We performed whole-cell patch clamp recordings of membrane current (11). In 20 mM external Na⁺, a family of typical Na⁺ currents was elicited (Fig. 1A) by depolarizing voltage pulses from 100 mV to potentials from 80 mV to +60 mV. As shown in the I-V relation (panel B), the currents reach a maximum amplitude at 20 mV and demonstrate a reversal potential near +60 mV. Such currents increased in amplitude in response to 1 µM isoproterenol or 10 µM dibutyryl cyclic AMP (32 ± 5%, n = 9; panel C, ), confirming the well-established response to phosphorylation (10). However, no ionic current was measurable when the external solution was switched to 0 Na⁺/10 mM Ca²⁺ (1 ± 2% of basal current, n = 7; panel C, ). That Na⁺ channels were still present and functional was verified by restoring Na⁺ to the external solution (panel C, ). Figure 1D shows the time course of the phosphorylation-mediated increase in peak I_Na and flux of Na⁺ () versus Ca²⁺ () through the modified rH1 Na channels. Similarly, there was no measureable Ca²⁺ flux through phosphorylated hH1 channels in 0 Na⁺/5 Ca²⁺ solution (Fig. 1E). Results were comparable with either the human or rat channels, excluding a possible species-specific response.

These experiments relied on expression of the  subunit alone. Such a test was motivated by the fact that, in rat myocardium, Na⁺ channels consist only of  subunits (12), making it unlikely that  subunits somehow contributed to the original observations of slip mode conductance in rat heart cells. Nevertheless, to exclude the possibility that  subunit coexpression is required, we performed further experiments in CHO cells that coexpress the  and 1 subunits (13, 14). In these experiments we quantified both major manifestations of ion selectivity: reversal potential (E_rev), and ion flux. These two reflections of selectivity are complementary. The reversal potential in solutions of mixed ion composition yields relative ion permeabilities (4), but E_rev measurements can be problematic methodologically because small changes, resulting from junction potential drift for example, are difficult to exclude. For Na⁺ currents, the very nature of the measurements necessitates that small inward and outward currents be quantified reliably, which is difficult to do in large cells (for example, cardiac myocytes) with a variety of ionic conductances. In principle, the use of TTX subtraction may help in distinguishing among various currents, but it does not prevent problems resulting from cumulative junction potential drift or endogenous TTX-sensitive Na⁺ currents. The alternative approach, that of measuring ion flux directly (as in Fig. 1), has various advantages. First, it is model-independent: if net current is carried by a given ion, then that ion must be permeant. Secondly, net flux is the parameter of physiological importance. In the case of slip mode, Ca²⁺ flux through Na⁺ channels is postulated to suffice to trigger excitation-contraction coupling (5). If that is the case, a sizable Ca⁺ current should be measurable through Na⁺ channels: there is no reason to rely solely on shifts of E_rev.

We took membrane current recordings from a representative CHO cell that coexpressed  + 1 subunits (Fig. 2A). The Na⁺ equilibrium potential was set to 0 mV by including 20 mM [Na⁺] in both the internal and external solutions (15). Membrane current was first recorded at baseline in the absence of cyclic AMP, but with 2 mM external [Ca²⁺]. The currents reversed at 0 mV (). The addition of 50 µM dibutyryl cAMP increased both inward and outward Na current, as expected from a simple increase in Na channel open probability. E_rev did not change despite the continuing presence of Ca²⁺ in the external solution (16) (). Subsequent removal of external Ca²⁺ increased the amplitude of the Na currents at negative potentials; this effect is expected from the known voltage-dependent block of Na channels by external Ca²⁺ (17), but is in the opposite direction to the change that would have been expected had the channels been permeable by Ca²⁺. Once again, the current reversed at 0 mV (). The single experiment shown in Fig. 2, A to C, was representative of five cells, whose mean current-voltage relations are shown (B). On removal of Ca²⁺ in the continued presence of external Na⁺, E_rev did not change (16). This stability differs from the shift of 5.1 mV (18), which would have been seen had the relative Ca²⁺/Na⁺ permeabilities equaled 1.25, as stated in the report (5). Inward currents disappeared (Fig. 2B) when the cells were bathed with an external solution containing 2 mM Ca²⁺ but no Na⁺ (). This observation further confirms the absence of an appreciable calcium conductance through Na channels. Results were indistinguishable between hH1 channels coassembled with human heart (h1) or rat brain (r1) 1 subunits. Confirmation that r1 in fact expresses functional subunits was derived from parallel experiments in which the same 1 cDNA increased current amplitude and shifted inactivation when coexpressed with Na channel  subunits (19). Because no E_rev shift was observed in mixed Na⁺/Ca²⁺ solutions, there is no basis for the notion that "slip mode" requires the modulatory presence of external Na⁺ ions.

Our results contradict findings in rat ventricular myocytes (5) and in HEK cells (20). We have used CHO cells, which are known to support cyclic AMP-mediated responses and which contain few endogenous conductances. The first reports of cyclic AMP-dependent upregulation of Na currents were from expression studies in CHO cells (10). HEK cells have the virtue of being readily transfected (21), but they do not consistently support cyclic AMP-dependent responses; several lines of evidence indicate that protein kinase activity is so high in the basal condition that kinase inhibitors must be added to reveal directionally appropriate responses (22). HEK cells also contain a variety of endogenous conductances that may interfere with the quantification of Na⁺ current reversal potentials. These include an endogenous TTX-sensitive voltage-dependent Na⁺ channel (21) and a dihydropyridine-sensitive Ca⁺ channel (23). The Ca⁺ channel literature raises the caution that endogenous channels may interact with exogenously expressed  subunits in an unanticipated manner, clouding the interpretation of multisubunit expression studies (24). HEK cells also possess a variety of endogenous K⁺ and Cl conductances (25), which may complicate attempts to measure small differences in Na⁺ current reversal potential.

We have not observed Na⁺ currents in nontransfected CHO cells, but such cells have occasionally been reported to express small endogenous TTX-sensitive Na⁺ currents (26, 27). To verify that our findings reflect the behavior of TTX-resistant cardiac Na⁺ channels, we measured I_Na in hH1 + 1 + GFP transfected CHO cells under drug-free conditions and in the presence of low- and high-dose TTX (n = 7). The lower concentration of 100 nM would be expected to block >95% of TTX-sensitive channels, but only suppressed the current by 10.7 ± 1.4% (n = 7). In contrast, a higher concentration appropriate to inhibit cardiac channels (5 µM TTX) blocked 62.1 ± 2.9% (n = 7) of the current in agreement with Krafte et al. (27). Our results indicate that at least 90% of current in transfected cells arises from the TTX-resistant hH1 (cardiac) Na⁺ current. This conclusion is further bolstered by the observation that nongreen cells in the same dishes (n = 2) and CHO cells transfected with GFP alone (n = 4) had no observable Na⁺ currents under identical recording conditions.

In summary, our experiments show that cardiac Na⁺ channels are up-regulated by cyclic AMP, but that such up-regulation is not accompanied by changes in their Na⁺/Ca²⁺ selectivity. Na⁺ channels are not measurably permeant to Ca²⁺ either in the basal state or after cyclic AMP-dependent stimulation. Our experiments were designed to investigate the molecular basis of the reported "slip mode"; neither the cardiac  subunit alone nor  + 1 subunits suffice to confer such a phenomenon. Furthermore, cyclic AMP-dependent upregulation occurs without an associated change in selectivity, which implies that "slip mode" does not reflect a direct consequence of phosphorylation of either subunit. For these reasons, we conclude that the reports of "slip-mode conductance" represent a technical artifact, possibly arising from suboptimal voltage control.

H. Bradley Nuss
Eduardo Marbán
Section of Molecular and
Cellular Cardiology,
Department of Medicine,
Johns Hopkins University
School of Medicine,
Baltimore, MD 21205, USA
E-mail: marban{at}jhmi.edu

REFERENCES AND NOTES

1. A. L. Hodgkin and A. F. Huxley, J. Physiol. 117, 500 (1952) .
2. L. J. Mullins, J. Gen. Physiol. 42, 817 (1959) [Abstract/Free Full Text] .
3. C. L. Armstrong in Membrane Transport, D. C. Tosteson, Ed., (Waverly, Baltimore, MD, 1989), pp. 261-273; B. Sakmann and E. Neher, Single-Channel Recording, (Plenum Press, New York, ed. 2, 1995). A notable exception is the apparent alteration of monovalent cation selectivity in certain K⁺ channels during slow inactivation, as observed by L. Kiss, J. LoTurco, and S. J. Korn [Biophys. J. 76, 253 (1999)].
4. B. Hille, Ionic Channels of Excitable Membranes (Sinauer, Sunderland, ed. 2, 1992).
5. L. F. Santana, A. M. Gómez, W. J. Lederer, Science 279, 1027 (1998) [Abstract/Free Full Text] .
6. Y. M. Shuba, B. Hesslinger, W. Trautwein, T. F. McDonald, D. Pelzer, Pflugers Arch. 415, 767 (1990) [ISI] [Medline] ; H. A. Fozzard, D. A. Hanck, J. C. Makielski, B. E. Scanley, M. F. Sheets, Experientia 43, 1162 (1987) [CrossRef] [ISI] [Medline] .
7. D. M. Bers, Excitation-Contraction Coupling and Cardiac Contractile Force, (Kluwer, Boston, 1993).
8. M. E. Gellens, et al., Proc. Natl. Acad. Sci. U.S.A. 89, 554 (1992) [Abstract/Free Full Text] .
9. R. B. Rogart, L. L. Cribbs, L. K. Muglia, D. D. Kephart, M. W. Kaiser, ibid. 86, 8170 (1989) [Abstract/Free Full Text].
10. A. Yoshida, M. Takahashi, S. Nishimura, H. Takeshima, S. Kokubun, FEBS Lett. 309, 343 (1992) [CrossRef] [ISI] [Medline] . M. Li, et al., Science 261, 1439 (1993) [Abstract/Free Full Text] .
11. Na⁺ channel  subunit cDNA was subcloned into a mammalian expression vector (GWIH) and co-transfected into CHO-K1 cells with GPF (pGreen Lantern-1, Gibco-BRL). The extracellular recording solution for the Ca²⁺ flux experiments was comprised of (in mM): 125 CsCl, 1 MgCl₂, 5 dextrose, 10 Hepes, pH 7.3 with CsOH, and 10 NaCl or 5 CaCl₂. Cells were dialyzed with an intracellular (pipette) solution containing (in mM): 135 CsCl, 10 TEACl, 10 Hepes, 5 EGTA, 4 MgATP, 0.33 MgCl₂, pH 7.3 with CsOH.
12. J. S. Yang, P. B. Bennett, N. Makita, A. L. George, R. L. Barchi, Neuron 11, 915 (1993) [CrossRef] [ISI] [Medline] ; S. A. Cohen and L. K. Levitt, Circ. Research 73, 735 (1993) [Abstract/Free Full Text].
13. The only accessory subunit expressed in cardiac muscle is 1 (14).
14. E. Marban, T. Yamagishi, G. F. Tomaselli, J. Physiol. 508.3, 647 (1998) .
15. Human and rat Na⁺ channel 1 subunit cDNA were subcloned into a mammalian expression vector (pGFPIrs) for bicistronic expression of the 1 subunit and GFP reporter [ L. L. Isom, et al., Science 256, 839 (1992) [Abstract/Free Full Text] ]; N. Makita, K. Sloan-Brown, D. O. Weghuis, H. H. Ropers, A. L. George, Genomics 23, 628 (1994) [CrossRef] [ISI] [Medline] . D. C. Johns, H. B. Nuss, E. Marban, J. Biol. Chem. 272, 31598 (1997) [Abstract/Free Full Text] . Coexpression of  and 1 Na⁺ channel subunits was accomplished by cotransfecting equal masses of each plasmid into CHO-K1 cells, which provided a molar excess of 1 subunit cDNA due to the differences in length of coding sequences. The control extracellular recording solution was comprised of (in mM): 125 CsCl, 20 NaCl, 2 CaCl₂, 1 MgCl₂, 5 dextrose, 10 HEPES, pH 7.3 with CsOH (). Dibutyryl-cAMP (50 µM) was present in all solutions except the initial control solution (, , and  in the figures). MgCl₂ was increased to 3 mM when CaCl₂ was removed from the bath () CsCl was increased to 145 mM to maintain the osmolarity of the 0 Na⁺ recording solution (). Cells were dialyzed with an intracellular (pipette) solution containing (in mM): 125 CsCl, 20 NaCl, 10 HEPES, 5 EGTA, 4 MgATP, 0.33 MgCl₂, pH 7.3 with CsOH.
16. Reversal potentials were determined by performing a linear regression of the ascending (linear) portion of each individual I-V and averaging the individual values of E_rev for each solution. The experimentally determined E_rev values equal: 1.3 ± 1.0 mV, n = 5 in 20 Na⁺, 2 Ca²⁺ (); 0.1 ± 0.6 mV, n = 5 in 20 Na⁺, 2 Ca²⁺, 50 µM db-cAMP (); 0.6 ± 0.5 mV, n = 5 in 20 Na⁺, 0 Ca²⁺, 50 µM db-cAMP (); 37.8 ± 1.6 mV, n = 4 in 0 Na⁺, 2 Ca²⁺, 50 µM db-cAMP (). Junction potential offsets for the external recording solutions measured with respect to the internal pipette solution were consistently between +0.8 to +1.2 mV. The calculated junction potentials for all four solutions at 22°C equals +0.7 mV with respect to the pipette solution (AxoScope, Axon Instruments, city, state). When the experimentally determined E_rev values (Fig. 2) are corrected for the measured junction potential offsets, the results equal: +0.1 mV for 20 Na⁺, 2 Ca²⁺ (); 1.3 mV for 20 Na⁺, 2 Ca²⁺, 50 µM db-cAMP (); 1.6 mV for 20 Na⁺, 0 Ca²⁺, 50 µM db-cAMP (); and 37.0 mV for 0 Na⁺, 2 Ca²⁺, 50 µM db-cAMP ().
17. M. F. Sheets, B. E. Scanley, D. A. Hanck, J. C. Makielski, H. A. Fozzard, Biophys. J. 52, 13 (1987) [Abstract/Free Full Text] .
18. A predicted shift in the reversal potential of 5.1 mV was calculated with the use of an explicit solution of E_rev for a channel conducting both divalent and monvalent ions using P_Na = 1 and P_Ca = 1.25. The equation was derived from the Goldman-Hodgkin-Katz constant field equation by D. L. Campbell, W. R. Giles, J. R. Hume, D. Noble, and E. F. Shibata [J. Physiol. 403, 267 (1988)].
19. r1 cDNA was cotransfected with the rat skeletal muscle µ1  subunit cDNA in HEK cells. The midpoint of steady-state inactivation measured per J. R. Balser, H. B. Nuss, D. Romashko, E. Marbán, and G. F. Tomaselli [J. Gen. Physiol. 107, 643 (1996)] shifted from 66.9 ± 1.3 mV (n = 3) to 59.8 ± 0.2 mV (N = 3, p < 0.01), in agreement with S. N. Wright, S.-Y. Wang, Y.-F. Xiao, and G. K. Wang [Biophys. J. 76, 233 (1999)]. Current density also increased with 1 coexpression consistent with H. B. Nuss, N. Chiamvimonvat, M. T. Pérez-García, G. F. Tomaselli, and E. Marbán [J. Gen. Physiol. 106, 1171 (1995)]; and T. Yamagishi, G. F. Tomaselli, and E. Marbán, unpublished data.
20. J. S. Cruz, L. F. Santana, W. J. Lederer, Biophys. J. 76, A196 (1999) .
21. C. Ukomadu, J. Zhou, F. J. Sigworth, W. S. Agnew, Neuron 8, 663 (1992) [CrossRef] [ISI] [Medline] .
22. E. Perez-Reves, W. Yuan, X. Wei, D. M. Bers, FEBS Lett. 342, 119 (1994) [CrossRef] [ISI] [Medline] ; K. W. Roche, R. J. O'Brien, A. L. Mammen, J. Bernhardt, R. L. Huganir, Neuron 16, 1179 (1996) [CrossRef] [ISI] [Medline] .
23. S. Berjukow, et al., Br. J. Pharmacol. 118, 748 (1996) [ISI] [Medline] .
24. A. E. Lacerda, E. Perez-Reyes, X. Wei, A. Castellano, A. M. Brown, Biophys. J. 66, 1833 (1994) [Abstract/Free Full Text] .
25. G. Zhu, Y. Zhang, H. Xu, C. Jiang, J. Neurosci. Methods 81, 73 (1998) [CrossRef] [ISI] [Medline] ; M. D. Ashen, B. O'Rourke, K. A. Kluge, D. C. Johns, G. F. Tomaselli, Am. J. Physiol. 268, H506 (1995) [Abstract/Free Full Text] .
26. P. H. Lalik, D. S. Krafte, W. A. Volberg, R. B. Ciccarelli, Am. J. Physiol. 264, C803 (1993) [Abstract/Free Full Text] ; R. Skryma, N. Prevarskaya, P. Vacher, B. Dufy, ibid. 267, C544 (1994) [Abstract/Free Full Text].
27. D. S. Krafte, et al., J. Mol. Cell. Cardiol. 27, 823 (1995) [CrossRef] [ISI] [Medline] . Krafte et al. found that TTX blocked hH1 Na⁺ channels stably expressed in CHO-K1 cells with an IC50 of 2.5 µM.
28. Supported by National Institutes of Health grant R01-HL-52768 (EM) and an American Heart Association Scientist Development Grant (HBN). We thank W. Catterall for providing the rat cardiac Na channel  subunit cDNA.

Santana et al. (1) observed a tetrodotoxin (TTX)-blockable Ca²⁺ current in rat ventricular myocytes after several pharmacological treatments [cyclic AMP, isoproterenol (ISO), or the cardiotonic steroids, ouabain and digoxin]. This TTX-blockable Ca²⁺ current was attributed to a Ca²⁺ permeability induced in classical cardiac Na⁺ channels that normally express little or none. Santana et al. described such treated Na⁺ channels as "promiscuous" (1). In the absence of any of these agents, we previously described (2) a TTX-blockable Ca²⁺ current (I_Ca(TTX)), also seen in rat ventricular myocytes (3). I_Ca(TTX) is generated by Na⁺ channels that are functionally distinct from those generating the classical cardiac Na⁺ current (I_Na). The question arises, then, of whether the induced Ca²⁺ current observed by Santana et al. might be identified with I_Ca(TTX). Possibly, protein kinase A (PKA)-dependent phosphorylation of classical Na⁺ channels (or whatever the nature of the modification produced by cardiotonic steroids) converts them to I_Ca(TTX) channels, and I_Ca(TTX) recorded in the absence of these agents arises from some basal pool of modified classical Na⁺ channels.

I_Ca(TTX) channels are unlikely to arise from PKA-dependent conversion of classical Na⁺ channels, because a substantial increase in I_Ca(TTX) produced by ISO was not accompanied by a reduction in I_Na, as would be required. Moreover, we find all of the TTX-blockable Ca²⁺ current seen in the presence of ISO flows through I_Ca(TTX) and not through classical Na⁺ channels. The enhanced TTX-blockable Ca²⁺ current seen in ISO must be attributed to an increase in I_Ca(TTX) and not to the induction of Ca²⁺ permeability in classical Na⁺ channels.

Because the inactivation kinetics of I_Ca(TTX) are slower than those of I_Na (2), the conversion of any significant fraction of classical Na⁺ channels into I_Ca(TTX) channels must necessarily slow the inactivation time course of the TTX-blockable current. In 1 mm external Na⁺ (Na⁺_O) plus 0.5 mm external Ca²⁺ (Ca²⁺_O), 1 µM ISO substantially increased the total TTX blockable current (Fig. 1A; mean increase: 73.2 ± 8.7% SEM; 19 determinations from four cells). This increase occured without a detectable change in the current time course (as shown by the scaled and superimposed traces of Fig. 1A, top row). ISO did not produce a significant change in the inactivation time constant (_h; single exponential fit; Fig. 1B, upper curve). The mean ratio of _h,ISO to _h,control was 0.988 (range: 0.826 to 1.208). These results are consistent with either a roughly parallel increase in the amplitudes of I_Na and I_Ca(TTX) or with an I_Ca(TTX) component that remains small relative to I_Na, but not with the simple conversion of I_Na into I_Ca(TTX) channels, because _h is slower for I_Ca(TTX). The mean values of _h for I_Ca(TTX) relative to that for I_Na ranged from 2.90-fold slower at 50 mV to 1.70-fold slower at 30 mV (2). However, the inactivation time course was unchanged by ISO, indicating that there was no detectable conversion of classical Na⁺ channels into I_Ca(TTX) channels. Therefore, it is unlikely that I_Ca(TTX) channels are just phosphorylated I_Na channels. Moreover, nearly all of the increase in current must be of I_Na, because I_Na at these potentials is considerably larger than I_Ca(TTX) (2).

ISO also increased I_Ca(TTX). In 1 mM Ca²⁺₀ and 0 Na⁺₀, 1 µM ISO increased the (TTX subtracted) current by 83.6 ± 10% SEM. (13 determinations from five cells), again without a change in current time course (Fig. 1A, scaled and superimposed traces, bottom row; different cell from top row). One cell (Fig. 1A, bottom row) expressed a particularly large amplitude I_Ca(TTX) component. The mean ratio of _h,ISO to _h,control in Ca²⁺₀ only was 0.913 (range: 0.712 to 1.250; Fig. 1B, bottom curve). In the absence of ISO, all the TTX-blockable Ca²⁺ current is generated by a single population of channels, I_Ca(TTX). The activation curve for the TTX-blockable current is well described by a single Boltzmann function in Ca²⁺₀ only, but requires the sum of two Boltzmann functions for a good description in Na⁺₀ plus Ca²⁺₀ (2). Similarly, the TTX-blockable current inactivates with a single exponential time course in Ca²⁺₀ only, but with two exponentials in Na⁺₀ plus Ca²⁺₀ (2). These data indicate that, in the presence of ISO, the TTX-blockable Ca²⁺ current is still generated by a single population of channels because a nearly twofold increase in current amplitude is not accompanied by any change in the inactivation time course. Under our experimental conditions (TTX subtracted currents; no more than 1 mM Na⁺₀), the increased current in ISO seen in Na⁺₀ plus Ca²⁺₀ must represent primarily increased Na⁺ current through classical Na⁺ channels and not Ca²⁺ current. The increase in TTX-blockable Ca²⁺ current reported by Santana et al. (1) may simply reflect an increase in I_Ca(TTX). They apparently did not attempt to identify which channel type generated the TTX-blockable Ca²⁺ current.

ISO substantially increases the classical I_Na without a change in its kinetics (Fig. 1A, upper traces). Thus, if ISO induces any appreciable Ca²⁺ permeability in classical Na⁺ channels, this slippage should be evidenced by the appearance of a faster inactivating component in the TTX-blockable Ca²⁺ current, but none was detected (Fig. 1A, lower traces). Therefore, our experiments are not consistent with a change in selectivity of classical Na⁺ channels induced by conditions that promote channel phosphorylation, but are in agreement with findings in guinea pig, rabbit, and canine ventricular myocytes, as well as in rat cardiac Na⁺ (SkM2) channels expressed heterologously in frog oocytes (4).

C. William Balke
Larry Goldman
Rajesh Aggarwal
Stephen R. Shorofsky
Departments of Physiology and Medicine,
School of Medicine,
University of Maryland, Baltimore,
Baltimore, MD 21201, USA

REFERENCES AND NOTES

1. L. F. Santana, A. M. Gómez, W. J. Lederer, Science 279, 1027 (1998) .
2. R. Aggarwal, S. R. Shorofsky, L. Goldman, C. W. Balke, J. Physiol. 505, 353 (1997) [CrossRef] [ISI] [Medline].
3. Isolated rat ventricular myocytes were obtained by a standard enzymatic dispersion technique described previously [ C. W. Balke and W. G. Wier, Proc. Natl. Acad. Sci. U.S.A. 89, 4417 (1992) [Abstract/Free Full Text] ]. Membrane currents were recorded using the whole-cell configuration of the patch clamp technique. Patch pipettes were as described previously (2). Series resistance compensation was used in all experiments. Currents were filtered at 2 kHz and digitized at 20 kHz. The external solution was composed of 140 mM tetraethylammonium chloride (TEA-Cl), 10 mM glucose, 10 mM Hepes, 10 mM CsCl, 1 mM MgCl₂ (pH adjusted to 7.3 with CsOH) and either 1 mM Na⁺ plus 0.5 mM Ca²⁺ or 1 mM Ca²⁺ only as noted. La³⁺ (10 µM) was included in all external solutions to suppress L-type Ca²⁺ currents. When indicated, isoproterenol (1 µM plus 10 µM EDTA to prevent oxidation) was added to this solution. The electrode filling solution contained 120 mM glutamic acid, 120 mM CsOH, 10 mM Hepes, 0.33 mM MgCl₂, 20 mM TEA-Cl, 4 mM adenosine 5'-triphosphate (Mg salt), and 5 mM EGTA (pH adjusted to 7.3 with CsOH). Temperature was 21 to 23°C. The holding potential was 100 mV. Depolarizing pulses varied between 60 and 25 mV, in increments of 5 mV. Analyses were performed on records subtracted with the use of 10 µM TTX.
4. K. Ono, T. Kiyosue, M. Arita, Am. J. Physiol. 256, C1131 (1989) [Abstract/Free Full Text] ; J. J. Matsuda, H. Lee, E. F. Shibata, Circ. Res. 70, 199 (1992) [Abstract/Free Full Text] ; K. Ono, H. A. Fozzard, D. A. Hanck, Pflügers Arch. 429, 561 (1995) [CrossRef] [ISI] [Medline]; W. Schreibmayer, et al., Receptors Channels 2, 339 (1994) [ISI] [Medline] . The studies using canine ventricular cells (Ono et al., 1995) were about the effects of forskolin. However, these forskolin effects are apparently not mediated through protein kinase A phosphorylation.
5. Animals used in the present study were maintained in accordance with the guidelines of the Institutional Animal Care and Use Committee of the University of Maryland School of Medicine and the Guide for the Care and Use of Laboratory Animals [Department of Health and Human Services publication No. (NIH) 85-23, revised 1985].

Response: The TTX-sensitive Na⁺ channel is selective for Na⁺ over other monovalent and divalent cations (1-5). Recently, however, we reported that the ion selectivity of cardiac Na⁺ channels could be dynamically modulated (6). Following phosphorylation by protein kinase A (PKA), the Ca²⁺ permeability of the Na⁺ channel (P_Ca) could increase relative to Na⁺ permeability (P_Na) so that the permeability ratio (P_Ca/P_Na) was greater than 1. Called "slip-mode conductance" of the Na⁺ channel, this behavior was shown to be functionally important for heart cells when it was activated. First, it was shown that Ca²⁺ influx through Na⁺ channels alone could trigger Ca²⁺-induced Ca²⁺-release (CICR); activating Ca²⁺ sparks and small [Ca²⁺]_i transients. Second, it was demonstrated that it was possible to evaluate the contributions of Ca²⁺ influx through Na⁺ channels under near-physiological conditions using action potential "clamp" experiments. These results demonstrated that I_Na could activate measurable Ca²⁺ influx, Ca²⁺ sparks or [Ca²⁺]_i transients only when slip-mode conductance was activated. We also carried out quantitative examination of I_Na in heart cells under conditions that made it possible to measure this current. Low [Na⁺] at cool temperatures kept I_Na to values that could be reliably measured (1 to 4 nA). With the use of this approach, we measured large and unambiguous positive shifts in the I_Na reversal potential, E_rev, of 9 mV to 10 mV, when slip-mode conductance was activated. This shift in E_rev suggested that an increase of P_Ca/P_Na from a very low value (0.0 to 0.1) to large values (1.2 to 1.4) occurred when slip-mode conductance was activated. In their comments, neither Nuss and Marbán nor Balke et al. have repeated these experiments in heart cells under conditions identical to those of our report (6). Consequently their comments do not directly address our report (6) and must be interpreted in light of the assumptions they have made. Balke et al. discuss a phenomenon that appears to be unrelated to slip-mode conductance of the Na⁺ channel. Nuss and Marbán, however, did raise a question that we too have been examining for the past year: Can slip-mode conductance of the cardiac Na⁺ channel be observed in an heterologous expression system? The experiments presented below, in contrast to those of Nuss and Marbán, show that Ca²⁺ can permeate Na⁺ channels in an heterologous expression system and thus provide strong evidence in support of our original findings and hypotheses (6).

If slip-mode conductance of the cardiac Na⁺ channel only relies on appropriate phosphorylation of the cardiac Na⁺ channel and the presence of Na⁺ channels, then it should be possible to reproduce the changes in I_Na and in Ca²⁺ flux that we have seen in heart using an heterologous expression system. A key factor in any such experiment is setting up the conditions in the correct manner. We chose to use HEK293 cells because of the low level of voltage-gated channels expressed in these cells (7). However, we did not know a priori what part of the heterotrimeric Na⁺ channel was responsible for slip-mode conductance. Although virtually all features of cardiac Na⁺ channels have been attributed to the  subunit and can be observed when this subunit of the Na⁺ channel is expressed in an heterologous system (2, 8), there are two other Na⁺ channel subunits, ₁ and ₂. Because of the importance of the  subunit, we examined it in HEK293 cells first. I_Na occurred when only the  subunit of the human isoform of the cardiac Na⁺ channel (hH1) was expressed in HEK293 cells (Fig. 1A). If slip-mode conductance of the Na⁺ channel could be produced when only the  subunit was expressed, then a shift of E_rev would be expected when PKA is activated. However under all conditions tested, E_rev remained at the Na⁺ equilibrium potential (E_Na), which was 0 mV under control conditions and 0.63 mV in Na⁺- dbcAMP (10). A small increase in I_Na magnitude following the addition of dbcAMP is consistent with published results (9), as is the block of I_Na by TTX (10 µM).

Under the ionic conditions of these experiments, an increase in P_Ca/P_Na would have led to inward current at E_Na and a positive shift in E_rev (10). Because neither was observed, we deduced that there was no significant change in P_Ca/P_Na (10), a finding similar to that observed by Grant et al. (11). How can these findings (Fig. 1A) be reconciled with the finding that Na⁺ channels in rat heart became permeable to Ca²⁺ following the activation of PKA (2)? Could our earlier results with rat heart cells (6) simply be wrong? That conclusion seemed unlikely, because multiple investigative methods were used to demonstrate slip-mode conductance of the cardiac Na⁺ channel in rat cardiac myocytes. We thus examined the possibility that a missing factor or channel subunit was responsible for the absence of any PKA-activated increase in P_Ca/P_Na when only the  subunit was expressed (see also Table 1). Because intact heart cells express  and both  subunits (12) and recent evidence suggests that both  and ₁ are associated with each other in heart (13) and in heterologous expression systems (14), we examined HEK293 cells that co-express  and ₁ subunits.

Table 1. Modulation of PCa/PNa. Control PKA activation Comment Conditions ENa (mV) Erev (mV) PCa/PNa ENa (mV) Erev (mV) PCa/PNa 1  : Nai=Nao=10 mM; Cao=2 mM; Cai=0; Monovalent replacement: Cs+ 0 0  0 0 +9 1.2 (2 Ca2+) Ref. (6). 2   in HEK293 cells; Nai=20 mM; Na0=20 mM (control); Nao=20.5 mM (dbcAMP); Cao=2 mM; Cai=100 nM; Monovalent replacement: Cs+ 0  0.23  0 0.63 0.82 0.04 Fig. 1A. 3   in HEK293 cells; Nai=Nao=20 mM (control); Nao=20.5 mM (dbcAMP); Cao=2 mM; Cai=100 nM; Monovalent replacement: Cs+ 0 0.18 0.04 0.63 4.00 0.79 Fig. 1B. 4   in HEK293 cells; Nai=Nao=20 mM (control); Nao=20.5 mM (dbcAMP); Cao=2 mM; Cai=100 nM; Monovalent replacement: Cs+ 0  0.02  0 0.63  0.05  0 Fig. 1C. 5   in HEK293 cells. ; Nai=Nao=10 mM; Cao=5 mM; Cai=0 nM; Monovalent replacement: Cs+; 10 mM TEA 0 0.94 0.04 1.2 7.18 0.32 Fig. 2B 6   in HEK293 cells. Nai=10 mM; Na0=0.5 mM (dbcAMP); Cao=5 mM; Cai=0 nM; Monovalent replacement: Cs+; 10 mM TEA  79 n.d. n.d.  76.2  37  0.11 Fig. 2B 7   in HEK293 cells. Nai=20 mM (control); Nao=20.5 mM (dbcAMP); Cao=2 mM; Cai=100 nM; Monovalent replacement: Cs+; 10 mM TEA 0 0.01  0 0.63 5.0 1.1 Fig. 2G 8   in HEK293 cells; Nai=Nao=20 mM (control); Nao=20.5 mM (dbcAMP); Cao=2 mM; Cai=100 nM; Monovalent replacement: NMG+. 0 0.53 0.11 0.63 6.35 1.48 Fig. 3D n.d., experiment not done.

We measured the current-voltage (IV) relationships for I_Na in HEK293 cells that co-express  and ₁ subunits of the Na⁺ channels under several experimental conditions (Fig. 1B). Under control conditions, E_rev = 0.23 mV, a value statistically indistinguishable from E_Na of 0 mV. After the addition of 500 µM Na⁺-dbcAMP, a significant increase of E_rev was observed (E_rev = 4.00 mV, P < 0.0001, n = 8). This increase in E_rev suggests that P_Ca/P_Na has increased from close to zero to 0.79 (10), indicating that Ca²⁺ was almost as permeable through Na⁺ channels as was Na^+. The altered E_rev was returned to the control level by the removal of extracellular Ca²⁺ (replaced by Mg²⁺), a result that also supports the conclusion that Ca²⁺ permeation underlies the shift of E_rev following the addition of dbcAMP. In the maintained presence of dbcAMP and with 2 mM [Ca²⁺]_o , TTX significantly reduced I_Na and also produced a significant negative shift in E_rev back to control conditions. We thus concluded that first, the observed TTX-sensitivity confirms the involvement of cardiac Na⁺ channels in the measured currents. Second, Na⁺ channels altered by PKA may be more sensitive to TTX when in slip-mode conductance than when they are not. If Na⁺ channels in slip-mode and those not in slip-mode were equally sensitive to TTX, then TTX would not have shifted E_rev back to E_Na. [This finding is identical to that observed in intact rat ventricular myocytes (6).] Third, protein-protein interactions between the  and ₁ subunits are important for proper Na⁺ channel function and provides a functional reason role for the ₁ subunit of the Na⁺ channel in heart. PKA-dependent phosphorylation underlies the activation of slip-mode conductance (Fig. 1C), which can be prevented by adding intracellular PKA inhibitory peptide (PKA-I, 100 µM).

Two control experiments (Fig. 1, D and E) indicated that neither the expression of the ₁ subunit alone nor a sham transfection of HEK293 cells altered the HEK293 cells to produce I_Na under control conditions or following exposure to dbcAMP. We thus conclude that there are no interfering voltage-gated currents in these cells to confound interpretation.

Ca²⁺ permeation through the cardiac Na⁺ channel is thus confirmed (Fig. 1) in support of our report (6). The  subunit alone is not enough (Fig. 1) to support this change in selectivity. Instead, it was found that the co-expression of an  Na⁺ channel subunit along with a  subunit was required for us to observe slip-mode conductance of the Na⁺ channel in an heterologous expression system. The  subunit is central to our understanding of ion permeation through Na⁺ channels because it contains the ion channel, the selectivity filter, the TTX binding site, and the PKA consensus phosphorylation sites. However, we have no structural information on how the subunit is altered to produce slip-mode conductance. Also we do not know what the ₁ subunit does and whether or not its action requires other factors. Intuition from nonselective channels, including mutated Na⁺ channels (15), suggests that larger changes in E_rev might be observed in bi-ionic conditions (for example, high Ca²⁺ outside and high Na⁺ inside with low Ca²⁺ inside and low Na⁺ outside). One finding (Fig. 2, A and B) was counterintuitive. HEK293 cells were made to express  and ₁ subunits (as was shown in Fig. 1), but with somewhat reduced intracellular and extracellular [Na⁺] (10 mM) and increased extracellular [Ca²⁺] (5 mM). Slip-mode conductance was readily activated by dbcAMP, leading to a positive shift of E_rev of more than 7 mV, more than was observed in Fig. 1. Unexpectedly, however, P_Ca/P_Na was only 0.32. Experiments (16) similar to those of Fig. 2, A and B, but with extracellular [Ca²⁺] at 2 mM led to a larger calculated P_Ca/P_Na of 0.95. Thus, although extracellular Ca²⁺ permeates the Na⁺ channel during slip-mode conductance, it also tends to block the conductance.