Slo1 Caveolin-binding Motif, a Mechanism of Caveolin-1-Slo1 Interaction Regulating Slo1 Surface Expression*

The large conductance, voltage- and Ca2+-activated potassium (MaxiK, BK) channel and caveolin-1 play important roles in regulating vascular contractility. Here, we hypothesized that the MaxiK α-subunit (Slo1) and caveolin-1 may interact with each other. Slo1 and caveolin-1 physiological association in native vascular tissue is strongly supported by (i) detergent-free purification of caveolin-1-rich domains demonstrating a pool of aortic Slo1 co-migrating with caveolin-1 to light density sucrose fractions, (ii) reverse co-immunoprecipitation, and (iii) double immunolabeling of freshly isolated myocytes revealing caveolin-1 and Slo1 proximity at the plasmalemma. In HEK293T cells, Slo1-caveolin-1 association was unaffected by the smooth muscle MaxiK β1-subunit. Sequence analysis revealed two potential caveolin-binding motifs along the Slo1 C terminus, one equivalent, 1007YNMLCFGIY1015, and another mirror image, 537YTEYLSSAF545, to the consensus sequence, φXXXXφXXφ. Deletion of 1007YNMLCFGIY1015 caused ∼80% loss of Slo1-caveolin-1 association while preserving channel normal folding and overall Slo1 and caveolin-1 intracellular distribution patterns. 537YTEYLSSAF545 deletion had an insignificant dissociative effect. Interestingly, caveolin-1 coexpression reduced Slo1 surface and functional expression near 70% without affecting channel voltage sensitivity, and deletion of 1007YNMLCFGIY1015 motif obliterated channel surface expression. The results suggest 1007YNMLCFGIY1015 possible participation in Slo1 plasmalemmal targeting and demonstrate its role as a main mechanism for caveolin-1 association with Slo1 potentially serving a dual role: (i) maintaining channels in intracellular compartments downsizing their surface expression and/or (ii) serving as anchor of plasma membrane resident channels to caveolin-1-rich membranes. Because the caveolin-1 scaffolding domain is juxtamembrane, it is tempting to suggest that Slo1-caveolin-1 interaction facilitates the tethering of the Slo1 C-terminal end to the membrane.

The large conductance, voltage-and Ca 2؉ -activated potassium (MaxiK, BK) channel and caveolin-1 play important roles in regulating vascular contractility. Here, we hypothesized that the MaxiK ␣-subunit (Slo1) and caveolin-1 may interact with each other. Slo1 and caveolin-1 physiological association in native vascular tissue is strongly supported by (i) detergent-free purification of caveolin-1-rich domains demonstrating a pool of aortic Slo1 co-migrating with caveolin-1 to light density sucrose fractions, (ii) reverse co-immunoprecipitation, and (iii) double immunolabeling of freshly isolated myocytes revealing caveolin-1 and Slo1 proximity at the plasmalemma. In HEK293T cells, Slo1-caveolin-1 association was unaffected by the smooth muscle MaxiK ␤1-subunit. Sequence analysis revealed two potential caveolin-binding motifs along the Slo1 C terminus, one equivalent, 1007 YNMLCFGIY 1015 , and another mirror image, 537 YT-EYLSSAF 545 , to the consensus sequence, XXXXXX. Deletion of 1007 YNMLCFGIY 1015 caused ϳ80% loss of Slo1caveolin-1 association while preserving channel normal folding and overall Slo1 and caveolin-1 intracellular distribution patterns. 537 YTEYLSSAF 545 deletion had an insignificant dissociative effect. Interestingly, caveolin-1 coexpression reduced Slo1 surface and functional expression near 70% without affecting channel voltage sensitivity, and deletion of 1007 YNMLCF-GIY 1015 motif obliterated channel surface expression. The results suggest 1007 YNMLCFGIY 1015 possible participation in Slo1 plasmalemmal targeting and demonstrate its role as a main mechanism for caveolin-1 association with Slo1 potentially serving a dual role: (i) maintaining channels in intracellular compartments downsizing their surface expression and/or (ii) serving as anchor of plasma membrane resident channels to caveolin-1-rich membranes. Because the caveolin-1 scaffolding domain is juxtamembrane, it is tempting to suggest that Slo1caveolin-1 interaction facilitates the tethering of the Slo1 C-terminal end to the membrane.
Large conductance, voltage-and Ca 2ϩ -activated potassium (MaxiK, BK) 4 channels play important roles in vascular, neuronal, and urinary functions. In vascular smooth muscle, MaxiK channel appears to be a unique signaling protein because of its ability to mediate the effects of several vasoconstricting as well as vasodilating agents. The ability of MaxiK protein to complete with high fidelity these opposite tasks calls for specific associations and subcellular compartmentalization with corresponding signaling partners (1). Recently, it has been appreciated that many signaling molecules are segregated primarily in specialized microdomains like caveolae (plasma membrane invaginations enriched with cholesterol and caveolin protein), thereby, optimizing signal transduction between agonists and specific effectors (2).
Three caveolin proteins have been identified, caveolin-1, -2, and -3. All of them seem to be expressed in smooth muscle (3,4). However, gene ablation experiments have shown that caveolin-1 plays a major role in the vasculature and pulmonary function. In this animal model, the disappearance of caveolae (and caveolin-1) uncouples MaxiK channel activity to Ca 2ϩ sparks (3), and there is an increased channel current density at the surface membrane (5). Yet, mechanisms favoring the interaction of MaxiK pore-forming ␣ subunit (Slo1) with caveolin-1 or explaining an increased expression in the absence of caveolin-1 are missing. Mammalian Slo1 possesses two potential motifs for caveolin-1 binding: one consensus ( 1007 YNMLCFGIY 1015 ) and another mirror image ( 537 YTEYLSSAF 545 ) to the consensus motif, XXXXXX (where is an aromatic amino acid and X is any amino acid) (6). The presence of these two sites makes Slo1 an excellent target for interaction with caveolin-1. We found for the first time that native Slo1 from rat aorta tightly associates with caveolin-1 sharing similar microdomains and displaying a close proximity at the surface of isolated vascular myocytes. We also found that the caveolin-1-binding motif 1007 YNMLCFGIY 1015 in Slo1 is the main mechanism underlying Slo1-caveolin-1 interaction and represents a new venue for channel surface expression control.
Antibody Properties-Anti-Slo1 recognized a single strong band of ϳ125 kDa in WB of transfected cells (Figs. 3 and 4; signals of dimers may also be detected); in native aorta, WB shows a strong signal of the expected molecular mass (ϳ125 kDa; weak signals that may correspond to dimers may also be detected) and a weaker ϳ75-kDa band. WB and immunocytochemistry signals were all absent by preadsorbing the Ab with the antigenic fusion protein (1 g of Ab:10 g of antigen/ml). Anti-caveolin-1 Abs (poly and monoclonal) recognized a single strong band of the expected size (ϳ21 kDa) in transfected cells (Figs. 3 and 4) and in aortic lysates and do not cross-react with expressed caveolin-2 or caveolin-3 (not shown). No signal was detected when polyclonal caveolin-1 Ab was preadsorbed with the antigenic peptide (0.25 g of Ab:2.5 g of peptide/ml, Santa Cruz Biotechnology). Anti-␤1 subunit polyclonal Ab recognized a single strong band of ϳ26 kDa (corresponding to the partially glycosylated form; without glycosylation the ␤1 molecular size is ϳ22 kDa) in transfected cells (8) (Fig. 3), and the signal practically vanished when the Ab was preadsorbed with the antigenic peptide (2 g of Ab:10 g of peptide/ml, Novus Biologicals). Secondary Abs gave practically no signals in immunocytochemistry and WB.
Digital Image Processing-Confocal sections were acquired every 0.1-0.2 m (z-axis) at 0.115 m/pixel (x-y plane) for aortic myocytes and every 0.25 m at 0.0575 m/pixel for transfected cells. Images were three-dimensionally blind deconvolved using AutoQuant, and pixel intensities were measured with Metamorph. Analysis for protein proximity index (PPI) was carried out with custom-made software (11). All conditions, including optical sectioning and exposures, were identical for a given experiment.
Patch Clamp-Macroscopic currents were measured in inside-out patches. To compare current densities, we used pipettes of similar resistances (ϳ2 M⍀) that should have similar diameters and normalized the currents to the resistance of each patch pipette. Voltage-dependent activation (fractional open probability (FPo) Ϫ voltage) curves were obtained by fitting the experimental data to a Boltzmann distribution: where; G ϭ conductance, G max ϭ limiting maximum conductance, V1 ⁄ 2 ϭ half-activation potential, z␦ ϭ effective valence, and F, R, and T have their usual thermodynamic meanings. Pipette and bath solutions were (mM): 105 potassium methanesulfonate, 5 KCl, 10 HEPES (free Ca 2ϩ ϭ 6.6 M), pH 7.0. Caveolin-1 SD peptide was reconstituted in 70% ethanol (stock 10 mM). Its effect was tested by first acquiring control currents in the presence of vehicle alone (0.07% ethanol and 0.01% bovine serum albumin).
Statistical Analysis-Data are mean Ϯ S.E. Tests for significance were with Student's t test (p) except in Fig. 2, where the normal approximation of the sign test (p sign test ) was used. p Ͻ 0.05 was considered significant; n, denotes the number of preparations. A minimum of three different animals was used in all experiments.

Association of Slo1 and Caveolin-1 in Native Vascular Tissue-
We first examined the possible association of Slo1 with caveolin-1 in native rat aorta by three independent methods: (i) comigration in sucrose density fractions after detergent-free extraction; (ii) reverse co-immunoprecipitation (co-IP), and (iii) co-localization in freshly dissociated myocytes.
Reverse Co-IP Supports Slo1 and Caveolin-1 Close Association in Native Vascular Tissue-The co-migration profile in Fig. 1A suggests the possible association of aortic Slo1 and caveolin-1 in a macromolecular complex. Indeed, reverse Co-IP experiments using aortic lysates support this idea. To maximize the specificity of associations, Co-IP was performed under high stringency in the presence of detergents (0.1% Nonidet P-40 and 0.25% sodium deoxycholate). To differentiate caveolin-1 signals (ϳ21 kDa) from huge signals of the IgG light chain (ϳ25 kDa), IP and WB analyses were performed with anti-caveolin-1 Abs generated in two different hosts. Typical WB analysis (Fig. 1B) shows the signals of Slo1 and caveolin-1 in tissue lysates (lane 1), and the products after IP without (blank, lane 2) and with Abs against Slo1 (lane 3) and caveolin-1 (lane 4). Regardless of the Ab used for IP, Slo1 and caveolin-1 were pulled down together. The specificity of the co-IP results was attested by the lack of signals in the blank. The mean values for co-immunoprecipitated caveolin-1 using anti-Slo1 Ab indicate that about 13 Ϯ 6% (n ϭ 3) of caveolin-1 was tightly associated with Slo1 in aorta (Fig. 1C). Likewise, 15% of Slo1 could be co-immunoprecipitated using anti-caveolin-1 Ab (n ϭ 2).
Co-localization of Slo1 and Caveolin-1 in Freshly Dissociated Aortic Myocytes-Because Slo1 is known to be abundant in smooth muscle, we directly analyzed the cellular distribution of Slo1 and caveolin-1 proteins in pre-cross-linked freshly dissociated aortic myocytes. These procedures were used to avoid any possible cellular remodeling due to enzymatic treatment or

Slo1 Channel and Caveolin-1 Complex
tissue culture (21). Single confocal images of doubly immunolabeled myocytes for Slo1 (green) and caveolin-1 (red) showed a high degree of overlap at the cell surface (Fig. 2, A-C). To eval-uate the degree of co-localization, we used a recently developed method to measure the correlation coefficient (CC) in discrete 2-m 2 areas of three-dimensional deconvolved images. The ratio of the number of areas with significant (p sign test Ͻ 0.05) positive correlation to the total number of areas examined is defined as the protein proximity index (PPI) (11). A representative example of the analysis (D-G) is shown for the cell in (A and B). Fig. 2D is the CC distribution histogram of 40 confocal planes (every 0.1 m, z-axis); in this case, the PPI for Slo1 to caveolin-1 was 0.6. Panel E shows the p sign test versus CC plot marking the data points with positive CC and p sign test Ͻ 0.05 (gray box). The PPI was practically the same at all cell depths (Fig. 2F). The localization precision of the method in the x-y plane was evaluated by shifting one of the planes every 115 nm in the x-axis, which demonstrated that the PPI decays to about half of its initial value by a shift of only ϳ400 nm (Fig. 2G). Image analysis of several preparations demonstrated a high degree of protein proximity at the plasma membrane with a mean PPI value of 0.58 Ϯ 0.07 (n ϭ 16, 5 preparations). Apparently there may be at least two cell populations as half of cells had a PPI Ͼ 0.5 (n ϭ 8, with 5 cells PPI Ն 0.9), whereas the other half had PPI Յ 0.5 (n ϭ 8). To validate our measurements, Fig. 2 (H-J) illustrates the labeling (H) and quantitative analysis (I and J) of a cell labeled for the same protein (caveolin-1) but with two different Abs. Four planes at the middle of the cell were analyzed. As expected, PPI was ϳ1, and the majority of data points had a positive CC value with p sign test Ͻ 0.05 (J, gray area). The mean PPI value was 0.98 Ϯ 0.01 (n ϭ 27 cells, two preparations).
Together the co-migration profile, the co-IP, and the significant degree of caveolin-1 and Slo1 co-localization in native aorta strongly support the idea that Slo1 can share similar plasma membrane domains with caveolin-1 and support the hypothesis that these two proteins may interact with each other. Thus, we designed experiments to test a potential protein-protein interaction and biological consequences.
Role of Smooth Muscle ␤1 Subunit in Slo1-Caveolin-1 Association in HEK293T Cells-Because in native vascular myocytes Slo1 is associated with its modulatory ␤1 subunit (22), we tested if Slo1 and caveolin-1 association is affected by coexpression of ␤1 subunit.    (6). Sequence analysis showed that Slo1 has a classic motif for caveolin-1 binding, 1007 YNMLCFGIY 1015 , and a potential caveolin-1-like binding motif, 537 YTEYLSSAF 545 , both within its C terminus. Thus, we proceeded to delete these motifs from hSlo1 and tested the impact of these modifications on hSlo1 and caveolin-1 association in co-IP experiments using co-transfected HEK293T cells. Fig. 4 (A and B) exemplifies the expression pattern of hSlo1 wild-type (WT), hSlo1 1007 YNMLCFGIY 1015 deletion mutant (hSlo1⌬1, ⌬1), and caveolin-1 expressed alone or in combination with WT or ⌬1 in cell lysates after WB analysis. All proteins were well expressed after transfection and were under the detection limit in lysates from untransfected cells. The arrows indicate the monomeric proteins of hSlo1 (ϳ125 kDa, A) and caveolin-1 (ϳ21 kDa, B). Fig. 4 (C and D) shows reverse co-IP of hSlo1-WT and hSlo1⌬1 cotransfected with caveolin-1. hSlo1-WT (Fig. 4C, top, lane 1) but not hSlo1⌬1 (Fig. 4C, top, lane 2) could efficiently co-IP with caveolin-1 when anti-caveolin-1 Ab was used for IP; the %hSlo1⌬1 that remained associated with caveolin-1 was 21 Ϯ 4% (n ϭ 4) (plot, Fig. 4C). Likewise, when anti-Slo1 Ab was used, the amount of co-immunoprecipitated caveolin-1 was greatly reduced to only 13 Ϯ 3% (n ϭ 3) when cells were co-transfected with hSlo⌬1 in comparison to hSlo1-WT (Fig. 4D). Similar results were obtained using anti-c-Myc Ab (recognizing hSlo1 N terminus) for IP (not shown). Negative controls (IP with no Ab) did not show any signals supporting the specificity of the interactions.
In summary, the results underscore the predominant role of 1007 YNMLCFGIY 1015 motif in mediating hSlo1-caveolin-1 interaction, because deletion of this motif in hSlo1 efficiently prevented their association by ϳ80 -85%, which can not be explained by misfolding of hSlo1.

Slo1 Channel and Caveolin-1 Complex
Caveolin-1 Controls hSlo1 Channel Surface and Functional Expression: Role of 1007 YNMLCFGIY 1015 -Caveolin-1 not only serves as a scaffold to keep together signaling molecules in caveolae but may also have functional consequences on their interacting protein partners by modifying their traffic in caveolar vesicles and/or inhibiting enzymatic protein activity (23,24). Thus, we wondered whether caveolin-1 interaction with hSlo1 could affect hSlo1 surface expression and/or its voltagedependent activation.
To allow simultaneous expression of hSlo1ϩcaveolin-1 in a given cell, a bicistronic vector containing both genes was used (Figs. 6 and 7). hSlo1 surface expression was monitored by immunolabeling live cells with anti-c-Myc Ab (against extracellular epitope, red) followed by permeabilization and double labeling of total hSlo1 protein with anti-Slo1 Ab (against C-terminal end epitope, green) (diagram in Fig. 6G) (10). The images reveal that hSlo1-WT surface expression was largely reduced in cells expressing hSlo1-WTϩcaveolin-1 (Fig. 6, A versus C).
The complete loss of surface functional expression induced by the deletion of the consensus caveolin-1 binding domain in hSlo1⌬1 is independent of the cell type used for expression, because similar results were attained in oocytes injected with hSlo1⌬1 cRNA (n ϭ 8, 2 oocyte batches).
The results, in particular the absence of hSlo1 surface expression when the hSlo1 1007 YNMLCFGIY 1015 consensus caveolin-1-binding motif was deleted, together with biochemical experiments support the view that caveolin-1 association with hSlo1 may serve as a down-regulator of the channel surface expression. Figs. 1 and 2 conclusively demonstrate that, in native aortic tissue, caveolin-1 and Slo1 can be in close proximity to each other. Quantitative analysis of protein proximity in freshly isolated myocytes revealed that there may be at least two populations of cells characterized by their degree of proximity between Slo1 and caveolin-1 with ϳ30% of cells showing a high degree of PPI (Ն0.9). Because dissociation of cells was performed after mild cross-linking, it is likely that these differences in protein proximity reflect true differences in types or metabolic states of vascular myocytes, rather than being a reflection of different degrees of cell transformation due to the cell isolation procedure, which can induce protein internalization (not shown). In this context, analyses of intimal and medial arterial myocytes in culture have revealed differences in contractile function and morphology supporting two different subpopulations of cells belonging to anatomically distinct regions of the rat aorta (25). Two differentially distributed Slo1 to caveolin-1 populations were also apparent from sucrose density fractionation because Slo1 was detected in both light (carbonate-resistant) and heavy (carbonate-solubilized) membrane fractions, whereas caveolin-1 was mostly present in light fractions. A bimodal distribution of Slo1 to caveolin-1 according to their solubilization properties also exists in human myometrial tissues (13). Supporting a subpopulation of aortic Slo1 in caveolin-1-rich microdomains are the co-IP results showing a fraction (ϳ20%) of Slo1 tightly associated with caveolin-1, which coincides with the percentage (ϳ30%) of double-immunolabeled cells showing a high degree of proximity between Slo1 and caveolin-1 proteins. The co-IP results also confirm the validity of the newly developed PPI method (11) to directly quantify the proximity of two proteins in native cells allowing the visual sorting of different populations of cells in regard to their macromolecular complexes.

Association of Slo1 Protein with Caveolin-1 in Native Aortic Smooth Muscle Cells-
Mechanism of hSlo1 and Caveolin-1 Association-Positive evidence for a functional correlation or association between Slo1 and caveolin-1 has been observed in other tissues (5,(13)(14)(15)26). Pull-down experiments using glutathione S-transferase-caveolin-1 protein, and lysates of HEK293T cells expressing hSlo1 indicate that these two proteins can associate; however, an indirect binding cannot be ruled out (14). Direct binding of caveolin-1 to its partners is thought to occur via the juxtamembrane caveolin-1 SD (caveolin 82-101 ) with specialized motifs, XXXXX and XXXXXX (where and X are aromatic and any amino acid, respectively) localized in their binding proteins at sites that are cytoplasmically exposed (6). The direct interaction of these two complementary motifs has been confirmed in signaling proteins like G i2␣ protein and receptor tyrosine kinases (e.g. epidermal growth factor and insulin

Slo1 Channel and Caveolin-1 Complex
receptors) (24,27). Here, we report for the first time that hSlo1 and caveolin-1 interaction requires the consensus caveolin-1binding motif 1007 YNMLCFGIY 1015 (XXXXXX) in hSlo1 as most (ϳ80%) of hSlo1-caveolin-1 association was lost following its deletion (Fig. 4). This could not be explained by a global misfolding or a major change in distribution pattern of the proteins (Fig. 5) even when hSlo surface expression was abolished (Fig. 6) as a significant amount of both hSlo1⌬1 and caveolin-1 co-localized intracellularly similarly to what was observed when hSlo1-WT and caveolin-1 were coexpressed (Fig. 5). A similar degree of interaction was reported for epidermal growth factor and insulin receptors XXXXX motifs with caveolin-1 SD using in vitro peptide competition assays (27). Mutations of the conserved residues (XXXXXX) to alanines further supported the role of 1007 YNMLCFGIY 1015 motif in hSlo1-caveolin-1 association, and highlighted the importance of neighboring Phe 1012 and Tyr 1015 in stabilizing caveolin-1 association by 15% each. Close by aromatic residues (X) have been shown to play a role in the association of glucagon-like peptide 1 receptor with caveolin-1 (28). To our surprise mutant Y1007A increased caveolin-1 association with hSlo1 suggesting that Tyr 1007 may have a destabilizing effect per se on hSlo1-caveolin-1 association. Because the consensus sequences were obtained in vitro using short peptide sequences (6), it is reasonable to imagine that in native proteins conformational requirements are also important in caveolin-1-target interactions, and that additional residues within or outside the consensus motif may also play a role. A role for residues other than the conserved aromatic residues has been observed for the glucagon-like peptide 1 receptor (28). In any event, the mutational/deletion analysis reported here support the view that 1007 YNMLCFGIY 1015 motif plays a preponderant role in hSlo1-caveolin-1 interaction. In addition to 1007 YNMLCFGIY 1015 , we also analyzed the role of a non-classic caveolin-1-binding motif, XXXXXX ( 537 YTEYLSSAF 545 ). We predicted that the site could bear the remaining 20% of hSlo-caveolin-1 interaction. Close to our prediction, deleting this site resulted in a modest (ϳ15%) but nonsignificant reduction in hSlo-caveolin-1 interaction. Interestingly, a sequence of this kind has been found to account for ␣-hemolysin-caveolin-1 interaction (29), which again points to the view that the structural context where the consensus sequence resides is also relevant for the efficiency of caveolin-1-target interactions. In any case, because caveolin-1 SD has a juxtamembrane localization, one can envision that its binding to its complementary motif in hSlo1 will "pin" the carboxylterminal end of hSlo1 to the plasma membrane in such a way that this hSlo1 region would tether to the membrane internal leaflet. In summary, these findings indicate that Slo1 can associate with caveolin-1 using mainly as mechanism the Slo1 caveolin-1-binding motif, 1007 YNMLCFGIY 1015 . Association of Slo1 with caveolin-1 SD (6) may provide an important structural means for Slo1 channel/signaling function.
New Role of Caveolin-1 in Controlling Surface Expression of hSlo1-Two main roles have been attributed to caveolin-1 in signal transduction: (i) to serve as a scaffold to keep together signaling molecules in caveolae and (ii) to serve as modulator of the activity of its associating partners (23,24). Consistent with caveolin-1 serving as scaffold to keep Slo1 together with other signaling molecules our results showed that, in native rat aorta, caveolin-1 and a population of Slo1 are co-localized at the surface membrane and tightly associated, thus segregating to membranes with similar sedimentation properties ( Figs. 1 and  2). However, our results in inside-out patches do not support a long term or direct role of caveolin-1 in modifying hSlo1 electrical properties as co-expression of caveolin-1 or perfusion of caveolin-1 SD peptide did not alter hSlo1 voltage-activation properties nor had an apparent effect on its kinetics. Our data are consistent with studies showing that the Ca 2ϩ /V sensitivities of Slo1 are similar in cerebral myocytes of caveolin-1 Ϫ/Ϫ and caveolin-1 ϩ/ϩ mice (5). Although caveolin-1 has an inhibitory effect on several associating partners (24), exceptions to this rule have been reported. Caveolin-1 SD increases the insulin receptor kinase activity (30), whereas caveolin-1 and caveolin-1 SD peptide have no effect on COX-2 enzyme activity, although caveolin-1 binds to it (31). Our present work extends these latter observations and supports the idea that caveolin-1 has no direct effect on hSlo1 channel voltage-activation properties.
Caveolins have also been reported to participate in the trafficking of two G-protein-coupled receptors to the plasma membrane, the angiotensin II type 1A receptor and the glucagon-like peptide 1 receptor (28,32). In marked contrast to an important role of caveolin-1 in targeting G-protein-coupled receptors to the plasma membrane (32), our data indicate that efficient hSlo1 surface expression does not require co-expression of caveolin-1, but rather co-expression of caveolin-1 results in a reduction of the number of hSlo1 channels reaching the cell surface. Using two independent measurements, patch clamp and immunocytochemistry (Figs. 6 and 7), we discovered that co-expression of caveolin-1 with hSlo1 results in a reduction by ϳ60 -70% in the number of channels reaching the surface. Our findings are consistent with recent reports showing that knocking down caveolin-1 protein by small interference RNA or gene ablation results in increased functional Slo1 (BK and K Ca ) current/channel density at the surface membrane (5,14). Thus, we propose that caveolin-1 has another previously undescribed role, which is its ability to constitutively down-regulate Slo1 protein surface expression.
Role of hSlo1 Caveolin-1-interacting Motif 1007 YNMLCF-GIY 1015 in hSlo1 Surface Targeting-We discovered that, besides serving for caveolin-1 interaction (Fig. 4), the 1007 YNMLCFGIY 1015 motif in the hSlo1 carboxyl-terminal end is also required for surface targeting; its deletion keeps the hSlo1 channel trapped in intracellular compartments without affecting its tetrameric assembly (Fig. 5, G and H). The hSlo1 intracellular carboxyl-terminal end is known to play a role in Slo1 surface expression (33)(34)(35). The 1007 YNMLCFGIY 1015 sequence has several important features that may explain why its absence causes channel trapping inside the cell: (i) it has a di-hydrophobic motif (underlined), which may act as endoplasmic reticulum export signal (36), (ii) it has a targeting tyrosinebased motif YXX (bold; X, any amino acid and , hydrophobic residue), which is a general sorting sequence that directs proteins to various cellular compartments, including the plasma membrane (37), and (iii) it is located in a strategic position near another two sequences containing di-hydrophobic motifs, YGKDFCKALK 1005 (in hSlo1, amino acids KD are replaced with DL, which add another dihydrophobic motif and a tyrosine-based motif) and 1047 DLIFCL 1052 , whose deletion is sufficient to prevent current development (38) and protein surface expression (35), respectively. 1047 DLIFCL 1052 has been proposed to act as an export signal. It is reasonable to envision that the spatial conformation conferred by all three motifs provides a combinatorial signal with high affinity for coat proteins involved in the anterograde traffic of the channel protein. This assumption would explain why deletion of any of these strings of amino acids in the context of the whole protein causes a defect in surface expression.
The fact that Slo1 truncations containing the first 20 and 118 amino acids of the carboxyl-terminal end, Slo1 1-343 and Slo 1-441 are reported to reach the membrane (39), regardless of the absence of motifs 996 YGDLFCKALK 1005 , 1047 DLIFCL 1052 (35,38), and 1007 YNMLCFGIY 1015 reported here, may be reconciled by the uncovering of alternate export signals in the truncation constructs, and support the idea that Slo1 traffic mechanisms require specific conformations of the C terminus.
Interestingly, deletion of 537 YTEYLSSAF 545 motif (hSlo⌬2), although correctly folded, also prevents Slo1 surface targeting. 5 However, the presence of this motif appears insufficient for Slo1 surface targeting (although it contains a dihydrophobic export signal); a truncation construct containing it but lacking C-terminal residues beyond amino acid 651, Slo1 1-651 , has been reported not to reach the surface membrane, although it is correctly folded (34). Consistent with a major role of residues beyond 651, in the normal Slo1 traffic to the membrane, cotransfection of Slo1 1-651 with its complementary C-terminal region, Slo 652-1113 (containing residues 996 -1015 and 1047-1052) rescued the protein to the plasma membrane. Thus, it seems reasonable to propose that residues 996 -1015 and 1047-1052 are relevant for normal channel surface expression.
Residues 996 -1015 and 1047-1052 are encoded by the constitutive exons 26 and 27 of hSlo1 that seem to be selected through evolution for sequences whose overall structure are required for proper membrane targeting of the channel protein.
How can we explain the findings that hSlo1-WTϩcaveolin-1-expressing cells show partial inhibition of channel density at the surface while hSlo1⌬1 shows full inhibition? Several mechanisms could explain these findings. One can speculate that caveolin-1 association with Slo1 1007 YNMLCFGIY 1015 motif allosterically inhibits the export machinery association with the channel and that its deletion causes local structural protein rearrangements further enhancing this inhibition. Alternatively and assuming that, 1007 YNMLCFGIY 1015 forms part of the channel "export domain" one can hypothesize that in cells expressing hSlo1-WTϩcaveolin-1, caveolin-1, and export-machinery proteins compete for the same "pocket" containing 1007 YNMLCFGIY 1015 ; whereas in hSlo1⌬1, neither caveolin-1 nor the export machinery can associate with the channel resulting in lack of Slo1 surface expression. Because hSlo1 is a tetrameric channel, the degree of surface expression in the presence of caveolin-1 would vary depending on the number of occupied sites in hSlo1. In fact, a small fraction of cells transfected with hSlo1-WTϩcaveolin-1 showed both extremes, full or null surface expression. Thus, a delicate equilibrium of the hSlo1 empty occupied "export domain" shared between caveolin-1 and export-machinery proteins would define hSlo1 fate. Still a combination of allosteric and competition mechanisms may add to the complexity of caveolin-1 modulation of hSlo1 surface expression. Structural information on hSlo1 and caveolin-1 is needed to pinpoint exact mechanisms at the level of single residues.
The robust Slo1 expression at the surface of native vascular myocytes calls for different timings in Slo1 and caveolin-1 synthesis (both proteins can reach the plasma membrane when expressed independently). Conditions favoring the simultaneous synthesis of caveolin-1 with Slo1, as in our in vitro experiments, would prevent Slo1 surface expression conceivably to a degree depending on the saturation of caveolin-1 binding domains in Slo1. We recently showed that ␤1 subunit is able to increase endocytosis of hSlo1 in HEK293T cells reducing hSlo1 surface expression by ϳ20% (40). Interestingly, caveolin-1-induced reduction of hSlo1 surface expression was not further enhanced by ␤1 (Fig. 6H) nor the degree of association between hSlo1 and caveolin-1 was affected by the regulatory subunit (Fig. 3). Taken together the results suggest that, in addition to retaining hSlo1 in intracellular compartments, once at the plasma membrane, caveolin-1 may stabilize hSlo1 as well.
Summarizing, the interaction of caveolin-1 with Slo1 may serve at least two non-exclusive roles in native tissues: (a) keeping part of Slo1 in caveolin-rich structures (caveolae) and (b) regulating Slo1 surface expression. One mechanism mediating these processes is likely the caveolin-1-binding motif of Slo1.