Structural effects of an LQT-3 mutation on heart Na+ channel gating

Abstract

Computational methods that predict three-dimensional structures from amino acid sequences have become increasingly accurate and have provided insights into structure-function relationships for proteins in the absence of structural data. However, the accuracy of computational structural models requires experimental approaches for validation. Here we report direct testing of the predictions of a previously reported structural model of the C-terminus of the human heart Na(+) channel. We focused on understanding the structural basis for the unique effects of an inherited C-terminal mutation (Y1795C), associated with long QT syndrome variant 3 (LQT-3), that has pronounced effects on Na(+) channel inactivation. Here we provide evidence that this mutation, in which a cysteine replaces a tyrosine at position 1795 (Y1795C), enables the formation of disulfide bonds with a partner cysteine in the channel. Using the predictions of the model, we identify the cysteine and show that three-dimensional information contained in the sequence for the channel protein is necessary to understand the structural basis for some of the effects of the mutation. The experimental evidence supports the accuracy of the predicted structural model of the human heart Na(+) channel C-terminal domain and provides insight into a structural basis for some of the mutation-induced altered channel function underlying the disease phenotype.

FIGURE 1

Inherited mutation Y1795C alters single channel properties of wild-type cardiac sodium channels. Cell-attached patch recordings are shown for wild-type (left), Y1795C (middle), and Y1795S (right) channels. Recordings were obtained in response to test pulses (−10 mV, 100 ms) applied at 2 Hz from −120 mV. Current recordings of individual and consecutive sweeps are shown to emphasize the effects of inherited mutations on channel opening kinetics. Ensemble currents (constructed by averaging 500 consecutive sweeps) are shown for each construct below the individual sweeps. Open time histograms were generated for each construct using 200-μs bins of all events recorded from 500 to 1000 sweeps. Patches used included three or fewer channels. MOT was estimated by the best-fit double exponential functions to each histogram. The fitted curves are illustrated as dashed lines in each panel. The resulting MOT estimates based on the extracted time constants are 0.76 ± 0.05 ms (WT, n = 4); 1.09 ± 0.8 ms (YC, n = 5); and 0.72 ± 0.04 ms (YS, n = 3); p = 0.01 for YC versus WT, YS versus WT (NS).

FIGURE 2

Distinct properties of Y1795C channels: effects on the kinetics of the onset of inactivation. Residue Y1795 (WT, Y, ○) was mutated to Cys (Y1795C, C, ♦), Ala (Y1795A, A, □), Phen (Y1795F, F, •), and Ser (Y1795S, S, ▪). (A) The plot shows the time for peak inward current elicited by the test voltages to decay to 50% of peak (t_1/2) plotted versus test pulse voltage for each construct. Data shown are mean ± SE: n = 8 (WT); n = 7 (C); n = 6 (S); n = 8 (A). (B) Summary plot of time to half decay (t_1/2) measured at +10 mV versus channel construct. Experimental numbers are as in A. ^*p < 0.01 versus WT; ^**NS versus WT.

FIGURE 3

Evidence for disulfide bond modification of inactivation in Y1795C mutant channels. Cells expressing WT channel (○, •), A1795 (□, ▪), or C1795 (⋄, ♦) mutant channels were treated (solid symbols) or not treated (open symbols) with DTT. (A) The upper row shows current traces recorded from cells before, during, and after washout of exposure to DTT (50 mM) in the external bath. Currents are shown for WT (left) and YC channels (right) recorded at +10 mV. (B). The mean (± SE) time to half current decay (t_1/2) is plotted versus test pulse voltage for each construct. The number of experiments was from six to nine for each construct. (C) The bars summarize t_1/2 measured at +10 mV in the absence (open) and presence (solid) of DTT (10 mM). ^*p < 0.01 compared with untreated and with WT and YA; ^**NS versus WT.

FIGURE 4

Structural model of the C-terminal domain in the SCN5A Na⁺ channel. (A) The linear sequence of the proximal region of the C-terminal tail of SCN5A. The arrows indicate positions of F7194 (green), Y1795 (red), and C1850 (blue). (B) Model-predicted C-terminal structure. The N-terminal half (residues 1773–1863) of the SCN5A C-terminal domain was predicted to contain two EF-hand structural motifs that are packed to adopt a folding topology similar to the EF-hands in calmodulin (Cormier et al., 2002). The helix barrels H1-H2 show the first EF-hand structural motif in the model; the H3-H4 helix barrels show the second EF-hand structural motif. The three-dimensional ball-and-stick models show the predicted positions of the residues Y1795C and C1850 in the structural model. The sulfur atoms in the cysteines are colored in yellow. These two atoms are separated by 4 Å in the model structure. Residue F1794 that appears from the back of H1 helix is also shown in ball-and-stick model. (C) The helix wheel for H1 helix shows schematically the amphipathic nature of the H1 helix. The residues that are expected to be involved in the formation of a disulfide bond in the Y1795C mutant are shown in green. The ionizable residues in H1 helix are colored in red.

FIGURE 5

Experimental tests of structural model predictions. The effects of DTT (10 mM) were measured on channels in which the possible cysteine pairs were probed by mutation. In each panel, the upper row illustrates currents measured (at +10 mV) in the absence and presence of DTT (10 mM). The lower row plots inactivation t_1/2 versus test pulse voltage in the absence (open) and presence (solid) of DTT 10 mM. (A). The Cys at position 1850 was replaced by Ala, and Tyr¹⁷⁹⁵ was replaced by Cys to generate construct C1850A_Y1795C. Plotted is mean ± SE data (n = 6 for each data set). (B) The Phe at position 1794 was replaced by Cys, and the Tyr at position 1795 was not changed to make construct F1794C_Y1795. Shown are mean ± SE data (n = 4–6 for each construct). At 10 mV, there is no significant difference between DTT-free or DTT-containing data (p > 0.3 in both cases).