Modulation of Cardiac Sodium Channel Gating by Protein Kinase A Can Be Altered by Disease-linked Mutation*

Mutations associated with sodium channel-linked inherited Long-QT syndrome often result in a gain of channel function by disrupting channel inactivation. A small fraction of channels fail to inactivate (burst) at depolarized potentials where normal (wild type) channels fully inactivate. These noninactivating channels give rise to a sustained macroscopic current. We studied the effects of protein kinase A stimulation on sustained current in wild type and three disease-linked C-terminal mutant channels (D1790G, Y1795C, and Y1795H). We show that protein kinase A stimulation differentially affects gating in the mutant channels. Wild type, Y1795C, and Y1795H channels are insensitive to protein kinase A stimulation, whereas “bursting” in the D1790G mutant is markedly enhanced by protein kinase A-de-pendent phosphorylation. Our results suggest that the charge at position 1790 of the C terminus of the channel modulates the response of the cardiac sodium channel to protein kinase A stimulation and that phosphorylation of residue 36 in the N terminus and residue 525 in the cytoplasmic linker joining domains I and II of the channel (cid:1) subunit facilitate destabilization of inactivation and thereby increase sustained current. Voltage-gated Na (cid:1) channels are integral membrane proteins (1, 2) that not only govern cell excitability but may determine the vulnerability of the heart to the development of abnormal rhythms Typically, Na (cid:1) channels open in response to membrane depolarization and rapidly enter an absorbing inactivation Electrophysiology— The membrane currents were measured using whole cell and single channel patch clamp procedures with Axopatch 200B amplifiers and the Pclamp 8 program (Axon Instruments, Foster City, CA). All of the measurements were obtained at room temperature (22 °C). Macroscopic whole cell Na (cid:1) current was recorded using the following solutions. The normal internal solution contained 50 mmol/ liter aspartic acid, 60 mmol/liter CsCl, 5 mmol/liter Na 2 -ATP, 11 mmol/ liter EGTA, 10 mmol/liter HEPES, 1 mmol/liter CaCl 2 , 1 mmol/liter and MgCl 2 , pH 7.4, adjusted with CsOH. In some experiments, the major anion in the pipette solution was replaced by fluoride: 10 mmol/liter NaF, 110 mmol/liter CsF, 20 mmol/liter CsCl, 10 mmol/liter EGTA, 10 mmol/liter HEPES, pH 7.4, adjusted with 25 mmol/liter CsOH. The external solution contained 130 mmol/liter NaCl, 2 mmol/liter CaCl 2 , 5 mmol/liter CsCl, 1.2 mmol/liter MgCl 2 , 10 mmol/liter HEPES, 5 mmol/ liter and glucose, pH 7.4, adjusted with CsOH. Dialysis with intracel- lular F (cid:5) was accompanied by hyperpolarizing shifts in the voltage dependence of inactivation not recorded for the aspartate/Cl (cid:5) intracel- lular solution (26); consequently holding potentials were changed from (cid:5) 100 to (cid:5) 150 mV for aspartate and F (cid:5) pipette solutions. The data were discarded when the series resistance was more than 7 M (cid:6) . Noninacti-vated sustained Na (cid:1) channel current ( I sus ) was determined by subtract- ing background currents measured in the presence of tetrodotoxin (TTX; 30 (cid:3) M , Molecular Probes) from TTX-free records. The amplitude of I sus was measured 150 ms after depolarization to (cid:5) 10 mV to avoid channel re-openings

Mutations associated with sodium channel-linked inherited Long-QT syndrome often result in a gain of channel function by disrupting channel inactivation. A small fraction of channels fail to inactivate (burst) at depolarized potentials where normal (wild type) channels fully inactivate. These noninactivating channels give rise to a sustained macroscopic current. We studied the effects of protein kinase A stimulation on sustained current in wild type and three disease-linked C-terminal mutant channels (D1790G, Y1795C, and Y1795H). We show that protein kinase A stimulation differentially affects gating in the mutant channels. Wild type, Y1795C, and Y1795H channels are insensitive to protein kinase A stimulation, whereas "bursting" in the D1790G mutant is markedly enhanced by protein kinase A-dependent phosphorylation. Our results suggest that the charge at position 1790 of the C terminus of the channel modulates the response of the cardiac sodium channel to protein kinase A stimulation and that phosphorylation of residue 36 in the N terminus and residue 525 in the cytoplasmic linker joining domains I and II of the channel ␣ subunit facilitate destabilization of inactivation and thereby increase sustained current.
Voltage-gated Na ϩ channels are integral membrane proteins (1,2) that not only govern cell excitability but may determine the vulnerability of the heart to the development of abnormal rhythms Typically, Na ϩ channels open in response to membrane depolarization and rapidly enter an absorbing inactivation state (3)(4)(5)(6). However, a small fraction of channels may fail to inactivate, a phenomenon termed "bursting." Bursting channels allow passage of Na ϩ at depolarized potentials, resulting in sustained macroscopic current (I sus ). A number of Na ϩ channel-linked mutations that underlie Long-QT syndrome (LQT-3) have been shown to promote channel bursting (7)(8)(9), and computational analysis has confirmed that this activity underlies action potential prolongation and increased risk of arrhythmia (10). Hence modulation of bursting in general and alteration in modulation by inherited mutations are of considerable interest.
The Na ϩ channel ␣ subunit forms the ion-conducting pore and contains multiple consensus sites for protein kinase A (PKA)-dependent 1 phosphorylation (11)(12)(13)(14)(15)(16). PKA-dependent phosphorylation of the ␣ subunit modulates Na ϩ channel activity in an isoform-and tissue-specific manner (17,18). Furthermore, the C terminus of the Na ϩ channel ␣ subunit modulates gating of brain and heart Na ϩ channels (19) including I sus cardiac Na ϩ channels (20,21). In this study we investigate the possibility that mutations in the heart Na ϩ channel C terminus may affect the functional response of the channel to PKA-dependent phosphorylation. We focused on the effects of PKA on I sus in wild type (WT) and three previously reported C-terminal mutations: D1790G (22), Y1795C (23), and Y1795H (23). We were particularly interested in the response of D1790G channels to PKA stimulation because this mutation has been reported to promote I sus by some investigators (24) but not by others (22), raising the possibility that I sus may be subject to regulation.
We indeed found marked differences in the response of the mutant channels to PKA stimulation. Bursting activity was insensitive to PKA in WT, Y1795C, and Y1795H channels but was significantly enhanced in a PKA-dependent manner in D1790G channels. We extended our inquiry to try to elucidate the mechanism of this differential response of the mutations by carrying out site-directed mutagenesis. Our analysis suggests that PKA-dependent modulation of bursting is enhanced by alteration of the negative charge at residue 1790 and may require the phosphorylation of both Ser 36 and Ser 525 in the channel ␣ subunit. The results support the idea that the C terminus of the Na ϩ channel plays an important role in modulating gating of the channel and suggest that multiple factors are likely to affect the cellular and systems phenotypes of disease-linked mutations of the Na ϩ ␣ subunit.
Electrophysiology-The membrane currents were measured using whole cell and single channel patch clamp procedures with Axopatch 200B amplifiers and the Pclamp 8 program (Axon Instruments, Foster City, CA). All of the measurements were obtained at room temperature (22°C). Macroscopic whole cell Na ϩ current was recorded using the following solutions. The normal internal solution contained 50 mmol/ liter aspartic acid, 60 mmol/liter CsCl, 5 mmol/liter Na 2 -ATP, 11 mmol/ liter EGTA, 10 mmol/liter HEPES, 1 mmol/liter CaCl 2 , 1 mmol/liter and MgCl 2 , pH 7.4, adjusted with CsOH. In some experiments, the major anion in the pipette solution was replaced by fluoride: 10 mmol/liter NaF, 110 mmol/liter CsF, 20 mmol/liter CsCl, 10 mmol/liter EGTA, 10 mmol/liter HEPES, pH 7.4, adjusted with 25 mmol/liter CsOH. The external solution contained 130 mmol/liter NaCl, 2 mmol/liter CaCl 2 , 5 mmol/liter CsCl, 1.2 mmol/liter MgCl 2 , 10 mmol/liter HEPES, 5 mmol/ liter and glucose, pH 7.4, adjusted with CsOH. Dialysis with intracellular F Ϫ was accompanied by hyperpolarizing shifts in the voltage dependence of inactivation not recorded for the aspartate/Cl Ϫ intracellular solution (26); consequently holding potentials were changed from Ϫ100 to Ϫ150 mV for aspartate and F Ϫ pipette solutions. The data were discarded when the series resistance was more than 7 M⍀. Noninactivated sustained Na ϩ channel current (I sus ) was determined by subtracting background currents measured in the presence of tetrodotoxin (TTX; 30 M, Molecular Probes) from TTX-free records. The amplitude of I sus was measured 150 ms after depolarization to Ϫ10 mV to avoid channel re-openings that occur in the voltage range in which activation and inactivation overlap (window current) (25,26). The voltage dependence of channel availability was determined by relative current amplitude after the application of conditioning pulses (500 ms) applied to a series of voltages once every 2 s. Boltzmann relationships were fit to the data using Origin (Microcal, Northhampton, MA) software to extract the voltage of half-maximal inactivation (V1 ⁄2 ) and slope factor (V k ) for this relationship.
For single channel experiments, the pipettes were coated with Sylgard (Dow Chemical Co., Midland, MI) to decrease noise and capacitance of the glass. Electrode resistance was typically 5-7 M⍀ when filled with single channel internal solution (110 mmol/liter NaCl, 10 mmol/liter HEPES, pH adjusted to 7.4). The external solution for single channel recordings was 140 mM KCl, 5 mM HEPES, 10 mM MgCl 2 , pH adjusted to 7.4. After establishing the cell-attached configuration (seal resistance Ͼ 10 G⍀), the membrane was held at a holding potential of Ϫ120 mV. Test pulses (Ϫ20 mV for 100 ms) were applied every 0.5 s. Single channel currents were filtered by a low pass filter in the clamp amplifier with a cut-off frequency of 5 kHz and digitized for storage on computer at a sampling frequency of 20 kHz. Capacity and leak currents were eliminated by digital subtraction of averaged null sweeps.
Bursting activity was defined as repetitive (more than three times) and long lasting (more than 30 ms) channel openings (see Fig. 6). Bursting probability (P b ) for less than 11 channel patches was calculated by the following equation.
where b, t, and n represent the number of non-bursting sweeps, total sweeps (1500 -3000), and channels, respectively. The number of channels included in each patch was estimated by counting the overlap of channel openings. The data are represented as the means Ϯ S.E. The statistical significance between control and treated group was evaluated by the unpaired Student's t test. Statistical significance of treated groups against control was first verified by analysis of variance (for multiple groups) and then evaluated by Dannet's t test; p Ͻ 0.05 was considered statistically significant. Okadaic acid, okadaic acid methylester (Calbiochem, San Diego, CA) and cyclic AMP (Sigma) were dissolved with internal solution and kept at Ϫ20°C. The concentration of the stock solution was 2 M for okadaic acid and okadaic acid methyl ester and 20 mM for cAMP. The protein kinase inhibitor peptide (PKI) was also purchased from Sigma and

RESULTS
To focus on possible changes in channel bursting induced by PKA, we measured currents in response to prolonged (150 ms) pulses to Ϫ10 mV, a voltage that is sufficiently positive to avoid measurement of channel reopening that may occur as a consequence of the overlap steady-state activation and inactivation curves (25,27). In whole cell recordings, bursting channels are reflected in sustained inward (I sus ) current under these conditions. As illustrated in Fig. 1, dialysis of cells with cAMP (200 M) has no effect on I sus recorded in cells expressing WT, Y1795H, or Y1795C channels but markedly increases I sus in cells expressing D1790G channels.
The results summarized in Fig. 1 suggest that cAMP-dependent phosphorylation may selectively increase I sus in D1790G channels. We next tested for activity of endogenous phosphatase activity that might exist and buffer phosphorylation-dependent changes in channel activity. Similar to the results summarized in Fig. 1, we found that dialysis of cells with the nonspecific phosphatase inhibitor okadaic acid (0.2 M) significantly increased I sus of D1790G channels but had no significant effect on I sus recorded in cells expressing WT, Y1795C, or Y1795H channels (Fig. 2). Similarly we found but did not illustrate that I sus recorded in cells expressing two other disease linked mutant channels (⌬KPQ 7 and S1103Y 28 ) was unaffected by okadaic acid dialysis (⌬KPQ, 0.40 Ϯ 0.06% (control, n ϭ 6), 0.37 Ϯ 0.08% (okadaic acid, n ϭ 6, n.s.); S1103Y, 0.09 Ϯ 0.01% (control, n ϭ 5), 0.09 Ϯ 0.02% (okadaic acid, n ϭ 5, n.s.)). Thus, the results of these experiments suggest that okadaic acid and/or dialysis of cells with cAMP can increase I sus specifically in cells expressing D1790G channels.
The effects of okadaic acid on I sus were due to dephosphorylation of a PKA-mediated reaction because 1) dialysis with the inactive methylester of okadaic acid did not have any effect and 2) concomitant dialysis of a PKA inhibitory peptide with okadaic acid ablated okadaic acid-enhanced I sus in cells expressing D1790G channels (Fig. 3). Further evidence supporting the effect of okadaic acid on D1790G channel current is due to inhibition of endogenous phosphatase activity is provided by the data in Fig. 4, which show that dialysis of cells with F Ϫ , which is also a nonspecific phosphatase inhibitor (29 -31) known also to activate G s (32,33), increases I sus in cells expressing D1790G channels (Fig. 4A, D1790G traces). This effect of F Ϫ is inhibited by concomitant dialysis with PKI (Fig.  4B, hatched bar) and is also specific for D1790G versus WT channels (Fig. 4, see wild type traces in A and bars in B).
Augmentation of I sus may occur as a result of a change in Na ϩ channel window current and/or via an increase in the activity of channels in an inactivation-deficient bursting mode (7,25). The voltage protocols used in the experiments summarized above were chosen to minimize possible effects on window currents; however, experiments summarized in Figs. 5 and 6 confirm that the effects of okadaic acid (and/or cAMP) on D1790G channel activity are a consequence of an increase in the bursting frequency of the expressed channels. An increase in sustained current that is caused by bursting should be detected over a broad range of voltages in response to a slow voltage ramp protocol. This is in contrast to window current, which occurs over a narrow voltage range, causing peak of current in response to a voltage ramp as illustrated in Fig. 5 for WT channels. The voltage dependence of WT channel current in response to the voltage ramp is clearly distinct from that of D1790G channels using the same protocol (Fig. 5, lower traces). In the case of D1790G channels, the non-zero current has a broad voltage dependence, and this current is more than doubled by dialysis with okadaic acid (0.2 M) plus cAMP (200 M).
Single channel recordings (Fig. 6) confirm an effect on channel bursting. In these experiments, D1790G channel activity measured in cell-attached patches is altered by exposure of cells to okadaic acid. In these experiments cells were incubated in okadaic acid (2 M) containing solution for 30 min, and then, in the continued presence of okadaic acid (2 M), single channel recordings were carried out. Okadaic acid significantly (p Ͻ 0.05) increased the frequency for which bursting was detected compared with recordings made under identical experimental conditions but in the absence of okadaic acid. These data are summarized as the bar graph in Fig. 6B, which compares bursting frequency in the absence (0.004 Ϯ 0.001%, control, n ϭ 9) and presence (0.011 Ϯ 0.003%, 2 M OA, n ϭ 10) of okadaic acid.
Taken together, these data strongly suggest that D1790G channels are modulated by cAMP-dependent in a manner that increases the frequency of bursting for these channels and that this effect appears specific for D1790G versus WT and all other (four) disease-linked mutant channels we investigated. We carried out additional analysis by mutagenesis to determine 1) whether or not the change in charge that occurs via the D1790G mutation contributes to the cAMP sensitivity of channel bursting and 2) whether or not previously identified consensus PKA phosphorylation sites might be involved in this process.
Residue 1790 falls within a region of the C-terminal tail of the ␣ subunit, which is predicted to be highly structured and enriched with negative charges (20). Because the naturally occurring mutation D1790G neutralizes a negative charge at residue 1790G, we tested the possibility that this alteration in charge may, at least in part, underlie the apparently unique responsiveness of D1790G channels to cAMP-dependent modulation. As such, we replaced the aspartate at this residue by a lysine (D1790K) that, instead of neutralizing the charge at this residue, changes its polarity. As is the case for D1790G channels, cells expressing D1790K channels do not exhibit marked I sus under control conditions (not shown) but share virtually an identical response to okadaic acid with D1790G channels (Fig.   FIG. 6. Okadaic acid promotes bursting of D1790G channels. A, cellattached single channel recordings of D1790G channels in the absence (left) and presence of external OA (2 M). The traces shown in both cases are consecutive records of single channel currents in response to voltage pulses (Ϫ20 mV for 100 ms) applied every 0.5 s from a Ϫ120 mV holding potential. The arrows indicate the initiation and termination of bursts of repetitive channel openings for each condition. B, the frequency of observation of bursting activity was calculated for each condition. The frequency of bursting was significantly increased by okadaic acid treatment (n refers to the number of patches obtained in different cells): control, 0.004 Ϯ 0.001%, n ϭ 9; with OA, 0.011 Ϯ 0.003%, n ϭ 10, *, p Ͻ 0.05. 7). Similar to currents for D1790G channels, D1790K I sus is more than doubled by okadaic acid (0.2 M) dialysis, and the effect is completely inhibited by concomitant dialysis with PKI (20 M) (Fig. 7). Importantly, replacing the aspartate at residue 1790 by another negatively charged amino acid, glutamate, has no effect on the sensitivity of channel bursting to cAMP. In fact both the basal bursting level and the lack of cAMP-dependent regulation of the D1790E channel are remarkably similar to those of the wild type (Asp 1790 ) channel. Thus, it appears that mutation-induced alteration in charge at position 1790 plays an important role in conferring a cAMP sensitivity to bursting of the expressed channels.
Mutagenesis, as summarized in Fig. 8, suggests that the effects of cAMP and okadaic acid on D1790G channels are mediated via PKA phosphorylation of specific sites on the ␣ subunit. Using the D1790G mutant as a backbone, we then used alanine mutations to analyze the roles of three consensus PKA sites, Ser 36 , Ser 525 , and Ser 528 , in the response of D1790G channels to cAMP (200 M) and okadaic acid (0.2 M). The response of D1790G channels to cAMP was not changed when Ser 528 (in the D1790G backbone) was mutated to alanine. Similarly, the individual mutations S36A and S525A in the D1790G backbone reduced but did not significantly alter cAMPdependent D170G channel bursting. However, mutation of both Ser 36 and Ser 525 to alanine within the D1790G backbone significantly reduced the effects of cAMP and odakaic acid on D1790G channel bursting (Fig. 8). In fact there is no significant difference between I sus measured in the presence of okadaic acid and cAMP for the D1790G double mutant (S36A, S525A) channels compared with D1790G channels without these mutations but in the presence of the inhibitory peptide PKI.

DISCUSSION
The results of this study strongly suggest that the D1790G mutation results in an increased sensitivity of channel bursting to PKA-dependent phosphorylation. Furthermore, the effects of cAMP on channel bursting appear to depend critically upon the negative charge at position 1790 of the C-terminal tail of the channel, because alteration of this charge by inherited (D1790G) or experimentally imposed (D1790K) mutation significantly increases PKA-dependent modulation of bursting, whereas replacement of Asp 1790 by a negatively charged glutamate residue has no effect either on basal bursting activity or on the response of the channel to PKA stimulation. Recording conditions that inhibit endogenous phosphatase activity promote bursting of D1790G channels, whereas in the absence of phosphatase inhibitions, bursting of these channels is not significantly different from wild type channels.
PKA-dependent Regulation of I sus Is Modulated by the Cterminal Domain-The Na ϩ channel ␣ subunit expressed in multiple tissues has been reported to be a substrate for both PKA and protein kinase C-dependent phosphorylation (11)(12)(13)(14)(15)(16)34), and various effects have been reported as consequences of phosphorylation of specific sites on the channel. Activation of protein kinase C has been reported to promote bursting of Na v 1.2 channels, but PKA stimulation has been reported previously to have little or no effect on bursting of WT or of ⌬KPQ mutant Na v 1.5 channels (35). Thus, this is the first report of significant PKA-dependent enhancement of Na ϩ channel bursting.
Attention has focused increasingly on a role of the Na ϩ channel C terminus in modulation of channel inactivation. Kinetic and voltage-dependent properties of brain (Na v 1.2) channel gating can be conferred upon heart (Na v 1.5) channels by exchange of C-terminal domains (19). A recent study showed that truncation of a distal region of the C-terminal domain of the heart channel (Na v 1.5) markedly stimulates the channel bursting (20). Our results are consistent with these studies and suggest not only an important role of the C-terminal domain in determining the effects of PKA on channel bursting but that discrete regions are involved in this regulation.
Diversity in Sodium Channel Structure May Contribute to Multiple Effects of PKA-Diversity in the sequences in the I and II cytoplasmic loops of Na v 1.2 (brain) and Na v 1.4 (skeletal muscle Na ϩ channels), which contain multiple PKA consensus phosphorylation sites, has been postulated to contribute distinct functional effects of PKA in these Na ϩ channel isoforms (12, 13, 18, 36 -38). PKA-dependent reduction of Na v 1.2 channel current was ablated or converted into a PKA-dependent increase in current by mutation of PKA consensus serines in the I-II linker. These effects were not seen when the I-II linker was replaced by the Na v 1.4 I-II linker (37). Similarly experiments provide evidence for a key role of the I-II linker in the response of heart (Na v 1.5) channels to cAMP (15).
In this study five putative PKA consensus sites (Thr 17 , Ser 483 , Ser 593 , Thr 979 , and Thr 1028 ) shown not to be involved in PKA-dependent modulation of Na ϩ channel (15) were not investigated. However three other consensus sites, including Ser 525 in the I-II loop, were investigated, and we did find a role for the I-II linker (S525) in the PKA-dependent control of bursting. However, interestingly, this is not the only cytoplasmic segment implicated in our work. We find an equal contribution of an N-terminal serine (Ser 36 ). Furthermore, the functional roles of these loops depend critically on the sequence of the C-terminal domain, suggesting that interactions among multiple cytoplasmic components of the channel contribute to the functional consequences of PKA-dependent modulation of the channel. Interestingly, similar interactions have been predicted previously for cytoplasmic loops and for direct interactions between the C-and N-terminal domains of rat skeletal (rSkM1) sodium channels (39). Clearly our data suggest a similar possibility for Na v 1.5 channels.
Insight from Structural Studies of the Na v 1.5 C-terminal Domain-A recent computational and experimental investigation of the secondary structure of the Na v 1.5 C-terminal domain provided evidence for a proximal region of the C terminus that is highly structured and enriched in negative changes (20). Residue Asp 1790 falls within this region, as does residue Tyr 1795 . More distal in the C-terminal domain is a putative helical positively charged region that, when truncated, results in promotion of channel bursting (20). PKA promotes bursting for Asp 1790 mutants (D1790G and D1790K) but not Tyr 1795 mutant channels (Y1795H and Y1795C), despite that fact that both residues 1790 and 1795 are included in first ␣ helix (H1) Independent mutation of residues 36 and 525 (36A and 525A) reduced mean bursting measured in the presence of OA, but these results were not significant when compared with WT D1790G channels. The single mutation S528A had no significant effect on channel bursting. The double mutation, S36A/S525A (36A525A), significantly reduced bursting compared with WT D1790G (*) and reduced current to levels that were not significantly different from WT D1790G recorded in cells dialyzed with OA plus PKI. The number of experiments for each construct was as follows: WT D1790G, n ϭ 20; WT D1790G plus PKI, n ϭ 10; S525A, n ϭ 14; S36A, n ϭ 12; S36A/S525A, n ϭ 14; S528A, n ϭ 10. *, p Ͻ 0.05, n.s., not significant between with PKI and S36A/S525A double mutation. of the C terminus (20). However, according to the structural model, these residues face opposite directions. Asp 1790 faces extraglobular and Tyr 1795 faces intraglobular regions. This model predicts that Asp 1790 can interact with other unknown proteins or internal loops of the Na ϩ channel ␣ subunit, but Tyr 1795 cannot because of the asymmetry of their orientation. Because deletion of the negative charge at 1790 residue promotes PKA-dependent modulation of modal gating, this negative charge may play an important role in stabilizing the inactivation state through electrostatic interactions with cytoplasmic loops of the channel. In rat skeletal Na ϩ channels, an electrostatic interaction between the N terminus (13-30, negative-rich) and C terminus (1716 -1737, positive-rich) has been suggested (39). The finding that the charge of residue 1790 appears to have a distinct role in PKA-dependent modulation of channel bursting suggests that its orientation or location in the structured proximal portion of the C-terminal domain may confer privileged access to key structures responsible for channel inactivation.
Na ϩ channel inactivation is due to rapid block of the inner mouth of the channel pore by the cytoplasmic linker between domains III and IV that occurs within milliseconds of membrane depolarization (3)(4)(5)(6), and bursting is an inactivationdeficient or unstable state (10). Modulation of bursting, or inactivation, thus suggests modification of these interactions. Our data implicate the N-terminal domain (Ser 36 ), the I-II linker (Ser 525 ), as well as the C-terminal domain (Asp 1790 ) as modulators of this process. Complex interactions between the cytoplasmic loops of the channel, which may be altered either by mutation and or phosphorylation state, provide a broad range of possibilities for the roles of sympathetic-mediated modulation of Na ϩ channels in the heart.
In summary, our data indicate that cAMP-dependent modulation of sodium channel gating can be affected by mutations in regions of the channel, such as the C-terminal domain, that do not directly contain PKA consensus sites and that the activity of endogenous kinases and phosphatases may contribute to channel activity recorded in heterologous expression systems.