Caveolin-1 Facilitates the Direct Coupling between Large Conductance Ca2+-activated K+ (BKCa) and Cav1.2 Ca2+ Channels and Their Clustering to Regulate Membrane Excitability in Vascular Myocytes*

Background: The contribution of caveolae to physiological interaction between two major ion channels, large conductance Ca2+-activated K+ (BKCa) and Ca2+ (Cav1.2) channels, is unknown in vascular myocytes. Results: The loss of caveola by caveolin-1 deficiency reduced BKCa-Cav1.2 coupling, Cav1.2 clustering, and membrane excitability regulation. Conclusion: Caveolin-1 provides platform for BKCa-Cav1.2 molecular complex. Significance: Caveolin-1-BKCa-Cav1.2 in caveola forms a novel Ca2+ signal domain for arterial tonus regulation. L-type voltage-dependent Ca2+ channels (LVDCC) and large conductance Ca2+-activated K+ channels (BKCa) are the major factors defining membrane excitability in vascular smooth muscle cells (VSMCs). The Ca2+ release from sarcoplasmic reticulum through ryanodine receptor significantly contributes to BKCa activation in VSMCs. In this study direct coupling between LVDCC (Cav1.2) and BKCa and the role of caveoline-1 on their interaction in mouse mesenteric artery SMCs were examined. The direct activation of BKCa by Ca2+ influx through coupling LVDCC was demonstrated by patch clamp recordings in freshly isolated VSMCs. Using total internal reflection fluorescence microscopy, it was found that a large part of yellow fluorescent protein-tagged BKCa co-localized with the cyan fluorescent protein-tagged Cav1.2 expressed in the plasma membrane of primary cultured mouse VSMCs and that the two molecules often exhibited FRET. It is notable that each BKα subunit of a tetramer in BKCa can directly interact with Cav1.2 and promotes Cav1.2 cluster in the molecular complex. Furthermore, caveolin-1 deficiency in knock-out (KO) mice significantly reduced not only the direct coupling between BKCa and Cav1.2 but also the functional coupling between BKCa and ryanodine receptor in VSMCs. The measurement of single cell shortening by 40 mm K+ revealed enhanced contractility in VSMCs from KO mice than wild type. Taken together, caveolin-1 facilitates the accumulation/clustering of BKCa-LVDCC complex in caveolae, which effectively regulates spatiotemporal Ca2+ dynamics including the negative feedback, to control the arterial excitability and contractility.

The calcium ion (Ca 2ϩ ) is a major second messenger responsible for variety of physiological responses, including neurotransmitter/hormone release, muscular contraction, cell proliferation, apoptosis, etc. It has been demonstrated that voltage-dependent Ca 2ϩ channel (VDCC) 2 and Ca 2ϩ effectors often accumulate spatially and form Ca 2ϩ microdomains (1,2) to increase the local Ca 2ϩ level and activate specific signal transduction events.
In smooth muscle cells (SMCs), it has been established that the L-type VDCC (LVDCC), ryanodine receptor (RyR), and large conductance Ca 2ϩ -activated K ϩ channel (BK Ca , K Ca 1.1) constitute the Ca 2ϩ microdomains. Two types of local Ca 2ϩ events, Ca 2ϩ hotspots (3) arising from depolarization-evoked Ca 2ϩ -induced Ca 2ϩ release (CICR) (4) and spontaneous Ca 2ϩ release (Ca 2ϩ sparks) (5), play a crucial role in the regulation of SMC contraction and relaxation. These two Ca 2ϩ events occur in the same distinct local areas in SMC so as to couple with BK Ca activity and, respectively, contribute to the action potential repolarization phase and spontaneous transient outward current (STOC) (6). Thus, BK Ca is thought to get activated mainly by Ca 2ϩ release from the sarcoplasmic reticulum through RyRs, which loosely couple with LVDCC on the plasma membrane in excitation-contraction coupling (7). In this study the relationship between LVDCC and BK Ca via RyR activation is termed as "loose coupling." Caveolae are composed of an omega-shaped structure on plasma membrane. Caveolin-1 is an essential factor for the properly formed caveolae structure and accumulates many types of signaling molecules (8,9). The BK Ca protein contains a binding motif to caveolin-1 and is often co-localized in caveolae (10,11). Several reports indicate that caveolae are involved in Ca 2ϩ hotspot and spark generation (12,13). Caveolin-1 knockout (KO) mice congenitally lack caveolae and exhibit many types of cardiovascular abnormalities (9,14,15).
The interaction of BK Ca with several functional molecules in addition to caveolin-1 has been identified (16). In some neurons it has been reported that BK Ca is directly activated by Ca 2ϩ influx through the tightly coupled VDCC (17)(18)(19)(20). On the other hand, in tsA-201 cells, which express LVDCC and BK Ca , the Ca 2ϩ entry through single LVDCCs rarely evokes BK Ca opening (21). These authors suggested that the BK Ca selectively interacts with N-type VDCC rather than L-type VDCC. To our knowledge, the direct molecular and/or functional interaction between BK Ca and VDCC has not been reported in muscles, including skeletal, cardiac, and smooth muscles, even though BK Ca co-localizes with Cav1.1 in the distrophin molecular complex (22) in skeletal muscle and can interact with caveolin-3 (23). In this study this direct coupling between LVDCC and BK Ca on the same plasma membrane is known as "tight coupling." The exact molecular mechanisms underlying direct physical interaction between BK Ca and Cav1. 2 have not yet been clarified. Subtypes of VDCCs acting as molecular partners for BK Ca depend upon the local tissues. The LVDCC encoded by Cav1.2 (␣1C) is highly expressed and serves as a predominant Ca 2ϩ entry pathway in SMCs (24). Because tissue-specific splice variants with differential functions have been found in both BK Ca (25) and Cav1.2 (26), it remains to be totally resolved whether BK Ca physically couples with Cav1.2 and the coupling is functionally significant in SMCs.
In the previous study we first demonstrated that BK␣ forms tetramer and directly interacts with caveolin-1 in living aortic SMCs using total internal reflection fluorescence (TIRF) microscopy. In the TIRF system, an evanescent wave excites fluorescent molecules within a 200-nm depth of a glass bottom. This enables visualization of the fluorescent-labeled molecules localized on the plasma membrane in living cells (27). The present study was undertaken to examine the possible BK Ca -Cav1.2 complex and the potential roles of caveolin-1 for this complex formation in vascular SMCs (VSMCs) by use of TIRF microscopy imaging methods. To demonstrate functional coupling between LVDCC and BK Ca , whole cell patch clamp recording using two different Ca 2ϩ chelators, EGTA and 1,2-bis(oaminophenoxy)ethane-N,N,NЈ,NЈ-tetraacetic acid (BAPTA) was performed (18,28). Results suggest that the BK Ca -Cav1.2 complex is mostly co-localized in the caveolae of VSMCs. Furthermore, data from KO mice denote the importance of caveolae as a platform of the ion channel complex formation that effectively regulates properly excitability and contractility of VSMCs.

EXPERIMENTAL PROCEDURES
Animals, Cell Isolation, and Culture-Caveolin-1 knock-out (KO) mice on the C57BL/6 background were obtained from The Jackson Laboratory (stock number 004585) (Bar Harbor, ME). Wild-type (WT) control mice (C57BL/6) were purchased (Japan SLC, Hamamatsu, Japan). All experiments were approved by the Ethics Committee of Nagoya City University and were conducted in accordance with the Guide for the Care and Use of Laboratory Animals of the Japanese Pharmacological Society. The superior mesenteric arteries were removed from male mice (8 -14 weeks) and cleansed of connective tissue in Ca 2ϩ -Mg 2ϩ -free Krebs solution containing 112 mM NaCl, 4.7 mM KCl, 1.2 mM KH 2 PO 4 , 25 mM NaHCO 3 , 14 mM glucose. From the third to the fifth branch were removed and incubated in Ca 2ϩ -Mg 2ϩ -free Krebs solution containing 0.4% collagenase (Amano Enzyme, Nagoya, Japan) and 0.1% papain (Sigma) for 45 min at 37°C. After incubation, these tissues were washed in Ca 2ϩ -Mg 2ϩ -free Krebs solution and dispersed mechanically. For electrophysiological experiments, a few drops of the cell suspension were placed in a recording chamber. For culture, single cells were suspended in DMEM supplemented with 10% heat-inactivated FBS, 20 units/ml penicillin, and 20 g/ml streptomycin (Sigma). After settling on glass-bottom dishes, cells were transiently transfected with fluorescent-labeled cDNA and cultured for 24 -48 h.
Electrophysiological Recording-Electrophysiological studies were performed using a whole cell voltage clamp technique with a CEZ-2400 amplifier (Nihon Kohden, Tokyo, Japan), an analogdigital converter (DIGIDATA 1320A; Molecular Devices, Sunnyvale, CA), and a pCLAMP software (Version 8.2; Molecular Devices) in vascular cells as described previously (29). To examine whether the functional coupling between BK Ca and Cav1.2 was tight or loose, we used two types of pipette solution containing 1) 140 mM KCl, 4 mM MgCl 2 , 10 mM HEPES, 2 mM Na 2 ATP, and 5 mM EGTA or 2) 140 mM KCl, 4 mM MgCl 2 , 10 mM HEPES, 2 mM Na 2 ATP, and 10 mM BAPTA (pH 7.2 with KOH). Here 10 mM BAPTA was used to inhibit both tight and loose coupling, and 5 mM EGTA was used to inhibit loose coupling. We defined the EGTA-resistant, BAPTA-sensitive current component as a BK Ca current directly activated by Ca 2ϩ influx through tightly coupled Cav1.2 channels. For simultaneous recording of STOC and Ca 2ϩ sparks, the pipette solution contained 140 mM KCl, 4 mM MgCl 2 , 10 mM HEPES, 2 mM Na 2 ATP, and 0.1 mM fluo-4 (Invitrogen) (pH 7.2 with KOH). The extracellular solution (normal HEPES-buffered solution) had an ionic composition of 137 mM NaCl, 5.9 mM KCl, 2.2 mM CaCl 2 , 1.2 mM MgCl 2 , 14 mM glucose, and 10 mM HEPES. The pH was adjusted to 7.4 with NaOH. For VDCC current measurements, the pipette solution contained 120 mM CsCl, 20 mM tetraethylammonium chloride (Tokyo Chemical Industry, Tokyo, Japan), 1 mM MgCl 2 , 10 mM HEPES, 2 mM Na 2 ATP, and 20 mM BAPTA (pH 7.2 with CsOH). The bath solution contained 92 mM NaCl, 5.9 mM KCl, 30 mM BaCl 2 , 1.2 mM MgCl 2 , 14 mM glucose, and 10 mM HEPES. The pH was adjusted to 7.4 with NaOH. Whole cell BK Ca currents and VDCC currents were activated from a holding potential of Ϫ60 mV by applying 150-ms voltage steps to voltages between Ϫ50 and ϩ40 mV in increments of 10 mV. To detect tight-coupling between BK Ca and Cav1.2, depolarizing stimuli from a holding potential of Ϫ60 to 0 or ϩ10 mV for 150 ms were applied. STOCs were measured at a steady membrane potential of Ϫ40 mV.
Single-molecule Imaging-Single-molecule imaging was performed using a TIRF imaging system with an objective lens (CFI Plan Apo TIRF 60ϫ/1.45 or CFI Apo TIRF 100ϫ/1.49, oil immersion; Nikon, Tokyo, Japan) as described previously (30). Data were collected with an EM-CCD camera and analyzed by AQUACOSMOS software (Hamamatsu Photonics, Hamamatsu, Japan). Cyan fluorescent proteins (CFP)-fused proteins were excited with a 405-nm blue diode, and YFP-and green fluorescent protein (GFP)-fused proteins were excited with a 488-nm argon laser (Coherent, Santa Clara, CA), respectively, and reflected off dichroic mirrors (427-441/490 -510 nm; Omega Optical, Brattleboro, VT). CFP/YFP emissions were collected through dichroic mirrors and dual band-pass filters (454 -479/523-567 nm; Omega Optical). A resolution of images was 71 per pixel (x-y) and less than 200 nm (z). TIRF images were collected at 465-ms exposure times and scanned every 1651 ms. Normal HEPES-buffered solution was used during recording. All experiments were carried out at room temperature (25°C).
Fluorescence Resonance Energy Transfer (FRET) Analysis-The efficiency of FRET (E FRET ) was evaluated based on the acceptor photobleaching method, in which the emission of the donor fluorophore is compared before and after the photobleaching of the acceptor (31). The fluorescence of YFP was photobleached using a mercury lamp (100 W, C-SHG1; Nikon) and a G-2A filter cube (Ex510 -560/DM575/BA590; Nikon) for 2.5 min. FRET efficiency (E FRET ) was calculated as the percentage increase in CFP emission after YFP photobleaching, as described previously (30).
Single-molecule GFP Bleaching-We counted subunits of BK Ca and Cav1.2 in membranes of HEK293 cells and arterial myocytes by observing bleaching steps of GFP fused to BK␣ and VDCC␣1C in a single particle, as described previously (30,32,33). HEK293 cells or primary-cultured myocytes were transiently transfected with 2 g of cDNA encoding BK␣-GFP or Cav1.2-GFP using Lipofectamine 2000. At 24 h after transfection, cells were fixed for 10 min with 4% paraformaldehyde in PBS, rinsed, and placed in PBS solution before the experiment. GFP was excited with a 488-nm laser and imaged with a B-2A filter cube (DM505/BA520; Nikon) and an objective (100ϫ/ 1.49). Images (256 ϫ 256 pixels; 1 pixel ϭ 107 nm) were acquired at 10 Hz for 100 -120 s. Fluorescence intensity in a region of interest (3 ϫ 3 pixels) was calculated by subtracting the background in 16 pixels around the region of interest. The number of bleaching steps was determined by eye from the fluorescence signal trace. The steps within each single trace were similar in amplitude but varied between different traces. This is a consequence of the Gaussian profile of the laser and complex topology of the plasma membrane, which results in different local illumination intensities of the evanescent field. We used the following criteria for discarding signals: (a) a signal exhibits an elliptical shape, (b) a signal is very close to other signals (Ͻ4 pixels), (c) a signal that fluorescent intensity fluctuates too much to be accurately determined bleaching steps, (d) a signal does not show complete bleaching.
Fluorescent Labeling of Freshly Isolated Myocytes-BK Ca and LVDCC in freshly isolated mesenteric arterial myocytes were labeled with polyclonal anti-BK␣ antibody (APC-107, Alomone Laboratories, Jerusalem, Israel) and DM-BODIPY (Ϫ)-dihydropyridine (Invitrogen). Cells were immunostained as described previously (34). Dissociated cells settled on glass-bottom dishes (Matsunami Glass Industry, Osaka, Japan) were labeled with 1:100 diluted antibodies for 12 h at 4°C after fixation and permeabilization. Then cells were washed and incubated with Alexa 405-conjugated anti-rabbit IgG goat antiserum (Invitrogen) for 1 h at room temperature. After washing, cells were loaded with 100 nM DM-BODIPY(Ϫ)-dihydropyridine for 5 min and subsequently washed in PBS. Fluorescently labeled cells were observed using a TIRF imaging system and AQUACOSMOS software (mentioned above). Alexa405 and BODIPY were excited with the blue diode and the argon laser, respectively. The emissions of Alexa405 and BODIPY were collected using CFP-HQ (DM450/BA460 -510; Nikon) and YFP-HQ (DM510/BA520 -560; Nikon) filter cubes, respectively.
Measurement of the Fluo-4 Signal-For simultaneous measurements of Ca 2ϩ sparks and STOCs in VSMCs, the Ca 2ϩ images were obtained using a TIRF imaging system and AQUA-COSMOS software (mentioned above) under the whole cellclamp mode. A myocyte was loaded with 100 M fluo-4 by diffusion from the recording pipette. An argon laser (488 nm) and a B-2A filter were used for excitation light and emission collection, respectively. Images were collected every 14 ms for 12 s. Resolution of images was 142 or 214 nm/pixel. Fluorescence intensity (F) in the region of interest was measured as an average from the pixels included in the area of a 2.14 ϫ 2.14 m square. The data are shown as ⌬F/F 0 (%), where F 0 is the basal F value obtained as the average intensity of the regions of interest acquired during the measurements, and ⌬F is the difference between F and F 0 .
To measure the averaged Ca 2ϩ concentration from whole cell area, myocytes were loaded with 10 M fluo-4 AM (Invitrogen), and fluorescent images were acquired using AQUA-COSMOS software. The minimum fluorescence intensity (F min ) and maximum fluorescence intensity (F max ) were obtained by applying Ca 2ϩ -free HEPES-buffered solution (CaCl 2 was removed from and 5 mM EGTA was added to normal HEPES-buffered solution) and 10 M ionomycin in normal HEPES-buffered solution containing 2 mM Ca 2ϩ , respectively. To induce a contraction in a myocyte, 40 mM KCl HEPES-buffered solution (40 KCl), which contained 102.9 mM NaCl, 40 mM KCl, 2.2 mM CaCl 2 , 1.2 mM MgCl 2 , 14 mM glucose, and 10 mM HEPES (pH 7.4 with NaOH), was applied. Ca 2ϩ elevation from base line was normalized to the maximum Ca 2ϩ change (i.e. the difference between F min and F max ). Images were collected every 2 s. A resolution of images was 267 nm/pixel.
Single Cell Shortening Measurements-Transmitted light images were acquired using the above-mentioned microscope and an objective lens (20ϫ, dry, Nikon). 40 KCl and/or 1 M paxilline (Pax) (Tocris Bioscience, Bristol, UK) were applied to myocytes. Cellular contraction was estimated as the decrease in cell area from the control condition. Images were collected every 1 s for 20 min. A resolution of images was 533 nm/pixel. Statistics-Pooled data are shown as the mean Ϯ S.E. Statistical significance between two groups was determined by Student's t test. Statistical significance among groups was evaluated by Tukey's test. Significant difference is expressed in the figures as p Ͻ 0.05 (*) or p Ͻ 0.01 (**).

BK Ca Activation by Ca 2ϩ Influx through Tightly Coupled
LVDCC in VSMCs-At first, membrane currents elicited by depolarization were measured in mesenteric artery SMCs from WT or KO mice using low Ca 2ϩ buffering pipette solution (50 M EGTA). A large part of the outward current upon depolarization was blocked by the addition of 1 M Pax (Fig. 1A) or 50 M Cd 2ϩ , indicating that BK Ca activation as one of the major outward currents is largely due to Ca 2ϩ influx through VDCC. In highly excitable SMCs such as those of bladder and vas deferens, the Ca 2ϩ source for BK Ca activation upon depolarization may also include Ca 2ϩ from the sarcoplasmic reticulum via Ca 2ϩ -induced Ca 2ϩ release (CICR) (4,35). This is also the case in mesenteric SMCs, and the outward currents upon depolarization showed irregular shapes and occurred concomitantly with local subcellular Ca 2ϩ transients (Fig. 1B). To prevent the component of BK Ca current activation due to CICR during depolarization, the following experiments were performed under much stronger Ca 2ϩ buffering conditions (36). Accordingly, we used two different Ca 2ϩ chelators, EGTA and BAPTA, to examine the possible tight-coupling between BK Ca and Cav1.2 in VSMCs (18). These Ca 2ϩ buffers have similar binding affinities for Ca 2ϩ , but the binding rate constant of BAPTA is ϳ150 times faster than EGTA (18,28). Thus, BAPTA is much more effective in preventing the diffusion of free Ca 2ϩ away from the entrance site in LVDCC on the plasma membrane. Based on these distinct characteristics, tight-coupling is effectively interfered by 10 mM BAPTA but not by 5 mM EGTA, whereas loose coupling is equally sensitive to 10 mM BAPTA and 5 mM EGTA (18).
In the presence of 5 mM EGTA in the pipette solution, a fast-activated and inactivated outward current remained in SMCs of the WT but not those of KO ( Fig. 2A). When the pipette solution contained 10 mM BAPTA, no such transient outward current was observed in either the WT or KO myocytes ( Fig. 2A). Thus, the fast inactivation recorded by the pipette solution containing 5 mM EGTA was presumably due to Ca 2ϩ removal by the slow Ca 2ϩ chelation by EGTA. The slope of the rising phase of the outward current was compared and is shown in Fig. 2B (in WT: 0.80 Ϯ 0.27 (pA/pF)/ms with EGTA (n ϭ 5) and 0.08 Ϯ 0.02 (pA/pF)/ms with BAPTA (n ϭ 5); in KO: 0.16 Ϯ 0.04 (pA/pF)/ms with EGTA (n ϭ 6) and 0.10 Ϯ 0.02 (pA/pF)/ms with BAPTA (n ϭ 8); p Ͻ 0.01 versus WT with EGTA). The time to the peak of the EGTA-resistant currents was 35.8 Ϯ 11.5 ms (n ϭ 5) in WT and 55.2 Ϯ 10.8 ms (n ϭ 6) in KO (p Ͼ 0.05). The peak amplitude of the EGTA-resistant outward currents was 243.3 Ϯ 38.4 pA (n ϭ 5) in the WT and 111.7 Ϯ 19.4 pA (n ϭ 6) in the KO myocytes (p Ͻ 0.05). Thus, the EGTA-resistant transient outward current was presumably activated by Ca 2ϩ influx via tight-coupled Cav1.2. This EGTAresistant current was significantly larger and activated faster in the WT than KO myocytes. Fig. 2C illustrates the I-V relationships of BK Ca currents detected as a component sensitive to 1 M Pax in WT (Fig. 2Ca)  and KO (Fig. 2Cb). In WT, the BK Ca current density resistant to 5 mM EGTA in the pipette solution was higher than that of the 10 mM BAPTA-resistant component at potentials positive to 0 mV. The summarized data at ϩ10 mV in WT show that the EGTA-resistant and BAPTA-resistant BK Ca currents were 1.7 Ϯ 0.3 pA/pF (n ϭ 5) and 0.3 Ϯ 0.3 pA/pF (n ϭ 5), respectively (p Ͻ 0.01, Fig. 2D). Furthermore, the components in KO were 0.8 Ϯ 0.2 pA/pF (n ϭ 6) and 0.3 Ϯ 0.2 pA/pF (n ϭ 8), respectively (p Ͼ 0.05) (Fig. 2D). It is notable that the amplitude of the EGTA-resistant component in the WT was significantly larger than that in the KO myocytes (p Ͻ 0.05).
In the next series of experiments, the functional expression of LVDCC and BK Ca in mesenteric arterial SMCs was compared between WT and KO. In mesenteric arterial SMCs, Cav1.2 was expressed so abundantly that substantial inward currents through LVDCC were detected at positive potentials to Ϫ20 mV and peaked at ϩ10 mV under the conditions, where outward currents were blocked. The LVDCC currents were detected as the component blocked by the addition of 50 M Cd 2ϩ . The peak LVDCC currents were recorded at ϩ10 or ϩ20 mV, and neither cell capacitance nor the peak LVDCC currents density at First, BK Ca and LVDCC in freshly isolated mesenteric SMCs were stained with an anti-BK␣ antibody and DM-BODIPY (Ϫ)dihydropyridine, respectively. In TIRF microscope images, fluorescent particles were detected by the Alexa405 and BODIPY dyes binding to these channels in the plasma membrane (Fig.  3A). A part of the particles from the BK Ca were co-localized with those from LVDCC in both WT and KO myocytes, but the ratio of co-localized particles against total BK Ca particles was significantly lower in KO than WT myocytes (WT: 45.8 Ϯ 1.9%, n ϭ 8; KO: 24.1 Ϯ 8.0%, n ϭ 7; p Ͻ 0.05; Fig. 3B). Next, BK␣ and VDCC␣1C, which were labeled with YFP and CFP at the C termini (BK␣-YFP and Cav1.2-CFP, respectively), were transiently co-expressed in SMCs isolated from mesenteric artery (Fig. 3, 4) and aorta, which were then primary-cultured. The transient expression of the labeled molecules was carefully performed by regulating the amount of cDNA to minimize the artifacts due to overexpression (see "Experimental Procedures"). Thus, the influence of transient expression of BK␣-YFP on the functional expression of BK Ca was examined in aortic myocytes. BK␣-YFP or YFP alone were transiently expressed in aortic myocytes from WT mice, and the functional expression of BK Ca was measured by the whole cell patch clamp recording. The pipette filling the solution contained 10 mM BAPTA to minimize the influence by changes in intracellular Ca 2ϩ concentration by Ca 2ϩ influx and release. The density of Pax-sensitive currents was 1.4 Ϯ 0.3 pA/pF at ϩ40 mV (n ϭ 4) in the control myocytes (i.e. expressing YFP alone; mean cell capacitance of 15.3 Ϯ 0.3 pF) and 1.1 Ϯ 0.2 pA/pF (n ϭ 6, p Ͼ 0.05 versus control) in BK␣-YFP-expressing myocytes (mean cell capacitance of 17.3 Ϯ 1.1 pF). The density of BK Ca current in aortic myocytes was not significantly changed by the expression of BK␣-YFP.
The detection of single molecule or the cluster of BK␣-YFP alone, Cav1.2-CFP alone, and BK␣-YFP/Cav1.2-CFP co-localization was performed as shown by the green, red, and yellow dots in the TIRF images, respectively ( Fig. 3C and supplemental Movies 1 and 2). The ratio of the co-localization against BK␣-YFP alone was compared between mesenteric arterial SMCs from KO and WT (Fig. 3D). The ratio of co-localization was significantly smaller in KO than WT (WT: 4.61 Ϯ 0.91, n ϭ 10; KO: 2.09 Ϯ 0.7, n ϭ 9; p Ͻ 0.05).
Molecular Assembly of Cav1.2 and BK Ca Complex and the Contribution of Caveolin-1-Furthermore, we applied singlemolecule fluorescence bleaching analyses (30,32,37) to clarify the number of GFP-tagged BK␣ and VDCC␣1C in the single fluorescent particles. Again, to minimize the artifacts of overexpression, the expression of GFP labeled BK Ca and/or Cav1.2 was kept relatively low.
At first the usefulness and accuracy of the system for singlemolecule fluorescence bleaching analyses were verified using HEK293 cells expressing Cav1.2-GFP. Single Cav1.2 as an ␣1Csubunit can form a functional channel on its own without other Ca 2ϩ channel subunits (37) (Fig. 5A). The majority of fluorescent particles displayed a single bleaching step (1 step, 79.3 Ϯ 5.3%; 2 steps, 18.4 Ϯ 5.2%; 3 steps, 2.4 Ϯ 2.4%; 4 steps, 0%, analyzed from 46 spots from 6 cells; Fig. 5B). A small part of spots bleached in two or three steps, and these probably arose from the rare co-localization of channels within a diffractionlimited area (37).
In the WT and KO myocytes, the fluorescent particles of BK␣-GFP exhibited mainly 1-4 bleaching steps (1 step, 35 (Fig. 5, C-E). These results confirmed our previous observation that only a part of BK Ca s contains BK␣-GFP as components of the tetramer with 1-3 native BK␣ molecules in mesenteric arterial myocytes under the experimental conditions in this study (30).
Notably, the proportion of the particles exhibiting four steps of Cav1.2-GFP bleaching in KO myocytes was significantly smaller than that in WT myocytes. Overall the number of bleaching steps in KO myocytes tended to be smaller than WT myocytes. These data may suggest that Cav1.2 preferentially formed homo-clusters. However, based on the finding that BK Ca forms a complex with Cav1.2 in caveolae, it is more likely that each BK␣ molecule of the tetramer formation in a BK Ca interacts with a single Cav1.2 to form hetero-clusters as a consequence.
Smaller STOC Frequency in KO Than in WT Myocytes-To examine the contribution of caveolae to the loose coupling between BK Ca and RyR, we performed simultaneous recordings of Ca 2ϩ sparks and STOCs at Ϫ40 mV in mesenteric arterial SMCs freshly isolated from WT and KO mice. Typical TIRF images are shown in supplemental Movies 3 and 4. Changes in the membrane currents and cytosolic Ca 2ϩ levels at the sites indicated by circles in the images were shown in Fig. 6A. The STOC frequency in the KO myocytes (1.1 Ϯ 0.2 Hz, n ϭ 10) was significantly smaller than the WT (2.4 Ϯ 0.4 Hz, n ϭ 5; p Ͻ 0.01) (Fig. 6B). The averaged STOC amplitude in the WT myocytes (25.1 Ϯ 1.7 pA, n ϭ 5) was similar to that in KO (29.7 Ϯ 3.5 pA, n ϭ 10; p Ͼ 0.05) (Fig. 6B). On the other hand, the Ca 2ϩ spark frequency (1.22 Ϯ 0.06 Hz, n ϭ 5) and amplitude (1.3 Ϯ 0.2%, n ϭ 5) in WT myocytes were comparable with those in KO (0.99 Ϯ 0.10 Hz, n ϭ 10; 1.8 Ϯ 0.3%, n ϭ 10, respectively; p Ͼ 0.05, Fig. 6C). These results suggest that the lack of caveolae attenuates the coupling between BK Ca and RyR and thereby reduces the efficacy to translate Ca 2ϩ spark signals to electrical STOC signals.

DISCUSSION
This report is, to our knowledge, the first to demonstrate that caveolin-1 facilitates molecular interaction between Cav1.2 and BK Ca and their accumulation as a molecular complex in caveola in living VSMCs. It is suggested that each of tetrameric BK␣ subunits, which form a functional BK Ca , can directly interact with Cav1.2 molecule to promote Cav1.2 clustering in the molecular complex in caveolae.
Several technologies, including analyses using patch clamp, biochemical procedures, mass spectrometry, and morphological analyses, have been applied to determine the molecular components and their molecular and/or functional couplings in Ca 2ϩ microdomains (1,2,20). To date, molecular interaction between two molecules among BK Ca , Cav1.2, and caveolin-1 has been reported in several types of cells including expression systems. We also showed that caveolin-1 directly interacts with BK Ca or Cav1.2 with FRET analysis. The caveolin-1 N terminus contains a caveolin-1 scaffolding domain (9) that interacts with various types of signaling molecules containing the caveolin-1   DECEMBER 20, 2013 • VOLUME 288 • NUMBER 51 binding motif (10,38). Caveolin-1 binding motifs are characterized as ⌽X⌽XXXX⌽ and ⌽XXXX⌽XX⌽, where ⌽ is an aromatic amino acid, and X is any amino acid (38). It has been reported that BK Ca has a caveolin-1 binding motif ( 1042 YNML-CFGIY 1050 ) within its C terminus and presumably accumulates in caveolae (10,11). Caveolin-3, which is mainly expressed in skeletal muscle, also directly interacts with BK Ca at the same binding site and presumably contributes to form the dystrophin microdomain (23).

BK Ca -Cav1.2 Complex in Caveolae
VDCC␣1C also contains the motif ( 1479 FDYLTRDW 1486 ) at the C terminus, and its association with caveolin-1 and caveolin-3 has been reported in several different tissues (39,40). Although the binding sites remain to be determined, the functional tight-coupling between BK Ca and Cav1.2 and their coimmunoprecipitation have been demonstrated in neurons, oocytes, and CHO cells (18). In this study the direct interaction between BK Ca and Cav1.2 was first visualized by FRET analyses in living cells. In addition, their functional coupling was demonstrated in mesenteric and aortic SMCs using two different Ca 2ϩ chelators, EGTA and BAPTA, in combination with the whole cell patch clamp recordings as has been shown in neu-rons (1,18,20). Because caveolin-1 deficiency resulted in caveolae disruption, a significant decrease in EGTA-resistant BK Ca current density, co-localization ratio, and FRET efficiency between BK␣-YFP and Cav1.2-CFP, it is clear that caveolin-1 facilitates the BK Ca -Cav1.2 interaction in VSMCs. It was, however, also suggested that caveolin-1 is not an essential factor for BK Ca -Cav1.2 tight coupling, because a small but nonetheless significant amount of coupling was observed in caveolin-1 KO. Caveolin-1 may indirectly promote BK Ca -Cav1.2 complex formation by offering a platform for both BK Ca and Cav1.2 to be accumulated in caveolae and prove higher probability to BK Ca -Cav1.2 complex formation in caveolae than in other cell membrane areas.
Single-molecule bleaching analyses revealed that single fluorescent spots derived from BK␣-GFP exhibited 1-4 bleaching steps in VSMCs. These findings confirm the previous observation that a tetrameric BK␣ assembly included 1-4 fluorescent protein-labeled BK␣s in VSMCs (30). On the other hand, a portion of the fluorescent particles of Cav1.2-GFP also exhibited more than one bleaching step in VSMCs, although only one VDCC␣1C subunit can act as a functional Ca 2ϩ channel in HEK293 cells. The possibility that the BK Ca -Cav1.2 complex may consist of tetrameric BK␣ assembly and four Cav1.2 channels has been speculated from the consequence of two-dimensional gel electrophoresis (18) and in a review (20). The result reported here that single fluorescent spots from Cav1.2-GFP exhibited multistep bleaching in mesenteric arterial myocytes provides the first direct evidence supporting this hypothesis (Fig. 8A). It has been also demonstrated that Cav1.2 can interact with nearby Cav1.2 channels via their C termini, and this interaction enables coupled gating of these channels in arterial myocytes (41). In the present study, however, the number of Cav1.2-GFP within single fluorescent particles analyzed by the step-bleaching in mesenteric arterial myocytes of KO appeared to be apparently smaller than WT. There are two limitations in this approach (37). (i) When a single fluorescent spot consists of a substantial number of multimers (Ͼ10), the detection of discrete steps becomes more difficult. In this case, the number of multimers can be estimated from the size of a single bleaching step and the total fluorescence, but the accuracy of this estimation may be limited. (ii) The other is that the GFP fluorescence emission occurs at the probability of ϳ80% in distinct fusion proteins when being excited continuously. This may result in the underestimation of the number.
In addition, SMCs from animal tissues, which expressed endogenous channels, were used in this study. In the preparation, fluorescent spots contained native subunits without GFP labeling that led to the underestimation of the subunit number within a single spot. This meant that the distribution histogram shown in Fig. 5E may be shifted to the left by the included native subunit (i.e. fluorescent spots contain more number of subunits than that estimated by bleaching steps counting). Taken together, it can be strongly suggested that caveolin-1 localizes/ accumulates Cav1.2 and BK Ca in caveolae and promotes effective coupling between these channels and their clustering, particularly that of Cav1.2 (Fig. 8B).
The tight-coupling between Cav1.2 and BK Ca and their accumulation in the caveolae of VSMCs may be much more signif- icant in terms of physiological impact in mesenteric arterial SMCs than aortic SMCs. The LVDCC density in mesenteric arterial myocytes is sufficiently high as to elicit action potentials due to sympathetic nerve stimulation (42). The fast outward current due to BK Ca activation via BK Ca -Cav1.2 tight coupling (EGTA-resistant BK Ca current) upon depolarization may contribute to early repolarization (20). It is notable that this fast outward current was not observed in KO myocytes, where BK Ca -Cav1.2 FRET efficiency was significantly reduced. The ensuing robust outward current due to CICR upon depolarization may contribute to the late repolarization and/or after hyperpolarization of an action potential (4), as has been also suggested in neurons (19). Alternatively, it is possible that the accumulation of efflux K ϩ in caveolae may induce a further activation rather than a decrease in Cav1.2 activity. This assumption seems, however, unlikely, as the membrane excitability and contraction in the mesenteric myocytes from KO mice were apparently increased more than in the WT. In fact, an application of 40 KCl to KO myocytes evoked a larger intracellular Ca 2ϩ elevation and contraction than in the WT. Moreover, in the presence of Pax, the muscular contraction in the WT was comparable with that in the KO myocytes. It is thus strongly suggested that the enhancement of BK Ca -Cav1.2 tight-coupling by caveolin-1 significantly contributes to the negative feedback regulation of membrane excitability and contraction in mesenteric arterial SMCs. RhoA and Rho kinase cause Ca 2ϩ -induced Ca 2ϩ sensitization in response to depolarization (43). Because caveolin-1 attenuates RhoA activity (44,45), caveolin-1 deficiency may activate RhoA and induce synergistic enhancement of tonic contraction. Thus, the augmentation of two pathways, (i) CICR and (ii) Ca 2ϩ -induced Ca 2ϩ -sensitization in KO cells, may enhance smooth muscle cell contraction.
So far the information about the relation between caveolae and STOC generation is controversial; both an increase (36) and decrease (14) in STOC frequency have been demonstrated upon caveolae disruption. In the present study data from a simultaneous recording of Ca 2ϩ sparks and STOCs revealed that the characteristics of the Ca 2ϩ sparks were unchanged, but the frequency of STOCs was decreased. In single myocytes, Pax induced a smaller contraction in KO than WT myocytes. This finding strongly suggests that cell excitability and contractility in the resting state are higher in KO than WT myocytes. It can be speculated that caveola deficiency makes BK Ca more inaccessible to the RyR (13), and this leads to the lower frequency of detectable STOCs. Thus, caveolae and caveolin-1 play obligatory roles in both tight and loose coupling in the Ca 2ϩ microdomain of SMCs. In addition to Ca 2ϩ sparks, Ca 2ϩ sparklets have been known as a quantal factor in Ca 2ϩ elevation by single LVDCC opening at rest (46,47). In cells where BK Ca -Cav1.2 tight coupling is present, STOCs due to Ca 2ϩ sparklets may concomitantly occur.
Guia et al. (48) have revealed that in coronary myocytes, Ca 2ϩ influx through LVDCC directly activated BK Ca at Ϫ30 mV in the presence of BayK8644, ryanodine, and cyclopiazonic acid using the cell-attached patch clamp technique. This means that the Ca 2ϩ sparklet may activate the tight-coupled BK Ca around the resting membrane potential and induce STOCs. However, these experiments were performed under limited conditions, where LVDCCs were well activated by BayK8644. Therefore, it is unclear whether Ca 2ϩ sparklets evoke or potentiate STOCs under physiological conditions. The major contributor to the STOCs generation in the resting state may be Ca 2ϩ sparks rather than Ca 2ϩ sparklets, because the amplitude of the BK Ca current due to a sparklet may be too small to explain the STOC amplitude in arterial myocytes.
In urinary bladder SMCs (49), Ca 2ϩ sparklets often trigger CICR and subsequent BK Ca current activation mainly via loose coupling, but a smaller component that is due to direct BK Ca activation is also included. 3 In the present study, however, the simultaneous measurement of Ca 2ϩ sparklet and STOCs was not systematically performed, and this remains to be determined.
Caveolin-1 acts as an endothelial nitric-oxide synthase inhibitor; accordingly, caveolin-1 ablation induces an elevation of the NO level (14). On the other hand, caveolin-1 deficiency causes remodeling of resistant vessels (50) and increases the 3 H. Yamamura and Y. Suzuki, unpublished data. A, BK Ca -Cav1.2 complex can consist of one BK Ca (a tetrameric BK␣ assembly) and four Cav1.2 channels. In caveolin-1-deficient (KO) myocytes, the interaction between BK Ca and Cav1.2 is attenuated by following reasons. The BK Ca -Cav1.2 complex, which is facilitated by the accumulation of these channels in caveolae by the interaction with caveolin-1, is reduced in KO myocytes. The number of Cav1.2 within the BK Ca -Cav1.2 complex is also reduced in KO myocytes. B, BK Ca can be activated by three steps of Ca 2ϩ increase. The first is Ca 2ϩ influx from tightly coupled Cav1.2. The second is Ca 2ϩ release from loosely coupled RyR, i.e. CICR. The third step is global Ca 2ϩ level elevation evoked by the Ca 2ϩ wave. We propose that large portions of the BK Ca -Cav1.2 complexes are localized in caveolae and that the complex of 1 BK Ca with 4 Cav1.2 cluster may contribute to initiate Ca 2ϩ local events in caveolae as the key membrane structure in Ca 2ϩ microdomain, which, however, includes the effective negative feedback mechanism for cytosolic Ca 2ϩ regulation. responsiveness to adrenergic stimulation by an elevation in protein kinase C activity (44). Furthermore the production of endothelium-derived hyperpolarization factor is impaired in the KO artery because the activities of connexin 43 and transient receptor potential vanilloid 4 (TRPV4) in the endothelium are attenuated (51). It has been also reported that the number of caveolae is reduced in endothelial cells of hypertensive rat aortae (52) and that the ratio of caveolin-1 dimer is decreased in mesenteric arterial myocytes from spontaneously hypertensive rats (53), i.e. hypertension disassembles caveolae. Our data from single isolated myocytes suggest that the lack of caveolae reduced the BK Ca -Cav1.2 complex and BK Ca activation by Ca 2ϩ influx via Cav1.2. The negative feedback mechanism for Cav1.2 activity regulation may, therefore, be reduced in KO myocytes. It is apparent that the effects of caveolin-1 deficiency on systemic arterial pressure are much more complex.
In conclusion, BK Ca activation upon depolarization takes place by three distinct steps of Ca 2ϩ increase (Fig. 8B). The first step is Ca 2ϩ influx through tightly coupled Cav1.2. The second step is local CICR via loosely coupled RyR in discrete sarcoplasmic reticula. The third step is global Ca 2ϩ level elevation due to CICR conduction in whole myocytes. It was demonstrated that large portions of the BK Ca -Cav1.2 complexes are accumulated in caveolae by the interaction with caveolin-1 serving as a microdomain that plays an obligatory role in the control of Ca 2ϩ signaling, excitability, and contractility in VSMCs.