Spliced isoforms of the cardiac Nav1.5 channel modify channel activation by distinct structural mechanisms

Abstract

Alternative splicing is an important cellular mechanism that fine tunes the gating properties of both voltage- and ligand-gated ion-channels. The cardiac voltage-gated sodium channel, Nav1.5, is subject to alternative splicing of the DI S3-S4 linker, which generates two types of channels with different activation properties. Here, we show that the gating differences between the adult (mH1) and neonatal (Nav1.5e) isoforms of Nav1.5 are mediated by two amino acid residues: Thr/Ser at position 207 and Asp/Lys at position 211. Electrophysiological experiments, in conjunction with molecular dynamics simulations, revealed that each residue contributes equally to the overall gating shifts in activation, but that the underlying structural mechanisms are different. Asp/Lys at position 211 acts through electrostatic interactions, whereas Thr/Ser at position 207 is predicted to alter the hydrogen bond network at the top of the S3 helix. These distinct structural mechanisms work together to modify movement of the voltage-sensitive S4 helix to bring about channel activation. Interestingly, mutation of the homologous Asp and Thr residues of the skeletal muscle isoform, Nav1.4, to Lys and Ser, respectively, confers a similar gating shift in channel activation, suggesting that these residues may fulfill a conserved role across other Nav channel family members.

Figure 1.

Alternative splicing at exon 6 selectively alters properties of channel activation. (A) Top: Topology model of Nav1.5. The polypeptide stretch encoded by exon 6, spanning from partway through the S3 segment to the bottom of the S4 segment in DI, is highlighted in the gray box. Bottom: Sequence alignment of exons 6b in mNav1.5 and 6a in mNav1.5e, with the black boxes highlighting the seven amino acid differences between them. (B) Sample current traces elicited by Nav1.5 during an activation protocol. Cells were stepped from −100 mV to a range of membrane potentials between −110 mV and +70 mV, in increments of 5 mV. (C) Activation profiles of mNav1.5 (black) and mNav1.5e (gray) revealed that mNav1.5e activates at potentials ∼10 mV more depolarized than mNav1.5. V_1/2 values of mNav1.5e and mNav1.5 were −16.8 ± 0.45 mV (n = 51) and −25.9 ± 0.55 mV (n = 36; P < 0.001), respectively. Slope factor (k) values were 9.0 ± 0.13 and 8.2 ± 0.16, respectively; P < 0.001. (D) Distributions of activation V_1/2 plotted against peak current amplitude reaffirm the 10-mV difference in the activation profiles of mNav1.5 and mNav1.5e. Correlation strength was moderate, with Pearson’s r of 0.40 and 0.45 for mNav1.5e and mNav1.5, respectively. Regression line y-intercepts of mNav1.5e and mNav1.5 were −14.2 ± 0.08 and −22.4 ± 0.13 mV (P < 0.001), respectively. Regression line slopes were parallel, at 0.48 ± 0.01 and 0.44 ± 0.01 mV/nA (P = 0.75), respectively, reflecting that the 10-mV difference between mNav1.5e and mNav1.5 was consistent across amplitudes. (E) Sample current traces elicited by Nav1.5 during a steady-state inactivation protocol. Cells were given a variable pre-pulse potential, ranging from −160 to −30 mV in increments of 5 mV, then stepped to −10 mV to evoke a current response. (F) Inactivation profiles of mNav1.5e and mNav1.5 show that the two splice variants do not differ in terms of steady-state inactivation. V_1/2 values were −82.0 ± 0.52 mV (n = 49) and −82.1 ± 0.82 mV (n = 34) for mNav1.5e and mNav1.5 (P = 0.91), respectively. k values were −7.2 ± 0.14 and −6.9 ± 0.16 for mNav1.5e and mNav1.5, respectively (P = 0.15). (G) mNav1.5e and mNav1.5 do not differ in terms of recovery from inactivation either. The fraction of Nav channels recovered was plotted against the interpulse interval and fitted to a double exponential function, yielding weighted τ values of 12.4 ± 0.41 ms (n = 46) and 12.7 ± 1.06 ms (n = 22) for mNav1.5e and mNav1.5, respectively (P = 0.75).

Figure 2.

The aspartate-lysine switch at position 211 accounts for part of the shift in channel activation. (A) Introducing D211K into mNav1.5 induces only a partial shift in the activation profile toward that of mNav1.5e. The activation V_1/2 of mNav1.5-D211K (red) was −20.1 ± 0.58 mV (n = 24), distinct from both mNav1.5 (P < 0.001) and mNav1.5e (P < 0.001). The slope factor k was 8.6 ± 0.10. (B) The y-intercept of the mNav1.5-D211K scatter measured −17.0 ± 0.19 mV, different from both mNav1.5 (P < 0.001) and mNav1.5e (P < 0.001). The regression line slope was 0.52 ± 0.03 mV/nA. (C) Replacing K211D in mNav1.5e also produces an intermediate phenotype. The activation V_1/2 of mNav1.5e-K211D (red) was −23.2 ± 0.50 mV (n = 30), distinct from mNav1.5e (P < 0.001) and mNav1.5 (P < 0.01). The slope factor k was 8.5 ± 0.14. (D) The y-intercept of the mNav1.5e-K211D scatter measured −18.5 ± 0.15 mV, different from both mNav1.5 (P < 0.001) and mNav1.5e (P < 0.001). The regression line slope was 0.48 ± 0.01 mV/nA.

Figure 3.

Most other amino acid exchanges serve negligible roles in setting channel activation following exon 6 splicing. (A) With the exception of Thr/Ser at position 207, none of the other amino acid exchanges in exon 6 can be introduced into mNav1.5 to render it mNav1.5e-like. The V_1/2 values of mNav1.5-T206V, mNav1.5-F209N, mNav1.5-V210I, mNav1.5-V215L, and mNav1.5-S234P were −25.9 ± 0.57 mV (n = 25), −25.2 ± 0.57 mV (n = 18), −25.2 ± 1.01 mV (n = 12), −27.0 ± 0.52 mV (n = 24), and −24.2 ± 0.52 mV (n = 20), respectively. None of these conditions differed significantly from mNav1.5 (P = 0.07). (B) Scatter dot plot showing the spread of V_1/2 values across the different possible amino acid swaps of positions 206, 209, 210, 215, and 234, engineered into mNav1.5. Individual V_1/2 values are shown as squares. The hollow circle represents the mean. Error bars reflect 1 SD, not SEM. (C) Apart from Thr/Ser at position 207, none of the other amino acid exchanges in exon 6 can be introduced into mNav1.5e to render it mNav1.5-like. The V_1/2 values of mNav1.5e-V206T, mNav1.5e-N209F, mNav1.5e-I210V, mNav1.5e-L215V, and mNav1.5e-P234S were −13.0 ± 0.47 mV (n = 24), −15.5 ± 0.68 mV (n = 21), −17.5 ± 0.64 mV (n = 24), −12.7 ± 0.51 mV (n = 21), and −14.7 ± 0.48 mV (n = 18), respectively. While some of these conditions differed significantly from mNav1.5e (P < 0.001), those were curiously all shifted to more depolarized potentials than mNav1.5e itself. (D) Scatter dot plot showing the spread of V_1/2 values across the different possible amino acid swaps of positions 206, 209, 210, 215, and 234, engineered into mNav1.5e. Individual V_1/2 values are shown as squares. The hollow circle represents the mean. Error bars reflect 1 SD, not SEM.

Figure 4.

The threonine-serine switch at position 207 also influences channel activation. (A) Introducing T207S into mNav1.5 also causes a shift in the activation profile toward that of mNav1.5e. The activation V_1/2 of mNav1.5-T207S (red) was −18.2 ± 0.64 mV (n = 17), distinct from mNav1.5 (P < 0.001) and apparently overlapping with mNav1.5e (P = 0.30). The slope factor k was 9.3 ± 0.15. (B) The y-intercept of the mNav1.5-T207S scatter measured −15.2 ± 0.21 mV, falling between mNav1.5 (P < 0.001) and mNav1.5e (P < 0.05). The regression line slope was steeper than usual, at 0.68 ± 0.04 mV/nA. (C) Replacing S207T in mNav1.5e produces an intermediate phenotype in terms of channel activation. The activation V_1/2 of mNav1.5e-S207T (red) was −20.3 ± 0.53 mV (n = 26), distinct from both mNav1.5e (P < 0.001) and mNav1.5 (P < 0.001). The slope factor k was 9.6 ± 0.14. (D) The y-intercept of the mNav1.5e-S207T scatter measured −17.7 ± 0.13 mV, different from both mNav1.5 (P < 0.001) and mNav1.5e (P < 0.001). The regression line slope was 0.40 ± 0.02 mV/nA.

Figure 5.

Combined, the Thr/Ser and Asp/Lys switches account for the full gating difference between mNav1.5 and mNav1.5e. (A) Introducing both T207S and D211K into mNav1.5 results in an activation profile that is indistinguishable from that of mNav1.5e. The activation V_1/2 of mNav1.5-T207S-D211K (red) was −16.2 ± 0.50 mV (n = 25), overlapping with mNav1.5e (P = 0.64) yet distinct from mNav1.5 (P < 0.001). The slope factor k was 9.3 ± 0.10. (B) The y-intercept of the mNav1.5-T207S-D211K scatter measured −13.3 ± 0.16 mV, coinciding with mNav1.5e (P = 0.43) but remaining distinct from mNav1.5 (P < 0.001). The regression line slope was 0.50 ± 0.02 mV/nA. (C) Replacing both S207T and K211D in mNav1.5e produces a phenotype similar to that of mNav1.5. The activation V_1/2 of mNav1.5e-S207T-K211D (red) was −27.6 ± 0.51 mV (n = 24), distinct from mNav1.5e (P < 0.001) yet identical to mNav1.5 (P = 0.10). The slope factor k was 8.8 ± 0.16. (D) The y-intercept of the mNav1.5e-S207T-K211D scatter measured −23.9 ± 0.17 mV, overshooting mNav1.5 in the negative direction (P < 0.05) yet still very distinct from mNav1.5e (P < 0.001). The regression line slope was 0.46 ± 0.02 mV/nA.

Figure 6.

The presence of a negative charge at position 211 is essential to the hyperpolarized gating profile of mNav1.5. (A) Introducing D211A in mNav1.5 results in an activation profile that is indistinguishable from the mutant mNav1.5-D211K, as if loss of the negative charge were more critical to the altered gating shift than gaining the positive charge. The activation V_1/2 of mNav1.5-D211A (blue) was −21.4 ± 0.75 mV (n = 14), similar to mNav1.5-D211K (red, P = 0.64). The slope factor k was 8.8 ± 0.19. (B) The y-intercept of the mNav1.5-D211A scatter measured −17.9 ± 0.07 mV, coinciding with mNav1.5-D211K (P = 0.92). The regression line slope was 0.46 ± 0.01 mV/nA. (C) Substituting K211A in mNav1.5e also produces a small but nonnegligible shift, taking on a phenotype between mNav1.5e-K211D and mNav1.5e, suggesting that the presence of the positive charge may matter to a limited degree. The activation V_1/2 of mNav1.5e-K211A (blue) was −19.8 ± 0.50 mV (n = 42), distinct from both mNav1.5e-K211D (red, P < 0.001) and mNav1.5e (P < 0.001). The slope factor k was 8.5 ± 0.18. (D) The y-intercept of the mNav1.5e-K211A scatter measured −15.7 ± 0.07 mV, distinct from both mNav1.5e-K211D (P < 0.001) and mNav1.5e (P < 0.05). The regression line slope was 0.47 ± 0.01 mV/nA.

Figure 7.

A constrained hydrogen bond donor is required at position 207 to yield the hyperpolarized activation parameters of mNav1.5. (A) Introducing T207A in mNav1.5 results in an activation profile that is also depolarized compared with mNav1.5, suggesting that the rigidity of the residue at this position is critical to the altered gating behavior. The activation V_1/2 of mNav1.5-T207A (blue) was −22.1 ± 0.56 mV (n = 26), distinct from mNav1.5 (P < 0.001). Admittedly, though, the shift is not as strong as that of mNav1.5-T207S (red, P < 0.001). The slope factor k was 8.3 ± 0.13. (B) The y-intercept of the mNav1.5-T207A scatter measured −18.9 ± 0.16 mV, distinct from both with mNav1.5 (P < 0.001) and mNav1.5-T207S (P < 0.01). The regression line slope was 0.42 ± 0.02 mV/nA. (C) Substituting S207A in mNav1.5e had virtually no effect, highlighting the importance of losing the sterically constrained threonine residue in causing the shift in channel activation. The activation V_1/2 of mNav1.5e-S207A (blue) was −16.8 ± 0.49 mV (n = 20), coinciding with mNav1.5e (P = 1.00). The slope factor k was 9.5 ± 0.11. (D) The y-intercept of the mNav1.5e-S207A scatter measured −14.5 ± 0.14 mV, indistinguishable from mNav1.5e (P = 0.95). The regression line slope was 0.37 ± 0.02 mV/nA.

Figure 8.

Molecular dynamics simulations of hNav1.5 and hNav1.5-D211K/T207S reveal disruptions within the VSD of DI in Nav1.5. (A) Minimum distance between D/K211 and POPC lipid head groups. The distances were calculated between the center of geometry of terminal side chain atoms (see http://github.com/bigginlab/nav-alt-splice for full definitions). (B) Representative snapshot of the S3–S4 linker containing D211 (dark blue) in the hNav1.5 VSD (left) and pronounced distortion with the D211K mutation (orange). The distance between E208 and R219 (cyan) at the top of the VSD increase in the presence of D211K (black arrow). (C) RMSD of the S3–S4 linker (magenta). (D) Snapshot of the decreased distance between E208 and R219 (cyan) due to the T207S mutation (dark green). (E) Minimum distance between the hydroxyl hydrogen in T/S207 and the carboxyl group oxygens in E208. (F) Dihedral angle between N, Cα, Cβ, and O (of the hydroxy group) atoms in T/S207. (G) Representative snapshots of T207 hNav1.5 conformations (dark blue) at the two main dihedral angle peaks (indicated by asterisks in F). The hydroxyl hydrogen interacts with the carbonyl backbone (single asterisks) of M208 and the carboxyl group of E208 (double asterisks).

Figure 9.

Overview of key charged residues present within the VSD. The VSD of D1 in mNav1.5. The VSD (white) contains four transmembrane helices (S1–S4, labeled in bold) with a number of positively (blue) and negatively (orange) charged residues that make up critical salt bridges during S4 translocation during mNav1.5 depolarization. Note that the S1 label has been removed and the helix shown as transparent for clarity.

Figure 10.

2-D joint plots of salt-bridge distances in the VSD. Minimum distances between two residues in the VSD. Distances were measured from the center of geometry of carboxyl oxygen atoms in glutamic/aspartic acid and the terminal −NH₂ atoms in the guanidino group. See http://github.com/bigginlab/nav-alt-splice for full atom selections. (A–C) Minimum distances between E208/R222 and E208/R219 for hNav1.5-T207S, hNav1.5-D211K, and hNav1.5-T207S-D211K, respectively. (D–F) Minimum distances between D152/R222 and D152/R219 for hNav1.5-T207S, hNav1.5-D211K, and hNav1.5-T207S-D211K, respectively. (G–I) Minimum distances between E161/R225 and E161/R222 for hNav1.5-T207S, hNav1.5-D211K, and hNav1.5-T207S-D211K, respectively.

Figure 11.

The ability for the DI S3–S4 linker to alter channel activation is conserved in mNav1.4. (A) Sequence alignment of mouse Nav1.4 and Nav1.5, in the adult splice form, in the regions of and surrounding exon 6. mNav1.5 and mNav1.4 are largely similar in sequence, with the exception of those amino acid differences shaded in gray. The black boxes highlight the Thr207 and Asp211 that were shown to be important in exon 6 splicing of Nav1.5 and that are also conserved in mNav1.4. (B) Sample current traces elicited by the mNav1.4 + β3 complex during an activation protocol. Cells were stepped from −100 mV to a range of membrane potentials between −110 mV and +70 mV, in increments of 5 mV. (C) Activation profiles of mNav1.4 + β3 (black) and the double mutant mNav1.4-T207S-D211K + β3 (gray) recapitulate the ∼10-mV shift in channel activation, as seen in Nav1.5. V_1/2 values of mNav1.4-T207S-D211K + β3 and mNav1.4 + β3 were 5.7 ± 0.50 mV (n = 25) and −2.3 ± 0.65 mV (n = 13, P < 0.001), respectively. The k values were 9.0 ± 0.24 and 8.8 ± 0.21, respectively (P = 0.51). (D) Distributions of activation V_1/2 plotted against peak current amplitude demonstrate the difference in activation profiles of mNav1.4-T207S-D211K + β3 and mNav1.4 + β3. Regression line y-intercepts of mNav1.4-T207S-D211K + β3 and mNav1.4 + β3 were 5.9 ± 0.18 and 0.48 ± 0.31 mV, respectively (P = 0.38). Regression line slopes were parallel, at 0.12 ± 0.02 and 1.4 ± 0.14 mV/nA (P = 0.11). (E) Sample current traces elicited by mNav1.4 + β3 during a steady-state inactivation protocol. Cells were given a variable pre-pulse potential, ranging from −160 to −30 mV in increments of 5 mV, and then stepped to +10 mV to evoke a current response. (F) Inactivation profiles of mNav1.4-T207S-D211K + β3 and mNav1.4 + β3 show that the two variants do not differ in terms of steady-state inactivation. V_1/2 values were −55.3 ± 0.36 (n = 21) and −55.8 ± 0.66 mV (n = 13) for mNav1.4-T207S-D211K and mNav1.4 + β3, respectively (P = 0.53). k values were −5.4 ± 0.15 and −5.3 ± 0.20 for mNav1.4-T207S-D211K and mNav1.4 + β3, respectively (P = 0.59). (G) mNav1.4-T207S-D211K + β3 and mNav1.4 + β3 do not differ in terms of recovery from inactivation either. The fraction of Nav channels recovered was plotted against the interpulse interval and fitted to a double exponential function, yielding weighted τ values of 1.04 ± 0.07 (n = 17) and 1.23 ± 0.18 ms (n = 12) for mNav1.4-T207S-D211K + β3 and mNav1.4 + β3, respectively (P = 0.34).