MAGI-1 interacts with Slo1 channel proteins and suppresses Slo1 expression on the cell surface

Abstract

Large conductance Ca²⁺-activated K⁺ (BK_Ca) channels encoded by the Slo1 gene (also known as KCNMA1) are physiologically important in a wide range of cell types and form complexes with a number of other proteins that affect their function. We performed a yeast two-hybrid screen to identify proteins that interact with BK_Ca channels using a bait construct derived from domains in the extreme COOH-terminus of Slo1. A protein known as membrane-associated guanylate kinase with inverted orientation protein-1 (MAGI-1) was identified in this screen. MAGI-1 is a scaffolding protein that allows formation of complexes between certain transmembrane proteins, actin-binding proteins, and other regulatory proteins. MAGI-1 is expressed in a number of tissues, including podocytes and the brain. The interaction between MAGI-1 and BK_Ca channels was confirmed by coimmunoprecipitation and glutathione S-transferase pull-down assays in differentiated cells of a podocyte cell line and in human embryonic kidneys (HEK)293T cells transiently coexpressing MAGI-1a and three different COOH-terminal Slo1 variants. Coexpression of MAGI-1 with Slo1 channels in HEK-293T cells results in a significant reduction in the surface expression of Slo1, as assessed by cell-surface biotinylation assays, confocal microscopy, and whole cell recordings. Partial knockdown of endogenous MAGI-1 expression by small interfering RNA (siRNA) in differentiated podocytes increased the surface expression of endogenous Slo1 as assessed by electrophysiology and cell-surface biotinylation assays, whereas overexpression of MAGI-1a reduced steady-state voltage-evoked outward current through podocyte BK_Ca channels. These data suggest that MAGI-1 plays a role in regulation of surface expression of BK_Ca channels in the kidney and possibly in other tissues.

large conductance Ca²⁺-activated K⁺ (BK_Ca) channels are activated by localized increases in intracellular Ca²⁺, membrane depolarization, and in some cell types by membrane stretch (35, 44). They are expressed in an unusually wide variety of excitable and nonexcitable cells, and therefore, have diverse cell physiological functions that depend on the source of Ca²⁺ that triggers their activation, the other channels that are present in the cell, their distribution on the cell surface, and the nature of interacting proteins that might be present. The transcripts that encode the pore-forming subunits of BK_Ca channels can be alternatively spliced at multiple sites especially in the cytosolic COOH-terminal (39), and additional variation at the NH₂-terminal emerges as a result of alternative promoter utilization (28). As a result, mammalian Slo1 channels occur in (≥20) transcript isoforms (39). The COOH-tail domain of the Slo1 channel molecule contains multiple sites for Ca² binding (17, 37, 50), as well as sites that affect interactions with cytoskeletal elements (23, 47, 54, 57, 61), proteases (20), and protein kinases (26, 56). Therefore, it is not surprising that splice variation in the COOH-terminal can result in channels with markedly different gating properties (1, 6, 9, 51) as well as markedly different patterns of steady-state expression on the cell surface (22, 36).

We have recently proposed that at least some Slo1 variants have protein-protein interaction motifs at their extreme COOH-terminus that can bind to proteins with PSD-95/Disks Large/Zonula Occludins (PDZ) domains (22). PDZ domains are protein interaction modules that bind cognate ligands with specific motifs that are usually, but not always, located at the extreme COOH-terminus (13). PDZ-binding domains occur in a number of other channel molecules. For example, the closely related Slo2 protein has a COOH-terminal PDZ-binding domain that facilitates interactions with PSD-95, a member of the membrane-associated guanylate kinase (MAGUK) family of molecular scaffolds expressed at postsynaptic specializations in brain (11, 55).

MAGUK family proteins recruit complexes are composed of signaling and trafficking molecule within cells (12), and many of the resulting complexes contain ion channels (19, 30, 31, 42). MAGUK proteins are often found at specialized intercellular contact regions, including synaptic junctions in neurons and tight junctions in epithelial cells (2, 41), and they may function at least in part to coordinate the interactions of these complexes with subjacent cytoskeletal elements. MAGUK proteins have a characteristic domain structure containing multiple PDZ domains, a Src homology 3 (SH3) or a related domain, and a guanylate kinase-like (GuK) domain. A subset of the MAGUK family of molecular scaffolds is composed of the membrane-associated guanylate kinase with inverted domain orientation (MAGI) proteins, which contain six PDZ domains, two WW domains, and one GuK domain (7, 10, 53). Three closely related MAGI proteins, known as MAGI-1, MAGI-2, and MAGI-3, occur with distinct tissue distributions (7, 8, 10). Recently, MAGI-1 has been shown to interact with the Kir4.1 channel in renal cells (53) as well as with acid-sensing ion channels (ASICs) in neurons (18).

We and others have previously demonstrated that one class of Slo1 variants, known as Slo1_VEDEC after its last five residues, tend to be retained in intracellular compartments but can be translocated to the plasma membrane after activation of appropriate signaling cascades (21, 22, 36). By contrast, two other classes of COOH-terminal splice variants, referred to as Slo1_QEERL and Slo1_EMVYR, show much higher constitutive levels of steady-state expression in the plasma membrane (21, 22). To better understand these phenomena, we searched for Slo1 binding partners that would interact with the COOH-terminal portions of these molecules using a yeast two-hybrid screen. MAGI-1 emerged in a screen using a bait derived from Slo1_VEDEC, although as will be seen in Fig. 4, MAGI-1 can associate with all three classes of Slo1 COOH-terminal variants. Here we describe the interaction of Slo1 with MAGI-1 and the role of this interaction in the regulation BK_Ca channels. We show that coexpression of MAGI-1 reduces steady-state surface expression of all Slo1 channel proteins in a heterologous expression system. MAGI-1 appears to have a similar effect on Slo1 channels in the differentiated cells of an immortalized podocyte cell line in which both of these proteins are expressed endogenously.

MATERIALS AND METHODS

Yeast two-hybrid screen.

This was performed as described previously (23). Briefly, an embryonic chick ciliary ganglion cDNA library was constructed using the Matchmaker yeast two-hybrid cDNA library construction and screening kit (Clontech, Mountain View, CA). The transcriptome was inserted into the pGADT7-Rec plasmid cDNA library vector by homologous recombination within the AH109 yeast strain. The library encodes fusion proteins were composed of expression motifs with an NH₂-terminal GAL4-activation domain (AD). The cDNA library cells were selected in SD/Leu⁻ agar. Yeast two-hybrid bait plasmids encoding the Slo1 COOH-terminal splice variants were generated by cloning restriction digested RT-PCR products into the pGBKT7 yeast two-hybrid bait vector (Clontech). The bait vector was designed to contain an NH₂-terminal region composed of amino acids 1–147 of the GAL4 DNA-binding domain and amino acids 1111–1171 of the Slo1_VEDEC isoform. The 5′ and 3′ oligonucleotides incorporated NcoI and BamHI restriction endonuclease sites, respectively. The bait was transformed into the yeast strain Y187. The cDNA library was screened by mating the bait and library yeast strains with subsequent growth selection upon the quadruple drop-out media SD/-Trp/-Leu/-Ade/-His. Resulting colonies representing putative interacting proteins were confirmed by MEL1 expression using media supplemented with the chromogenic substrate 5-bromo-4-chloro-3-indolyl-α-d-galactopyranoside. Colonies representing expression of putative interacting proteins were confirmed by selection upon the quadruple drop-out media SD/-Leu/-Trp/-Ade/-His/X-α-Gal agar. After selection, the pGADT7 plasmids encoding fragments of putative interacting proteins were isolated by standard methods and transformed into Escherichia coli Top Ten cells (Invitrogen, Carlsbad, CA) and cDNAs encoding putative interacting proteins were sequenced. cDNA encoding portions of the second PDZ domain and all of the third PDZ domains of MAGI-1a and MAGI-1c were obtained in this screen.

Plasmid constructs and siRNA.

Expression plasmids encoding NH₂-terminal (ectofacial) Myc-tagged versions of Slo1_VEDEC, Slo1_QEERL, and Slo1_EMVYR were kindly provided by Dr. Min Li of the Dept. of Neuroscience at Johns Hopkins University. An expression plasmid encoding an NH₂-terminal HA-tagged form of MAGI-1a was kindly provided by Dr. Kevin Patrie of the Dept. of Internal Medicine at the University of Michigan. Plasmids encoding glutathione S-transferase (GST) were generated by cloning restriction-digested PCR products into the pGEX-KG bacterial expression vector (Amersham Biosciences, Piscataway, NJ) as described previously (23). The fidelity of all DNA constructs was confirmed by DNA sequencing. We obtained a small interfering RNA (siRNA) directed against MAGI-1a from Santa Cruz Biotechnology (Santa Cruz, CA). A negative control siRNA composed of scrambled sequence was obtained from the same vendor.

Cell culture and transfection.

These procedures were described in detail previously (24). Briefly, a human embryonic kidney cell line (HEK-293T) was transiently transfected with plasmids encoding one of the three Slo1 isoforms in the presence and absence of plasmids encoding MAGI-1a. For experiments on endogenous proteins, a podocyte cell line was obtained from Dr. Peter Mundel (Mount Sinai School of Medicine, New York, NY). Cells were maintained at 33°C in RPMI 1640 medium (GIBCO, Carlsbad, CA) supplemented with 10% FBS and 100 U/ml penicillin-streptomycin, with or without recombinant mouse γ-interferon (Sigma, St. Louis, MO), in humidified 5% CO₂ incubators. Removal of γ-interferon and a temperature switch to 37°C induced podocyte differentiation. We did not carry out tests to exclude the presence of mycoplasmal contamination. In some electrophysiological experiments, podocytes were transfected in 24-well plates using Lipofectamine-2000 (Invitrogen) with a final DNA concentration of 10 μg/ml of HA-MAGI-1a and a plasmid encoding green fluorescent protein (GFP), or with GFP plasmid alone. Podocytes and HEK-293T cells were used for physiology 48 h after transfection. The transfection efficiency of plasmids encoding full-length proteins in HEK-293T cells was >90% but was very low in podocytes (<1–2%), which precluded biochemical analysis of transfected cells in that system. Methods for transfecting podocytes with small siRNAs were described previously (24). Cells were used for biochemistry and electrophysiology 48 h after transfection.

Coimmunoprecipitation, GST pull-down assays, and immunoblot analysis.

These procedures were done as described in detail in Kim et al. (23, 24). Briefly, for coimmunoprecipitation, NH₂-terminal Myc-tagged Slo1_VEDEC, Slo1_QEERL, or Slo1_EMVYR, and NH₂-terminal HA-tagged MAGI-1a were expressed in HEK-293T cells. Experiments on endogenous proteins were carried out in differentiated podocytes. Cells were lysed and cell extracts (500–700 μg of protein) were incubated in the presence of primary antibodies anti-Myc (Cell Signaling Technology, Danvers, MA), anti-HA (clone HA-7, Sigma), anti-MAGI-1 (Abcam, Cambridge MA), anti-Slo1 (Millipore, Bellerica, MA), or IgG for 4 h at 4°C, followed by the addition of 20 μl of protein A/G agarose (Santa Cruz Biotechnology) for 12 h. Pellets were washed four times, boiled for in SDS sample buffer, and subjected to SDS-PAGE on 10% gels. Cell-extracted protein (50–100 μg) was used as control in each experiment. Blots were blocked, washed, and then incubated with the primary antibody overnight at 4°C and washed again, and the membrane was incubated with horseradish peroxidase-conjugated secondary antibody for 1 h at room temperature. The proteins were visualized using a chemiluminescent substrate (SuperSignal West Pico, Pierce Biotechnology, Rockford, IL). In these experiments, a sample of cell lysate was saved and used to visualize electrophoretic mobility of the interacting proteins and is labeled as “Input” in figures. It is not intended for quantification of the amounts of interacting protein present. For GST pulldown, the GST-Slo1 fusion proteins expressed from the pGEX-KG vector were expressed and extracted from the BL21 strain of E. coli. After that, 100 μg of each eluted GST-Slo1 fusion protein were bound to different 15-μl aliquots of glutathione-sepharose 4B beads (GE Healthcare Biosciences, Piscataway, NJ) in ice-cold PBS. Samples composed of 300 μg of lysate from podocytes or transfected HEK-293T cells were analyzed for interaction with the different fusion proteins. These binding reactions were carried out in 0.5% Triton X-100 in PBS (PBST) at 4°C with gentle rotation overnight. The bound proteins were eluted into SDS sample buffer by heating to 100°C for 5 min. The entire sample was loaded onto 9% SDS-PAGE gels and then analyzed by immunoblot.

Confocal microscopy.

These procedures were described in detail previously (24, 25). To monitor surface expression of Slo1, HEK-293T cells grown on poly-d-lysine-coated coverslips were transiently transfected with Myc-tagged Slo1_VEDEC, Slo1_QEERL, or Slo1_EMVYR with or without HA-MAGI-1a. Cells were subsequently exposed to fluorescein-conjugated goat anti-Myc (Abcam) (1:500) in Opti-MEM medium for 1 h at 37°C to label surface Slo1 channels. Cells were then washed in PBS, fixed by 30 min exposure to 4% paraformaldehyde in PBS, rinsed in PBS, blocked with 10% normal goat serum, and then permeabilized in PBS containing 0.5% Triton X-100. They were then incubated with mouse anti-Myc antibody (1:1,000) for 1 h (Cell Signaling Technology, antibody 9B11) and then exposed to Alexa-568-conjugated anti-mouse IgG (Invitrogen/ Molecular Probes) (1:1,000) for 1 h to label intracellular Slo1 channels. The cells were then rinsed in PBS and mounted using Vectashield (Vector Laboratories, Burlingame, CA). Laser excitation intensities and detector sensitivities were held constant to facilitate comparison of surface expression of Slo1. To assess colocalization of Slo1 and MAGI-1 in podocytes, cells were fixed in 4% paraformaldehyde, blocked, and permeabilized in PBST and exposed to a monoclonal mouse anti-Slo1 and rabbit anti-MAGI-1. The monoclonal antibody against Slo1 (clone L6/60) was obtained from the UC Davis/NINDS/NIMH NeuroMab Facility, supported by National Institutes of Health (NIH) Grant U24-NS050606. Secondary antibodies were Alexa Fluor 488-conjugated anti-rabbit and Alexa Fluor 594-conjugated anti-mouse (both from Invitrogen/Molecular Probes, 1:1,000 dilution). An Olympus Fluoview FV1000 confocal microscope equipped with multiline argon (457 nm, 488 nm, 514 nm) and green helium neon (543 nm) lasers was utilized for data acquisition with an Olympus ×60/NA1.42 oil immersion objective. Stacks of images from different Z-planes were obtained using a step size of 200 nm, and fluorescent signals from different channels were scanned sequentially. Images that show colocalization of proteins are from the same optical section. For analyses of Slo1 surface expression, fluorescent intensity from the fluorescein-conjugated goat anti-Myc and Alexa-568-conjugated anti-mouse IgG was quantified by measuring signal intensity in intracellular compartments and on the cell surface of individual HEK-293T cells using Photoshop v5.5. Data were analyzed from a single optical section of 200-nm thickness from 50 cells in each group, and the ratio of surface signal intensity to intracellular signal intensity was computed for each cell, such that each cell is its own control. The TIFF images were not processed before measurement of signal intensities. In control experiments, we observed that no signal was obtained using preabsorbed primary antibodies (data not shown).

Cell-surface biotinylation.

Cell-surface biotinylation assays were performed as described in detail previously (21, 22, 23). Briefly, intact cells were treated with a membrane-impermeable biotinylation reagent sulfo-N-hydroxy-succinimidobiotin (Pierce Biotechnology) (1 mg/ml in PBS buffer) for 1 h on ice with gentle shaking. The reaction was stopped, cells were lysed, and biotinylated proteins from the cell surface were recovered from lysates by incubation with immobilized streptavidin-agarose beads (Pierce Biotechnology). A sample of the initial cell lysate was also retained for analysis of total proteins. These samples were separated on SDS-PAGE, and proteins were quantified by immunoblot analysis. Protein bands in immunoblots were quantified by densitometry using ImageJ software (NIH, Bethesda, MD). These and all other biochemical experiments were repeated at least three times. For quantification, we set the band with the highest density as 100%, and we normalized all other bands against that value to calculate mean maximum density.

Electrophysiology.

These recordings were carried out as described in detail previously (21, 22, 23) using an Axopatch 1D amplifier (Molecular Devices). HEK-293T cells do not endogenously express BK_Ca channels but express them at high levels after transient transfection. For whole cell recordings from HEK-293T cells, the bathing solution contained (in mM) 150 NaCl, 0.08 KCl, 0.8 MgCl₂, 5.4 CaCl₂, 10 glucose, and 10 HEPES, and the pH was adjusted to 7.4 with NaOH. The pipette solution contained (in mM) 145 NaCl, 2 KCl, 6.2 MgCl₂, 10 HEPES, and 5.0 N-(2-hydroxyethyl)ethylenediamine-N,N′,N′-triacetic acid (H-EDTA); pH 7.2. The free Ca²⁺ concentration in this solution was titrated to a concentration of 5 μM using a calibrated calcium electrode. These ionic conditions were chosen to provide sufficient intracellular Ca²⁺ for activation of BK_Ca channels by depolarizing step pulses while at the same time keeping the resulting macroscopic currents sufficiently small to avoid saturation of the patch-clamp amplifier. However, the potassium equilibrium potential was physiological (−85 mV). We have previously shown that these currents are completely blocked by application of the specific inhibitor paxilline (25). For experiments on podocytes, the bath solution contained (mM) 150 NaCl, 5.4 KCl, 0.8 MgCl₂, 5.4 CaCl₂, and 10 HEPES; pH 7.4. Pipette solutions contained 10 mM NaCl, 125 mM KCl, 6.2 mM MgCl₂, and 10 mM HEPES, pH 7.2, and 5 μM free Ca²⁺ buffered with 10 mM H-EDTA, as determined with the calcium electrode. Cell capacitance in individual podocytes ranged from 100 to 250 pF. Voltage activated whole cell currents in podocytes when recording pipettes contain no added CaCl₂ and 10 mM EGTA (data not shown). In podocytes and HEK-293T cells, whole cell currents were evoked by a series of eight 450-ms depolarizing steps from a holding potential of −60 mV. Data analyses were as described previously (60). Macroscopic BK_Ca currents in podocytes have unusually slow activation kinetics, possibly due to the presence of β₄-subunits, as we have described in detail previously (24). They are also blocked completely by paxilline (24).

Statistics.

All quantitative data are presented as means ± SE. Data were analyzed by Student's unpaired t-test, since each group has its own control. Throughout this, P < 0.05 was regarded as significant.

RESULTS

The extreme COOH-terminal of Slo1 occurs in multiple variants as a result of alternative splicing in the principle neurons of the embryonic chick ciliary ganglion (21, 22) as well as in mouse podocytes of the renal glomerulus (24). We recently proposed that the Slo1 COOH-terminals may function as PDZ domain binding motifs that play a role in regulating steady-state surface expression of BK_Ca channels (22). To explore this idea, we performed yeast two-hybrid screens of an embryonic chick ciliary ganglion cDNA library using baits derived from all three COOH-terminal Slo1 variants. An apparent interaction with a cDNA library clone encoding MAGI-1 emerged from the Slo1_VEDEC bait. A schematic diagram of the domain structure of MAGI-1 is shown in Fig. 1A. MAGI-1 occurs in three splice variants (MAGI-1a, MAGI-1b, and MAGI-1c), which diverge primarily in the COOH-termini downstream of the fifth PDZ-domain, although MAGI-1b also has a small insert in the vicinity of the second PDZ domain (8). The avian cDNA that we found contained a portion of the second PDZ domain and the entire third PDZ domain found in MAGI1a and MAGI-1c, and it lacked the insert found in mammalian MAGI-1b. Full-length avian MAGI-1 cDNAs are not available, and therefore, we proceeded to confirm the interaction of mammalian MAGI-1 with mammalian Slo1 by several independent methods. To carry out pull-down assays, we prepared GST-Slo1 fusion proteins that correspond to the three mammalian Slo1 COOH-terminal splice variants as described previously (23, 24). These were incubated with lysates of podocytes or lysates of HEK-293T cells transiently overexpressing a construct encoding a HA-tagged MAGI-1a fusion protein. The GST-Slo1fusion proteins and their interacting proteins were then isolated by affinity chromatography and separated on SDS-PAGE, and interacting proteins were identified by immunoblot analysis. In experiments on podocyte cell lines we used a commercially available antibody against MAGI-1 that does not discriminate between different MAGI-1 splice variants (Fig. 1B), whereas for HEK-293T cells MAGI-1 was detected using an antibody against the HA tags (Fig. 1C). With both designs, we observed that all three GST-Slo1 fusion proteins were able to interact with full-length MAGI-1 protein. This interaction was not seen with GST itself. This implicates the extreme COOH-terminals of all three Slo1 variants as one of the Slo1 channel domains that drives the interaction with MAGI-1.

Fig. 1.

Membrane-associated guanylate kinase with inverted orientation protein-1 (MAGI-1) interacts with three different COOH-terminals of Slo1 proteins. A: schematic diagram of the domain structure of MAGI-1. Top bar shows region of the cDNA that was obtained by yeast two-hybrid screen. B: glutathione S-transferase (GST) pull-down assay on lysates prepared from differentiated cells of a podocyte cell line using GST fusion proteins that included the COOH-termini of all three Slo1 variants. Endogenously expressed MAGI-1 was detected by immunoblot in samples prepared with all three fusion proteins but not in a sample prepared using GST. In this experiment, immunoblot analysis was carried out using anti-MAGI-1a. C: a similar result was obtained using lysates of human embroynic kidney (HEK)293T cells expressing HA-tagged MAGI-1a. In this case, MAGI-1a was detected using anti-HA.

The Slo1-MAGI-1 interaction was also observed by coimmunoprecipitation using multiple combinations of antibodies (Fig. 2). In podocytes the interaction between endogenously expressed proteins was detected when initial immunoprecipitation was carried out using a commercially available pan-Slo1 antibody, and the interaction was detected with anti-MAGI-1 (Fig. 2A, left) and also when immunoprecipitation was carried out with anti-MAGI-1, and the interaction was detected with anti-Slo1 (Fig. 2A, right). This interaction was also observed in HEK-293T cells transiently coexpressing any one of the three Myc-tagged Slo1 variants (Slo1_VEDEC, Slo1_QEERL, or Slo1_EMVYR) together with full-length HA-tagged MAGI-1a (Fig. 2B). In those experiments, we used antibodies against HA to detect or immunoprecipitate MAGI-1a, and we used antibodies against Myc to detect or immunoprecipitate the Slo1 variants. The interaction could be detected regardless of which antibody was used for the initial immunoprecipitation. These experiments confirmed that full-length MAGI-1a can form a complex with all three COOH-terminal variants of Slo1.

Fig. 2.

MAGI-1 interacts with full-length Slo1 channels. A: coimmunoprecipitation of endogenously expressed proteins in differentiated cells of a podocyte cell line. Left: detection of MAGI-1 in a precipitate prepared using an antibody against Slo1 but not in one prepared with an IgG. Right: same interaction using a reciprocal immunoblot procedure. B: coimmunoprecipitation of Slo1 isoforms and MAGI-1a in HEK293T cells transiently expressing HA-tagged MAGI-1 along with one of the three full-length Slo1 isoforms, as indicated. Each of the Slo1 channels carried an ectofacial NH₂-terminal Myc tag. Interactions could be detected when the initial precipitation was carried out with an antibody against the Myc tags on the Slo1 channels (top) or against the HA tags on MAGI-1a (bottom). No signals were detected when immunoprecipitation was carried out with IgG.

Confocal microscopy revealed at least partial colocalization of Slo1 and MAGI-1 in the differentiated cells of a podocyte cell line (Fig. 3), although colocalization was not equally strong in all regions of the cells. There appeared to be a greater degree of coexpression in the nuclear/perinuclear regions of podocytes, with somewhat weaker signals toward the periphery of the cells. In this regard, it is notable that MAGI-1c has been reported to be heavily expressed in the nuclear fractions of a kidney cell line (8). There are also isolated patches of Slo1-MAGI-1 coexpression around the surface of the cells, but these were never observed along the entire cell surface. More extensive colocalization of Slo1 and MAGI-1 was seen throughout the interior of HEK-293T cells transiently overexpressing these proteins (data not shown), although one cannot exclude that the high overlap in their distribution in that system is a consequence of abnormally high expression levels.

Fig. 3.

Partial colocalization of MAGI-1a and Slo1 in differentiated podocytes. Confocal signal from Slo1 (green fluorescence) and MAGI-1 (red fluorescence) and a merged signal (bottom) in a single optical section. Note that the most intense areas of colocalization are in perinuclear regions (middle) and in occasional patches that occur around the periphery of cells (right). Middle and right panels are expanded views of the regions inside the white boxes in the left panels.

We next carried out experiments to ascertain the functional significance of Slo1-MAGI-1 interactions. Studies in other systems have shown that MAGI family scaffold proteins can affect the trafficking and localization of their binding partners (e.g., 7, 52). To determine whether MAGI-1 produces similar effects on Slo1, we transiently cotransfected HEK-293T cells with the Myc-tagged Slo1 isoforms in the presence or absence of HA-tagged MAGI-1a and observed the steady-state surface expression of Slo1 channels by cell-surface biotinylation assays (Fig. 4) and confocal microscopy (Fig. 5). Data from cell-surface biotinylation assays were quantified by densitometry from five repetitions of the experiment. In the confocal assays, we examined the surface expression of Slo1 channels in HEK-293T cells in the presence or absence of MAGI-1a by labeling surface Slo1 channels in live cells with FITC-conjugated anti-Myc antibody (green fluorescence). The cells were then fixed, blocked, permeabilized, and labeled with a nonconjugated anti-Myc to obtain signal from intracellular Slo1 channels (red fluorescence). The distribution of channels was then analyzed by confocal microscopy, using the same laser excitation intensities for acquisition of all images. We quantified surface-to-total signal intensity ratio from each of ∼50 cells in each group. In both the cell-surface biotinylation assays and the confocal assays we observed that coexpression of MAGI-1a led to a significant decrease in the steady-state cell surface expression of all three Slo1 isoforms. Total Slo1 expression levels in whole cell lysates were not significantly different in cells coexpressing MAGI-1a with any of the Slo1 isoforms compared with controls (Fig. 4).

Fig. 4.

Coexpression of MAGI-1a reduces steady-state surface expression of full-length Slo1 channels in HEK293T cells. A: example of a cell-surface biotinylation assay carried out on HEK293T cells transiently expressing one of the three Slo1 isoforms in the presence or absence (Con) of MAGI-1a, as indicated. B: results of densitometric analyses of five repetitions of this experiment. Data were analyzed using the 130-kDa (monomeric) Slo1 signal because it yields the most reproducible results. Bar graphs show data from HEK293T cells expressing one of the three full-length Slo1 isoforms by itself (C) or in cells also expressing MAGI-1a (M). MAGI-1a had no effect on total Slo1 expression in the lysates.

Fig. 5.

Coexpression of MAGI-1a reduces steady-state surface expression of Slo1 channels in HEK293T cells. Confocal assay showing Myc-tagged Slo1 channels on the cell surface (green fluorescence) and total Myc-Slo1 expression in the same cells after fixation and permeabilization. Surface and total Slo1 expression were examined using different antibodies against the Myc tags. Images are from single optical sections, and the laser excitation intensities were kept constant throughout these experiments. Surface Slo1 signal is reduced in cells coexpressing MAGI-1a compared with cells expressing Slo1 by itself. Note higher levels of constitutive surface expression of the Slo1_QEERL and Slo1_EMVYR forms compared with the Slo1_VEDEC forms. Right bar graphs show data from 50 cells in each group showing that MAGI-1a reduces surface expression of all three Slo1 variants but has no effect on total expression. *P < 0.05 by Student's unpaired t-test.

The effect of MAGI-1a on surface expression of Slo1 had functional consequences for cells expressing these constructs. To address this, we made whole cell recordings from HEK-293T cells using recording pipettes containing 5 μM free Ca²⁺ buffered with H-EDTA as described previously (23, 24, 25). As with our previous reports, we observed large currents through BK_Ca channels in cells expressing the Myc-tagged Slo1_QEERL and Slo1_EMVYR forms and substantially smaller currents in cells expressing Myc-tagged Slo1_VEDEC. In all three cases, MAGI-1a coexpression resulted in a consistent and dramatic reduction in the BK_Ca channels generated from each of the Slo1 COOH-terminal isoforms (Fig. 6). Coexpression of MAGI-1a did not have any obvious effect on the activation kinetics or voltage dependence measured in HEK-293T cells expressing any of the Slo1 splice variants (data not shown). Together with the data from biochemistry, confocal microscopy, and whole cell recordings, these results indicate that MAGI-1a can reduce steady-state surface expression of Slo1 channels in the plasma membrane of cells in a heterologous expression system.

Fig. 6.

Coexpression of MAGI-1a reduces whole cell current through large conductance Ca²⁺-activated K⁺ (B_KCa) channels in HEK293T cells transiently expressing one of the three Slo1 variants. Whole cell recordings were made using recording pipettes filled with an H-EDTA-buffered salt solution containing 5 μM free Ca²⁺ and families of currents were evoked by a series of depolarizing step commands from a holding potential of −60 mV. A: examples of typical currents. B: bar graphs showing means ± SE of current densities compiled from >25 cells in each group. Currents were measured at the end of step commands to +80 mV. Coexpression of MAGI-1a results in smaller currents, regardless of which Slo1 isoforms are expressed. Currents in cells expressing Slo1_VEDEC are always smaller than in cells expressing either of the other variants.

However, the question arises as to whether endogenous MAGI-1 has a similar effect in a more normal cellular context. We examined this in differentiated cells derived from an immortalized podocyte cell line. To address the role of MAGI-1 in BK_Ca channel trafficking, we reduced endogenous MAGI-1 expression by applying a specific MAGI-1 siRNA in cultured podocytes. MAGI-1 is endogenously expressed in those cells where it interacts with a number of other proteins, including megalin (48), nephrin (14), and the actin-binding proteins synaptopodin and α-actinin (49). The effectiveness of the knockdown was assessed using immunoblot anaylsis, which showed that MAGI-1 expression was reduced just greater than 50% in podocytes transfected with MAGI-1 siRNA compared with cells transfected with control siRNA, based on densitometric quantification of immunoblots (Fig. 7A). We were never able to achieve anything close to a complete knockout of MAGI-1 expression. We then examined the effects of these treatments on the surface expression of Slo1. With cell-surface biotinylation assays, we observed a significant increase in the steady-state surface expression of endogenous Slo1 channels but no quantitative difference in the amounts of total Slo1 in podocytes transfected with MAGI-1 siRNA (Fig. 7B). These treatments also caused a statistically significant increase in the amplitudes of whole cell currents in podocytes measured using recording electrodes filled with a solution containing free Ca²⁺ buffered to 5 μM (Fig. 7C).

Fig. 7.

Knockdown of MAGI-1 expression in podocytes causes an increase in steady-state surface expression of BK_Ca channels. A: immunoblot analysis showing reduction in MAGI-1 expression in podoytes using a specific small interfering RNA (siRNA). Bar graph shows densitometric analysis of three repetitions of this experiment. Average knockdown is >50% of total MAGI-1 expression. B: cell-surface biotinylation assay showing that MAGI-1 siRNA causes a significant increase in steady-state surface expression of Slo1 but no effect on total Slo1. Bar graph shows densitometric analysis of three repetitions of this experiment. In this experiment, immunoblots were done with a pan-Slo1 antibody. C: whole cell recordings (example shown on left) reveal an increase in BK_Ca current in podocytes treated with MAGI-1 siRNA. Right bar graph shows means ± SE calculated from currents measured at the end of a step command to +80 mV with n = 25 cells in each group. *P < 0.05 by Student's unpaired t-test.

In a converse experiment, we examined whether overexpression of HA-tagged MAGI-1a influences the surface expression of endogenous Slo1 in podocytes (Fig. 8). Because of the relatively low transfection efficiency for full-length cDNA constructs, this could only be examined by means of whole cell recordings (which can be done in single cells in which transfection could be ascertained by the presence of green fluorescence). We observed that overexpression of HA-MAGI-1a together with GFP in podocytes resulted in a significant decrease in macroscopic outward current compared with that observed in podocytes transfected with GFP alone (Fig. 8). Collectively, these data are consistent with a model in which MAGI-1a plays an inhibitory role in the regulation of surface expression of podocyte Slo1 channels.

Fig. 8.

Overexpression of MAGI-1a in podocytes causes a decrease in current through BK_Ca channels in podocytes. Cells were transiently transfected with green fluorescent protein (GFP) and MAGI-1a or GFP alone (control cells), and whole cell recordings were made from fluorescent cells. Cells overexpressing GFP and MAGI-1a had a significant decrease in current density compared with cells overexpressing GFP. Data were from currents measured at the end of a step command to +80 mV. *P < 0.05 by Student's unpaired t-test with 25 cells in each group.

DISCUSSION

In this study, we demonstrate that MAGI-1, a widely distributed member of the MAGUK family of scaffolding proteins (8, 29), interacts with Slo1, the pore-forming subunit of BK_Ca channels. We also show that this interaction suppresses the trafficking of Slo1 proteins to the cell surface in a heterologous expression, as well as in a renal-derived cell line where these proteins are expressed endogenously. Although this interaction was initially suggested on the basis of interactions with one specific Slo1 COOH-terminal splice variant, we observed this effect on all three classes of vertebrate Slo1 COOH-terminal variants that are present in the data base.

We have previously proposed that the terminal residues of Slo1 channels may function as PDZ domain-binding motifs. Consistent with that, the fragment of MAGI-1 that we obtained in the yeast two-hybrid screen contained the entire third PDZ domain of MAGI-1a and MAGI-1c. However, interaction with Slo1 could also occur at any of the other PDZ domains, at the WW domains, or at the GuK domain. In this regard, we have previously shown that the β₁-subunits of voltage-activated Ca²⁺ channels (Ca_vβ1) contain a GuK domain that binds to a conserved site close to the so-called “calcium bowl” domains found in all Slo1 isoforms (60). That interaction can occur even in the absence of the pore-forming subunits of Ca²⁺ channels and it causes marked modulation of the gating properties of Slo1. The effects are characterized by slowing of the activation kinetics and an apparent reduction in the Ca²⁺ sensitivity of the Slo1 channels. This was not observed in the present study after coexpression of MAGI-1a. In addition, the interaction with Ca_vβ1 has no effect on the steady-state surface expression of Slo1 in HEK-293T cells (60), in marked contrast to the effects of MAGI-1a described here. Slo1 channels also have a noncanonical proline-rich SH3 domain-binding motif that contributes to the interactions with cortactin (54) and Ca_vβ1 (60). This is significant because SH3 domains and WW domains appear to have similar binding surfaces (38) that can explain their very similar binding predilections (5, 38). Based on our studies of Ca_vβ1, it is possible that interactions with a MAGI-1a WW domain could stabilize a MAGI-1-Slo1 complex but are unlikely by themselves to lead to retention of Slo1 in intracellular compartments.

Our present data do not address the mechanism whereby MAGI-1a suppresses constitutive trafficking of Slo1 to the cell surface; indeed this is not a universal consequence of its interactions with other ion channel proteins. To the contrary, the COOH-terminal of acid-sensing ion channel-3 (ASIC-3) contains a PDZ-binding motif that enables it to interact with MAGI-1, and coexpressing these proteins in COS-7 cells causes an increase in proton-evoked currents through ASIC-3 with no effect on their kinetics (18), suggesting that MAGI-1b increases trafficking of ASIC-3 channels to the cell surface. This may also occur with Kir5.1/Kir4.1 channel heteromers expressed on the basalateral surfaces of tubular epithelial cells. These channels also interact with MAGI-1a, and disruption of that interaction by covalent modification of MAGI-1a causes the Kir5.1/Kir4.1 channels to move into perinuclear components (53). Thus, while interactions with MAGI-1 appear to regulate the distribution and surface expression of ion channels, the direction of the movement caused by the interaction may vary on the cell type and channel under consideration.

Broad classes of models to explain these phenomena can be suggested on the basis of the other proteins that interact with MAGI-1. In podocytes, MAGI-1 interacts with nephrin (14), α-actinin-4 (49), megalin (14, 48), ZO-1 (29), and synaptopodin (49), suggesting a link between MAGI-1, its binding partners, and the underlying actin cytoskeleton. Moreover, in other cell types MAGI-1 has been demonstrated to associate with various GTPase-associated proteins (40, 43, 52) that are likely to play a role in regulating cytoskeletal dynamics and possibly BKCa trafficking (61). Finally, PDZ domains have recently been shown to bind phosphinositides, which could provide a direct mechanism for regulating binding and unbinding of the cognate ligands (58) and thus their cellular distribution.

In this regard, we and others have shown that direct interactions with actin cytoskeleton are essential for normal trafficking of Slo1 channels to the cell surface (61). In addition, indirect interactions mediated by actin-binding proteins such as cortactin are required for stretch-sensitive gating of Slo1 channels (54), a mode of gating that occurs in podocytes (44) and many other cells. Finally, we have shown that interactions with nephrin are required for normal surface expression of functional BK_Ca channels in podocytes (24) and that BK_Ca channels form complexes with other podocyte channels such as TRPC6 (25). From this, it appears that MAGI-1 participates in the formation of very large and dynamic multiprotein complexes and that perturbing the stoichiometry of these complexes alters the distribution of ion channels in the cell. It is also possible that large polyvalent scaffolding molecules such as MAGI-1 allow for recruitment of signaling molecules such as mitogen-activated protein kinases (59), which can regulate BK_Ca activity and distribution (32, 34, 45, 46). Therefore, MAGI-1 may function as a docking site for a number of proteins that bind to the other PDZ, WW, or GuK domains of MAGI-1. The formation of such a large complex could provide a mechanism to regulate BK_Ca channel activity and trafficking. It may also be part of a mechanism that allows coupling between the processes of cell adhesion, dynamic changes in the cytoskeleton, and changes in associated domains of the plasma membrane.

The fact that we obtained MAGI-1 from a yeast two-hybrid screen of a neuronal cDNA library deserves some comment. We have established that growth factors regulate trafficking of neuronal BK_Ca channels in the cells from which our cDNA was constructed (3, 4, 33). MAGI-1 is expressed in the brain (8, 29), as is a closely related molecule known as MAGI-2 (also known as synaptic scaffolding molecule or S-SCAM) (15). Almost nothing is known about the role of MAGI-1 in the brain, although based on the data presented here, it is quite reasonable to suggest that it plays a role in trafficking of BK_Ca channels. However, there is a substantial literature on MAGI-2, which plays a role in trafficking of a wide range of molecules, notably N-methyl-d-aspartic acid receptors (15), dendritic HCN channels (27), and postsynaptic density proteins (16). MAGI-2 has also been proposed to play a role in clustering of synaptic molecules and remodeling of dendritic spines (7). Given this, it is tempting to propose that podocyte foot processes have dynamic properties similar to those of dendritic spines.

In summary, we demonstrated a novel interaction between MAGI-1 and Slo1, the pore-forming subunits of BK_Ca channels. Binding of MAGI-1 to these channels reduces the steady-stage surface expression of BK_Ca channels in heterologous expression systems, whereas depletion of MAGI-1 by siRNA causes an increase in the steady-state surface expression of endogenous BK_Ca channels in podocytes. This interaction may play a role in other epithelial cells, as well as in the nervous system.

GRANTS

This study was supported by a Grant to Enhance Research Effectiveness from the University of Houston Division of Research.