Syntrophin γ2 Regulates SCN5A Gating by a PDZ Domain-mediated Interaction*

SCN5A encodes the α subunit of the cardiac muscle and intestinal smooth muscle mechanosensitive Na+ channel. Mechanosensitivity in the intestine requires an intact cytoskeleton. We report, using laser capture microdissection, single cell PCR, and immunohistochemistry, that syntrophins, scaffolding proteins, were expressed in human intestinal smooth muscle cells. The distribution of syntrophin γ2 was similar to that of SCN5A. Yeast two-hybrid and glutathione S-transferase pull-down experiments show that SCN5A and syntrophin γ2 co-express and that the PDZ domain of syntrophin γ2 directly interacts with the C terminus of SCN5A. In native cells, disruption of the C terminus-syntrophin γ2 PDZ domain interaction using peptides directed against either region result in loss of mechanosensitivity. Co-transfection of syntrophin γ2 with SCN5A in HEK293 cells markedly shifts the activation kinetics of SCN5A and reduces the availability of Na+ current. We propose that syntrophin γ2 is an essential Na+ channel-interacting protein required for the full expression of the Na+ current and that the SCN5A-syntrophin γ2 interaction determines mechanosensitivity and current availability.

Mechanosensitivity is a requirement for the survival of an organism (1,2). At a cellular level, ion channels often serve as the unitary element that underlies mechanosensitivity (3,4). Mechanosensitive ion channels are of particular relevance in organs constantly submitted to movement, such as the gastrointestinal tract and the heart (5,6). Contractility in both cardiac muscle and gastrointestinal smooth muscle is initiated by membrane electrical events as a result of changes in ionic conductances (7,8). In the heart, the upstroke of the action potential is mediated via opening of a tetrodotoxin-insensitive Na ϩ channel, the ␣ subunit of which is encoded by SCN5A (9). Mutations in SCN5A can result in clinically significant cardiac arrhythmias (8,10). SCN5A is also expressed in human intestinal circular smooth muscle and the native Na ϩ current is mechanosensitive (11)(12)(13). Mechanosensitivity appears to be dependent on the actin cytoskeleton, since disruption of the actin cytoskeleton by cytochalasin D or gelsolin abolishes mechanosensitivity (13). This suggests that the actin cytoskeleton is required to transmit force to the ion channel. Syntrophins are suggested to be a link between the actin cytoskeleton and membrane-associated proteins including ion channels, enzymes, and receptors, since actin filaments are not known to directly interact with these proteins (14 -17). Syntrophins are a multigene family of homologous proteins (18 -22). Five syntrophins, ␣, ␤1, ␤2, ␥1, and ␥2, have been characterized (22,23). Each syntrophin is encoded by a separate gene but shares a common domain organization. Each syntrophin contains two tandem pleckstrin homology domains at the N terminus, a single PDZ domain, and a highly conserved C terminus syntrophin-unique region (22,23). The PDZ domains of syntrophins ␣, ␤1, and ␤2 but not ␥1 are known to interact with SCN5A and with SCN4A, a skeletal muscle Na ϩ channel, via the C terminus sequence motif (E(S/T)XV) (14,15,24). The objective of this study was to investigate the interaction between SCN5A and syntrophins in intestinal smooth muscle and the functional consequences of such an interaction on mechanosensitivity. Our hypothesis was that syntrophins couple SCN5A with the actin cytoskeleton, providing a mechanism for mechanical regulation of voltage-dependent ion channel gating.

Preparation of Single Human Jejunal Circular Smooth Muscle
Cells-The Mayo Foundation Institutional Review Board approved the use of human tissue obtained as surgical waste tissue during gastric bypass operations performed for morbid obesity. The method for dissociation of smooth muscle cells from human jejunal circular smooth muscle strips was as previously described (11,12). Briefly, the mucosa, submucosa, and longitudinal muscle of the jejunum were removed from the specimen by sharp dissection. The circular smooth muscle layer was then cut into small pieces and incubated with enzyme to release single smooth muscle cells used for patch clamp studies and single cell reverse transcription (RT) 1

-PCR.
Poly(A) RNA Isolation and c-DNA Library Preparation-Procedures for RNA isolation and the procedure for preparation of c-DNA libraries were as described previously (11,12).
PCR and Single Cell RT-PCR-All PCR amplifications were performed using GeneAmp 2400 PCR systems (PerkinElmer Life Sciences) using standard procedures (12). The protocol for single cell RT-PCR was as previously outlined (12). Spindle-shaped single smooth muscle cells were collected directly into PCR tubes containing tRNA and proteinase K. RT was performed using a mixture of random hexamer and oligo(dT) primers following the instructions of the manufacturer (PerkinElmer Life Sciences). The product of the RT reaction was then amplified for syntrophins using gene-specific primers that were specifically designed to flank a region that contained introns. All PCR products were purified and sent to the Mayo Molecular Core Facility for automated DNA sequencing.
Laser Capture Microdissection-Surgical waste tissue was fixed in ice-cold acetone according to the protocol described previously (12). The same number of spots of tissue containing about 1500 smooth muscle cells from the circular muscle layer or the longitudinal muscle layer were collected using the PIX II cell laser capture microdissection system (Arcturus Engineering Inc., Santa Clara, CA) with the 7.5-m spot size. The caps with collected cells were then immediately placed into sterile 0.5-ml microcentrifuge tubes containing 300 l of RNA STAT-60 reagent (Tel-TEST Inc., Friendswood, TX) for isolation of total RNA. After washing with 75% ethanol, the RNA pellet was resuspended in nuclease-free water (Ambion Inc., Austin, TX) and used for the RT-PCR.
Immunohistochemistry-Pieces of human jejunum (approximately 1 ϫ 1 cm) were prepared for immunohistochemistry as previously described (12). Briefly, cryostat sections (12 m thick) were mounted onto glass slides, air-dried, fixed for 10 min in either cold acetone or 4% paraformaldehyde, and rinsed in phosphate-buffered saline (PBS). Sections were incubated with 10% normal donkey serum and 0.3% Triton X-100 for 1 h to block nonspecific absorption sites and then incubated overnight at 4°C with anti-syntrophin ␥2 rabbit polyclonal antibody (diluted 1:200 in 5% normal donkey serum; a kind gift from Dr. Vincenzo Nigro). The specificity of this antibody for syntrophin ␥2 has been previously shown by Piluso et al. (22). After several rinses in PBS, the sections were incubated for 1 h with donkey, anti-rabbit IgG conjugated to CY3 (1:100 dilution in 2.5% normal donkey serum), rinsed in PBS, and coverslipped in glycerol-PBS containing an anti-fade reagent.
Yeast Two-hybrid Assay-Yeast two-hybrid assays were performed using HybriZAP-2.1 two-hybrid system (Stratagene, La Jolla, CA). The cDNA fragments encoding the C terminus of SCN5A (amino acids 1915-2015, CTSCN5A) and the C terminus with last 10 amino acids truncated (amino acids 1915-2005, CTSCN5A-10) were amplified by PCR and cloned into EcoRI and SalI restriction sites of the pBD-GAL4 vector to serve as baits in the yeast two-hybrid experiments. Two full-length splice variants of syntrophin ␥2, one with an intact PDZ domain and the other lacking the PDZ domain, were inserted into pAD-GAL4 vector (pADϩ␥2 and pADϩ␥2-PDZ, respectively). The nucleotide sequences of the DNA inserts were confirmed by sequence analysis to verify that inserts did not contain mutations. The inserts were expressed as fusion proteins with the DNA binding domain and DNA activating domain of GAL4. Several small scale yeast transformations were performed using the lithium acetate method with 40% polyethylene glycol. The plasmids CTSCN5A and CTSCN5A-10 were cotransformed with the plasmids pADϩ␥2 or pADϩ␥2-PDZ into the YRG-2 yeast strain containing HIS3 and lacZ double reporter genes. After transformation, the yeast was plated on selective plates lacking tryptophan, leucine, and histidine to show activation of the GAL4inducible reporter gene HIS3 through protein-protein interaction. The colonies that grew on the selective plates were either due to the leaky expression of HIS3 or to the specific interaction between proteins resulting in expression of the HIS3 gene. To distinguish between leaky expression and specific protein interactions, we used filter lift assays to detect the expression of a second reporter gene, lacZ. The colonies on the selective plate were lifted onto a filter and assayed for ␤-galactosidase activity using 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside (X-gal) as a substrate.
GST Pull Downs-The bacterial expression vector pGEX-5X-1 (Amersham Biosciences) was used to produce a GST fusion protein in Escherichia coli. The C-terminal fragments of SCN5A with and without the truncation were fused in frame into pGEX-5X-1 to produce plasmids pGEXϩCT5A and pGEXϩCT5A-10. To express GST fusion proteins, pGEXϩCT5A and pGEXϩCT5A-10 were transformed into BL21 (DE3) (Stratagene, La Jolla, CA). Each bacterial culture was induced by isopropyl-1-thio-␤-D-galactopyranoside at 0.1 mM and allowed to express protein at 28°C for 4 h. Then the cells were pelleted and resuspended in PBS and 10 mg/ml lysozyme. After incubation for 30 min on ice, protease inhibitor mixture (Calbiochem) was added, and the cells were rocked at 4°C for 10 min. To destroy the DNA and RNA, the samples were treated with DNase and RNase. The protein extract was then extracted by centrifugation at 3000 ϫ g for 30 min and incubated with glutathione beads (Amersham Biosciences).
The complete coded sequences of the splice variants of syntrophin ␥2 with and without the PDZ domain were linked into the expression vector pCMV-FLAG, which tags the FLAG epitope at the N terminus, to make plasmids pCMV-␥2 and pCMV-␥2-PDZ. After the constructs were verified by sequencing their inserts, the pCMV-␥2 and pCMV-␥2-PDZ constructs were transfected into HEK 293 cells to express the FLAGtagged syntrophin ␥2 proteins. The cells were lysed 24 h later in lysis buffer, and the supernatants were used for GST pull-down experiments.
100 l (ϳ200 g) of HEK 293 cell extracts containing the FLAGtagged syntrophin ␥2 proteins were incubated with 20 -25 l (ϳ10 g) of GST ϩ CT5A/GST ϩ CT5A-10 bead-bound fusion proteins and, as a control, the GST-alone bead-bound fusion protein. After extensive washing with lysis buffer, the samples were subjected to SDS-PAGE and transferred to polyvinylidene difluoride membrane. The membrane was then incubated with anti-FLAG M2 monoclonal antibody (Sigma). After washing, horseradish peroxidase-conjugated anti-mouse IgG was used. The immunoreactive bands were visualized by ECL according to the manufacturer's instructions (Amersham Biosciences).
Expression and Purification of Fusion Protein-The pGEX-5X-1 plasmid constructs for the PDZ domain of syntrophin ␥2 were generated by using the same method described above for the C terminus of SCN5A. The constructs were verified by sequencing, and the plasmids encoding GST alone and GST plus PDZ were then introduced into BL21 cells for expression. After induction by isopropyl-1-thio-␤-D-galactopyranoside, the cell extracts were incubated with glutathione-agarose beads for affinity purification. Following washing, GST and GST plus PDZ were eluted with 10 mM reduced glutathione in 50 mM Tris-HCl (pH 8.0). For patch clamp analysis, the elution was dialyzed in 10 mM HEPES with 50 mM CsCl. The purity of the proteins was determined by SDS-12% polyacrylamide gel electrophoresis. The bands of the fusion proteins were of the expected size with high purity. Protein concentrations were estimated using the Bio-Rad protein assay kit.
Peptide Synthesis-A peptide corresponding to the last 10 amino acids (SPDRDRESIV) of the C terminus of the human SCN5A sequence and a control peptide containing the same amino acids but in random sequence (PRRSVSDDEI) were synthesized by the Mayo Peptide Synthesis Facility of the Mayo Proteomics Research Center. The peptides were purified by reverse phase high performance liquid chromatography using a Vidak C-18 column. Purity was Ͼ95% as assessed by amino acid analysis and analytical high performance liquid chromatography. Mass weight of the peptide was verified by electrospray ionization mass spectrometry on a Sciex 165B (Concord, Canada).
Plasmid Constructs and Mammalian Cell Transfection-The pcDNA3 expression vector (Invitrogen) with human SCN5A (hH1c) was used in the Na ϩ channel expression experiments. The truncation vector containing the full-length SCN5A minus the last 10 amino acids was made by PCR. The cDNAs for syntrophin ␥2 were produced by PCR and inserted into the pcDNA3 vector using EcoRI and SalI. All constructs were verified by sequencing. LipofectAMINE™ 2000 Reagent (Invitrogen) was used to transfect green fluorescent protein, pEGFP-C1, (Clontech), sodium channel SCN5A, and syntrophin ␥2 HEK293 cells (ATCC, Manassas, VA). Transfected cells were identified by fluorescence microscopy and patch-clamped.
Electrophysiological Recordings-Whole cell patch clamp recordings were made using standard patch clamp techniques. Whole cell recordings were obtained using Kimble KG-12 glass pulled on a P-97 puller (Sutter Instruments, Novato, CA). Electrodes were coated with R6101 (Dow Corning, Midland, MI) and fire-polished to a final resistance of 3-5 megaohms. Currents were amplified, digitized, and processed using a CyberAmp 320 amplifier, a Digidata 1200, and pCLAMP 8 software (Axon Instruments, Foster City, CA). Whole cell records were sampled at 5 kHz and filtered at 2 kHz with an eight-pole Bessel filter. 70 -75% series resistance compensation (lag of 60 s) was applied during each recording. The cell capacitance (C m ) was 40 -100 picofarads in human jejunal circular smooth muscle cells and 5-30 picofarads in HEK 293 cells. The access resistance (R a ) was 5-10 megaohms. All records were obtained at room temperature (21°C).
For human jejunal circular smooth muscle cell recordings, cells were held at Ϫ100 mV and stepped to Ϫ80 through ϩ35 mV at 5-mV intervals for 50 ms. The interval from the start of one depolarization to the next was 1 s. SCN5A-overexpressed Na ϩ currents at a Ϫ100-mV holding potential were much larger in HEK 293 than in native cells. Therefore, for HEK-293 cell records, cells were held at Ϫ80 mV to reduce maximal peak inward current. Transfected cells were then pulsed to the same voltages as described above. Steady state inactivation in transfected cells was determined using a pulse protocol where cells held at Ϫ80 mV, stepped to Ϫ110 through Ϫ60 mV in 5-mV intervals for 3 s to reach a steady state of inactivation, briefly stepped to Ϫ110 mV for 10 ms (to standardize transients), and finally stepped to Ϫ40 mV. Current was measured at Ϫ40 mV. The interval from the start of one depolarization to the next was 4 s.
The pipette solution contained 145 mM Cs ϩ , 20 mM Cl Ϫ , 2 mM EGTA, 5 mM HEPES, and 125 mM methane sulfonate for most whole cell recordings. In peptide experiments, the terminal 10-amino acid SCN5A peptide or the control jumbled sequence peptide was added to this intracellular solution to a concentration of 1 mM. The bath solution contained 149.2 mM Na ϩ , 4.7 mM K ϩ , 159 mM Cl Ϫ , 2.5 mM Ca 2ϩ , and 5 mM HEPES (normal Ringers solution) with an osmolarity of 290 -300 mosM. All chemicals other than peptides were obtained from Sigma.
Bath perfusion was used to assess mechanosensitivity as previously described (25,26). The bath was perfused at 10 ml/min for 30 s to create shear stress and activate the mechanosensitive Na ϩ channel according to a previously established protocol (13).
Data Analysis-Electrophysiological data were analyzed using PCLAMP 8 software, custom macros in Excel (Microsoft, Redmond, WA), or SigmaPlot 2001 for Windows (SPSS Science Marketing, Chicago, IL). Voltages were adjusted for junction potentials using JP-CalcW. Statistical comparisons were performed by a two-tailed paired Student's t test, and p Ͻ 0.05 was used for statistical significance. Tau of inactivation values were determined by fitting a standard two-exponential decay curve between the points at the peak and at 50 ms. Time-to-peak was measured as the difference between the time when the pulse started and the time at maximal peak inward current. Steady state activation and inactivation curves were fitted with a three-parameter sigmoid (Boltzmann) function.

Identification of Syntrophins in Human Jejunal Circular
Smooth Muscle Cells-To determine which syntrophins are expressed in intestinal smooth muscle, we used gene-specific primers designed against the five known syntrophins (␣, ␤1, ␤2, ␥1, and ␥2) to amplify cDNAs from dissociated human jejunal circular smooth muscle cell libraries. Products of the expected size were detected for syntrophin ␣, ␤1, and ␤2 using the appropriate oligonucleotide primers (Fig. 1A). PCR amplification for syntrophin ␥2 produced three bands. Sequence analysis showed that all three cDNA fragments were different splice variants of syntrophin ␥2. Syntrophin ␥1 was not present in human jejunal circular smooth muscle (Fig. 1A), consistent with its limited expression to the brain (22). RT-PCR was also carried out on aliquots of freshly dissociated smooth muscle cells and cDNA bands of the expected size for syntrophin ␣, ␤1, ␤2, and ␥2 obtained (data not shown), again suggesting that these syntrophins were expressed in intestinal smooth muscle together with the known expression of syntrophins in cardiac and skeletal muscle (22,23).
To determine the anatomical location of syntrophins within the human intestinal smooth muscle layers, we used laser capture microdissection to collect, separately, smooth muscle cells from the jejunal circular muscle layer and longitudinal muscle layer. Total RNA was extracted from the harvested cells, and reverse transcription was carried out using random primers. An aliquot from the reverse transcription was used for PCR amplification with one specific primer pair for either ␣, ␤1, ␤2, or ␥2 in each tube. Single bands of the expected size for syntrophin ␣ and ␤2 were observed in both circular and longitudinal smooth muscle (Fig. 1B). Syntrophin ␥2 was detected only in circular smooth muscle, and ␤1 was detected only in longitudinal smooth muscle (Fig. 1B). The housekeeping gene glyceraldehyde-3-phosphate dehydrogenase was amplified from the RT products and used as an internal control.
To further confirm the localization of syntrophin ␥2 to the circular muscle layer of the human intestine, we immunolabeled human jejunal sections with a rabbit polyclonal antibody directed against syntrophin ␥2 (a kind gift from Dr. Vincenzo Nigro). As shown in Fig. 1C, strong syntrophin ␥2 immunoreactivity was observed in the circular smooth muscle layer but FIG. 1. Expression of syntrophins in human intestinal smooth muscle. A, gene-specific primers designed against syntrophin ␣, ␤1,␤2, ␥1, and ␥2 were used for PCR amplification from cDNA libraries. Single cDNA bands were obtained from syntrophin ␣, ␤1, and ␤2 primers but not ␥1. Three bands were observed on syntrophin ␥2 amplification. Sequence analysis showed that all three cDNA fragments were different transcripts of syntrophin ␥2. B, to determine anatomical localization of syntrophins in intestinal muscle layers, ϳ1500 human jejunal smooth muscle cells from circular muscle and longitudinal muscle were collected by laser capture microdissection and syntrophin message amplified by RT-PCR on multiple aliquots with gene-specific primers. Bands for syntrophin ␣, ␤2, and ␥2 were present in circular muscle and bands for syntrophin ␣, ␤1, and ␤2 in longitudinal muscle. C, immunolabeling for syntrophin ␥2. Human jejunal sections were immunolabeled with an anti-syntrophin ␥2 antibody. Immunopositive smooth muscle cells were present in the circular but not longitudinal muscle layer. D, cellular localization of syntrophins. Primers designed to span introns to exclude genomic DNA contamination were used for two or three smooth muscle cell PCR amplification. Products of the expected size for syntrophin ␣, ␤2, and ␥2 were three freshly dissociated human jejunal circular smooth muscle cells. Product identity was confirmed by band sequencing. not in the longitudinal muscle layer. Control experiments omitting the primary antibody showed no immunoreactivity in the circular muscle layer (data not shown). Since several different cell types reside in the intestinal smooth muscle layers, single cell RT-PCR was used to localize syntrophins to smooth muscle cells. Smooth muscle cells were dissociated from strips of circular smooth muscle and identified according to their spindle shape. Two or three cells were collected and placed in each tube. Syntrophin ␣, ␤2, and ␥2 were successfully amplified with gene-specific primer sets designed to span an intron (Fig. 1D) to exclude genomic sequence. The correct size band was recovered from the agarose gel using conventional techniques, and results were confirmed by sequence analysis. There were no products in negative controls (4 l of bath solution aspirated just above a smooth muscle cell) (Fig. 1D). These results were duplicated in three additional experiments.
Isolation and Characterization of Syntrophin ␥2 Splice Variants-Similar to the layer-specific expression of syntrophin ␥2, SCN5A is expressed in human intestinal circular smooth muscle cells but does not appear to be strongly expressed in longitudinal cells (12). This cell-specific colocalization between expression of SCN5A and syntrophin ␥2 led us to focus on syntrophin ␥2. PCR amplification for syntrophin ␥2 (Fig. 1A) showed three bands, suggesting different transcripts of syntrophin ␥2. To isolate the full-length coding sequence for each transcripts, we designed primers to cover the whole open reading frame. RT-PCR amplification using human jejunal muscle tissue produced a 1.7-kb cDNA fragment, which was subcloned into pCR2.1 plasmid vector using the TOPO-TA protocol (Invitrogen). Sequence analysis showed at least five splice variants of syntrophin ␥2 in human jejunal muscle (Fig. 2). Splice variant 1 was identical to the published sequence of syntrophin ␥2 with 17 exons (accession number NM_018968). Splice variant 2 had a 27-bp in-frame deletion in exon 9 that would result in loss of the protein kinase C phosphorylation site. A similar deletion has been described in human brain tissue (22). In splice variant 3, exons 3-6 were spliced out, resulting in the nearly complete elimination of the PDZ domain, and a 222-bp in-frame insertion was present between exons 11 and 12. In another splice variant, variant 4 in Fig. 2, a 256-bp insertion with a stop codon was inserted between exons 9 and 10, resulting in loss of the pleckstrin homology and the syntrophinunique domains and loss of the ATP/GTP-binding site. The fifth splice variant had a 46-bp insertion with a stop codon inserted between exons 14 and 15, resulting in loss of the ATP/GTPbinding site and the syntrophin-unique domains.
Direct Interaction between SCN5A and Syntrophin ␥2-To determine whether syntrophin ␥2 and SCN5A directly interact in human intestinal smooth muscle and to determine whether the interaction was mediated via the PDZ binding domain on syntrophin, we performed yeast two-hybrid and GST pull-down experiments. The last 100 amino acids of SCN5A with (CTSCN5A) and without (CTSCN5A-10) the last 10 amino acids (the last 10 amino acids include the PDZ binding domain) were inserted into the pBD-GAL4 vector and used as baits (Fig.  3). Splice variant 1 of syntrophin ␥2 with an intact PDZ domain and splice variant 3 lacking the PDZ domain were fused into pAD-GAL4 vector and used as prey. In the yeast YRG-2 cells, the reporter gene His was activated only by the interaction between the splice variant 1 of syntrophin ␥2 with PDZ domain and the intact C terminus of SCN5A (Fig. 3b). Other pairings of syntrophin ␥2 with SCN5A did not activate His, indicating that the last 10 amino acids of SCN5A mediated the binding of SCN5A to the PDZ domain of syntrophin ␥2. The specificity of this interaction was confirmed by testing activation of second reporter gene lacZ (Fig. 3c). In addition, syntrophin ␥2 and the last 100 amino acids of SCN5A both did not self-activate when transformed into yeast with empty bait or prey vectors (data not shown).
The interaction between the C terminus of SCN5A and the PDZ domain of syntrophin ␥2 was verified using GST pulldown assays. Splice variants 1 and 3 of syntrophin ␥2 were expressed in HEK 293 cells. Affinity-purified GST, GST plus CT5A, and GST plus CT5A-10, immobilized on glutathione-Sepharose beads, were incubated with cell lysates containing the syntrophin ␥2 splice variants 1 and 3. Fig. 4 shows that only GST plus CT5A trapped splice variant 1 of syntrophin ␥2 (lane 2) and that there was no interaction between GST plus CT5A with or without the last 10 amino acids (aa) (lanes 5 and 6) and the splice variant 3 of syntrophin ␥2, again suggesting a specific interaction between the PDZ domain of syntrophin ␥2 and the last 10 aa of SCN5A. GST alone did not bind to either splice variant of syntrophin ␥2 (lanes 1 and 4).
The specificity and functionality of the interaction between the C terminus of SCN5A and the PDZ domain of syntrophin ␥2 were tested on the native Na ϩ current in human intestinal circular smooth muscle cells. Freshly dissociated cells were patch-clamped, and a 10-aa peptide (1 mM) corresponding to the last 10 aa of SCN5A was introduced via the patch pipette into the cells. The peptide completely abolished the perfusioninduced increase in Na ϩ current. The peptide was allowed to diffuse into the cells for 10 min after breaking in, and the Na ϩ current was activated by perfusion. Perfusion did not increase peak inward Na ϩ current (5 Ϯ 2% increase, n ϭ 8, p Ͼ 0.05 compared with preperfusion) (Fig. 5a). In contrast, in control cells without the peptide, perfusion increased peak inward Na ϩ current by 27 Ϯ 3% (data not shown). Introduction of a control jumbled peptide had no effect on this perfusion-induced increase in inward Na ϩ current (27 Ϯ 6% increase in current, n ϭ 6, p Ͻ 0.05 compared with preperfusion). Furthermore, intro-duction of a GST-bound 98-aa sequence (10 nM) corresponding to the PDZ domain of syntrophin ␥2 also blocked the perfusioninduced increase in inward Na ϩ current (6 Ϯ 2% increase in current, n ϭ 6, p Ͼ 0.05 compared with preperfusion) (Fig. 5b). Introduction of GST alone into the human intestinal smooth muscle cells did not block the perfusion-induced increase in current (19 Ϯ 4% increase in current, n ϭ 8, p Ͻ 0.05 compared with perfusion without GST). These results suggest that a direct interaction between the last 10 aa of SCN5A and the PDZ domain of syntrophin ␥2 is required to maintain mechanosensitivity of the Na ϩ channel.
Modulation of Gating of SCN5A by Syntrophin ␥2-The above data indicate a specific interaction between syntrophin ␥2 and SCN5A and that syntrophin ␥2 is required for mechanosensitivity of SCN5A. Syntrophins do not appear to be required for the sarcolemmal localization of sodium channels (27,28). Therefore, to further delineate the functional consequence of the interaction between the two proteins, we transfected HEK293 cells with either SCN5A alone or with SCN5A and syntrophin ␥2. Co-transfection of syntrophin ␥2 with SCN5A shifted in a positive direction the voltage-dependent activation of SCN5A by 8.5 mV. (Fig. 6, a and b). V1 ⁄2 for SCN5A alone was Ϫ43.4 Ϯ 0.3 mV (n ϭ 9) and shifted to Ϫ34.9 Ϯ 0.3 mV (n ϭ 9), p Ͻ 0.05, when syntrophin ␥2 was co-transfected with SCN5A. As a result, maximal peak inward Na ϩ current shifted from Ϫ23 Ϯ 1 to Ϫ15 Ϯ 2 mV. At Ϫ40 mV, peak inward Na ϩ current decreased from Ϫ737 Ϯ 94 pA to Ϫ281 Ϯ 67 pA when syntrophin ␥2 was co-transfected with SCN5A (n ϭ 8, p Ͻ 0.05). However, maximal peak inward Na ϩ current was unchanged (Ϫ1169 Ϯ 174 pA, n ϭ 7 for SCN5A alone; Ϫ1163 Ϯ 278 pA, n ϭ 8, p Ͼ 0.05 for SCN5A with syntrophin ␥2). Time to peak current increased at all voltages tested (Fig. 6c) with a change from 0.87 Ϯ 0.03 to 1.53 Ϯ 0.056 ms at maximal inward Na ϩ current (Ϫ20 mV, n ϭ 6, p Ͻ 0.05). The slope of the activation curve was also changed (Fig. 6b), with a k value of 5.1 Ϯ 0.2 mV for SCN5A alone and a k value of 6.5 Ϯ 0.2 for FIG. 3. Direct interaction between the PDZ domain of syntrophin ␥2 and the last 10 amino acids of SCN5A in vivo. a, schematic diagram of baits and preys used in the yeast two-hybrid system analysis (CTSCN5A, last 100 aa of SCN5A; CTSCN5A-10, C terminus lacking the last 10 aa of SCN5A; Syn-␥2 1 , syntrophin-␥2 splice variant 1 with an intact PDZ domain; Syn-␥2 3 , syntrophin-␥2 splice variant 3 lacking a PDZ domain). b, expression of the reporter gene HIS3. Each pair of constructs as indicated in A was co-transfected into the YRG-2 yeast strain. Yeast transformants were then selected on selective plates and tested for expression of reporter gene HIS3. Strong expression of HIS3 only occurred when the last 10 aa of SCN5A and the PDZ domain of syntrophin ␥2 were both present. c, ␤-galactosidase activity. Colonies that grew on the selective plates were transferred onto the filter papers and assayed for ␤-galactosidase activity, confirming that SCN5A and syntrophin ␥2 interact and that the interaction occurs through the C terminus of SCN5A and the PDZ domain.

FIG. 4. Direct interaction between the PDZ domain of syntro-
phin ␥2 and the last 10 amino acids of the C terminus of SCN5A in vitro. A, the full-length syntrophin ␥2 with a PDZ domain (Syn ␥2 1 ) and the syntrophin ␥2 without a PDZ domain (Syn ␥2 3 ) were transfected into HEK 293 cells. Approximately 200 g of cell lysate was then incubated with GST, GST plus CT5A (last 100 aa of the C terminus of SCN5A), or GST plus CT5A-10 (last 100 aa of the C terminus except for the very last 10 aa) beads. After washing, the proteins bound to the beads were resolved by 10% SDS-PAGE and identified by Western blots using the anti-FLAG antibody as the probe. Specific binding was observed only between GST plus CT5A and syntrophin ␥2 1 (with a PDZ domain). B, lanes 1 and 2 were loaded with 1% of the cell lysates compared with 10% of the extracts for lanes 3 and 4.
SCN5A and syntrophin ␥2. Inactivation was fit with 2 taus. Co-expression of syntrophin ␥2 resulted in a slower first tau (fast inactivation) with no change noted for the second tau (slow decay, Fig. 6c, n ϭ 6, p Ͻ 0.05). No effect was noted on the kinetics of steady-state inactivation (Fig. 6b); therefore, the net result of the observed changes was a reduction in the overlap of the activation and inactivation relationships, resulting in a reduced window current (29) (Fig. 6b) compared with the window current observed with SCN5A alone. Truncation of the last 10 aa of SCN5A (n ϭ 6, Fig. 7a) or cotransfection of the splice FIG. 5. SCN5A C terminus peptide and syntrophin ␥2 PDZ domain peptide block perfusion-induced increase in peak Na ؉ current. a, control Na ϩ current obtained from a human jejunal circular smooth muscle cells using the pulse protocol in the inset 10 min after breaking in with 1 mM C terminus peptide in the pipette solution and lack of activation of the Na ϩ current by perfusion. b, control Na ϩ current obtained 10 min after breaking in with 1 mM C terminus scrambled peptide in the pipette solution and activation of the Na ϩ current by perfusion. c and d, mean current-voltage relationships for the effects of perfusion in the presence of the C-terminal peptide and the control scrambled peptide, respectively. e, mean peak inward Na ϩ currents. f, Coomassie Blue-stained recombinant purified GST and GST plus PDZ shown on 15% SDS-PAGE. g, control Na ϩ current 10 min after breaking in with 10 nM GST-PDZ peptide in the pipette solution and lack of activation of the Na ϩ current by perfusion. h, control Na ϩ current obtained 10 min after breaking in with just the GST peptide and activation of the Na ϩ current by perfusion. i and j, mean current-voltage relationships for the effects of perfusion in the presence of GST-PDZ domain peptide and GST, respectively. k, mean peak inward Na ϩ currents. variant of syntrophin ␥2 lacking a PDZ domain (syntrophin ␥2 3 , n ϭ 6, Fig. 7b) had no significant effect on Na ϩ channel gating, suggesting that the effects of syntrophin ␥2 on SCN5A current were again mediated by a specific interaction between the PDZ domain of syntrophin ␥2 and the last 10 amino acids of SCN5A. DISCUSSION The main finding of this study is that mechanosensitivity of the human circular smooth muscle and cardiac Na ϩ channel is dependent on a specific interaction between the PDZ domain of syntrophin ␥2 and the last 10 amino acids of the C terminus of SCN5A, the ␣ subunit of both the tetrodotoxin-resistant cardiac muscle and the native intestinal smooth muscle Na ϩ channel. The mechanisms that underlie ion channel mechanosensitivity are complex and vary according to the ion channel studied (3,30). Potential mechanisms for ion channel mechanosensitivity include a direct interaction between the transmembrane portion of the channel and the lipid bilayer, suggesting that this is an unavoidable consequence of inserting a channel into the membrane (30 -32), activation of mechanosensitive signaling cascades that subsequently activate ion channels via phosphorylation or other post-translational modification (3), or force transmission via protein-protein interactions between the channel and the cytoskeleton that alter the channel open probability (3,4,6,33,34). Our data suggest that for SCN5A, the latter mechanism appears to be a central one for the mechanosensitivity observed upon perturbing the cell's membrane, since mechanosensitivity of the native Na ϩ channel was completely lost when the interaction between the cytoskeleton and the C terminus of SCN5A was disrupted. The expression of SCN5A and syntrophin ␥2 in intestinal muscle was similar, with immunohistochemical and molecular evidence to suggest that both are expressed in human intestinal circular but not longitudinal smooth muscle. The co-expression of SCN5A and syntrophin ␥2 suggests that the interaction may be specific to syntrophin ␥2 and not be generalizable to all syntrophins with PDZ domains. This is supported by the data from native cells with block of mechanosensitivity when the specific aa sequence of the PDZ domain of syntrophin ␥2 is introduced into the cell. The sequences of PDZ domains are known to be highly conserved among all five known syntrophins (22,23). However, the recently identified syntrophin ␥1 has a PDZ domain similar to that of other syntrophins and yet does not bind SCN4A and SCN5A, suggesting different specificity of PDZ domains of syntrophins (24). Moreover, the Na ϩ channel-syntrophin interaction is not necessarily dependent on PDZ domains, since brain Na ϩ channels, which lack the consensus motif E(S/T)XV at their C termini, required to bind PDZ domains, still copurified with syntrophin (14), suggesting that multiple interactions occur between syntrophins and Na ϩ channels and that the specific interaction between the C terminus of SCN5A and the PDZ domain of syntrophin ␥2 may only be only an absolute requirement for mechanosenstivity.
A link between Na ϩ channel activity and actin cytoskeleton has been proposed previously. Treatment of cardiac myocytes with cytochalasin-D to inhibit actin polymerization reduces peak Na ϩ current and slows inactivation (35,36). Disruption of the cytoskeleton also alters Na ϩ channel properties in skeletal muscle, epithelial tissue, and leukemia cells (28,(37)(38)(39). The data presented here suggest that, for SCN5A, the likely link between SCN5A and the actin cytoskeleton are the syntrophins. Syntrophins localize associated proteins to the membrane by binding with dystrophin (40,41). In genetically modified mice lacking syntrophin ␣ or the PDZ domain of syntrophin, neuronal nitric-oxide synthase is absent from the sarcolemma, suggesting dependence on syntrophin for sar-FIG. 6. Effect of co-transfection of syntrophin ␥2 with SCN5A. a, inward Na ϩ current at Ϫ40 mV for SCN5A alone and SCN5A plus syntrophin ␥2. b, steady state activation and inactivation curves for SCN5A alone and SCN5A plus syntrophin ␥2, showing the right shift in activation and the smaller window current when syntrophin ␥2 was co-transfected. c, time to peak (activation) was slower at all voltages tested when syntrophin ␥2 was co-transfected with SCN5A. Fast inactivation was also slower at all voltages tested, whereas slow decay was unchanged (see "Results" for details).
colemmal localization of neuronal nitric-oxide synthase (27,42). Similarly, the membrane localization of water channel, aquaporin-4, also requires syntrophin (43). Aquaporin-4 is not localized to the membrane when syntrophin ␣ is absent. However, syntrophin is not always required for the membrane localization of its associated proteins. The skeletal muscle sodium channel SCN4A is known to bind syntrophin ␣, ␤1, and ␤2, but localization to the membrane does not appear to depend on syntrophins (27,28). Also, the absence of syntrophin ␣ does not have an apparent effect on the distribution or level of FIG. 7. Effect of truncation of SCN5A and of loss of the syntrophin ␥2 PDZ domain on SCN5A kinetics. a and d, whole cell current traces at Ϫ40 mV for SCN5A, SCN5A without the last 10 aa cotransfected with syntrophin ␥2, and SCN5A co-transfected with syntrophin ␥2 without the PDZ domain. b and e, steady-state activation and inactivation curves. c and f, activation and inactivation kinetics. Both truncation of SCN5A and absence of the PDZ domain of syntrophin ␥2 resulted in loss of the kinetic changes seen when the full-length SCN5A was co-expressed with syntrophin ␥2. expression of the Na ϩ channel. In mdx mice, the distribution of Na ϩ channels remains identical despite the lack of dystrophin and reduction of sarcolemmal density of syntrophins (28). These observations suggest that syntrophins do not play a key role in the sarcolemmal localization of Na ϩ channel. Rather, the data presented here indicate that syntrophins directly alter SCN5A gating behavior. Thus, syntrophin ␥2 may be an essential Na ϩ channel-interacting protein required to recapitulate the full phenotype of the Na ϩ current.
The changes in activation and inactivation kinetics observed on cotransfection of syntrophin ␥2 with SCN5A would have significant effects on the Na ϩ channel whole cell current. A shift in the voltage of activation for SCN5A coupled with slowing of activation will result in a smaller contribution of the peak Na ϩ current to the action potential. Prolongation of fast inactivation would be expected to increase available Na ϩ current over a few milliseconds after the peak, but the reduced window current would tend to reduce late Na ϩ current over the window voltage range (29). No effect on slow decay was seen in our study. However, a previous study has reported that the last 8 aa of the C terminus of SCN5A slows inactivation with little effect on activation (44). During the gastrointestinal slow wave the membrane potential of smooth muscle is within the window current of SCN5A suggesting that the channel contributes to the regulation of membrane potential and of intracellular Na ϩ . Changes in the window current, as seen with syntrophin ␥2, may therefore alter membrane potential and intracellular Na ϩ homeostasis. Mutations in SCN5A are associated with changes in gating behavior and clinical disease (10). Mutations in SCN5A that prolong slow decay result in an enhancement of inward plateau current and a vulnerability to potentially fatal heritable arrhythmias like long QT syndrome subtype 3 and rarely sudden infant death syndrome (45). Mutations that result in a decrease in Na ϩ current lead to Brugada's syndrome and Lenegre disease (10). The changes in SCN5A current kinetics observed on co-transfection of SCN5A in HEK cells with syntrophin ␥2 may therefore lead to more than one mechanism for cardiac arrhythmias dependent on the relative expression of syntrophin ␥2 variants. It is presently unclear if syntrophins can directly interact with membrane proteins and the actin cytoskeleton or if dystrophin is always required. However, considering that only 65% of long QT syndrome is genotyped presently on the known long QT-causing genes (46), molecular and functional characteristics of syntrophin ␥2 described herein makes this Na ϩ channel-interacting protein an attractive pathogenomic target for unexplained, nongenotyped sudden cardiac death.
In summary, mechanosensitivity of the Na ϩ channel encoded by SCN5A requires a specific interaction between the C terminus of SCN5A and the PDZ domain of syntrophin ␥2. Coexpression of SCN5A and syntrophin ␥2 results in a smaller Na ϩ window current, suggesting the possibility that syntrophins may directly regulate the cardiac muscle and intestinal smooth muscle Na ϩ current. These results also raise the possibility that mutations in the PDZ domain of syntrophin ␥2 or differential expression of the identified splice variants of syntrophin ␥2 may contribute to clinically significant arrhythmias.