Multiple Loss-of-Function Mechanisms Contribute to SCN5A-Related Familial Sick Sinus Syndrome

Background To identify molecular mechanisms underlying SCN5A-related sick sinus syndrome (SSS), a rare type of SSS, in parallel experiments we elucidated the electrophysiological properties and the cell surface localization of thirteen human Nav1.5 (hNav1.5) mutant channels previously linked to this disease. Methodology/Principal Findings Mutant hNav1.5 channels expressed by HEK293 cells and Xenopus oocytes were investigated by whole-cell patch clamp and two-microelectrode voltage clamp, respectively. HEK293 cell surface biotinylation experiments quantified the fraction of correctly targeted channel proteins. Our data suggested three distinct mutant channel subtypes: Group 1 mutants (L212P, P1298L, DelF1617, R1632H) gave peak current densities and cell surface targeting indistinguishable from wild-type hNav1.5. Loss-of-function of these mutants resulted from altered channel kinetics, including a negative shift of steady-state inactivation and a reduced voltage dependency of open-state inactivation. Group 2 mutants (E161K, T220I, D1275N) gave significantly reduced whole-cell currents due to impaired cell surface localization (D1275N), altered channel properties at unchanged cell surface localization (T220I), or a combination of both (E161K). Group 3 mutant channels were non-functional, due to an almost complete lack of protein at the plasma membrane (T187I, W1421X, K1578fs/52, R1623X) or a probable gating/permeation defect with normal surface localisation (R878C, G1408R). Conclusions/Significance This study indicates that multiple molecular mechanisms, including gating abnormalities, trafficking defects, or a combination of both, are responsible for SCN5A-related familial SSS.


Introduction
Sick sinus syndrome (SSS) was first described nearly 40 years ago by Lown as a complicating arrhythmia following cardioversion [1]. Today SSS (also called sinus node dysfunction, SND) refers to abnormalities in sinus node impulse formation and propagation, including sinus bradycardia, sinus pause/arrest, chronotropic incompetence, and sinoatrial exit block [2]. SSS is frequently associated with conduction system disease in the heart and various supraventricular tachyarrhythmias, such as atrial fibrillation and atrial flutter. When associated with supraventricular tachyarrhythmias, SSS is often termed tachy-brady syndrome [2]. SSS accounts for approximately 50% of the million permanent pacemaker implants per year worldwide [3]. SSS may be associated with underlying cardiac disease conditions including any form of acquired heart disease (e.g. coronary artery disease, cardiomyopathies and valvular heart disease), and occurs following surgical injury. However, SSS most commonly occurs in the elderly in the absence of an apparent accompanying heart disease [4].
Recently, a number of studies have linked genetic defects in ion channels, including human Na v 1.5 (hNa v 1.5), to familial SSS. To date, fourteen SCN5A mutations have been associated with this disease [14,15]. Although SCN5A-related familial SSS is rare, reports have conclusively demonstrated that hNa v 1.5 plays essential roles in the excitation of atrial and ventricular myocytes, and in impulse generation and propagation by the sinus node [14,[16][17][18][19][20]. Clinical data on SCN5A-related SSS are strongly supported by several experimental results. Transcription and expression of Na v 1.5 mRNA and protein respectively in the human peripheral sinus node was demonstrated recently by Chandler et al. [21]. Na + currents were recorded in sinus node pacemaker cells from both mammalian animals [14,22] and humans [23]. Importantly, reduced Na + channel expression in the hearts of heterozygous SCN5A +/2 mice resulted in sinus bradycardia, slowed sinoatrial conduction, and sinoatrial exit block; these phenomena are also observed in SSS patients [24].
Most of these SSS-related mutant channels have been expressed by different heterologous host cells including mammalian cell lines and Xenopus oocytes, and several loss-of-function features have been identified [22]. However, electrophysiological recordings have not yet been combined with biochemistry techniques that enable quantitative estimation of the localization of mutant channel proteins at the plasma membrane. Together, these data would enable discrimination between electrical defects of correctly targeted mutant channels and impaired subcellular localization. Furthermore, none of the mutant channels has been expressed by parallel systems in the same lab. Previous studies, for example results for BrS mutation T1620M [25], suggest that lossof-function properties depend on the expression system chosen. Similarly, minor defects were observed for the familial SSSrelated mutant channel D1275N when expressed by Xenopus oocytes [17]; however, it has been speculated that more severe defects might be observed upon expression by a mammalian cell line [26,27].
In this study, we have investigated the molecular mechanisms underlying loss-of-function of hNa v 1.5 mutant channels associated with familial SSS. The first aim was to investigate both the electrophysiological properties and the plasma membrane localization of thirteen mutant channels in HEK293 cells: This would enable us to discriminate between protein targeting and electrophysiological defects. The second aim was to compare the electrophysiological properties of nine selected mutant channels expressed by both HEK293 cells and Xenopus oocytes: This would identify any dependence of results upon these commonly used expression systems. Our results provide novel insights into the multiplicity of molecular mechanisms underlying SCN5A-related SSS, and hence a better understanding of genotype-phenotype correlations for this rare channelopathy.
A feature common to all group 1 mutants was a significant shift of steady-state inactivation towards hyperpolarized potentials (Table 1, Fig. 2C). This is likely to result in reduced channel availability in patients' cardiomyocytes. The largest shift of the mid-inactivation potential was observed for R1632H (220.7 mV). Shifts of 210.5 mV, 210.6 mV, and 28.9 mV were recorded for L212P, DelF1617, and P1298L respectively. The left-shift of steady-state inactivation was accompanied either by an unchanged steady-state activation (P1298L, R1632H) or by a left-shift of steady-state activation (215.0 mV for L212P, 26.8 mV for DelF1617; Table 1, Fig. 2B). Left-shifts of steady-state activation towards hyperpolarized potentials may partially antagonize loss-offunction caused by reduced channel availability. Recovery from inactivation by group 1 mutants was unchanged (L212P, P1298L), slightly accelerated (DelF1617), or dramatically decelerated (R1632H; Fig. 2D). Full recovery of R1632H at 2120 mV took 5 seconds. This suggests that most R1632H channels are inactivated at physiological membrane potentials and normal heart rates.
When considering inactivation time constants as a function of the membrane voltage ( Fig. 2E and 2F), we observed a second feature common to all group 1 mutants: The voltage dependency  of the inactivation process was reduced, resulting in a faster current decay and hence loss-of-function at less depolarized membrane potentials. In addition, DelF1617 and R1632H displayed inactivation defects at more depolarized membrane voltages and hence a potential gain-of-function mechanism (Fig. 2F). For L212P, lower inactivation time constants at more negative potentials can be at least partially explained by the pronounced negative shift of the mid-activation potential by 215 mV (Fig. 2E).
In contrast to the clear hyperpolarizing shift of steady-state inactivation seen with group 1 mutant channels, availability of group 2 mutants was either unchanged (E161K, D1275N) or only slightly reduced (shift by 25.6 mV for T220I; Table 1, Fig. 3C). Steady-state activation curves were shifted by +19.8 mV and +3.1 mV towards depolarizing potentials for E161K and D1275N, respectively ( Table 1, Fig. 3B). Positive shifts were not observed for group 1 mutants. In addition to the reduced current densities, the positive shifts of steady-state activation can be expected to further impair excitability of cardiomyocytes [28]. Steady-state activation was unchanged for T220I only. Recovery from inactivation for group 2 mutants was unchanged (E161K), slightly decelerated (T220I), or accelerated (D1275N; Table 1, Fig. 3D). We found slightly reduced and unchanged voltage dependency of channel inactivation time constants for T220I and D1275N, respectively. For E161K, slower channel inactivation at potentials from 240 to 0 mV can be at least partially explained by the clear positive shift by nearly 20 mV of the mid-activation potential (Fig. 3B).

Electrophysiological properties of hNa v 1.5 mutant channels in Xenopus oocytes
In some previous reports, differences between electrophysiological properties reported for the same hNa v 1.5 mutant channel were attributed to the use of different expression systems [25].
Compared to Xenopus oocytes, HEK293 or CHO expression systems are often considered more reliable because of their mammalian origin and because of their cultivation at body core temperature. Xenopus oocytes injected with mutant cRNA are incubated for several days at 18uC, which may stabilize mutant channels resulting in a partial restoration of functional surface expression. In order to investigate whether similar electrophysiology results would be obtained from mutant channels expressed by Xenopus oocytes, we selected nine out of the thirteen SSS-related mutations, injected oocytes with wild-type and mutant cRNA, and measured whole-cell currents using the two-microelectrode voltage clamp technique.
For all group 3 mutant channels tested (T187I, R878C, G1408R, W1421X), we were unable to detect any Na + current even after injecting undiluted cRNA, supporting our data from HEK293 cells. Neither the lower incubation temperature nor a high cRNA concentration could restore functional expression. For the other channel variants, classified upon expression by HEK293 cells as group 1 mutants (L212P, P1298L) or group 2 mutants (E161K, T220I, D1275N), we also observed data similar to those obtained from HEK293 cells (Fig. 4, Table 2). Only a few results differed from those obtained from transfected HEK293 cells: First, normalized peak currents obtained from oocytes were significantly smaller for E161K, D1275N and even for P1298L, which was classified as a group 1 mutant according to results from HEK293 cells (Figs. 1 and 4). For E161K, currents were clearly above background but too small to ensure a reliable evaluation of channel kinetics. Second, the mid-inactivation potential for D1275N was shifted towards hyperpolarized potentials only for oocytes, suggesting a contribution of reduced channel availability to the patients' phenotype. Third, recovery from inactivation by L212P was decelerated in oocytes but not in HEK cells, indicating that only the oocyte expression system revealed this potential lossof-function mechanism. Fourth, minor differences between the two expression systems were as follows: a slightly altered slope of the steady-state activation curve for D1275N channels in oocytes only; a slightly altered slope of the steady-state inactivation curve for P1298L channels in HEK293 cells only; and a slight increase of the fast recovery time constant for T220I channels in HEK293 cells (compare Tables 1 and 2).
In conclusion, during comparisons between SSS-related mutant channels and wild-type hNa v 1.5, most of the electrophysiological parameters obtained from oocytes were not different from respective results obtained from HEK293 cells (Tables 1 and 2). Notably, for three out of the four mutant channels that showed expression system-dependent properties, more severe loss-offunction features were associated with oocytes: This suggests that results from oocytes may account better for cardiac excitation abnormalities observed in familial SSS.  Evaluation of the plasma membrane targeting using cell surface biotinylation experiments We performed cell surface biotinylation experiments in conjunction with Western blotting using HEK293 cells transfected with the thirteen familial SSS-related mutants. For each mutant, the fraction of correctly targeted channel to the total cellular channel pool was determined and normalised with the corresponding fraction determined for wild-type from the same blot.
Each calculation was performed in (at least) triplicate to enable the determination of statistically significant differences. Representative Western blots are shown in Fig. 5; normalized cell surface signals are shown alongside peak current densities in Fig. 6.
As expected from our whole-cell current measurements of HEK293 cells, all group 1 mutants were present at the plasma membrane in quantities similar to that for wild-type hNa v 1.5. For group 2 mutants, which generated smaller whole-cell currents ( Fig. 1B), we noticed either reduced (D1275N, E161K) or unchanged (T220I) cell surface signals when compared to wildtype hNa v 1.5 (Fig. 6). For D1275N, the normalized cell surface signal was reduced to 68.0%. Interestingly, this reduction in channel number was in close agreement with the reduction of the peak current density to 66.3%. For E161K, the cell surface signal was reduced to ,70%, but peak current amplitudes were reduced to ,30% (Fig. 6). Consequently, the reduction in current density cannot solely be explained by the reduction in channel number at the plasma membrane. T220I showed no change in cell surface expression despite having a reduction in peak current to 61.6% (Fig. 6). Together, these data suggest that altered single channel properties are responsible for reduced whole-cell currents for both E161K and T220I, a hypothesis that remains to be tested.
For the non-functional group 3 mutants we noticed two distinct channel subtypes. First, normal surface localization was observed for R878C and G1408R, suggesting that these mutations affected channel gating or permeation properties rather than intracellular channel maturation and trafficking. Second, very small fractions of the mutant T187I and the shortened variants W1421X, K1578fs/ 52, and R1632X were detected at the plasma membrane. Consequently, these mutants are largely trafficking-deficient.

Discussion
This study evaluated cell surface localization of thirteen hNa v 1.5 mutant channels previously linked to familial SSS, and correlated these data with the electrophysiological properties of the mutant channels. Whereas several electrophysiological parameters for some of the mutant channels have been reported by other groups, little or no information was available on the contribution of possible maturation/trafficking defects to losses-of-function for the mutants. Hence, this study provides new mechanistic insight into the pathogenesis of SCN5A-related familial SSS.

Loss-of-function mechanisms
This study revealed four distinct loss-of-function mechanisms for the mutant hNa v 1.5 channels (Fig. 7): First, mutants R878C and G1408R were correctly transported to the plasma membrane, but did not form functional channels, suggesting a complete gating/ permeation defect. Our data on R878C and G1408R are in good agreement with previous electrophysiological, GFP-labeling and immunohistochemical studies that demonstrated surface localization of non-functional channels [29,30]. Second, several mutant channels (L212P, T220I, P1298L, DelF1617, and R1632H) with normal presence at the plasma membrane displayed affected kinetics (Tables 1 and 2). For example, R1632H exhibited normal surface localization with peak current densities comparable to those for wildtype hNa v 1.5 (Fig. 1A); however, the strongly reduced steady-state availability in combination with a dramatically decelerated recovery is likely to result in non-functional channels under physiological conditions (Table 1, Fig. 2). This mutation was first reported by Benson et al. who found similar electrophysiological defects in tsA201 cells [16]. Third, some mutant channels (T187I and all truncated variants) were non-functional because of almost completely deficient maturation/trafficking (Figs. 1 and 6). Previous studies on these mutant channels also reported non-functional expression, but did not specify trafficking-deficiencies [16,31]. For the truncated variants, intracellular protein retention was not surprising. Interestingly, we consistently observed a much weaker signal from the total cellular extracts for W1421X (see the example in Fig. 5), suggesting that the shortened channel proteins were subjected to enhanced intracellular degradation. However for T187I, only a single amino acid exchange led to a near-complete intracellular retention, whilst levels of the intracellular T187I pool were comparable to wild-type hNa v 1.5 Figure 4. A comparison of normalized peak current amplitudes for wild-type and mutant hNa v 1.5 channels expressed by two different expression systems. Whole-cell currents were recorded by the patch clamp technique (HEK293 cells) and by the two-microelectrode voltage clamp technique (oocytes) as described in ''Methods''. L212P and T220I exhibited similar relative expression levels in both systems. With E161K, D1275N, and P1298L, whole-cell currents, normalized to hNa v 1.5 values, were significantly smaller for the oocyte system (* indicates p,0.01). Channels classified as group 3 mutants upon expression by HEK293 cells (T187I, R878C, G1408R, W1421X) did not produce detectable Na + inward currents in the oocyte system, even upon injection of undiluted cRNA (data not shown). doi:10.1371/journal.pone.0010985.g004  5 and 6). In a previous study on T187I, functional expression could neither be restored by mexiletine treatment nor by coexpression of the b1 subunit [31]. A screen for known protein motifs in the affected S2-S3 linker in domain I was unsuccessful. We noticed that the distal half of this linker that included threonine 187 is highly conserved among voltage-gated Na + channels. It is possible that this linker region, which is located intracellularly, contains an unknown ER export signal. Fourth, loss-of-function can also be attributed to both a partial trafficking defect and alterations of channel kinetics, as observed for E161K and D1275N (Tables 1 and 2). For E161K, the reduction of the cell surface biotinylation signal was much less pronounced than the reduction of the whole-cell current (Fig. 5), suggesting impaired single channel properties even for correctly targeted mutant channels.

Association of loss-of-function of hNa v 1.5 mutants to familial SSS and to other cardiac excitation disorders
Classifications of hNa v 1.5 mutant channels according to current densities or trafficking properties are less helpful when explaining genotype-phenotype correlations in SCN5A-related familial SSS because such reflections do not account for a possible age-dependent onset of the disease, for other accompanying cardiac rhythm disorders, or for additional risk factors that contribute towards SSS.
In Table 3, we summarize the distinct SCN5A genotypes, the corresponding patients' phenotypes, and major biochemical and electrophysiological properties of the thirteen hNa v 1.5 mutant channels investigated in this study. A strong correlation between a defective hNa v 1.5 channel and familial SSS was found in patients carrying compound heterozygous SCN5A mutations (P1298L/ G1408R, DelF1617/R1632H, and T220I/R1623X) [16]. These patients experienced an early onset of SSS (Table 3), and no other excitation disorders have been reported to date. In each of these three cases, one of the mutants was characterized by mild loss-offunction (P1298L, DelF1617, and T220I), whereas the second mutant, produced from the other allele, can be considered as either severely affected (R1632H) or non-functional (G1408R, R1623X) [16]. This indicates that sinus node function is affected at an early developmental stage when the total Na + current is reduced to less than 50% of its maximal value. The parents of the patients, carrying only one of the six mutations and hence producing a total of at least 50% of the maximum Na + current from the wild-type plus the affected allele, were either asymptomatic or presented with a mild cardiac rhythm disorder (first degree AV block) [16]. Wild-type hNa v 1.5 channel activity plus the residual Na + channel activity of some of the mutants were sufficient to maintain normal cardiac function. However, it must be mentioned that more severe phenotypes were observed in unrelated families carrying the heterozygous mutations G1408R, T220I, or DelF1617 (Table 3) [29,32,33]. Moreover, the heterozygous mutation R1623X caused a combination of BrS and SSS in another patient [31]. In contrast to carriers of compound heterozygous SCN5A mutations, the R1623X carrier was already 65 years old when he experienced recurrent syncope due to sinus arrest. Similarly, several carriers of other mutations (T187I, R878C, W1421X, K1578fs, and E161K) investigated in this study also presented with other cardiac rhythm disorders, such as BrS, CCD, or AV block. Most of these patients were adults when they developed signs of SSS (Table 3). Consequently, a strong Na + current reduction, arising from SCN5A haploinsufficiency, can result in sinus node dysfunction, but the disease phenotype is accompanied by other cardiac excitations disorders, and onset of symptoms occurs at a later ontogenetic stage. Notably, dozens of other severe loss-of-function mutations in SCN5A are known to cause BrS, CCD, or both [28]. However, only a few mutations have been related to sinus node dysfunction, suggesting that as yet unknown factors other than Na + current reduction and aging contribute to the development of SSS in carriers of T187I, R878C, W1421X, K1578fs, and E161K.
The importance of other risk factors in SCN5A-related SSS becomes obvious when considering the striking genotype-phenotype disassociation for L212P and D1275N carriers. Atrial standstill exhibited by the L212P carrier, a three years old boy, occurred only in combination with a connexin 40 (Cx40) polymorphism [34]. The father of the index patient, an L212P carrier lacking this Cx40 variant, exhibited normal sinus rhythm [34]. The normal cardiac phenotype of this L212P carrier is in good agreement with the normal trafficking and the mildly affected electrophysiological properties of L212P channels (Table 3). Similar observations were reported for D1275N [17]: The study by Groenewegen et al. suggested a rare Cx40 polymorphism as a prerequisite for atrial standstill in D1275N carriers [17]. However, D1275N was correlated in unrelated families to sinus node dysfunction, arrhythmia, and ventricular dilation [26], to atrial fibrillation, SSS, CCD, and DCM [32], and to atrial arrhythmia and CCD [27]. The reasons for this spectrum of disease manifestations are unknown, and it is hard to imagine that the minor changes of channel kinetics associated with D1275N are responsible for this variety. Recently, we found that the common SCN5A polymorphism H558R enhanced trafficking of D1275N [35]. Consequently, it is reasonable to assume that several factors, such as polymorphisms in other cardiac genes, gender differences, modulatory proteins of hNa v 1.5 [36], and/or ageing and fibrosis, are important determinants in SCN5A-related channelopathies, including familial SSS.

Limitations of this study
This study possesses a number of limitations. First, caution should be taken when extrapolating the findings from this in vitro Figure 6. Correlation between cell surface expression and peak current densities for the thirteen SSS-associated mutant channels. Normalized cell surface signals were obtained by dividing the ''cell surface'' signal by the corresponding ''total'' signal; both signals being obtained from the same Western blot (see Fig. 5). A ratio obtained for a mutant was further normalized by dividing with the ratio obtained for wild-type hNa v 1.5 again from the same blot. The mean double-normalized cell surface signals shown were calculated from at least three independent biotinylation experiments. * indicates p,0.05, ** indicates p,0.01. doi:10.1371/journal.pone.0010985.g006 study of HEK293 cells and oocytes to the more complex in vivo conditions. A bridge to the human in vivo situation might be provided by genetically modified mouse models that could possibly provide further information. Second, this study was unable to examine other contributing factors, such as the fibrosis, the remodeling of ion channels, and the autonomic dysfunction that occur as a result of ageing. Third, the effects of different b subunits upon the biophysical properties and intracellular trafficking of the Na + channel a subunit have not been explored in this study. Accessory b subunits are well-known modulators of voltage-gated Na + channels, and it is possible that the different b subunits exert distinct effects on the SSS-associated mutant channels investigated in this study.   showing the locations of the thirteen SSS-associated mutations investigated here. Compared to wild-type hNa v 1.5, the mutant channels produced either similar (group 1, white), reduced (group 2, light grey) or no (group 3, dark grey) current. (B) Schematic representation of a sinus node cell and the cellular mechanisms involved in SCN5A-related familial SSS. Early onset of familial SSS occurred (a) when two alleles were affected (compound mutations P1298L/G1408R, DelF1617/R1632H, and T220I/R1623X) [16], or (b) when other risk factors were present (for example, a Cx40 polymorphism combined with L212P; Table 3) [34]. Notably, when wild-type hNa v 1.5 channels are expressed from the normal allele, most of the mutants also caused other cardiac disorders like BrS, AV block, CCD or DCM, as observed for carriers of T187I, W1421X, K1578fs/52, and E161K (see Discussion and Table 3). In most if not all of these cases, SSS onset relatively late. doi:10.1371/journal.pone.0010985.g007  Observed only in HEK293 cells.

c)
Reduced voltage dependency of channel inactivation, resulting in faster current decay at less depolarized potentials and slower decay at more depolarized potentials (Fig. 2F).
R1623X alone caused BrS plus SSS in another unrelated patient [31].

e)
Atrial standstill in this patient was related to a Cx40 polymorphism [34].

Expression in Xenopus laevis oocytes
For in vitro transcription, the coding regions of all constructs were recloned into the pSP64Poly(A) vector (Promega, WI, USA). The preparation of oocytes from Xenopus laevis, in vitro transcription and cRNA injection were performed according to established procedures [37]. For current comparisons, the cRNA concentration for each variant was adjusted to ,0.04 mg/mL (40 and 60 nL per oocyte) by dilution. After 3 days of incubation at 18uC in Barth medium, the whole-cell hNa v 1.5 peak current amplitude was usually between 0.5 and 6.0 mA. Measurements were repeated for at least 5 different batches of oocytes. Whole-oocyte Na + currents were recorded with the two-microelectrode voltage clamp technique using an OC725C amplifier (Warner Instruments, Hamden, CT, USA), as previously described [37]. The glass microelectrodes were filled with 3 mol/L KCl. The microelectrode resistance was between 0.2 and 0.5 MV. The bath solution contained (in mmol/L): NaCl 96, KCl 2, CaCl 2 1.8, HEPES 10, pH 7.2 (KOH).

Pulse protocols and evaluation of electrophysiological data from HEK293 cells and oocytes
Steady-state activation was determined by applying test potentials from 2120 to 50 mV in 5 mV or 10 mV increments at a pulsing frequency of 0.5 Hz. Data were fitted to the Boltzmann equation G Na = [1+exp(V 1/2 2V)/k] 21 where V 1/2 is the half-maximal voltage, k is the slope factor, and the Na + conductance G Na is defined by G Na = I Na,norm /(V2V rev ). Here, I Na,norm is the normalized peak current, and V rev is the reversal potential.
Steady-state inactivation was determined with a double pulse protocol consisting of 500 ms prepulses from the holding potential of 2120mV (or 2140mV for R1632H only) to voltages between 2140 and 220 mV (or 2160 and 220mV for R1632H only), followed by a constant test pulse of 20 ms duration to 220 mV at a pulsing frequency of 0.5 Hz. The amplitude of peak current I Na during the test pulse was normalized to the maximum peak current I Na,max and plotted against the prepulse potential. Data were fitted with the Boltzmann function I Na = I Na,max [1+exp(V 1/2 2 V)/k] 21 , where V is the test potential, and V 1/2 and k are as defined above.
Recovery from inactivation was determined with a double pulse protocol consisting of 500 ms prepulses to 220 mV, followed by variable recovery intervals at 2120 mV, and a constant 20 ms test pulse to 220 mV at a pulsing frequency of 0.25 Hz (or 0.15 Hz for R1632H only). The normalized peak current amplitude (I Na,norm = I Na /I Na,max ) elicited by the test pulse was plotted against the recovery interval. Data were fitted with 2 exponentials: I Na,norm = A f [12exp(2t/t f )]+A s [12exp(2t/t s )], where t is the recovery time interval, t f and t s represent fast and slow time constants respectively, and A f and A s represent fractional amplitudes of fast and slow recovery components, respectively.

Cell surface biotinylation and Western blotting
Cell surface biotinylation was performed as described previously [39]. Briefly, cell membrane proteins were labelled by incubation (4uC, 30min.) with 1 mg/mL EZ-Link TM sulfo-NHS-LC-Biotin (Pierce, IL, USA). Cells were lysed by RIPA buffer (0.5% sodium deoxycholate, 0.1% SDS, 1% Triton X-100, 50 mmol/L Tris, 150 mmol/L NaCl, and 1 mmol/L EDTA) in the presence of cocktail protease inhibitors (Roche, IN, USA). Labeled cellmembrane proteins were precipitated by incubating (4uC, overnight) cell lysate (200mL) with streptavidin-agarose beads (150mL, Amersham NJ, USA). The beads were recovered by centrifugation (15000g, 15min., 4uC), and the supernatant was retained. The beads were washed with RIPA buffer. Biotinylated cell membrane proteins were eluted from the beads into Laemmli buffer by rotation. Both the cell lysate (0.6%) and the purified biotinylated cell surface proteins (50%) were analysed by SDS-PAGE (6% acrylamide) followed by Western blotting as previously described [39,40]. The anti-Na v 1.5 antibodies (Alomone, Jerusalem, Israel) ASC-013 and ASC-005 detected full-length and truncated channels respectively. Antibody against the endoplasmic reticulum protein calnexin was used in simultaneous control experiments to verify that no cytosolic protein had been biotinylated during the procedure. Results are presented as mean 6 standard error of the mean (SEM). Statistical analyses were performed using the oneway ANOVA test with significance being assumed for p,0.05.