A common cardiac sodium channel variant associated with sudden infant death in African Americans, SCN5A S1103Y

Abstract

Thousands die each year from sudden infant death syndrome (SIDS). Neither the cause nor basis for varied prevalence in different populations is understood. While 2 cases have been associated with mutations in type Valpha, cardiac voltage-gated sodium channels (SCN5A), the "Back to Sleep" campaign has decreased SIDS prevalence, consistent with a role for environmental influences in disease pathogenesis. Here we studied SCN5A in African Americans. Three of 133 SIDS cases were homozygous for the variant S1103Y. Among controls, 120 of 1,056 were carriers of the heterozygous genotype, which was previously associated with increased risk for arrhythmia in adults. This suggests that infants with 2 copies of S1103Y have a 24-fold increased risk for SIDS. Variant Y1103 channels were found to operate normally under baseline conditions in vitro. As risk factors for SIDS include apnea and respiratory acidosis, Y1103 and wild-type channels were subjected to lowered intracellular pH. Only Y1103 channels gained abnormal function, demonstrating late reopenings suppressible by the drug mexiletine. The variant appeared to confer susceptibility to acidosis-induced arrhythmia, a gene-environment interaction. Overall, homozygous and rare heterozygous SCN5A missense variants were found in approximately 5% of cases. If our findings are replicated, prospective genetic testing of SIDS cases and screening with counseling for at-risk families warrant consideration.

Figure 1

A polymorphism in SCN5A encodes a variant cardiac sodium channel. (A) Denaturing HPLC waveform and direct sequencing of wild-type and S1103Y variant. The chromatogram shows the additional peak resulting from heteroduplex assembly of S1103 and Y1103 amplicons. Forward sequence of homozygous S1103 and Y1103 SIDS cases shows the C3308A change that leads to the substitution of serine (S) by tyrosine (Y). (B) Topology of the cardiac sodium channel encoded by SCN5A shows the cytoplasmic location of the S1103Y missense change (red), the 4 homologous membrane domains (DI–DIV), the pore-forming (P) loops, and the voltage-sensing segments (+).

Figure 2

Y1103 channels show a change in inactivation gating at low pH. Whole-cell currents in HEK-293 cells expressing SCN5A S1103 (open squares) or SCN5A Y1103 (filled squares). Top: Sample traces with S1103 channels. Arrows indicate points of measure. Scale bar: 1 nA. Filter and sampling frequencies were 5 and 20 kHz, respectively. (A) Normalized conductance/voltage (G/G_max) relationships for S1103 and Y1103 channels show no change with a shift of internal pH (pHi) from pH 7.4 to pH 6.7 (n = 4–13). Curves fitted to a Boltzmann relationship (y = 1/{1 + exp[V – V_1/2/Vs]}, where Vs is slope factor). Inset: Current/voltage relationships. Values for V_1/2 are reported in Supplemental Table 3. (B) Y1103 channels showed a shift in voltage-dependent steady-state inactivation at pHi 6.7, whereas S1103 channels did not. Normalized peak current values (I/I_max) are plotted against prepulse potential (mV). Curves fitted as in A. At pH 6.7, V_1/2 for S1103 is –84 + 1 mV (n = 7) and for Y1103 is –77 + 2 mV (n = 8; P < 0.05; 2-population Student’s t test). Values are mean ± SEM. (C) Time constants for recovery from inactivation were the same for S1103 and Y1103 channels at physiological pH but were speeded when Y1103, but not S1103, was exposed to pHi 6.7 (n = 11). Curves are exponential fits to the recovery time course, I(t)/I_max = Af[1 – exp(–t/τf)] + As[1 – exp(–t/τs)], where A is the amplitude of fast (f) and slow (s) components and t is time. Values for τ are reported in Supplemental Table 3.

Figure 3

Cells expressing Y1103 channels show sustained current at a low pHi. Whole-cell currents were recorded in HEK-293 cells expressing S1103 or Y1103 channels as indicated with pipette and bath solutions (see Methods) with cells held at –100 mV and stepped to –30 mV for 100 ms at a rate of 0.5 Hz. Scale bars represent 300 pA and 20 ms. (A) Representative currents recorded from cells at the indicated pHi. (B) Ensemble averages of 500 sweeps recorded from 10 cells with the indicated channel and pHi. Late current was 5.23% ± 0.85% of peak (n = 10 cells). Peak currents were not significantly different and were –1311 ± 143 pA and 1252 ± 139 pA, for S1103 at normal and low pH, respectively, and 1267 ± 173 pA and 1195 ± 245 pA for Y1103 at normal and low pH, respectively (n = 10 cells per condition).

Figure 4

Single Y1103 channels show abnormal late reopenings. Single channels studied in inside-out, off-cell patches excised from HEK-293 cells. Currents were stimulated every 2.5 seconds by a 50-ms depolarizing pulse to –30 mV from a holding potential of –120 mV. Data were recorded at filter and sampling frequencies of 5 and 50 kHz, respectively. Pipettes were filled with bath solution described in Methods. Cells were perfused with the pipette solution described in Methods. For each cell, null sweeps (with no channel activity) were identified, averaged offline, and subtracted from data sweeps before analysis. For display purposes, data were refiltered offline using a 2-kHz Bessel filter. (A) With depolarization, single S1103 channels opened, inactivated rapidly, and did not reopen. This behavior was unaltered when internal pH was lowered from 7.4 to 6.7. Null traces at pH 7.4 and 6.7 were 50% ± 2.7% (n = 15 patches, 1,367 sweeps) and 51% ± 3.4% (n = 12 patches, 1,028 sweeps), respectively. In contrast, Y1103 channels behaved like S1103 channels at pH 7.4 but showed late reopenings at pH 6.7. Null traces were 51% ± 4.5% at pH 7.4 (n = 8 patches, 742 sweeps) and 51.7 ± 2.5% (n = 10 patches, 942 sweeps) when pH was reduced to 6.7. (B) Ensemble average traces (n = 100–150 sweeps) for the indicated channels and conditions. Y1103 channels failed to remain in the inactivated state when exposed to internal pH 6.7. Scale bars: 0.5 pA; 10 ms.

Figure 5

Abnormal reopenings are suppressed by mexiletine. Single channels were studied in inside-out, off-cell patches as in Figure 3. (A) Dose-response curves for phasic block by mexiletine of peak macroscopic currents with S1103 (open squares) and Y1103 (filled squares) channels studied in whole-cell mode. Each point shows the average percent reduction in peak current by the dose indicated after a series of repetitive depolarizing stimuli (n = 3–7 cells). Cells were depolarized to –30 mV from a holding potential of –100 mV for 10 ms at 2.5 Hz to mimic a rate of 150 bpm. Inset: Representative S1103 current traces evoked by a pulse in drug-free solution (control) and the first (pulse 1) to show tonic block and fiftieth (pulse 50) to show phasic block by 10 μM mexiletine (mex). Values for tonic and phasic block by mexiletine, propranolol, and amiodarone are reported in Supplemental Table 4. (B) Single Y1103 channels studied in inside-out off-cell patches showed that late reopenings of variant channels were fully suppressed by 5 μM mexiletine. Null traces in the absence of drug (see Figure 3B) at pH 7.4 and 6.7 were 51% ± 4.5% (n = 8 patches, 742 sweeps) and 52 ± 2.5% (n = 10 patches, 942 sweeps), respectively; null traces at pH 6.7 with drug were 55% ± 3.5% (n = 6 patches, 561 sweeps). The therapeutic blood level of mexiletine is 0.8–2.0 μg/ml (3.7–9.3 μM). Scale bars: 2 ms, 200 pA (A); 0.3 pA, 10 ms (B).