| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J. Biol. Chem., Vol. 279, Issue 21, 22331-22346, May 21, 2004
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
||
From the
Department of Molecular, Cellular, and Developmental Biology and Neuroscience Research Institute, University of California, Santa Barbara, California 93106, the Department of Cell Biology, The Scripps Research Institute, La Jolla, California 92037, and the ¶Department of Physiology and Biophysics, University of Washington, Seattle, Washington 98195
Received for publication, January 12, 2004 , and in revised form, March 2, 2004.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
1-, 
1-, and 2-syntrophin, dystrophin, and dystrobrevin, interact with Kir2 channels, as demonstrated by immunoaffinity purification and affinity chromatography from skeletal and cardiac muscle and brain. Affinity pull-down experiments revealed that Kir2.1, Kir2.2, Kir2.3, and Kir4.1 all bind to scaffolding proteins but with different affinities for the dystrophin-associated protein complex and SAP97, CASK, and Veli. Immunofluorescent localization studies demonstrated that Kir2.2 co-localizes with syntrophin, dystrophin, and dystrobrevin at skeletal muscle neuromuscular junctions. These results suggest that Kir2 channels associate with protein complexes that may be important to target and traffic channels to specific subcellular locations, as well as anchor and stabilize channels in the plasma membrane. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
The strong inward rectifier potassium (Kir2.x)1 channel family of potassium channels contribute to a wide variety of physiological functions such as the modulation of cell excitability, repolarization of the action potential, and determination of the cellular resting potential (1-3). These functions are likely to be influenced by intracellular interactions with regulatory proteins.
The three most abundant strong inward rectifier potassium channels Kir2.1, Kir2.2, and Kir2.3 contain a protein-protein interaction motif at their C terminus that can bind to scaffolding proteins that contain PDZ domains. This PDZ binding motif is only found within a subset of Kir channels that also includes Kir3.2c and Kir4.1. Via the channel C-terminal class I PDZ binding motif (amino acids S(E/A)I), Kir2 channels have been reported to interact with four scaffolding/trafficking proteins: PSD-95, SAP97, Chapsyn 110, and Veli-2 (also known as Lin-7b or Mals2) (4-8). PSD-95, SAP97, and Chapsyn 110 are members of the membrane-associated guanylate kinase (MAGUK) family, a group of proteins that act as molecular scaffolds to recruit signaling and trafficking molecules to ion channels and receptors (9, 10). PSD-95 has been demonstrated to promote Kir2.3 channel clustering in the plasma membrane and to influence channel surface expression (5). SAP97 also binds to Kir2 channels (6) and has been implicated in receptor trafficking and recruitment of signaling complexes with other receptors (11). Another PDZ-containing protein, Veli-2, can bind to Kir2.3 and was reported to decrease channel internalization from the plasma membrane (7).
In addition to these previously identified Kir2 channel-interacting proteins, a large number of other proteins contain class I PDZ domains. We hypothesized that novel channel interactions may exist with previously unidentified channel-binding proteins, and we sought to identify these potential channel regulators. Not only might proteins bind directly to channels and influence their function, but scaffolding proteins may serve as a link to recruit multiprotein complexes to the channels. PSD-95, SAP97, Veli-2, as well as other scaffolding proteins are composed of modular domain structures capable of assembling with additional proteins that may influence Kir2 channel activity, trafficking, membrane stability, and localization.
We adopted a proteomics approach to identify proteins in the brain and heart that complex with Kir2 channels via the channel C-terminal PDZ binding motif. Our goal was to identify proteins that interact directly with the channels, as well as multiprotein complexes that are associated with the channels. Affinity chromatography and co-immunoprecipitation coupled with immunoblotting techniques have been useful for discovery of individual interacting proteins, but protein identification involves considerable trial and error, and macromolecular complexes are difficult to analyze by this method. An inherent difficulty in immunoblotting is that it is unlikely to reveal truly novel interactions, because potential interacting proteins must be selected by some criteria from among previously identified proteins. In addition, limitations arise from the lack of specific antibodies and the need to individually test for association with each candidate protein. An alternative approach for identification of proteins is mass spectrometry. A recent study utilized affinity chromatography combined with matrix-assisted laser desorption ionization time-of-flight spectrometry to determine peptide masses of the 5-hydroxytryptamine receptor-associated proteins separated by two-dimensional SDS-PAGE (12). Another study employed affinity chromatography and co-immunoprecipitation together with tandem mass spectrometry to identify NMDA receptor-adhesion protein signaling complexes (13). Recent advances in high pressure liquid chromatography directly coupled with tandem mass spectrometry (HPLC-MS-MS) have led to technology that allows the identification of individual proteins within complex protein mixtures (14-17).
In this study, we developed an affinity chromatography technique to specifically purify proteins associated with the C-terminal binding PDZ motif of Kir2.2. Proteins associated with the C terminus of Kir2.2 were identified by HPLC-MS-MS, N-terminal Edman degradation, and immunoblotting. We identified eight members of the MAGUK family of proteins (SAP97, PSD-95, Chapsyn-110, SAP102, CASK, Dlg2, Dlg3, and Pals2), two isoforms of Veli (Veli-1 and Veli-3), Mint1, and the actin-binding LIM protein (abLIM) as Kir2.2-associated proteins. Several of these proteins were investigated in detail in a parallel study in which we demonstrated that SAP97, CASK, Veli-3, and Mint1 form a multiprotein complex with Kir2 channels in brain and heart and showed that a CASK-containing complex is involved in channel trafficking (18). Here, we also show for the first time that components of the dystrophin-associated protein complex (DAPC), including 1-, 1-, and 2-syntrophin, dystrophin, and dystrobrevin, associate with the C terminus of Kir2 channels in muscle and brain. Immunofluorescent localization studies demonstrated that Kir2.2 colocalizes with the DAPC components syntrophin, dystrophin, and dystrobrevin at skeletal muscle neuromuscular junctions.
|
EXPERIMENTAL PROCEDURES |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
cDNA Constructs—cDNAs encoding C-terminal regions of Kir channels were generated by PCR and cloned in-frame to the 3'-end of GST in the bacterial expression vector pGEX2T (Amersham Biosciences). The C-terminal segments of the fusion constructs coded for Kir2.1 (mouse Kir2.1 amino acids 372-428), Kir2.2 (rat Kir2.2 amino acids 362-427), Kir2.2
3 (rat Kir2.2 amino acids 362-424), Kir2.3 (human Kir2.3 amino acids 390-445), and rat Kir4.1 (amino acids 329-379). cDNAs encoding full-length Kir2.2 and Kir2.23 were subcloned into pcDNA1 as described previously (18). cDNAs encoding 1-, 1-, and 2-syntrophin were subcloned into pGW1.
Preparation of Affi-Gel-GST-Kir2.2 Affinity Matrix—BL21 Escherichia coli expressing either GST, GST-Kir2.2, or GST-Kir2.23 fusion proteins (6) were pelleted and resuspended in 40 ml of HNG (20 mM Hepes, pH 7.4, 1 mM EDTA, 1 mM EGTA, 1 mM MgCl2, 150 mM NaCl, 10% glycerol, 1 mM PMSF, 1x protease inhibitor mixture) plus 1% Triton X-100 on ice. Cells were lysed with a French press. The lysate was centrifuged at 16,000 x g for 10 min. Two milliliters of packed glutathione-agarose beads was added to the supernatant, and the mixture was incubated 30 min at 4 °C while rotating. Beads were pelleted, and the supernatant was removed, followed by washing the beads three times, once with HNG plus 1% Triton X-100 and twice with HNG. Bound GST fusion proteins were eluted from beads by the addition of 20 mM glutathione (Sigma). Eluted proteins were then dialyzed against 200 volumes 1x PBS overnight at 4 °C. To prepare affinity columns, 5 mg of purified GST fusion protein was incubated with 1 ml of packed Affi-Gel-10 (Bio-Rad) overnight at 4 °C according to manufacturer's instructions. To remove uncross-linked fusion protein, the Affi-Gel-10/GST fusion protein matrix was transferred to a column and washed consecutively with 20 volumes of Buffer C (50 mM Hepes, pH 7.6, 125 mM NaCl, 20% glycerol, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol), 20 volumes of Buffer C with 2.5 M urea, 20 volumes of Buffer C with 4 M urea, 10 volumes of Buffer D with 1% Triton X-100, and finally 10 volumes of Buffer D with 1 mM PMSF, 0.02% sodium azide and stored as 50/50 slurry at 4 °C.
Affinity Chromatography for Silver Staining, Western Blotting, and Mass Spectrometry—50 µl of packed Affi-Gel-GST fusion protein matrix (5 mg/ml) was added to 10 ml of 1% Triton X-100 solubilized rat brain or heart protein (
10 mg/ml) and incubated in batch overnight at 4 °C while rotating. At 4 °C, matrix was loaded onto a column and washed with 1000 column volumes of Buffer D wash (Buffer D, 1% Triton X-100, 0.5% IGEPAL CA-630 (Nonidet P-40), 500 mM NaCl, 1 mM PMSF) followed by 1000 column volumes of Buffer D with 1% Triton X-100, 1 mM PMSF. Columns were moved to room temperature, and bound proteins were eluted with 400 µl of Buffer D, 1% Triton X-100, supplemented with 2 mg/ml competitive peptide (ERPYRRESEI) for 1 h, followed by elution with Buffer D containing 1% SDS. Eluted proteins were precipitated with 20% trichloroacetic acid, washed with -20 °C acetone, and dried with centrifugation under vacuum. Dried proteins were either used directly for mass spectrometry analysis (see below) or resuspended in 1x SDS sample buffer and separated by SDS-PAGE. Gels were either silver-stained (Silver Stain Plus, Bio-Rad) or transblotted to Hybond ECL nitrocellulose membranes (Amersham Biosciences) and incubated with primary antibodies followed by peroxidase-conjugated secondary antibodies. Immunoreactive proteins on membranes were detected by incubation with Supersignal West Dura enhanced chemiluminescent substrate (Pierce) followed by exposure to x-ray film.
Affinity Chromatography for N-terminal Microsequencing—Brain extract was subjected to the purification scheme described above, but scaled up ten times. Purified proteins were separated by SDS-PAGE and transferred to Hybond-P (Amersham Biosciences) polyvinylidene fluoride (PVDF) membrane. Proteins transferred to PVDF were stained with Coomassie Blue and air-dried, the desired protein bands were excised, and N-terminal protein microsequencing was performed by automated Edman degradation (Proseq Inc.).
Multidimensional High Pressure Liquid Chromatography Coupled with Electrospray Ionization Tandem Mass Spectrometry—The trichloroacetic acid-precipitated sample was reduced and alkylated using Tris 2-carboxyethyl phosphine and iodoacetamide according to the methods of McCormack et al. (14). The sample was then sequentially digested with endoproteinase Lys-C (Roche Diagnostics) and sequencing grade soluble trypsin (Promega) to produce a large collection of peptide fragments (14). The resulting peptide mixture was then separated and analyzed by multidimensional protein identification technology, which features a series of HPLC columns (reverse phase/strong cation exchanger/reverse phase) followed by electrospray ionization tandem mass spectrometry as described previously (15, 19), with modifications as described in McDonald et al. (17). Tandem mass spectra were searched against the rat, mouse, and human translated genomic databases using the SEQUEST software algorithm (20, 21). Search results were filtered and grouped using the DTASelect program, which generates a file in which peptides are matched to a genetic locus (22) and identifications confirmed through manual evaluation of spectra. Eluted proteins common to both the control (Kir2.23) and experimental (Kir2.2) columns were subtracted using the Contrast algorithm, which allows the comparison of two datasets from different mass spectrometry runs (22). Common contaminants such as the GST fusion protein, keratin, hemoglobin, HSP70, and tubulin were also subtracted as nonspecific binding proteins.
GST Fusion Protein Pull-down Assay—Pull-down assays with GST fusion proteins were performed as described (6) using 10 mg of heart extract (see above) and 10 µg of GST fusion protein per pull-down.
COS-1 Cell Protein Expression and Co-immunoprecipitation—COS-1 cells were co-transfected with 1-, 1-, or 2-syntrophin with Kir2.2 or Kir2.23 utilizing FuGENE 6 (Roche Applied Science) and used for co-immunoprecipitation as described (18).
Immunoaffinity Purification—Syntrophin and associated proteins were purified from rat hindlimb muscle using mouse monoclonal antisyntrophin 1351 immunoaffinity chromatography as previously described (23).
Antibodies and Antisera—Primary antibodies used include affinitypurified rabbit polyclonal anti-Kir2.2 (6, 24), mouse monoclonal anti-SAP97, anti-Mint1 antibody, and anti-dystrobrevin (BD Transduction Laboratories), mouse monoclonal anti-CASK antibody, anti-PSD-95 family antibody (pan-MAGUK; clone K28/86.2), and rabbit polyclonal anti-GluR1 (Upstate Biotechnology), rabbit polyclonal anti-SAP97 (PA-741, Affinity Bioreagents), rabbit polyclonal anti-Veli/MALS-1, -2, and -3 (Zymed Laboratories Inc.), and mouse monoclonal anti-syntrophin 1351 (25), mouse monoclonal anti-dystrophin (MANDRA1, Sigma), mouse monoclonal anti--dystroglycan (Novocastra), horseradish peroxidase-linked anti-GST antibody (Santa Cruz Biotechnology). The rabbit anti-abLIM antiserum generated against a GST-abLIM fusion protein (26) (generous gift from Dr. Tiansen Li) and rabbit anti-Pals2 antiserum generated against a GST-Pals2 fusion protein (27) (generous gift from Dr. Ben Margolis) were pre-absorbed against GST-Kir2.2 fusion protein linked to glutathione-agarose prior to use in immunoblotting. Anti-rabbit IgG and anti-mouse IgG secondary antibodies conjugated to horseradish peroxidase (Amersham Biosciences) were used where appropriate.
Immunofluorescence Microscopy—Mice were euthanized with CO2, and sternomastoid muscles were dissected and immediately flash frozen in isopentane. Muscle was mounted with Histoprep. 16-µm frozen sections were cut with a cryostat (Leica CM 1850) and transferred to charged microscope slides (Fisher Scientific). Sections were fixed with 100% methanol 10 min at -20 °C then incubated in blocking buffer (3% BSA, 1% donkey serum, 1% goat serum in 1x PBS) overnight at 4 °C. Sections then were incubated with primary antibodies diluted in blocking buffer overnight at 4 °C. Sections were washed 5 times 30 min with PBS at room temperature. Sections were incubated with secondary antibodies diluted in blocking buffer (1:50 donkey anti-rabbit fluorescein isothiocyanate (Jackson ImmunoResearch); 1:200 goat anti-IgG1 Alexa Fluor 647; 1:200 -bungarotoxin, Alexa Fluor 594 conjugate (Molecular Probes)) overnight at 4 °C, followed by washing 5 times 30 min with PBS at room temperature. Coverslips were then mounted on sections with Prolong (Molecular Probes). Images were captured using an Olympus Fluoview 500 laser scanning confocal microscope, and images were merged and displayed with Adobe Photoshop.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
3) (Fig. 1A). Comparing proteins isolated by these columns should allow one to identify proteins that are specifically associated with the Kir2.2 C-terminal PDZ binding motif (ESEI; Fig. 1A). We incubated these matrices with detergent-solubilized rat cerebellum extracts, washed the columns extensively, and eluted bound proteins by competition with an excess of the competitive peptide, which corresponds to final ten C-terminal amino acids of Kir2.2. Eluted Kir2.2-associated proteins were then analyzed and identified by a number of different methods: 1) subjecting the eluted protein mixture directly to multidimensional high pressure liquid chromatography coupled with electrospray ionization tandem mass spectrometry (HPLC-MS-MS) followed by protein data base searching with identified peptide masses, 2) SDS-PAGE followed by transfer to PVDF membrane and identification of candidate proteins by microsequencing using N-terminal Edman degradation, and 3) SDS-PAGE followed by transfer to nitrocellulose and probing the membrane with antibodies to candidate proteins (Fig. 1B).
|
3 and eluted with the competitive peptide. Eluted proteins were then separated by SDS-PAGE and silver-stained. Proteins that were specifically bound and eluted from the Kir2.2 column with the competitive peptide could easily be detected (Fig. 1C, silver stain, lane 4, arrowheads). These specific Kir2.2-binding proteins were not detected in eluates from the control GST or Kir2.23 columns (Fig. 1C, silver stain, lanes 2 and 3). As a separate control for the specificity of the eluting peptide, a scrambled peptide encoding the same amino acids as the competitive peptide, but in a random order, was used to elute protein isolated by the Kir2.2 column. None of the specific proteins eluted by the competitive peptide were eluted by the scrambled peptide (Fig. 1C, silver stain, lane 5). This shows that the competitive peptide specifically elutes proteins associated with the C terminus of Kir2.2.
To demonstrate that the Affi-Gel-Kir2.2 matrix could efficiently enrich known Kir2.2-interacting proteins (4, 6), eluted fractions from columns were subjected to Western blotting with anti-SAP97 antibodies. SAP97 was significantly and specifically enriched in competitive peptide eluates as was PSD-95, which is also recognized by the antibody (Fig. 1C, peptide elution, lane 4). No SAP97 or PSD-95 could be detected in peptide eluates from the GST or Kir2.23 columns (Fig. 1C, peptide elution, lanes 2 and 3). The scrambled peptide eluates from the Kir2.2 column contained barely detectable amounts of SAP97 (Fig. 1C, peptide elution, lane 5). Following peptide elution, the columns were then incubated with 1% SDS to elute the remaining bound proteins. No SAP97 or PSD-95 was detected in eluates from the control GST or Kir2.23 columns (Fig. 1C, SDS elution, lanes 2 and 3). Residual SAP97 bound to Kir2.2 not completely eluted by the competitive peptide was detectable in SDS eluates of this column (Fig. 1C, SDS elution, lane 4). SDS eluates from the Kir2.2 column eluted with scrambled peptide contained large amounts of SAP97 and PSD-95 (Fig. 1C, SDS elution, lane 5), indicating that the scrambled peptide could not elute Kir2.2-associated proteins. Analysis of proteins in the cerebellar extract not bound during incubation with column matrices indicated that the Kir2.2 matrix efficiently depleted the pool of SAP97 in these detergent-solubilized extracts (Fig. 1C, flowthrough fraction, lanes 4 and 5). This analysis of the Kir2.2 chromatography matrix coupled with specific elution by the competitive peptide demonstrates the ability of this method to be used for purification and identification of Kir2 channel-associated proteins.
Analysis of Kir2.2 Affinity-purified Brain Proteins by HPLC-MS-MS—Recent advances in HPLC-MS-MS and development of computer software for the analysis of large peptide fragment data sets, coupled with the huge increase in available protein sequence databases, has greatly improved our ability to analyze purified protein mixtures and complexes (see "Experimental Procedures") (14-16, 19). Affi-Gel-Kir2.2 and Kir2.23 affinity chromatography was performed with detergentsolubilized rat brain extracts followed by specific elution with the competitive peptide. Fig. 2A depicts the silver-stained SDS-PAGE of the proteins analyzed by mass spectrometry. Many bands are clearly visible, specifically eluting from the Kir2.2 column and not present in the Kir2.23 column (Fig. 2A, arrows). There are also a number of additional bands associated with both columns, most likely corresponding to proteins that bound non-specifically to the columns.
|
|
Both Veli-1 and Veli-3 were identified with 29.2% and 27.9% sequence coverage, respectively. The Veli proteins contain a single N-terminal L27 domain followed by a single class I PDZ domain. In addition to known interactions of CASK, Dlg2, Dlg3, and Pals2 with the Veli L27 domain, the C terminus of the NMDA receptor NR2B can interact with the Veli PDZ domain (38). Interestingly, we also identified the actin-binding LIM protein, which could provide a link of the channel to the actin cytoskeleton.
Kir2 Channel-associated Proteins Identified by N-terminal Edman Degradation—To complement the HPLC-MS-MS, we identified proteins isolated with the Kir2.2 columns by purifying-associated proteins from large amounts of rat brain extract (from 20 rat brains) and directly obtaining amino acid sequences. Purified proteins were separated by SDS-PAGE, transblotted to PVDF membrane, and stained with Coomassie Blue. Excised bands were N-terminally sequenced by modified automated Edman degradation. Microsequencing of a prominent band at 140 kDa revealed the N-terminal sequence: MPVRKQDTQXA. A search of the protein data base with this sequence matched it to the N-terminal sequence of the MAGUK protein SAP97, a Kir2-interacting protein we had previously identified (6). Sequencing of a band running at 60 kDa revealed the N-terminal sequence: MPVAA(T). A search of the protein data base with this sequence matched it to the N-terminal sequence of Dlg2. The microsequencing data confirm the presence of two Kir2-associated proteins identified by mass spectrometry.
Immunoblot Analysis of Kir2.2 Affinity-purified Brain Proteins—Another method to confirm that the Kir2-associated proteins that were identified by mass spectrometry are indeed positive identifications was Western blot analysis of peptide eluates from both the Kir2.2 and Kir2.23 columns with available antibodies (Fig. 2B). In all cases that we tested, the immunoblot results confirmed the protein identifications from mass spectrometry. The 140-kDa protein SAP97 was specifically eluted from the Kir2.2 columns in great abundance as expected. It is likely that more than one splice variant is represented (several bands are observed in the input), but these variants were not resolved in the Kir2.2 eluate because of the abundance of purified SAP97 protein and their similar molecular masses. To identify Chapsyn-110, SAP102, and PSD-95 (also identified in Fig. 1C), we used an antibody that recognizes all four PSD-95-related MAGUK proteins (pan-MAGUK). Four bands could be resolved on the Western blot, with SAP97 migrating as a broad band at 140 kDa, Chapsyn-110 migrating at 110 kDa, SAP102 migrating at 100 kDa, and PSD-95 migrating at 95 kDa. The lower band at 75 kDa is probably a PSD-95 breakdown product that is often observed with a PSD-95-specific antibody (data not shown). Two splice variants of CASK were distinguished with the monoclonal anti-CASK antibody (Fig. 2B). Veli-1 and Veli-3 were present in abundance, whereas Veli-2 was barely detectable (Fig. 2B). These data are consistent with the absence of Veli-2 identification via mass spectrometry. The presence of abLIM at 75 kDa (probably the m-abLIM isoform (26)) was detected specifically in the eluate from the Kir2.2 column. Using antiserum to Pals2, a band at 60 kDa corresponding to Pals2 was specifically eluted from the Kir2.2 column, supporting the mass spectrometry data. The upper bands on this blot in eluates from both the Kir2.2 and Kir2.23 columns are cross immunoreactivity with GST fusion protein dimers (GST-Pals2 was used to make antiserum (27)). The AMPA receptor, GluR1, which is abundant in brain and interacts with SAP97 (39), was not detected in these eluates by Western blot or mass spectrometry. This further supports the specificity of the affinity chromatography and the uniqueness of the complement of proteins isolated with the Kir2.2 column. Dlg2 and Dlg3 were not analyzed by Western blotting due to lack of available antibodies.
We made probable assignments of the Kir2.2-specific band identities observed on the silver-stained gels. Combining the predicted molecular mass data (Table I) with known mobilities of the proteins identified by Western blot (Fig. 2B), approximate locations of proteins were aligned with specifically eluted bands (Fig. 2A, arrows). These protein band assignments are complicated by posttranslational modifications, unique protein folding characteristics, and alternative splicing forms, making some of the bands difficult to resolve on an SDS-PAGE. Fig. 2A shows that most of the proteins identified by mass spectrometry and immunoblotting can be correlated with silver-stained bands in the correct vicinity of their molecular masses.
Analysis of Kir2.2 Affinity-purified Heart Proteins by HPLC-MS-MS—Non-neuronal tissues contain unique types of channel-associated proteins that are not present in brain. Kir2 channels are particularly abundant in muscle, including cardiac muscle, and are likely to associate with proteins that are distinct from those in brain. To compare Kir2 channel-associated proteins identified in brain with proteins associated with the channels in cardiac tissue, we elected to perform Kir2.2 affinity chromatography using detergent-solubilized heart samples. Isolated heart proteins were then subjected to mass spectrometry/multidimensional protein identification technology analysis as described above for purified brain proteins.
The purified heart proteins from Kir2.2 or Kir2.23 columns were eluted with competitive peptide followed by a second elution with SDS. A fraction of the eluates were resolved by SDS-PAGE and silver-stained. Proteins were detected that specifically were eluted from the Kir2.2 column and were not present in eluates from the control Kir2.23 column (Fig. 3A). Unlike the Kir2.2-interacting proteins purified from brain, specific cardiac Kir2.2-binding proteins were eluted both by the competitive peptide (Fig. 3A) and by the subsequent elution with SDS (Fig. 3B). The proteins eluted by the peptide competition and SDS were then analyzed and identified by mass spectrometry as described above. Again, nonspecific binding proteins were subtracted in the analysis, which defines specific proteins as those eluted from the Kir2.2 column, but not the Kir2.23 column.
|
|
|
1, 1, and 2 isoforms of syntrophin, dystrophin, and -dystrobrevin (Tables II and III). The DAPC forms a structural link between the actin cytoskeleton and extracellular matrix and is particularly important in muscle cells where alterations in dystrophin can lead to muscular dystrophy (44). Syntrophins are PDZ and pleckstrin homology domain-containing proteins that have been demonstrated as part of the DAPC. Two syntrophin-interacting proteins in the complex include dystrophin and dystrobrevin, both identified here as Kir2 channel-associated proteins.
Immunoblot Analysis of Kir2.2 Affinity-purified Heart Proteins—To confirm that the Kir2-associated heart proteins that were identified by mass spectrometry are indeed positive identifications, we performed Western blots on peptide as well as SDS eluates from both the Kir2.2 and Kir2.23 columns with antibodies to the proteins (Fig. 3C). By immunoblot we confirmed the identifications of SAP97, CASK, Veli-3, Mint1, three isoforms of syntrophin (monoclonal antibody 1351 recognizes 1-, 1-, and 2-syntrophin), -dystrobrevin-1, -dystrobrevin-2, dystrophin, and the dystrophin gene product Dp71. We detected the presence of Mint1 protein in peptide eluates, a new finding, because it was previously reported only to be neuronal (42, 43). We detected Veli-3, but not Veli-1, in Western blots using isoform-specific anti-Veli antibodies, although both were identified by mass spectrometry (Fig. 3C and data not shown). These data agree with previous studies where the lower molecular mass Veli-3 protein is the predominant Veli expressed in heart (38, 40). None of these proteins eluted from the Kir2.23 columns, which indicates that the association with all of these proteins is mediated by the Kir2.2 PDZ binding motif.
To test for the presence of Kir2-associated proteins in brain that may have been missed by mass spectrometry, affinity-purified brain proteins were analyzed with antibodies to the DAPC and Mint1. When we probed these eluates with antibodies directed against protein components of the DAPC, we detected two isoforms of syntrophin, two isoforms of -dystrobrevin, and two Dp71 isoforms of dystrophin (Fig. 4). Upon immunoblot analysis, Mint1 protein was readily detected in Kir2.2 column eluates from brain (Fig. 4). Mint1 is known to form a complex with CASK and Veli in brain (40, 41), and our recent studies demonstrate that this complex is associated with Kir2 channels in brain and heart (18). No DAPC proteins or Mint1 were detected in the control Kir2.23 brain eluates, demonstrating that the intact PDZ binding motif is essential for their association with Kir2.2 (Fig. 4). Therefore, Kir2 channels also associate with the DAPC and Mint1 in brain.
|
50 amino acid tails of Kir2.1 (PDZ binding motif: SEI), Kir2.2 (PDZ binding motif: SEI), Kir2.3 (PDZ binding motif: SAI), and Kir4.1 (PDZ binding motif: SNV). These GST fusion proteins were used in affinity pull-down assays with Triton X-100-solubilized heart and skeletal muscle extracts. In heart extracts, we readily detected syntrophin, -dystrobrevin-2, and the Dp71 isoform of dystrophin in association with Kir2.2, Kir2.3, and Kir4.1 fusion proteins (Fig. 5, lanes 4, 6, and 7), but not the controls GST and Kir2.23 (Fig. 5, lanes 3, 1, and 5). Association of syntrophin, -dystrobrevin-2, and Dp71 with the Kir2.1 construct was detected with longer exposure times (data not shown). Additionally, the Kir4.1 construct pulled down the full-length isoform of dystrophin (Fig. 5, lane 7). To compare Kir channel association with other interacting proteins, we probed the same immunoblots for SAP97, CASK, and Veli-3. Kir2.1, Kir2.2, and Kir2.3 fusion proteins all associated with these three proteins (Fig. 5, lanes 2, 3, and 5); however, the Kir4.1 construct did not pull down detectable SAP97, CASK, or Veli-3 (Fig. 5, lane 7). We obtained similar results for affinity interaction assays using skeletal muscle extracts (data not shown).
|
1-, 1-, and 2-Syntrophin Associate with Kir2.2 When Coexpressed in a Cellular Environment—Because syntrophins contain class I PDZ domains capable of interacting with C-terminal ligands on sodium and water channels (45, 46), we predicted that syntrophins and Kir2 channels should interact when expressed together in vivo. To determine whether Kir2.2 interacts with 1-, 1-, and 2-syntrophin in a cellular environment, we expressed both proteins alone and in combination in COS-1 cells. Using an antibody against the N terminus of Kir2.1/2.2, we immunoprecipitated Kir2.2 from COS-1 cell extracts. All three syntrophin isoforms co-immunoprecipitated with Kir2.2 (Fig. 6A, lanes 2, 5, and 8). When 1-, 1-, or 2-syntrophin was co-expressed with a mutant Kir2.2 construct with the last three C-terminal amino acids deleted (Kir2.23), no syntrophin co-immunoprecipitation could be detected (Fig. 6A, lanes 3, 6, and 9). Additionally, 1-, 1- or 2-syntrophin did not co-immunoprecipitate if Kir2.2 was not transfected (Fig. 6A, lanes 1, 4, and 7). Expression levels of Kir2.2, Kir2.23, 1-, 1-, and 2-syntrophin were comparable between transfections as indicated by immunoblotting a fraction of the immunoprecipitation inputs (Fig. 6A, lower panels). This demonstrates that full-length Kir2.2 can associate with 1-, 1-, or 2-syntrophin in cells and the C terminus of Kir2.2 encoding the PDZ binding motif is required for association.
|
-dystroglycan, co-purified with syntrophin, as shown previously (Fig. 6B, lower panels and Ref. 23). This demonstrates that Kir2 channels associate with syntrophin and the DAPC in skeletal muscle.
Kir2.2 Co-localizes with Syntrophin, Dystrophin, and Dystrobrevin at Neuromuscular Junctions—The DAPC is well known for its role in skeletal muscle where the complex provides mechanical stabilization of the plasma membrane during contraction and is involved in the formation and maintenance of the neuromuscular junction. In the absence of the normal complement of dystrophin, the sarcolemma may become damaged during contraction, leading to muscular dystrophy (44). Because the C terminus of Kir2.2 could readily associate with proteins of the DAPC, localization of Kir2.2 and the DAPC were examined in skeletal muscle where syntrophin, dystrophin, and dystrobrevin are known to be highly concentrated at neuromuscular junctions (NMJs). Immunofluorescence microscopy demonstrated that Kir2.2 was localized at NMJs in mouse sternomastoid muscle (Fig. 7, left panels). Neuromuscular junctions were identified by labeling with -bungarotoxin, a nicotinic acetylcholine receptor antagonist that labels the receptorrich crests of the postsynaptic folds (Fig. 7, right panels). Kir2.2 distribution partially co-localized with antibody labeling for syntrophin, dystrophin, and dystrobrevin (Fig. 7, middle panels), all of which were abundant at the NMJ (Fig. 7). Colocalization with the DAPC proteins syntrophin, dystrophin, and dystrobrevin suggests that Kir2.2 interaction with the DAPC may be important to target, stabilize, or anchor Kir2.2 at the NMJ.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONEXPERIMENTAL PROCEDURESRESULTSDISCUSSIONREFERENCES |
|---|
1, 1, and 2 syntrophin, dystrophin, Dp71, -dystrobrevin-1, and -dystrobrevin-2) associate with Kir2 channels proteins in heart, skeletal muscle, and brain. Because most of these proteins are composed of modular protein-protein interaction domains (Fig. 8), we conclude that Kir2 channels are associated with large complexes of proteins that may perform a variety of unique functions in channel trafficking and localization.
|
3 control column, which was identical to the Kir2.2 column, but lacked the three C-terminal amino acids. Further specificity was obtained by eluting the columns with a competitive peptide encoding the final ten amino acids of Kir2.2. There are many other proteins expressed in brain and heart that contain class I PDZ domains that might have interacted with the Kir2 channel C termini (48), but we did not detect specific binding by this method. For example, only the Veli-1 and Veli-3 isoforms were purified from brain, even though Veli-2 is also expressed in this tissue (18, 38, 49). In early studies, we predicted that Kir2 channels might be linked to GluR1 AMPA receptors, because both proteins interact with SAP97 (39). However, under these purification conditions we were not able to detect any GluR1 associated with Kir2.2. Taken together, these data indicate that the proteins purified on the Kir2.2 columns are specific binding partners for Kir2.2. A variety of experiments shows that the proteins are functional interacting partners of the Kir2 channels in native tissues. In this study, and the parallel study (18) we describe the functional interactions of many of these proteins. We find that these proteins are associated with the channels when expressed in COS cells (SAP97, CASK, Mint1, Veli, PSD-95, and syntrophin), that they co-immunoprecipitate with the channels in native tissues (SAP97, CASK, and syntrophins), are colocalized with the channels in cultured cells and native tissues (SAP97, PSD-95, Veli, Mint1, syntrophin, dystrophin, and dystrobrevin), and functionally regulate channel trafficking or membrane clustering (CASK and PSD-95) (5, 18).2 These associations will be described in more detail below with the individual proteins. SAP97, CASK, Mint1, and Veli are investigated in depth in a parallel study (18).
How are the proteins associated with the channel? Of these Kir2-associated proteins, nine proteins contain class I PDZ domains (PSD-95, SAP97, SAP102, Chapsyn-110, Veli-1, Veli-3, 1-syntrophin, 1-syntrophin, and 2-syntrophin), and we predict these interact directly with Kir2 C termini. Indeed our parallel study demonstrates the ability of SAP97, Veli-1, and Veli-3 to directly interact with the Kir2.2 C terminus (18). These proteins with a class I PDZ domain likely act as the bridge to link the channel with other identified channel-interacting proteins that lack a class I PDZ domain.
PSD-95 is a well-studied membrane-associated MAGUK protein with a class I PDZ domain. PSD-95 is known for its roles in clustering ion channels and receptors (5, 32, 34), post-synaptic targeting, and localization due to dynamic palmitate modification (50-53), involvement in glutamatergic synaptic development (54), and learning and memory formation (55). PSD-95 and its homologous family members SAP97, SAP102, and Chapsyn-110 all contain three PDZ domains, followed by an SH3 domain and GK domain (Fig. 8), with the second PDZ domain essential for interaction with the C-terminal PDZ binding motifs of ion channels and receptors. We find that coexpression of Kir2.2 channels with PSD-95 causes co-clustering of the channels and PSD-95 in the plasma membrane; this effect is not observed with control Kir2.23 channels or in the absence of PSD-95.2 Recent evidence indicates that Kir2.3, like PSD-95, is localized at synapses in the olfactory bulb (8). The ability of PSD-95 to cluster Kir2 channels, and the similar distribution of Kir2.3 and PSD-95, suggest that Kir2 channels expressed in neurons may be clustered or stabilized at postsynaptic sites by PSD-95. Both SAP102 and Chapsyn-110 are expressed in cell-type specific manners (56-58), have been demonstrated to interact with PSD-95, and may also be involved in neuronal channel targeting and clustering (59-61).
SAP97, a MAGUK protein closely related to PSD-95, seems to be involved in very different functions. Unlike PSD-95, SAP97 is ubiquitously expressed, is not tightly membrane associated (62), and when expressed in heterologous cells does not cluster NMDA receptors (33), Kv1 channels (35), or Kir2 channels2 at the plasma membrane. Instead, SAP97 seems to interact with receptors and channels early during their biosynthesis and may be involved in targeting and trafficking proteins, as well as assembly of signaling complexes. Interaction of SAP97 with the GluR1 AMPA receptor (39) was demonstrated to participate in the recruitment of a protein kinase-signaling complex, which may be involved in synaptic plasticity (11). SAP97 also was shown to interact with GluR1 early in their biosynthetic pathway while the receptors are in intracellular compartments (63), implicating SAP97 in receptor targeting. In epithelial cells, the N-terminal domain of SAP97 is involved in targeting SAP97 to basolateral membranes (64, 65), and its localization is influenced by interactions with CASK (37).
CASK also is classified as a MAGUK protein, but one that contains a single class II PDZ domain followed by an SH3 domain and GK domain (Fig. 8). Because CASK contains a class II PDZ domain, it is unlikely to interact directly with the Kir C terminus, and indeed we have shown that it can be recruited to the channel C terminus by SAP97 or Veli (18). Both SAP97 and CASK were demonstrated to interact in epithelial cells (36, 37), and we have recently demonstrated that they form a stable complex with Kir2 channels in brain and heart (6, 18). Indeed, SAP97 and CASK co-localize with Kir2.2 and Kir2.3 channels (7, 18), and a dominant-negative CASK construct results in Kir2.2 channel mislocalization in epithelial cells suggesting that a CASK-containing complex is involved in targeting Kir2.2 channels (18). Because SAP97 and CASK are ubiquitously expressed, perhaps these proteins act together to influence Kir2 channel function in many different tissues. Indeed, targeted gene disruptions of SAP97 or CASK in mice result in the similar phenotype of cranial dysmorphogenesis and cleft palate (66, 67) providing genetic evidence for their functional association. CASK is known to form a complex with the proteins Veli and Mint1, and we found that all of these proteins were purified by Kir2.2 affinity chromatography (Tables I and II and Figs. 2, 3, 4). The evolutionarily conserved tripartite complex composed of CASK, Veli, and Mint1 is implicated in ion channel/receptor targeting and trafficking in both Caenorhabditis elegans epithelial cells (47) and mammalian neurons (68). Because the CASK, Mint1, and Veli complex associates with a variety of channels and receptors (12, 47, 68), the complex may play a general role in targeting and trafficking receptors and channels containing PDZ binding motifs.
The Veli proteins have a relatively simple domain structure with an N-terminal L27 domain fused to a Class I PDZ domain (Fig. 8). The L27 domain has been shown to interact with N-terminal sequences of CASK-like MAGUK proteins and the PDZ domain interacts with NR2B NMDA receptors (38) and Kir2 channels (7, 18). As well as forming a complex with CASK, Mint1, and SAP97 (18), Veli proteins are involved in neuronal NMDA receptor trafficking and are implicated in the retention of channels and transporters in the plasma membrane, including the BGT-1 GABA transporter in epithelial cells and Kir2.3 channels expressed in Xenopus oocytes (7, 69). Thus, it appears that Veli proteins may have different functional roles depending on the cellular context.
Mint1 is a protein that contains two PDZ domains near its C terminus as well as other functional domains (Fig. 8). Mint1 has been shown to interact with a variety of neuronal proteins, including the motor protein KIF17 (68), suggesting an interesting function to link cargo proteins to motor proteins for transport. Our finding that Mint1 associates with the Kir2.2 column in heart was unexpected. Previous reports of mRNA distribution characterized Mint1 as a protein expressed exclusively in neurons (42, 43). However, mass spectrometry identified two proteolytic peptides from Mint1 (Table II), and the identification was confirmed by immunoblot (Fig. 4). This discovery implicates functional roles in cargo transport for Mint1 in non-neuronal tissues.
Three of the Kir2-interacting proteins from brain, Dlg2, Dlg3, and Pals2, are members of a CASK-like MAGUK protein family. Little is known about the function of these MAGUKs. All three proteins have the same general domain organization as CASK with two L27 (L27N and L27C) domains at the N terminus followed by a class II PDZ domain, an SH3, and GK domains (Fig. 8). All three proteins can interact with Veli proteins via their L27C domain (27, 37, 40). The homology that these proteins display with CASK indicates that CASK, Dlg2, Dlg3, and Pals2 may associate with Kir2 channels in a similar manner. They probably do not interact directly with the channels, because the PDZ domains of these proteins will not recognize class I PDZ binding motifs, but like CASK, may be recruited by SAP97 or Veli (18). Dlg2 and Dlg3 recently were shown to bind to SAP97 (70), and these proteins contain L27 domains, which may interact with Veli proteins (27, 37). No direct binding partners for the PDZ domains of Dlg2, Dlg3, and Pals2 have been described to date. It is possible that these proteins may be able to substitute for CASK, but target channel complexes differently. Additionally, the N-terminal domains of Dlg2, Dlg3, or Pals2 are unique, which increases the possible diversity of protein complexes with the channels. It is likely that these MAGUKs may assemble other, as yet unidentified proteins into a Kir2-associated complex.
An interesting Kir2-associated protein in brain is the actinbinding LIM protein, abLIM, which exists in three different isoforms, short, medium, and long. The protein has a multidomain structure composed of N-terminal LIM domains, a dematin-like domain, fused to a domain homologous to villin (26). LIM domains are cysteine-rich zinc finger motifs first identified in homeodomain proteins involved in development and cellular differentiation, whereas other LIM-containing proteins have been found associated with the cytoskeleton (71). The dematin-like domain/villin domain was demonstrated to facilitate binding of abLIM to F-actin (26). The association of this protein with Kir2 channels may link the channels to the actin cytoskeleton. It is unclear at this point how abLIM associates with the channels, but it is probably through another protein, because abLIM does not contain a PDZ domain.
An intriguing group of novel Kir2-interacting proteins are the members of the dystrophin-associated complex (1-, 1-, and 2-syntrophin, dystrophin, and dystrobrevin). This is the first identification of members of the dystrophin complex associated with Kir2 channels. These proteins were identified by HPLC-MS-MS in heart extracts and confirmed in heart, skeletal muscle, and brain by immunoblotting. The 1, 1, and 2 syntrophins are part of a family of proteins containing five isoforms that are known for their interactions with the dystrophin-associated protein complex (dystrophin, utrophin, and dystrobrevin). The syntrophins are characterized by their domain structure, which can contain up to two pleckstrin homology domains, a class I PDZ domain, and a syntrophin-unique domain (Fig. 8) (44). The syntrophin-unique domain mediates binding of syntrophin to the dystrophin complex. In addition to its interaction with Kir2 channels, the PDZ d
