cAMP-dependent Phosphorylation of Two Sites in the α Subunit of the Cardiac Sodium Channel

The voltage-sensitive Na+ channel is responsible for generating action potentials in the heart which are critical for coordinated cardiac muscle contraction. Cardiac Na+ channels are regulated by cAMP-dependent phosphorylation, but the sites of phosphorylation are not known. Using mammalian cells expressing the rat cardiac Na+ channel (rH1) α subunit and site-specific antibodies, we have shown that the α subunit of rat heart Na+ channel is phosphorylated selectively by cAMP-dependent protein kinase (PKA) in vitro and in intact cells. Analysis of the sites of phosphorylation by two-dimensional phosphopeptide mapping and site-directed mutagenesis of fusion proteins revealed that the cardiac α subunit is phosphorylated selectively in vitro by PKA on Ser526 and Ser529 in the intracellular loop connecting homologous domains I and II (LI-II). These two residues were phosphorylated in intact cells expressing the rH1 α subunit when PKA was activated. Our results define a different pattern of phosphorylation of LI-II of cardiac and brain Na+ channels and implicate phosphorylation of Ser526 and Ser529 in the differential regulation of cardiac and brain Na+ channels by PKA.

The voltage-sensitive Na ϩ channel is responsible for generation of action potentials in the heart. Cardiac function depends on the amplitude, timing, and voltage dependence of Na ϩ current through Na ϩ channels. Biochemical and molecular cloning studies have shown that rat heart contains a cardiac-specific ␣ subunit designated rH1 (1)(2)(3). Denervated skeletal muscle expresses a form of the Na ϩ channel ␣ subunit (Skm2) which is identical in amino acid sequence to the cardiac isoform (4). Immunoblotting experiments with anti-peptide antibodies have indicated that the cardiac ␣ subunit is approximately 230 -240 kDa in apparent molecular mass (1,2). The ␣ subunit of the rat heart channel isoform is sufficient to form functional voltage-sensitive Na ϩ channels when expressed in mammalian cells (5) or Xenopus oocytes (4, 6 -9).
The ␣ subunit of the purified rat brain Na ϩ channel is phosphorylated on four residues in the intracellular loop connecting domains I and II (L I-II ) 1 by cAMP-dependent protein kinase (protein kinase A, PKA) in vitro, in intact neurons, and in mammalian cells transfected with the ␣ subunit (10 -13). Brain Na ϩ channel activity is regulated by phosphorylation of these sites by PKA (14 -16). PKA phosphorylation of the Na ϩ channel ␣ subunit decreases peak Na ϩ current with no alteration in the time course of channel inactivation (15) and can be reversed by treatment with a mixture of catalytic subunits of phosphatases 1 and 2A (15).
PKA has been reported to cause diverse effects on cardiac Na ϩ channel activity in various heart cell preparations. Receptor-activated modulation of cardiac Na ϩ current is believed to be mediated by both a fast direct G-protein pathway and by the slower indirect phosphorylation pathway (17). Treatment of mammalian myocytes with cAMP-elevating agents or cell-permeant cAMP analogs has been shown to cause either an increase (17)(18)(19) or decrease in inward Na ϩ currents (20,21). Na ϩ currents in Xenopus oocytes expressing the Skm2 ␣ subunit are augmented by treatment with these agents (8,22). Ono et al. (21) showed that cAMP shifted the voltage dependence of both activation and inactivation of Na ϩ channels in guinea pig, dog, and canine myocytes in the hyperpolarizing direction, providing a mechanism through which increases in intracellular cAMP could result in either an increase or decrease of Na ϩ current depending on the selected holding and test potentials in different experiments. The complex effects of activation of PKA on Na ϩ channel function suggest that it may phosphorylate multiple sites that cause different functional effects.
Identification of residues phosphorylated by PKA is critical to understanding the molecular mechanisms whereby phosphorylation modifies cardiac Na ϩ channel activity. In the present study we have determined that the ␣ subunit is phosphorylated selectively by PKA on two sites in L I-II intracellular loop in vitro and in intact cells. These sites are distinct from the sites of phosphorylation of brain Na ϩ channels providing a rationale for the complex differential regulation of these two Na ϩ channels by PKA.

EXPERIMENTAL PROCEDURES
Materials-Protein A-Sepharose was purchased from Sigma. [␥-32 P]ATP (3,000 Ci/mmol) was obtained from DuPont NEN. Sequencing grade modified porcine trypsin was purchased from Promega Corp. (Madison, WI). Chromogram sheets and x-ray film were purchased from Eastman Kodak. Molecular weight markers for SDS-polyacrylamide gel electrophoresis (PAGE) were purchased from Novex (San Diego).
SDS-PAGE and Immunoblotting-Denatured 32 P-labeled Na ϩ channel protein was analyzed by SDS-PAGE in 6% Laemmli gels (23) using a Mini-Protean II gel system (Bio-Rad). Phosphorylated ␣ subunit was localized by autoradiography for subsequent phosphopeptide analysis. 32 P-Labeled fusion proteins were analyzed on 10 -20% Tricine gradient polyacrylamide gels (Novex, San Diego, CA). For phosphopeptide mapping analysis, 32 P-labeled fusion proteins were analyzed on 7.5% Laemmli gels (23) using a Mini-Protean II gel system. Electroblotting to nitrocellulose membranes was carried out in a Novex apparatus for 150  min at 195 mA (constant current) with 25 mM Tris, 192 mM glycine, 20% (v/v) methanol (pH 8.3) as the transfer buffer. Unbound sites were blocked for 2 h at room temperature with 5% (w/v) skim milk powder in 10 mM Tris, 0.15 M NaCl (pH 7.4) (TBS-milk). Membranes were then incubated with antibodies in blocking buffer 120 min at room temperature followed by three 10-min washes with TBS-milk containing 0.05% Tween 20. Immunoreactive bands were visualized using the Amersham ECL immunoblotting detection system with horseradish peroxidaselinked protein A as the affinity reagent.
Two-dimensional Phosphopeptide Analysis-Two-dimensional phosphopeptide mapping was carried out essentially as outlined previously (24) in the 1% (NH 4 ) 2 CO 3 (pH 8.9) system except that digestion was carried out with 1.2 g of sequencing grade trypsin.
Phosphoamino Acid Analysis-Phosphorylated Na ϩ channel ␣ subunit was trypsinized in situ as outlined above for two-dimensional tryptic phosphopeptide mapping. Subsequent acid hydrolysis and electrophoresis were carried out as outlined previously (24).
Construction and Expression of Recombinant Glutathione S-Transferase (GST) Fusion Proteins-Cardiac Na ϩ channel fusion proteins were generated from pcDNA-3/rH1 (5) using the polymerase chain reaction. Fragments of L I-II from the cardiac Na ϩ channel ␣ subunit were amplified by polymerase chain reaction and cloned into the pGEX-3X expression vector (Pharmacia Biotech Inc.) to obtain in-frame recombinant proteins containing GST. Fusion proteins were constructed, verified, expressed, and purified as described previously (25). In some ␣ subunit fusion proteins, specific serine residues were mutated to alanine residues using polymerase chain reaction-based mutagenesis (26).
Culturing and DCl-cBIMPS Treatment of SNa-rH1cells-SNa-rH1 cells were maintained in culture as described previously (27). Cells were harvested by solubilization in 1% Triton X-100 as described previously (12) except that the phosphatase inhibitor calyculin A (0.05 mM) was included in the culture medium for 20 min prior to harvesting. Cells were incubated at 37°C in the absence or presence of the cAMP analog DCl-cBIMPS (100 M) for 10 min and harvested as described above in the presence of the following phosphatase inhibitors: ␤-glycerol phosphate (20 mM), sodium pyrophosphate (50 mM), p-nitrophenyl phosphate (1 mM), NaFl (50 mM), and calyculin A (0.05 mM).
Preparation of Phosphospecific Anti-peptide Antibodies-Synthetic peptides were synthesized by the solid phase method of Merrifield (28) corresponding to residues 521-531 (peptide A) and 526 -535 (peptide B), respectively, of the rat cardiac Na ϩ channel ␣ subunit primary sequence (see Fig. 3) (3). Peptides were synthesized by Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry except that the NH 2 -terminal amino acid was introduced as a t-butoxycarbonyl derivative, and the serine of interest (Ser 526 in peptide A; Ser 529 in peptide B) was incorporated with its side group unprotected. The peptides were phosphorylated globally at the unprotected serine residue by reaction with dibenzyl-N,N-diisopropylphosphoramidite as per Novabiochem instructions. Phosphopeptides were purified by reversed phase high performance liquid chromatography on a Waters DeltaPak C18 column (25 ϫ 300 mm, 15 m, 300 Å), and their identities and complete phosphorylation were confirmed by mass spectrometry. Antisera against the peptides were prepared as outlined previously (29). The anti-peptide antibodies were purified by affinity chromatography on phosphopeptide columns and assayed by enzyme-linked immunosorbent assay.

RESULTS
Phosphorylation of Cardiac Na ϩ Channel Isolated from Cultured Cells Expressing the ␣ Subunit of Rat Heart Na ϩ Channel-To determine if the rat heart Na ϩ channel is a substrate for PKA in vitro, Chinese hamster lung cells (line 1610) expressing the ␣ subunit of rat heart Na ϩ channel (SNa-rH1, 5) were solubilized with 1% Triton X-100, and the ␣ subunit was immunoprecipitated with either anti-SP19, an anti-peptide antibody that recognizes a well conserved segment of Na ϩ channel ␣ subunits (1), or nonimmune IgG. Immunoprecipitated ␣ subunit was then incubated under phosphorylating conditions in the presence of [␥-32 P]ATP and PKA. Proteins were analyzed by SDS-PAGE, and 32 P-labeled phosphoproteins were visualized by autoradiography as described under "Experimental Procedures." Fig. 1A shows that anti-SP19 specifically immunoprecipitated a single protein with an apparent molecular mass of approximately 240 kDa which is a substrate for PKA. Preimmune IgG failed to immunoprecipitate this phosphoprotein, confirming its specificity (Fig. 1A).
Two-dimensional phosphoamino acid analysis (Fig. 1B) showed that PKA phosphorylated the ␣ subunit in vitro solely on serine residues. Phosphorylation of the ␣ subunit was examined further by two-dimensional tryptic phosphopeptide mapping as described under "Experimental Procedures." Following autoradiography, two major phosphopeptides (I and II) were visualized ( Fig. 2A). Overexposure of these phosphopeptide maps indicates the presence of three reproducible minor phosphopeptides (III-V) whose migration positions are illustrated in Fig. 2B. These phosphopeptides are not clearly visible in this photograph but are visible in the original overexposed autoradiogaphs (also see Figs. 5 and 7). A minor phosphopeptide near phosphopeptide III was also visualized in overexposed maps. The presence of this peptide varied from experiment to experiment and was distinct from phosphopeptide III. Together these results suggest that the cardiac Na ϩ channel ␣ subunit is phosphorylated by PKA in vitro on multiple serine residues.
Phosphorylation of GST Fusion Proteins Containing L I-II -The rat brain Na ϩ channel ␣ subunit is phosphorylated on four residues clustered in L I-II both in vitro and in intact cells in response to activation of PKA (12,13). Although these residues are not conserved in the rat heart isoform of the channel (3), several potential consensus sequences for PKA recognition are found in this intracellular loop. Lack of an efficient purification procedure for cardiac Na ϩ channels precludes the identification of phosphorylation sites by conventional protein microsequencing techniques. We therefore constructed and expressed GST fusion proteins containing portions of L I-II of the rat heart Na ϩ channel ␣ subunit (Fig. 3A). NaFH1 corresponds in sequence to amino acids 436 -645 of the full-length cardiac ␣ subunit predicted by cDNA cloning (3). NaFH1 was purified by chromatography on glutathione-Sepharose, incubated under phosphorylating conditions in the presence of PKA and FIG. 1. Phosphorylation and phosphoamino acid analysis of cardiac Na ؉ channel ␣ subunits immunoprecipitated from intact cells. Panel A, Sna-rH1 cells were solubilized in Triton X-100, and Na ϩ channel ␣ subunits were immunoprecipitated with either anti-SP19 or nonimmune IgG. Isolated channels were then phosphorylated in the presence of PKA and [␥-32 P]ATP and analyzed by SDS-PAGE on a 6% Laemmli gel. 32 P-Labeled ␣ subunits were visualized by autoradiography. Molecular weight markers are represented as M r ϫ 10 Ϫ3 . Panel B, 32 P-labeled ␣ subunits were excised from the wet gel and processed for phosphoamino acid analysis as described under "Experimental Procedures." Acid-hydrolyzed samples were subjected to twodimensional electrophoresis on thin layer cellulose plates at pH 1.9 (first dimension), and at pH 3.5 (second dimension). The migration positions of phosphoserine (S), phosphothreonine (T), and phosphotyrosine (Y) standards are designated by the broken circles and arrows.
[␥-32 P]ATP, and analyzed by SDS-PAGE and autoradiography as described under "Experimental Procedures." Fig. 4A shows that the NaFH1 was a good substrate for PKA. A faster migrating phosphoprotein is present in this preparation and likely represents a proteolytically cleaved fragment of fulllength NaFH1. GST alone is not a substrate for PKA (data not shown). 32 P-Labeled NaFH1 was then subjected to two-dimensional tryptic phosphopeptide mapping as described under "Experimental Procedures" (Fig. 4B). Two major and one minor phosphopeptide were visualized whose position in two-dimensional maps corresponded with phosphopeptides I, II, and III derived from tryptic digestion of 32 P-labeled ␣ subunit isolated from intact cells ( Fig. 2A). Overexposed autoradiographs also contain two other minor phosphopeptide spots whose map position corresponds to phosphopeptides IV and V from tryptic digestion of 32 P-labeled ␣ subunit isolated from intact cells (Fig. 2B). These results are consistent with the conclusion that PKA phosphorylates the same residues in NaFH1 and in rat cardiac Na ϩ channel ␣ subunit isolated from intact cells. Evidently, the rat cardiac Na ϩ channel ␣ subunit is phosphorylated by PKA in vitro solely on residues contained in L I-II , as observed previously for the brain Na ϩ channel ␣ subunit.
L I-II of the cardiac Na ϩ channel ␣ subunit contains eight candidate Ser residues within the PKA consensus sequence KRXXS, RXXS, or RXS (Fig. 3B), which are all distinct from phosphorylation sites in the brain Na ϩ channel. To locate the region that is phosphorylated by PKA, two new fusion proteins were constructed which divide NaFH1 into two parts. NaFH2 and NaFH3 correspond in sequence to amino acids 436 -511 and 491-645, respectively. Equal amounts of purified NaFH2 and NaFH3 were incubated under phosphorylating conditions in the presence of PKA and [␥-32 P]ATP and analyzed by SDS-PAGE and autoradiography as described above. NaFH2 was not a substrate for PKA (data not shown). NaFH3, however, was a good substrate in vitro for PKA (Fig. 5A). A faster migrating phosphoprotein is present in this lane and likely represents a proteolytically cleaved fragment of full-length NaFH3. When 32 P-labeled NaFH3 was subjected to two-dimensional tryptic phosphopeptide mapping, two major and one minor phosphopeptide were visualized (Fig. 5B) whose migration positions corresponded to phosphopeptides I, II, and III, respectively, in the peptide map derived from the full-length ␣ subunit isolated from intact cells ( Fig. 2A) and NaFH1 (Fig.  4B). Overexposed maps of 32 P-labeled NaFH3 also contained the minor phosphopeptides IV and V (data not shown). These results suggest that the rat heart Na ϩ channel ␣ subunit isolated from intact cells is phosphorylated by PKA on serine residues located between amino acids 512 and 645 in L I-II .
To narrow further the region phosphorylated by PKA, NaFH3 was split into two fusion proteins (NaFH4 and NaFH5) which correspond in sequence to amino acids 491-569 and 530 -645, respectively (Fig. 3A). Fig. 6 shows that when equivalent amounts of the fusion proteins NaFH4 and NaFH5 were incubated under phosphorylating conditions in the presence of PKA and [␥-32 P]ATP, only NaFH4 was phosphorylated. When 32 P-labeled NaFH4 was subjected to two-dimensional tryptic phosphopeptide mapping, two major and one minor phosphopeptide were visualized whose migration positions corresponded to phosphopeptides I, II, and III in the peptide map derived from the full-length ␣ subunit (data not shown). Overexposed maps of 32 P-labeled NaFH4 also contained the minor phosphopeptides IV and V (data not shown). These results further narrow the region of the heart Na ϩ channel phosphorylated by PKA to residues 511-530 in L I-II . This region contains five serine residues; residues 520, 526, and 529 conform to the RXS consensus motif for PKA recognition (Fig. 3B).
Individual Serine Residues Phosphorylated by PKA-To determine which of these serine residues is phosphorylated by PKA, three additional fusion proteins were generated, corresponding in sequence to NaFH3 but with the serines at positions 520 (NaFH-S520A), 526 (NaFH-S526A), and 529 (NaFH-S529A) mutated to alanine residues. Each fusion protein was FIG. 2. Two-dimensional tryptic phosphopeptide analysis of cardiac Na ؉ channel ␣ subunits isolated from intact cells. Panel A, cardiac Na ϩ channel ␣ subunits were isolated, phosphorylated, and analyzed by SDS-PAGE as described in Fig. 1. 32 P-Labeled ␣ subunits were located by autoradiography, excised from the wet gel, and processed for tryptic phosphopeptide analysis as described under "Experimental Procedures." Tryptic phosphopeptides were separated in two dimensions by high voltage electrophoresis (pH 8.9) followed by thin layer chromatography. Arrows designate the direction of electrophoresis (ϩ) and chromatography (C). Phosphopeptides visualized by autoradiography were assigned a Roman numeral based on their relative migration position. Panel B, these autoradiographs also contained phosphopeptides designated III-V whose migration positions are shown by the broken circles. These phosphopeptides are not clearly visible in this photograph but are visible in the original overexposed autoradiographs (see also Figs. 5 and 7). purified by chromatography on glutathione-Sepharose, incubated under phosphorylating conditions in the presence of PKA and [␥-32 P]ATP, and analyzed by SDS-PAGE and autoradiography as described under "Experimental Procedures." All three of these mutant fusion proteins were substrates for PKA (data not shown). When 32 P-labeled NaFH-S520A was subjected to two-dimensional tryptic phosphopeptide mapping, the resultant phosphopeptide map was identical to that generated from full-length ␣ subunit (data not shown). This result indicates that Ser 520 in NaFH3 and in the full-length cardiac Na ϩ channel ␣ subunit is not phosphorylated by PKA in vitro. When 32 P-labeled NaFH-S526A was subjected to two-dimensional tryptic phosphopeptide mapping the resultant map contained only one major phosphopeptide, phosphopeptide II (Fig. 7). Overexposed maps of NaFH-S526A did contain the minor phosphopeptide IV; however, the minor phosphopeptides III and V were absent (data not shown). The two-dimensional tryptic map of an additional fusion protein, NaFH-S525/6A, in which Ser 525 and Ser 526 were both mutated to alanine residues, was identical to the maps of NaFH-S526A (data not shown). The two-dimensional tryptic phosphopeptide maps of two other mu-tant fusion proteins, NaFH-S560/1A and NaFH-S594/5A, were also identical to NaFH3 and the intact ␣ subunit (data not shown). Together, these results indicate that phosphopeptides I, III, and V in the intact Na ϩ channel ␣ subunit require the presence of Ser 526 . When 32 P-labeled NaFH-S529A was subjected to two-dimensional tryptic phosphopeptide mapping, the resultant phosphopeptide map contained phosphopeptides III and V (Fig. 7). These results indicate that phosphopeptides I, II, and IV derived from NaFH3 or the intact Na ϩ channel ␣ subunit require the presence of Ser 529 . Thus, our data show that the principal in vitro PKA phosphorylation sites on these fusion proteins are Ser 526 and Ser 529 . The two-dimensional phosphopeptide maps of mutant fusion proteins suggest that phosphopeptide I is phosphorylated on both Ser 526 and Ser 529 , phosphopeptides II and IV on Ser 529 , and phosphopeptides III and V on Ser 526 .
To confirm this conclusion, an additional fusion protein was generated, corresponding in sequence to NaFH3 but with the serines corresponding to residue 525, 526, and 529 mutated to alanine residues (NaFH-S525/6/9A). The ability of PKA to phosphorylate NaFH-S525/6/9A was then compared with its ability to phosphorylate the fusion protein NaFH-S560/1A, which contains Ser 526 and Ser 529 (Fig. 8). NaFH-S560/1A, like NaFH3, is a good substrate for PKA (data not shown). Equivalent amounts (Fig. 8C) of purified NaFH-S560/1A and NaFH-S525/6/9A were incubated under phosphorylating conditions for 1 min (Fig. 8A) or 10 min (Fig. 8B) in the presence of PKA and [␥-32 P]ATP and analyzed by SDS-PAGE and autoradiog- FIG. 4. Phosphorylation and tryptic phosphopeptide mapping of the fusion protein NaFH1. Panel A, purified NaFH1 was phosphorylated in the presence of PKA and [␥-32 P]ATP, analyzed by SDS-PAGE on a Novex 10 -20% Tricine gel, and 32 P-labeled fusion protein was located by autoradiography. Molecular weight markers are represented as M r ϫ 10 Ϫ3 . Panel B, phosphorylated NaFH1 was analyzed on a 7.5% Laemmli SDS gel and located by autoradiography. 32 P-Labeled NaFH1 was excised from the wet gel and processed for tryptic phosphopeptide analysis as described under "Experimental Procedures." Tryptic phosphopeptides were separated in two dimensions by high voltage electrophoresis (pH 8.9) followed by thin layer chromatography. Arrows designate the direction of electrophoresis (ϩ) and chromatography (C).
FIG. 5. Phosphorylation and tryptic phosphopeptide mapping of the fusion protein NaFH3. Panel A, purified NaFH3 was phosphorylated in the presence of PKA and [␥-32 P]ATP, analyzed by SDS-PAGE on a Novex 10 -20% Tricine gel, and 32 P-labeled fusion protein was located by autoradiography. Molecular weight markers are represented as M r ϫ 10 Ϫ3 . Panel B, phosphorylated NaFH3 was analyzed on a 7.5% Laemmli SDS gel and located by autoradiography. 32 P-Labeled NaFH3 was excised from the wet gel and processed for tryptic phosphopeptide analysis as described under "Experimental Procedures." Tryptic phosphopeptides were separated in two dimensions by high voltage electrophoresis (pH 8.9) followed by thin layer chromatography. Arrows designate the direction of electrophoresis (ϩ) and chromatography (C). .5% Laemmli gels, located by autoradiography, and subjected to two-dimensional mapping as described under "Experimental Procedures." Tryptic phosphopeptides were separated in two dimensions by high voltage electrophoresis (pH 8.9) followed by thin layer chromatography. Arrows designate the direction of electrophoresis (ϩ) and chromatography (C). Broken circles designate positions of absent phosphopeptides. raphy as described under "Experimental Procedures." NaFH-S525/6/9A was a very poor substrate for PKA compared with NaFH-S560/1A (Fig. 8, A and B). No phosphorylation is detected after 1 min, and after 10 min a barely detectable amount of 32 P is incorporated into NaFH-S525/6/9A. This result is consistent with the conclusion that Ser 526 and Ser 529 are the major sites of in vitro phosphorylation of these Na ϩ channel fusion proteins.
Phosphorylation of Ser 526 and Ser 529 of the Cardiac Na ϩ Channel ␣ Subunit in Intact Cells in Response to Activation of PKA-We next determined whether Ser 526 and Ser 529 were phosphorylated in intact cells by PKA in response to changes in the concentration of intracellular cAMP. To address this question we utilized the technique of "back-phosphorylation." Cells expressing the cardiac Na ϩ channel ␣ subunit were incubated in the absence or presence of the cell permeant cAMP analog DCl-cBIMPS to activate intracellular PKA. Cells were then solubilized, and the Na ϩ channel ␣ subunit was isolated by immunoprecipitation and incubated under phosphorylating conditions in the presence of PKA and [␥-32 P]ATP as described under "Experimental Procedures." In this experiment, a decrease in the in vitro PKA-mediated incorporation of 32 P into the ␣ subunit from treated cells is proportional to cAMP-stimulated phosphorylation of the ␣ subunit in situ in intact cells. PKA-mediated in vitro incorporation of 32 P into the ␣ subunit was reduced by 65% Ϯ 2% (n ϭ 3) in cells pretreated for 10 min with 100 M DCl-cBIMPS (Fig. 10, A and B). To assess the effect of stimulation with DCl-cBIMPS on the PKA-mediated 32 P incorporation into specific Ser residues, we generated tryptic phosphopeptide maps of the ␣ subunit isolated from treated and untreated cells. Fig. 11A shows the phosphopeptide map of the ␣ subunit derived from untreated cells which contains the major phosphopeptides I and II seen in previous maps ( Fig.  2A). Diminished back-phosphorylation of phosphopeptides I and II was observed in the phosphopeptide map derived from DCl-cBIMPS-treated cells (Fig. 11B). Fig. 11A also shows that the minor variable phosphopeptide near the postion of phosphopeptide III was present in this experiment and was not diminished by the treatment of cells by DCl-cBIMPS. These results indicate that this phosphopeptide is not phosphorylated in intact cells by PKA in response to treatment with a membrane-permeant derivative of cAMP. Together these results indicate that both Ser 526 and Ser 529 are phosphorylated by PKA in intact cells in response to increased intracellular levels of cAMP. The magnitude of diminution phosphorylation of Ser 526 and Ser 529 in this experiment was approximately 65 and 75%, respectively. Thus, treatment with DCl-cBIMPS caused a substantial increase in the endogenous phosphorylation of both Ser 526 and Ser 529 in the ␣ subunit of the cardiac Na ϩ channel in intact cells.

DISCUSSION
Cardiac Na ϩ Channels Are Regulated by cAMP-dependent Phosphorylation-Cardiac voltage-sensitive Na ϩ channels are modulated by activation of ␤-adrenergic receptors acting through both direct and indirect pathways (17)(18)(19)(20)(21)(22). The effect of adrenergic agents on cardiac Na ϩ channel activity is significant, suggesting that it may be an important physiological mechanism in the heart. Recent mutagenesis work by Schreibmayer et al. (8) has shown that activation of ␤ 2 receptors in Xenopus oocytes expressing the ␣ subunit of the rat cardiac Na ϩ channel modulates Na ϩ channel activity. These authors mutated five potential PKA phosphorylation sites in the ␣ subunit but did not abolish the observed effect of the kinase. Of the five potential PKA sites mutated, one was located in the amino terminus, two were located in the intracellular loop connecting homologous domains II and III, and two were located in the intracellular loop connecting homologous domains I and II. Neither of these sites in L I-II corresponded to Ser 526 or Ser 529 . Thus, previous experiments have not succeeded in identifying sites that are required for the physiological effect of phosphorylation by PKA, possibly because the consensus sites that have been examined by mutagenesis are not actually phosphorylated in vivo.
Phosphorylation of Cardiac Na ϩ Channel ␣ Subunit Isolated from Mammalian Cells-Our results show that cardiac Na ϩ channels immunoprecipitated from mammalian cells expressing the rat cardiac Na ϩ channel ␣ subunit are phosphorylated by PKA in vitro. These results are in agreement with those of Gordon et al. (1) and Cohen and Levitt (2), which showed that Na ϩ channel immunoprecipitated from solubilized rat cardiac membranes is a substrate for PKA in vitro. PKA phosphorylation of rat cardiac ␣ subunit occurs exclusively on serine residues, and the ␣ subunit is phosphorylated by PKA in intact cells in response to increases in intracellular levels of cAMP.
Principal Sites of Phosphorylation of the Cardiac Na ϩ Channel ␣ Subunit-Since brain Na ϩ channel activity is modulated by phosphorylation of the ␣ subunit in L I-II (12,13,15,16), we focused on this region as a possible target for PKA phosphorylation in the rat cardiac Na ϩ channel. This loop contains 11 consensus sequences for PKA with the motifs of KRXXS, RXXS, and RXS but does not contain any serine residues within the canonical RRXS motif. A fusion protein containing most of L I-II was a substrate for PKA in vitro, and the tryptic phosphopeptide map of 32 P-labeled NaFH1 was similar to the map obtained from 32 P-labeled ␣ subunit isolated from intact cells, indicating that the same residues were being phosphorylated in NaFH1 as in the full-length channel. These results demonstrate that in vitro, PKA phosphorylates the cardiac Na ϩ channel ␣ subunit solely on residues contained in the intracellular loop connecting domains I and II.
Using a series of truncated fusion proteins within L I-II we narrowed the phosphorylation sites to the sequence between residues 511 and 530. This region contains eight Ser residues including three (Ser 520 , Ser 526 , and Ser 529 ) in RXS consensus motifs for PKA recognition. Comparison of tryptic phosphopeptide maps of wild type and mutant fusion proteins showed that Ser 529 was required for the generation of phosphopeptides I, II, and IV, whereas Ser 526 was required for the generation of phosphopeptides I, III, and V. A mutant fusion protein (NaFH-S525/6/9A) in which Ser 525 , Ser 526 , and Ser 529 were all mutated to alanine residues was a very poor substrate for PKA, consistent with the conclusion that Ser 526 and Ser 529 are the major sites of phosphorylation. The low level of residual phosphorylation of NaFH-S525/6/9A is not surprising since this fusion protein still contains 17 other Ser residues including four in RXS consensus motifs for PKA recognition. It is likely, therefore, that in the absence of the preferential phosphorylation sites (Ser 526 and Ser 529 ) PKA phosphorylates these other Ser residues to a very low extent. Consistent with this conclusion, the two-dimensional tryptic phosphopeptide map of the low level of phosphopeptides derived from phosphorylation of NaFH-S525/6/9A does not contain spots that coincide with maps of NaFH3 and full-length ␣ subunit (data not shown). Thus, all of our results support the conclusion that Ser 526 and Ser 529 are the principal sites of phosphorylation in the ␣ subunit of the cardiac Na ϩ channel.
Origin of Multiple Phosphopeptides-We detected two major and three minor phosphopeptides in our tryptic two-dimen- FIG. 10. cAMP-dependent phosphorylation of rat cardiac Na ؉ channel ␣ subunit in intact cells. Panel A, SNa-rH1 cells expressing the rat cardiac Na ϩ channel ␣ subunit were incubated in the absence (control, CONT) or presence of the PKA activator DCl-cBIMPS, solubilized in 1% Triton X-100, and the ␣ subunit was isolated by immunoprecipitation as described under "Experimental Procedures." Isolated ␣ subunit was then back-phosphorylated in the presence of PKA and [␥-32 P]ATP and analyzed by SDS-PAGE on a 6% Laemmli gel. 32 P-Labeled ␣ subunit was visualized by autoradiography. Molecular weight markers are represented as M r ϫ 10 Ϫ3 . Panel B, 32 P incorporated into the cardiac Na ϩ channel ␣ subunit was quantitated by liquid scintillation counting; 100% equals 32 P incorporated into the ␣ subunit of untreated cells. Results are the average of three separate experiments Ϯ S.E.
FIG. 11. Two-dimensional tryptic mapping of back-phosphorylated Na ؉ channel ␣ subunit isolated from control and DCl-cBIMPS-treated cells. 32 P-Labeled back-phosphorylated cardiac Na ϩ channel ␣ subunits isolated from control (panel A) and DCl-cBIMPStreated SNa-rH1 cells (panel B) were analyzed by SDS-PAGE on a 6% Laemmli gel, located by autoradiography, and processed for tryptic phosphopeptide mapping as described under "Experimental Procedures." Tryptic phosphopeptides were separated in two dimensions by high voltage electrophoresis (pH 8.9) followed by thin layer chromatography. Arrows designate the direction of electrophoresis (ϩ) and chromatography (C). sional maps which apparently are generated from phosphorylation of only two Ser residues, Ser 526 and Ser 529 . Trypsin cleaves at the carboxyl side of Lys and Arg residues. The amino acid sequence surrounding Ser 526 and Ser 529 contains Arg at positions 533, 544, and 545. Tryptic cleavage may occur at one or more of these positions generating multiple peptides. In addition, the Arg at position 522 is followed by a Pro residue at position 523. Since trypsin does not cleave Arg-Pro bonds efficiently (30), multiple peptides could be generated by incomplete cleavage at this site.
Our data show that mutation of either Ser 525 or Ser 529 caused the loss of phosphopeptide I. This can be explained based on the specificity of trypsin cleavage. Trypsin does not cleave efficiently after Arg residues in the sequence RXS when the serine is phosphorylated (30). If Ser 526 and Ser 529 were both phosphorylated, trypsin would not cleave after Arg 524 or Arg 527 efficiently (Fig. 3B), so it is plausible that phosphopeptide I represents a doubly phosphorylated peptide including Ser 526 and Ser 529 . When Ser 526 is mutated to Ala to prevent phosphorylation, trypsin can cleave after Arg 524 more efficiently. Therefore, phosphopeptides I, III, and IV are absent from tryptic maps of NaFH-S526A. When Ser 529 is mutated to Ala to prevent phosphorylation, trypsin can cleave after Arg 527 more efficiently. Therefore phosphopeptides I, II, and IV are absent from tryptic maps of NaFH-S529A. Phosphopeptides I and II are the major tryptic fragments in the maps of phosphorylated ␣ subunit isolated from mammalian cells and are present in roughly equal intensity. The fact that phosphopeptide V, which is only generated when Ser 526 but not Ser 529 is phosphorylated, is only a minor component in maps of the full-length ␣ subunit suggests that Ser 529 is phosphorylated extensively and therefore the preferential site of phosphorylation of the ␣ subunit by PKA. Studies using synthetic peptides have shown that a hydrophobic residue immediately following the phosphorylated residue is an important determinant of high reactivity of PKA (31,32). Ser 529 is followed immediately by Ile, whereas Ser 526 is followed by an Arg residue. This is consistent with the conclusion that Ser 529 is the preferred substrate for PKA phosphorylation.
Comparison with Physiological Studies-As mentioned above, Schreibmayer et al. (8) mutated two Ser residues in L I-II , as well as several residues elsewhere, and found no effect on regulation of Na ϩ channel function by protein phosphorylation. The sites in L I-II correspond to Ser 485 and Ser 594 in our studies (Fig. 3B). Thus, their results are consistent with ours in that we do not detect cAMP-dependent phosphorylation of either Ser 485 or Ser 594 in vitro or in intact cells in response to cAMP stimulation. In contrast, our results show that the rat cardiac Na ϩ channel ␣ subunit is phosphorylated in vitro and in intact cells in a cAMP-dependent manner on Ser 526 and Ser 529 and therefore suggest that Ser 526 and/or Ser 529 may play a role in the cAMP-dependent regulation of cardiac Na ϩ channel activity. Identification of these phosphorylation sites will allow the design of mutagenesis experiments to determine the precise role of phosphorylation of these amino acid residues in the modulation of cardiac Na ϩ channel activity.