| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* Centre for Integrative Physiology, School of Biomedical Science, University of Edinburgh, Edinburgh, UK;
Division for Molecular and Cellular Pharmacology, Medical University Innsbruck, Innsbruck, Austria; and
Pharmakologie und Toxikologie, Pharmazeutisches Institut der Universität Tübingen, Tübingen, Germany
2Correspondence: Centre for Integrative Physiology, School of Biomedical Science, University of Edinburgh, Edinburgh EH8 9XD, Scotland, UK. E-mail: mike.shipston{at}ed.ac.uk
ABSTRACT
Calcium-activated potassium (BK) channels play a central role in regulating multiple physiological processes, from the control of blood flow to neuronal excitability. Coordinated regulation of BK channel activity by changes in actin cytoskeleton dynamics has been implicated in several of these processes and related disease states such as epilepsy and stroke. However, how BK channels interact with the actin cytoskeleton is essentially unknown. Here we demonstrate noncanonical Src homology domain 3 (SH3) binding site motifs in the intracellular C terminus of the BK channel pore-forming 
-subunit that are conserved from fish to humans. These noncanonical motifs target multiple SH3 domain cellular signaling proteins to BK channels, including the SH3 adapter protein cortactin (EMS1). We demonstrate that cortactin provides a molecular bridge between BK channels and the cortical actin cytoskeleton in cells. Disruption of the SH3-mediated interaction prevents the regulation of BK channel activity controlled by changes in actin cytoskeletal dynamics. Targeting of cortactin to BK channels via a novel, noncanonical SH3 domain binding motif has important implications for the coordination of BK channel function in normal physiology and disease.—Tian, L., Chen, L., McClafferty, H., Sailer, C. A., Ruth, P., Knaus, H-G., Shipston, M. J. A noncanonical SH3 domain binding motif links BK channels to the actin cytoskeleton via the SH3 adapter cortactin.
Key Words: KCNMA1 • Src homology 3 domain • macromolecular signaling complex
LARGE CONDUCTANCE CALCIUM- and voltage-activated potassium (BK) channels are expressed in most cells of the body and play an important role in a diverse range of physiological processes ranging from control of blood flow (1
, 2)
and micturition (3)
to the control of neuronal excitability and neurotransmitter release (4
5
6
7
8)
. Indeed, a number of disorders, including hypertension (1
, 2)
, epilepsy (9)
, incontinence (3)
, and sexual dysfunction (10)
, may result from perturbations of BK channel function. A key feature of BK channels that allows them to control such diverse physiological processes is their ability to integrate multiple cellular signaling pathways (11
, 12)
, including changes in intracellular free calcium levels, cellular REDOX potential (13)
, hypoxia (14
, 15)
, reversible protein phosphorylation (16
17
18)
, and changes in cortical actin cytoskeletal dynamics (19
20
21
22
23)
. Regulation of BK channels by these pathways is dependent on cellular context, and assembly of BK channel signaling complexes is likely to be dynamic and cell specific (17
, 24
, 25)
.
Increasing evidence supports the hypothesis that the large intracellular C-terminal domain of the BK channel, pore-forming -subunit acts as a signaling organizer by allowing the BK channel to interact with multiple signaling pathways via protein-protein interactions (17
, 24
, 25)
. However, in contrast to many other ion channels that bind to adapter proteins to assemble into macromolecular signaling complexes, the BK channel is largely devoid of canonical protein-protein interaction domains. Indeed, how the majority of proteins reported to assemble as a complex with the BK channel interact with channel is unknown (17
, 24
, 25)
. To date, motifs identified and partially characterized in vertebrate BK channel -subunits are a leucine zipper domain (26)
, caveolin-targeting domain (19)
, and a tyrosine kinase immunoreceptor tyrosine-based activation motif (27)
.
The Src homology domain 3 (SH3) is one of the most ubiquitous protein interaction modules, with >200 protein members in the human genome (28
, 29)
. Typically, SH3 domains bind to canonical proline-rich domains, including a "core" PXXP motif in the target protein. These canonical motifs are further classified as type 1 (+X
PXP) or class 2 (PXPX+), where and + are usually hydrophobic and arginine residues, respectively (28
29
30)
. However, as for other protein-protein recognition domains, considerable diversity and degeneracy of SH3 domain binding motifs have been identified in a number of proteins (28
, 29
, 31)
.
Intriguingly, the intracellular C terminus of vertebrate BK channels contains a proline-rich region (Fig. 1
A, PRD) between the two regulator of potassium conductance (RCK) domains. Although no canonical PxxP motifs can be identified within the entire open reading frame of the murine -subunit, manual sequence analysis revealed two adjacent RxxPxxxP motifs (SBM1 and SBM2, Fig. 1A
) in the PRD. These motifs deviate from the canonical class 1 SH3 domain binding motif with an additional residue between the core proline residues. We thus asked whether SH3 domain proteins may interact directly with the C terminus of the BK channel via these degenerate motifs and whether such assembly is of functional relevance for BK channel regulation.
|
MATERIALS AND METHODS
Cell culture
HEK 293 cells were maintained in Dulbecco’s modified Eagle medium (Invitrogen, The Leek, Netherlands) with 10% FBS as described previously (32)
. HEK 293 cells grown on glass coverslips were transfected with the appropriate cDNA (1 µg/ml), using Lipofectamine 2000 (Invitrogen), and used for biochemical and electrophysiological recordings between 1 and 3 days after transfection. Primary hippocampal neuronal cultures were prepared from C57Bl6/N mice (Charles River, Wiga, Germany) at embryonic day 17 as described previously (33
, 34)
. Briefly, pups were decapitated, hippocampi isolated and dissociated in 0.25% trypsin for 15 min at 37°C by trituration using a flame-polished Pasteur pipette. Neurons were plated on poly-L-lysine-coated glass coverslips in 60 mm cell culture dishes at a standard density of 150000–200000 cells per dish in minimum essential medium (MEM) containing 10% heat-inactivated horse serum. Cells were allowed to attach for 3 h in an atmosphere of 5% CO2 and 37°C before transferring the coverslips neuron-side-down into dishes containing a monolayer of astroglial cells. Cocultures were maintained in serum-free MEM with N2 supplements (Invitrogen) and 0.1 mM sodium pyruvate. To prevent further glial proliferation cytosine arabinoside (4 µM) was added 3 days after plating. Neurons were used for immunocytochemistry 3 wk after plating.
Construction of expression plasmids
The epitope-tagged murine ZERO and e22 splice variants in the mammalian expression vector pcDNA3 have been described previously (32
, 35)
. The thioredoxin fusion proteins of the murine BK channel C terminus spanning amino acids V553 to H753 were generated by cloning of a polymerase chain reaction (PCR) -amplicon from the e22 or zero channel variant into the pBAD/TOPO ThioFusion vector (Invitrogen) that contains a N-terminal thioredoxin moiety and a C-terminal –V5 and –His6 motif. The central proline residues, within the noncanonical SH3 domain binding motifs (P656 of SBM1 and P667 of SBM2), were mutated to alanine by Quickchange site-directed mutagenesis (Stratagene, San Diego, CA, USA). Plasmids expressing glutathione S-transferase (GST) fusions of the SH3 domains of cortactin, CRKL, and amphyphysin were generous gifts from Xi Zhan (University of Maryland School of Medicine), John Groffen (Childrens Hospital of Los Angeles) and Mike Cousin (University of Edinburgh), respectively. GST fusions of mona/Gads were generous gifts of Roland Bourette (Centre de Genetique Moleculaire et Cellulaire, Universite Claude Bernard). Full-length mona/Gads was ligated between the BamH1 and EcoRI sites of pcDNA3.1 for mammalian expression. Full-length CRKL in the mammalian expression vector pCMV-SPORT6 was from RZPD, Heidelberg, Germany (IRAKp961M1089Q2). Recombinant thioredoxin fusion and GST fusion proteins were expressed in Escherichia coli BL21 RIL and purified using standard approaches as described previously (26)
. Mutations and sequence integrity of all constructs were confirmed by sequencing of both strands.
Electrophysiology
Single channel patch-clamp recordings of BK channels were made at room temperature (20–24°C) essentially as described (35
, 36)
. Briefly, inside-out patch experiments were performed in equimolar potassium gradients with the intracellular face of the channel exposed to solution containing in mM: 140 KCl, 5 NaCl, 2 MgCl2, 10 HEPES, 30 Glucose (Glc), 1 ATP, 1 BAPTA, pH 7.3 with free calcium buffered to 0.1 µM. The external face of the channel (in excised inside-out or cell-attached mode) was exposed to solution containing in mM: 140 KCl, 5 NaCl, 2 MgCl2, 10 HEPES, 30 Glc, 0.1 CaCl2, pH 7.4. Channel activity was recorded using an Axopatch 200B patch-clamp amplifier (Axon Instruments, USA) and analyzed using either pClamp 9 (Axon Instruments, Union City, CA, USA) or Win EDR (Version 2.3.9, Dempster, J., University of Strathclyde, Glasgow, UK). To determine the effect of actin-disrupting agents in isolated patches, BK channel activity was determined at +40 mV in patches containing less than eight channels. In patches containing two to three channels, channel amplitude and open probability (Po) were estimated from all-points amplitude histograms, determined from samples of 60 s duration, fitted with Gaussian functions. In patches containing three to eight channels, Po was determined from the total current (I), the integral of the current recorded for 60 s: where I = Nf.Po.i, with Nf being the number of functional channels in the patch, Po the open probability, and i being the single channel current amplitude. To produce the best estimate of Po, the maximum number of functional channels observed in a given patch was determined by exposing patches to >10 µM free calcium and +80 mV depolarization (36)
. Under these conditions, the channel variants used in this study are maximally activated. Cytochalasin D was dissolved in DMSO with a final bath concentration of DMSO of <0.01%, which had no effect on channel activity.
Immunoprecipitation and Western blot
HEK 293 cells and mouse brain membranes were solubilized at 4°C in lysis buffer (LB) containing (in mM): 150 NaCl; 50 HEPES, pH 7.5; 1.5 MgCl2; 10 sodium pyrophosphate; 20 NaF; 1 EDTA; 5 EGTA; 10% v/v glycerol; 1% Triton-X-100 and protease inhibitor cocktail (Roche, Basel, Switzerland). Lysates were spun and supernatants precleared for 30 min with protein-G beads (Sigma, St. Louis, MO, USA). Lysate (1 mg) was then incubated overnight at 4°C with the respective antibody (Ab) and 40 µl protein-G beads. Primary antibodies for immunoprecipitation were as follows: 5 µg rabbit anti-BK(1118–1132) (37)
, 1.25 µg anti-hemagglutinin (HA) (mouse monoclonal 12CA5, Roche or 2 µg rabbit polyclonal, Zymed Laboratories, San Francisco, CA, USA), 5 µg anti-cortactin (mouse monoclonal clone 4F11, Upstate Biotechnology or rabbit anti-cortactin, H-191, Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA), 5 µg anti-Gads (rabbit polyclonal, Upstate Biotechnology, Lake Placid, NY, USA), 2 µg anti-CRKL (rabbit polyclonal C-20, Santa Cruz Biotechnology). Negative control immunoprecipitations were performed in parallel and included use of beads alone, 10 µg nonimmune rabbit serum as immunoprecipitating Ab or Ab in the presence of competing antigen where available. In addition, for HEK293 cotransfection experiments, immunoprecipitations of mock transfected cells or cells in which BK channels lacked the C-terminal HA tag were also used as negative controls. Representative negative controls are shown in the respective figures. Bound complexes were washed a minimum of five times with LB eluted with SDS sample buffer at 45°C for 10–15 min, separated through a 10% SDS-PAGE gel, and electroblotted to polyvinidylfluoride (PVDF) membrane. Blots were probed overnight with the respective Ab at 4°C. Rabbit polyclonal antibodies and their dilutions used for Western blot were as follows: anti-BK(913–926) (38)
, 1/1000; anti-HA (Zymed Laboratories), 1/1000; anti-cortactin (H-191, Santa Cruz Biotechnology), 1/250; anti-Gads (Upstate Biotechnology), 1/500; anti-CRKL (C-20, Santa Cruz Biotechnology), 1/200; antiactin (Sigma-Aldrich, Munich, Germany), 1/250. Blots were then incubated with horseradish peroxidase-conjugated anti-rabbit IgG secondary Ab (1/5000 dilution, Diagnostics Scotland, Edinburgh, UK) for 1 h at room temperature. Signals were detected using enhanced chemiluminescence (ECL). A minimum of three independent coimmunoprecipitation experiments were performed for each construct or experiment.
GST pull-down and overlay assays
Commercial arrays (Panomics TranSignalTM SH3 domain arrays I, II, and III) of human GST-SH3 domain fusion proteins spotted in duplicate were obtained from Panomics Inc.(Fremont, CA, USA). For overlay assays as in Fig. 2
B, recombinant, purified GST fusions of the respective SH3 domains (
10 µg) were transferred to PVDF membrane with a minimum of three independent assays per construct or experiment. The commercial arrays, or overlay assays, of GST fusion proteins were blocked for 1 h in PBS containing 1% Triton X-100 (PBS-T) and 5% skimmed milk powder. Blots were incubated overnight at 4°C with the respective thioredoxin-channel fusion protein, at a concentration of 30 µg ml–1, in PBS-T and extensively washed. Interaction was detected by immunoblotting for the –V5 epitope in the thioredoxin fusion protein, using a mouse monoclonal anti-V5 Ab (1/5000 dilution, Invitrogen), and processed as for Western blots using horseradish peroxidase-conjugated anti-mouse secondary Ab. For GST fusion pull-down assays, GST fusion proteins (35 µg) of the isolated SH3 domains were bound to 60 µl of glutathione Sepharose beads (Pharmacia Biotech, Piscataway, NJ, USA) in 1 ml of PBS-T or LB for 1 h at 4°C. At least three independent pull-down assays were performed for each construct or experiment. Beads were washed extensively in PBS-T or lysis buffer and then incubated with the respective thioredoxin fusion protein (5 µg) overnight at 4°C with rotation in 1 ml of PBS-T or lysis buffer. Beads were precipitated by centrifugation and extensively washed with PBS-T or lysis buffer. Precipitates were heated at 100°C for 5 min in SDS-PAGE loading buffer, separated by SDS/PAGE, and immunoblotted for the anti-V5 epitope of the respective thioredoxin-channel fusion protein as above.
|
Immunohistochemistry and imaging: HEK 293 cells
Cell surface labeling of the N-terminal Flag tag epitope of BK channels in nonpermeabilized HEK 293 cells was performed as described previously (32)
using a 1/100 dilution of mouse monoclonal anti-Flag Ab (M2, Sigma) and 1/1000 Alexa-594 conjugated anti-mouse IgG (Molecular Probes, Eugene, OR, USA). Endogenous cortactin was detected following cell permeabilization essentially as described using 1/200 dilution of a rabbit anti-cortactin Ab (H-191, Santa Cruz Biotechnology) and 1/1000 Alexa-488 conjugated secondary anti-rabbit Ab (Molecular Probes). Cells were stained with TOPRO before mounting on microscope slides using Mowiol. Negative controls for nonspecific labeling included staining with secondary antibodies alone, staining of mock-transfected HEK 293 cells, or HEK293 cells expressing BK channels lacking the N-terminal Flag tag. Confocal images were acquired on a Zeiss LSM510 laser scanning microscope, using a 63x oil Plan Apochromat (numerical apertune=1.4) objective lens in multitracking mode to minimize channel cross-talk.
Hippocampal neurones
Immunocytochemistry, imaging, and characteristics of the mouse monoclonal anti-BK Ab (clone L6/60) used were performed as described (33)
. Briefly, neurons were fixed in ice-cold methanol for 10 min at –20°C, rehydrated, and washed in 10 mM PBS, pH 7.2 (PBS) for 15 min. Cells were blocked in 6% normal goat serum (NGS) in PBS containing 0.2% BSA (BSA, IgG-free, Sigma-Aldrich) and 0.2% Triton X-100 for 30 min at 20°C. Primary antibodies (0.5 µg ml–1, monoclonal anti-BK(m)(690–1196) clone L6/60 (NeuroMab, UC Davis); 1.5 µg ml–1 rabbit anti-cortactin Ab (Santa Cruz Biotechnology, H-191) were applied overnight at 4°C. Cells were washed and incubated with a fluorochrome-conjugated secondary Ab for 1 h (1/4000 dilution of Alexa 594 goat anti-mouse IgG and Alexa 488 goat anti-rabbit IgG; Molecular Probes). The cells were mounted in p-phenylene-diamine-glycerol and neurons were analyzed on an Axioplan microscope equipped with a 63x (1.4 numerical apertune, NA) objective lens (Zeiss, Oberkochen, Germany). Images were recorded using a CCD camera and processed with Metaview software (Universal Imaging Corp., Downingtown, PA, USA) and Adobe Photoshop.
RESULTS
Manual sequence analysis revealed that the two putative RxxPxxxP SH3 domain binding motifs (SBM1 and SBM2) are conserved from fish to man (except in Xenopus, where SBM2 is degenerate) but are not conserved in Drosophila or C. elegans, suggesting a vertebrate specialization in BK channels (Fig. 1A
). SBM1 and SBM2 are located immediately downstream of mammalian site of splicing 2. At this site of splicing, an additional putative SH3 domain binding motif, PxxPxxP, is expressed in the murine e22 alternatively spliced insert (Fig. 1A
).
Noncanonical SH3 domain binding motifs target SH3 domains to BK channels
As a first step to address whether these motifs (SBM1, SBM2, and the e22-specific motif) may function as ligands for SH3 domains, we screened a commercial library of >100 GST fusion proteins of isolated SH3 domains from human proteins (Panomics TranSignalTM SH3 domain arrays I, II, and III). The library was screened by an overlay assay (Fig. 1B
) using a soluble thioredoxin fusion protein, spanning amino acids V553 to H753 of the murine BK channel -subunit, including the alternatively spliced exon 22 and the proline-rich domain (Fig. 2A
). We identified more than a dozen robust interactions with isolated SH3 domains from multiple adapter and cellular signaling proteins. The majority of these interactions were also observed using a thioredoxin fusion protein of the zero variant BK channel in which exon 22 is excluded, and thus lacks the PxxPxxP motif of the e22 variant (Fig. 1B
, Fig. 2A
). This suggests that the majority of SH3 domain interactions observed are likely to be conserved across multiple species. Furthermore, as the PxxPxxP motif of murine exon 22 is not conserved in humans or rats (32)
, interactions via this motif are likely to be murine specific. Neither the e22 nor zero variant thioredoxin fusion proteins interact with GST alone (Fig. 1B
, Fig. 2B
). Furthermore, thioredoxin alone did not interact with any of the GST fusions of SH3 domains in these arrays (see Fig. 1B
, Fig. 2C
). Taken together, these data suggested that the proline-rich domain, spanning the putative SH3 domain binding motifs SBM1 and SBM2, is an important site for interaction with a family of SH3 domain proteins.
The three most robust interactions in the primary screen for both the e22 and zero fusion proteins were with SH3 domains from adapter proteins. These were 1) the first SH3 domain of the chicken tumor virus no.10 regulator of kinase like adapter protein (CRKL) (39)
; 2) the SH3 domain of the actin binding protein, cortactin (CTTN) (40
, 41)
; and 3) the N-terminal SH3 domain of the monocyte adaptor protein (mona), a member of the Grb2 family of adapter proteins (also referred to as the Grb2-related adaptor downstream of Shc, Gads, or GRAP2) (42
, 43)
. CRKL and mona/Gads are adapter proteins that can assemble multimolecular signaling complexes via interactions of their cognate SH3 and SH2 domains (39
, 42
, 43)
. For example, CRKL is involved in coordinating signaling events involved in cell adhesion, migration, and phagocytosis (39)
. CRKL is also essential for neural crest development in mice with deletion of CRKL phenocopy neurocristopathies of DiGeorge syndrome (44)
. Mona/Gads plays an important role in coordinating T cell receptor signaling and is implicated in monocyte and macrophage development (42
, 43)
. Cortactin directly links several proteins to the cortical actin cytoskeleton and plays an important role in coordinating the dynamics of actin reorganization (40
, 41)
.
We verified the interaction of these isolated SH3 domains using independent overlay (Fig. 2B
) and GST pull-down (Fig. 2C
) assays. For overlay assays, bacterially expressed GST fusions of the SH3 domains were immobilized on PVDF membrane and probed with the thioredoxin fusion proteins as for commercial arrays (Fig. 2B
). In agreement with commercial arrays, the e22, and zero variant thioredoxin fusion proteins interact with the SH3 domains of CRKL, cortactin (Fig. 2B
), and mona/Gads (not shown) but not with GST or other SH3 domains, such as the SH3 domain of amphyphysin (Fig. 2B
), which were negative in the primary screen of the commercial arrays. To investigate further the site of interaction, we exploited GST pull-down assays using the GST fusion of the CRKL SH3 domain (Fig. 2C
) that displayed the most robust interaction in the overlay assays (Fig. 1B
, Fig. 2B
). We first asked whether deletion of the entire PRD domain abolished interaction of the BK channel fusion protein with the GST-CRKL-SH3 domain. Truncation of the e22 variant BK channel fusion protein N-terminal to SBM1 and SBM2 (e22-
Q637 construct) abolished the interaction with the GST-CRKL–SH3 domain (Fig. 2C
), suggesting that the PRD domain is required for interaction. Similar data were observed with mona/Gads and cortactin SH3 domains (not shown). To test the hypothesis that the SBM1 and SBM2 motifs within the PRD are required for interaction of SH3 domains with BK channels, we mutated the central prolines of both the SBM1 and SBM2 domains to alanine. Mutation of proline 656 (e22P656 construct) or proline 667 (e22P667 construct) alone had no significant effect on the ability of the e22 fusion protein to interact with GST-CRKL-SH3. However, mutation of both proline residues (e22P656:667A construct) abolished interaction with GST-SH3 domains in vitro (Fig. 2C
) as for the truncation mutant, e22-Q637. Under identical conditions, the GST-CRKL-SH3 domain did not interact with thioredoxin alone. The double proline mutant e22P656:667A also failed to interact with the isolated CRKL, cortactin, and mona/Gads SH3 domains in overlay (Fig. 2B
) or pull-down assays (not shown).
Assembly of SH3 domain proteins with BK channels in cells requires the noncanonical binding motifs
To examine whether functional SBM1 and SBM2 motifs are required to target full-length SH3 domain adapter proteins to BK channels in cells, we performed coimmunoprecipitation experiments with HA-tagged, full-length BK channels expressed in HEK 293 cells. Expression of either full-length e22-HA or zero-HA BK channel -subunits in HEK 293 cells resulted in reciprocal coimmunoprecipitation with endogenous cortactin (Fig. 2D, E
). This interaction was abolished using the e22P656:667A-HA channel mutant that expresses at levels equivalent to those of the e22-HA or zero-HA variant (Fig. 2D, E
). As HEK293 cells do not express detectable levels of CRKL or mona/Gads protein (by Western blot, see Fig. 2D
or immunoprecipitation; data not shown), we coexpressed recombinant full-length CRKL or mona/Gads with the respective full-length BK channel constructs. Exogenous CRKL or mona/Gads was robustly immunoprecipitated by the respective antibodies from HEK 293 cells under our conditions (data not shown). In agreement with the in vitro data, both mona/Gads and CRKL reciprocally coimmunoprecipitated with either the e22-HA or zero-HA channels (Fig. 2D, E
). The coimmunoprecipitation of mona/Gads was abolished using the e22P656:667-HA BK channel mutant. For CRKL, coimmunoprecipitation was significantly reduced compared with e22-HA or zero-HA; however interaction was not abolished, suggesting that CRKL may be targeted in multiple ways to the BK channel complex in transfected HEK 293 cells. In support of this hypothesis, the BK channel C terminus contains a putative SH2 domain phosphotyrosine binding motif (pYxxP) for the SH2 domain of CRKL (39)
The SH3 domain adapter protein cortactin is a molecular bridge between the actin cytoskeleton and BK channels
To address whether interaction of SH3 domain proteins with the intracellular C terminus of the BK channel is important for channel behavior, we explored the functional role of BK channel interaction with cortactin. The activity and/or localization of several ion channels (45
46
47
48
49
50)
, including BK channels (19
20
21
22
23)
, can be dynamically regulated through alterations in the actin cytoskeleton. For example, in hippocampal neurones leptin activates BK channels through disruption of the actin cytoskeleton (20)
and actin-destabilizing agents activate BK channels in a variety of systems, including smooth muscle from myometrium (19)
, coronary artery (23)
, and cultured vascular endothelial cells (51)
. We thus examined whether cortactin may provide the molecular link between actin dynamics and changes in BK channel activity.
BK channels heterologously expressed in HEK 293 cells localize at the plasma membrane with endogenous cortactin (Fig. 3
A). Both e22-HA and zero-HA BK channel variants coimmunoprecipitate as a complex with both endogenous cortactin and actin in HEK 293 cells (Fig. 3B
). In channels in which the proline residues within the SH3 domain binding motifs were mutated to alanine (e22P656:667A-HA channel mutant), no coimmunoprecipitation of either cortactin or actin with the channel was observed. The double proline (P656:667A) mutation had no significant effect on channel trafficking to or insertion into the plasma membrane (see Fig. 5
and data not shown). Further, heterologous expression of BK channels did not modify the peripheral localization of endogenous cortactin compared with untransfected HEK293 cells (not shown). Thus, with the spatial limitations of colocalization analysis at the light microscopy level (200 nm), the double proline (P656:667A) mutation did not abolish the apparent colocalization of the channel with cortactin at the cell periphery in immunofluorescence assays (not shown). These data suggest that transport, or membrane insertion, of the channel is not dependent on an intact SH3 domain binding and that BK channels do not promote cortactin recruitment or redistribution at the cell periphery. Rather, BK channels in the plasma membrane may interact with the cortical actin cytoskeleton using the endogenous cortactin as a molecular bridge.
|
|
We next addressed whether BK channels assemble with cortactin in native cells. In hippocampal neurones, endogenous BK channels colocalize in discrete puncta in the proximal dendrites with endogenous cortactin (Fig. 4
A). Furthermore, native brain BK channels coimmunoprecipitate with both cortactin and actin, suggesting that a relatively stable complex can exist in native cells (Fig. 4B
). Taken together, this strongly supports the hypothesis that cortactin provides a direct molecular bridge between the BK channel and the cortical actin cytoskeleton in both heterologous systems and native cells, and that this interaction is mediated via the noncanonical SH3 domain binding motifs (SBM1 and SBM2) of the channel.
|
An actin-cortactin-BK channel complex regulates BK channel activity
To examine whether the actin-cortactin-BK channel interaction affects BK channel function, we assayed the effect of actin disruption on BK channel activity in HEK 293 cells (Fig. 5)
. Application of the actin-disrupting reagent cytochalasin D (CD, ref. 52
) to the intracellular face of inside-out patches from HEK293 cells expressing either the zero-HA or e22-HA splice variant BK channels resulted in a robust increase of channel activity by 69.0 ± 15.3% (n=11) and 74.7 ± 12.9% (n=8) of pretreatment control, respectively. Similar activation was observed in cell-attached patches: zero channels were activated by 80.3 ± 9.3% (n=12); this effect was not observed with the vehicle control (<0.01% DMSO, data not shown). The effect of CD was abolished by pretreatment of inside-out patches with the actin stabilizing drug phalloidin (52)
. The mean percentage activation elicited by CD, in the presence of phalloidin, was –4.7 ± 11.3% (n=6) and 1.2 ± 5.4 (n=9) for patches expressing the e22-HA or zero-HA variant, respectively. Phalloidin alone had no significant effect on channel activity (not shown). In contrast, CD had no significant effect on the SH3 domain binding motif channel mutant e22P656:667A-HA, mean activation was 0.4 ± 7.9% (n=8). However, the mean single channel open probability of the e22P656:667A-HA mutant (mean Po was 0.109±0.026), under basal (pretreatment) conditions, was significantly (P<0.01, Student’s t test) elevated compared with the e22-HA variant under identical conditions (the mean Po for e22-HA was 0.050±0.010, n=11). This was accompanied by a modest left shift in the apparent half maximal voltage for activation of 8 mV determined at 0.1 µM free calcium for the e22P656:667A-HA variant compared with e22-HA. The 2-fold increase in basal Po of the e22P656:667A-HA channel mutant was similar to the activation observed on actin cytoskeletal disruption by CD on e22-HA BK channels. The simplest interpretation of this increase in Po, coupled with the loss of CD activation observed with the e22P656:667A-HA variant, under identical recording conditions is that the loss of interaction of the e22P656:667A-HA channel mutant with endogenous cortactin prevents assembly with the actin cytoskeleton. Taken together, these data suggest that cortactin is a major adapter protein linking the actin cytoskeleton to BK channels at the plasma membrane through interaction of the cortactin SH3 domain with the noncanonical motifs in the BK channel C terminus.
DISCUSSION
Increasing evidence supports the hypothesis that the large intracellular C-terminal domain of the BK channel, pore-forming -subunit acts as a signaling organizer by allowing the BK channel to interact with multiple signaling pathways via protein-protein interactions (17
, 24
, 25)
. However, the BK channel C terminus is largely devoid of canonical protein-protein interaction motifs, and how the majority of proteins assemble with the channel complex is not known. Here, we demonstrate that noncanonical Src homology domain 3 (SH3) binding motifs target multiple SH3 domain adapter proteins to the intracellular C terminus of vertebrate BK channel -subunits. Furthermore, we show that the SH3 domain adapter protein cortactin acts as a molecular and functional bridge between changes in cortical actin cytoskeletal dynamics and regulation of BK channel activity.
Noncanonical SH3 domain binding motifs target multiple proteins to the BK channel complex
Identification of noncanonical SH3 domain binding motifs in the intracellular C terminus of the BK channel, pore-forming subunits considerably expands the potential diversity of proteins that may interact directly with the channel (17
, 24
, 25)
. Indeed, we identified more than a dozen distinct SH3 domain interactions in our primary screen, suggesting that proteins in addition to cortactin, CRKL, and mona/Gads may be directly coupled to BK channel, pore-forming subunits. As BK channels are tetramers of pore-forming subunits, this may allow multiple SH3 domain-dependent signaling pathways to be assembled within the same channel complex. Moreover, as many of the proteins we identified are adapter proteins that contain other common protein-protein interaction motifs such as SH2 domains (e.g., CRKL and mona/Gads), this may provide a mechanism for other protein families (e.g., SH2 domain proteins) to assemble as a complex with the BK channel. As these interactions are mediated via noncanonical SH3 domain binding motifs our data thus have important implications for other signaling pathways dependent on SH3 domain interactions that may be mediated via similar noncanonical binding domains. Indeed, increasing evidence suggests that several key pathways are dependent on noncanonical SH3 domain binding motif interactions (28
, 29
, 31)
. In addition to these channel-centric views of BK channel macromolecular assembly, assembly of BK channels with distinct SH3 domain proteins may provide an important link between BK channel integration of voltage and calcium signaling events and regulation of diverse, downstream signaling cascades (24
, 53
, 54)
. Clearly, further work is essential to address this hypothesis.
Cortactin is a molecular bridge between BK channels and the cortical actin cytoskeleton
One of the most robust interactions observed in our primary screen was for the SH3 adapter protein cortactin. Cortactin directly binds the actin cytoskeleton and is involved in coordinating the spatial organization of many transmembrane proteins (40
, 41)
. Here we provide the first molecular and functional evidence that cortactin directly links the actin cytoskeleton to BK channels, and that this interaction is important for channel regulation. Intriguingly, disruption of the actin cytoskeleton by leptin activates BK channels and suppresses epileptiform-like activity in hippocampal neurones (55)
. Furthermore, BK channel activators (56)
, as well as the actin-severing protein gelsolin (57)
, are neuroprotective in models of stroke, suggesting that assembly of the actin cytoskeleton with BK channels is an important determinant of neuronal function. Whether the interaction between cortactin and BK channels is itself dynamically regulated remains to be explored. In this regard, cortactin can be phosphorylated by both tyrosine and serine/threonine kinases, and several other proteins may interact with the SH3 domain of cortactin. The phosphorylation status of cortactin may play a role in regulating protein-protein interactions mediated via its SH3 domain (40
, 41
, 58)
.
As the actin cytoskeleton coordinates the assembly and/or targeting of other signaling proteins, it is likely that the interaction of BK channels with the cortical actin cytoskeleton coordinates assembly of other signaling pathways to the BK channel complex. In support of this idea, BK channels targeted to caveolae in smooth muscle cells coimmunoprecipitate with actin (19)
, and recent data suggest that BK channels may exist in a complex with the actin binding protein dystrophin and the dystrophin-associated protein complex in striated muscle (59)
. Furthermore, other proteins that exist in complexes with the cytoskeleton, including ß-catenin (60)
and microtubule-associated protein 1a (61)
, have been reported to interact directly with the BK channel C terminus. Thus, BK channels may utilize multiple adapter proteins to couple to the cytoskeleton, providing a mechanism for differential subcellular or tissue-specific targeting. While the functional relevance of these interactions is poorly understood, taken together with our data that cortactin acts a molecular bridge between the channel and actin, these data strongly suggest that interaction of BK channels with the cytoskeleton plays an important functional role.
Cell-specific assembly of BK channel complexes mediated via noncanonical SH3 domains?
Both CRKL (39)
and mona/Gads (42
, 43)
contain multiple protein-protein interaction domains (two SH3 domains and one SH2 domain in each protein). While the functional relevance of BK channel interaction with these proteins remains to be elucidated, both proteins act as molecular scaffolds to assemble networks of divergent signaling pathways. For example, CRKL allows integration of signaling pathways activated by growth and differentiation factors in a variety of cell types (39)
and is essential for neural crest development (44)
. In contrast, mona/Gads expression is restricted to hematopoietic cells where mona/Gads coordinates signaling cascades involved in macrophage development (42
, 43)
. Taken together, this suggests that distinct SH3 adapter proteins may allow BK channels to be coupled to distinct signaling cascades in a cell-specific manner.
An important physiological feature of SH3 domain mediated protein-protein interactions is their moderate binding affinity and relative promiscuity for SH3 domain binding motifs (28
, 29)
. Coupled with the fact that they are one of the most ubiquitous protein interaction motifs, with >200 proteins in the human proteome (28
, 29)
, this would provide a mechanism for dynamic regulation and assembly of distinct BK channel macromolecular signaling complexes determined by cell type, subcellular localization, and/or activity of signaling networks. This would facilitate nonlinear, dynamic, and combinatorial regulation of BK channel activity by multiple signaling pathways and contribute to the ability of BK channels, encoded by a single gene, to control a diverse array of physiological processes.
In conclusion, we demonstrate that noncanonical SH3 domain binding motifs in the intracellular C terminus of the BK channel, pore-forming -subunit targets multiple signaling proteins to the BK channel complex. A major challenge for the future is to characterize the BK channel macromolecular signaling complexes in different tissues and how these complexes are spatiotemporally regulated according to the physiological requirement of the tissue. Elucidation of the functional role of SH3 domain protein interactions with the BK channel should provide significant insight into how BK channel complexes are organized and regulated in health and disease.
ACKNOWLEDGMENTS
We thank members of the respective laboratories for useful discussions. This work was supported by the Wellcome Trust.
FOOTNOTES
1 These authors contributed equally to this work. 
Received for publication April 3, 2006. Accepted for publication August 22, 2006.
REFERENCES
-subunits. Proc. Natl. Acad. Sci. U. S. A. 102,17870-17876This article has been cited by other articles:
|
|
 |
|
  E. Y. Kim, K.-J. Choi, and S. E. Dryer Nephrin binds to the COOH terminus of a large-conductance Ca2+-activated K+ channel isoform and regulates its expression on the cell surface Am J Physiol Renal Physiol, July 1, 2008; 295(1): F235 - F246. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. J. Bell, K. Y. Miyashiro, J.-Y. Sul, R. McCullough, P. T. Buckley, J. Jochems, D. F. Meaney, P. Haydon, C. Cantor, T. D. Parsons, et al. From the Cover: Cytoplasmic BKCa channel intron-containing mRNAs contribute to the intrinsic excitability of hippocampal neurons PNAS, February 12, 2008; |
