (n = 3, 12 of 12 cells). In contrast, permeabilized cardiomyocytes exhibited remarkable longitudinal labeling similar to mitochondria (Fig. 6 E and F). The coincidence in BK_Ca-DEC and MitoTracker is illustrated in the overlay (Fig. 6 G and H). Greater than 90% of the BK_Ca-DEC transduced cells examined exhibited this exceptional mitochondrial localization (n = 4, 14 of 15 cells), which resembled that observed in native cells (Fig. 2). In fact, image analysis demonstrated a PPI for BK_Ca-DEC to MitoTracker of 0.51 ± 0.04 (n = 4 experiments; 14 cells) which is close to the PPI of native BK_Ca to MitoTracker-loaded mitochondria. These results indicate that BK_Ca-DEC, but neither insertless BK_Ca nor BK_Ca-SV27, contains sequences recognized by ventriculocytes’ machinery for mitochondrial delivery.

Fig. 6.

The expression of DEC exon favors BK_Ca mitochondrial targeting in transduced adult ventriculocytes. (A) In nonpermeabilized cells, no significant BK_Ca-DEC signal was observed. (B) The same ventriculocyte readily loaded with MitoTracker. (C) Overlay of A and B. (D) Zoom-out of the square in C. (E) In permeabilized cells, a profuse intracellular BK_Ca-DEC labeling that resembled MitoTracker labeling (F) was observed. (G) Overlay with remarkable colocalization of BK_Ca-DEC and MitoTracker. (H) Region at higher magnification (square in G). The cartoon shows BK_Ca-DEC intracellular location.

Kcnma1 Is Essential for the Cardioprotective Action of NS1619.

mitoBK_Ca has been proposed to protect the heart from ischemic insult because a general BK_Ca opener, NS1619, increases mitochondrial K⁺ uptake and promotes heart protection against ischemia/reperfusion (1). Because our previous results indicate that mitoBK_Ca is a product of the Kcnma1 gene, we first investigated whether the hearts from Kcnma1^−/− animals subjected to ischemia/reperfusion lost their characteristic functional improvement in response to NS1619. To test this point, isolated hearts from WT and Kcnma1^−/− mice were preconditioned for 10 min without (−, vehicle alone) or with (+) NS1619 (10 µM); this period was followed by an 18-min period of ischemia and ∼1 h of reperfusion, as depicted in Fig. 7A.

Fig. 7.

BK_Ca protects the heart from ischemic injury. (A) Ischemia/reperfusion protocol. (B and C) Function traces of hearts preconditioned with vehicle (DMSO, control) or with NS1619 (10 µM) in WT and Kcnma1^−/− mice. (D) NS1619 significantly improved mean RPP in WT but not in Kcnma1^−/− mice. (E and G) WT hearts preconditioned with NS1619 exhibited less infarct size (white) compared with the control. (F and G) In Kcnma1^−/−, infarct size was not reduced with NS1619. (H–J) Mitochondrial Ca²⁺ uptake. NS1619 preconditioning increased the amount of Ca²⁺ needed to induce a large Ca²⁺ release in WT but not in Kcnma1^−/−samples. Black arrows, addition of mitochondria. Blue arrows, 40 nmol Ca²⁺ pulses. Arrowheads, massive release of Ca²⁺. *P < 0.05 vs. control (Ctrl); CRC, Ca²⁺ retention capacity.

Typical cardiac function traces for the left ventricular-developed pressure as a function of time demonstrate that preconditioning the heart from WT animals with NS1619 causes, as expected, an improvement in cardiac function during reperfusion compared with the decreased function observed in the control hearts that were preconditioned with vehicle alone (Fig. 7B). In contrast, Kcnma1^−/− hearts lost their ability to be protected by NS1619 (Fig. 7C). Accordingly, the mean values for heart rate pressure product (RPP) (Fig. 7D) and the maximum rates of rise (dP/dt max) and fall (dP/dt min) (Fig. S3), which were measured as the mean of the last 10 min of reperfusion, indicated a significant improvement because of NS1619 in WT animals (n = 4), a property that was lost in Kcnma1^−/− animals (n = 3–4).

Similarly, myocardial infarction measured at the end of the reperfusion period was less pronounced after preconditioning WT animals with NS1619. Fig. 7E illustrates typical heart cross-sections stained with 2,3,5-triphenyltetrazolium chloride, in which the pale areas that correspond to nonviable tissue are smaller with NS1619 preconditioning than in vehicle-treated hearts (control). In contrast, hearts from Kcnma1^−/− were not protected by NS1619 and exhibited significant damage similar to that observed in the control hearts (Fig. 7F). Individual and mean infarct size values are given in Fig. 7G for WT (n = 4 each) and Kcnma1^−/− (n = 3 each).

Next, we tested whether NS1619 preconditioning could also benefit mitochondria function and if this effect was retained in Kcnma1^−/−. To this end, we measured the Ca²⁺ retention capacity of mitochondria isolated at 10 min of reperfusion. NS1619 (10 µM) treatment resulted in an increase in the number of Ca²⁺ pulses (Fig. 7H, blue arrows) needed to cause a massive Ca²⁺ release (Fig. 7H, arrowheads). This beneficial effect was abrogated in mitochondria from Kcnma1^−/− animals (Fig. 7 I and J) (n = 4–5 each). Overall, these experiments clearly demonstrate that Kcnma1 gene expression is a prerequisite for NS1619-mediated cardioprotection from ischemia/reperfusion injury, and lead us to propose that mitoBK_Ca is a target of NS1619 in the heart.

Discussion

In this work, we uncovered the origin of mitoBK_Ca as the product of the Kcnma1 gene, the same gene that encodes the plasma membrane BK_Ca, and demonstrated that splicing of the DEC exon is a mechanism that allows the protein to be targeted to cardiac mitochondria.

The strategies used to reach these conclusions were: (i) use of Percoll-purified mitochondria from isolated ventriculocytes or the left ventricle to assess the molecular mass of mitoBK_Ca through immunoblotting with antibodies raised against the plasma membrane BK_Ca, for protein sequence analysis of BK_Ca immunoprecipitates, and for visualization of mitoBK_Ca clusters at superresolution; (ii) use of isolated cardiomyocytes for immunolabeling of native mitoBK_Ca, for isolation of mRNAs for exon scanning, for quantification of splice inserts, and for transduction of BK_Ca isoforms to monitor subcellular localization; and (iii) use of Kcnma1^−/− mice for negative controls and to demonstrate loss of the protective action of a BK_Ca opener, NS1619, against ischemic injury.

Cardiac mitoBK_Ca and Kcnma1 Gene.

The finding that the mitoBK_Ca primary subunit has a size of ∼140 kDa is in contrast with an early study, reporting a smaller size protein of ∼55 kDa (1), and a more recent report of a larger protein with mass of ∼125 kDa (2); in both studies the protein was detected in mitochondrial lysates from isolated cardiomyocytes. In all instances, including the present study, mitochondrial lysates were examined with polyclonal antibodies (from Alomone Labs and Affinity Bioreagents) against the C-terminal end of the plasma membrane BK_Ca α-subunit encoded by the Kcnma1 gene. Because in some immunoblots, we have also observed a signal of ~50 kDa, it is possible that the observed smaller fragments, including the fragment near 100 kDa reported here, result from mitoBK_Ca proteolysis. This assumption is supported by the fact that BK_Ca is known to suffer C-terminal cleavage during purification, producing fragments of similar size (13). Conclusive evidence for the mitoBK_Ca α-subunit being a protein similar to its plasma membrane counterpart—encoded by the Kcnma1 gene—was given by sequence analysis of immunopurified BK_Ca from inner mitochondrial membranes (mitoplasts) purified from the left ventricle and from the brain. In both instances, peptides throughout the soluble C terminus were identified, whereas in brain mitoplast samples, the identified peptides spanned the whole BK_Ca protein from its N terminus to the end of its C terminus (Fig. 1 and Fig. S1).

Similar to our results (Fig. 4), a previous transcript analysis of ventricular myocytes reported expression of all 27 constitutive exons of the Kcnma1 gene (7), thereby supporting the view that mitoBK_Ca is indeed encoded by this gene. However, in those studies, expression of transcripts containing spliced exons, as we report here, was not detected, and thus expression of the insertless cDNA in H9c2 cells did not target BK_Ca to mitochondria.

Our conclusion that mitoBK_Ca is encoded by the Kcnma1 gene implies that all of the essential functional cassettes of the protein are conserved in both mitochondrial and plasma membrane versions. In fact, the previously reported biophysical and pharmacological properties of mitoBK_Ca from cardiomyocytes and other cell types resemble those of plasma membrane BK_Ca (3).

Mapping Cardiac BK_Ca to Mitochondria and Splice Variation as a Targeting Mechanism.

To our knowledge, BK_Ca activity has not been reported at the plasma membrane of adult ventriculocytes, which is consistent with our immunocytochemistry results. In contrast, we found profuse BK_Ca immunosignals coincident with MitoTracker-labeled mitochondria in these cells, a property that was confirmed beyond doubt using purified mitochondria and nanoimaging (Figs. 2 and 3). Compared with labeling cells, labeling-purified mitochondria had the advantage that we could rule out contamination and interference from signals originating from other intracellular organelles. The selective BK_Ca mitochondrial distribution vs. plasma membrane in adult ventriculocytes is not a property of neonatal cardiac cells (H9c2) in which BK_Ca can be visualized in both subcellular compartments (3). Dual targeting to the plasma membrane and mitochondria appears to be a common property of BK_Ca in other cell types.

In this work, we demonstrate that BK_Ca targeting to ventriculocytes mitochondria requires the inclusion of DEC exonic sequences at the C-terminal end of mitoBK_Ca. Thus, our work provides further evidence for a role of the C-terminal end in targeting proteins to mitochondria, unlike the canonical view of mitochondrial targeting by N-terminal sequences (14, 15). In agreement with the functional relevance of DEC sequences in mitochondria targeting, from the three splice inserts found in our ventriculocyte mRNA analysis, the DEC insert was the most abundant, with an abundance equal to the level of total BK_Ca. STREX transcripts were second in abundance to DEC transcripts, but the function of STREX is yet to be discovered. Nevertheless, our results demonstrate that the sole inclusion of DEC in BK_Ca is sufficient for mitochondrial delivery, which rules out an essential role for STREX in this function.

In contrast with our findings in cardiac myocytes, where BK_Ca-DEC is readily targeted to mitochondria, in COS cells a BK_Ca-DEC variant with additional splicing-derived sequences, NATRMTRMGQA, is found predominantly in the endoplasmic reticulum but also at the plasma membrane (16). These findings suggest that additional sequences may disrupt the targeting of BK_Ca-DEC to mitochondria or that cell-specific mechanisms are necessary to complement DEC in targeting BK_Ca to mitochondria. Whether the inclusion of STREX or SV27 modifies BK_Ca-DEC targeting to cardiomyocytes mitochondria is yet to be established. However, data using CHO cells indicate that SV27 does not have a major effect as a BK_Ca-DEC variant with additional splice inserts including SV27 can still be observed in mitochondria of CHO cells (17).

Cardiac BK_Ca and Cardioprotection.

Previous studies regarding the role of BK_Ca in protecting the heart from ischemic insult were based on pharmacology (1, 6, 18, 19). Because the most commonly used BK_Ca opener, NS1619, is able to inhibit Ca²⁺ currents when used at ∼30 µM (20), there was always the open question of whether NS1619 acts via BK_Ca or through some other mechanism (21). In our current study, we demonstrated that the protective effect of 10 µM NS1619 on the preischemic reperfused heart and on mitochondria isolated from these hearts was abolished in Kcnma1^−/− mice (Fig. 7), thereby indicating that NS1619-mediated cardioprotection requires expression of BK_Ca. At present, we do not have an explanation of why the Kcnma1^−/− mice exhibited reduced infarct size after ischemia/reperfusion under basal conditions. Nevertheless, the cardioprotective effects of NS1619 in the heart correlate with the activation of mitoBK_Ca (1) and thus, our studies using Kcnma1^−/− mice support the view that activation of cardiomyocyte mitoBK_Ca is one mechanism that protects the heart against ischemia-reperfusion injury. However, additional mechanisms triggered by NS1619, including the opening of BK_Ca in nonmyocyte heart cells, cannot be ruled out.

Materials and Methods

Animals.

Three-month-old male C57BL/6NCrL mice and Sprague–Dawley (SD) rats were used. Kcnma1^−/− mice were from Andrea Meredith (University of Maryland School of Medicine, Baltimore) and locally bred. All protocols received University of California, Los Angeles approval.

Antibodies and Procedures.

Antibodies and procedures are given in SI Materials and Methods.

Statistics.

The mean ± SEM for at least three individual experiments is given. A two-tailed Student t test value of P < 0.05 was considered significant.