A hERG mutation E1039X produced a synergistic lesion on IKs together with KCNQ1-R174C mutation in a LQTS family with three compound mutations

Congenital long QT syndrome (LQTS) caused by compound mutations is usually associated with more severe clinical phenotypes. We identified a LQTS family harboring three compound mutations in different genes (KCNQ1-R174C, hERG-E1039X and SCN5A-E428K). KCNQ1-R174C, hERG-E1039X and SCN5A-E428K mutations and/or relevant wild-type (WT) cDNAs were respectively expressed in mammalian cells. IKs-like, IKr-like, INa-like currents and the functional interaction between KCNQ1-R174C and hERG-E1039X channels were studied using patch-clamp and immunocytochemistry techniques. (1) Expression of KCNQ1-R174C alone showed no IKs. Co-expression of KCNQ1-WT + KCNQ1-R174C caused a loss-of-function in IKs and blunted the activation of IKs in response to isoproterenol. (2) Expression of hERG-E1039X alone and co-expression of hERG-WT + hERG-E1039X negatively shifted inactivation curves and decelerated the recovery time from inactivation. (3) Expression of SCN5A-E428K increased peak INa, but had no effect on late INa. (4) IKs and IKr interact, and hERG-E1039X caused a loss-of-function in IKs. (5) Immunocytochemical studies indicated that KCNQ1-R174C is trafficking defective and hERG-E1039X is defective in biosynthesis/degradation, but the abnormities were rescued by co-expression with WT. Thus, KCNQ1-R174C and hERG-E1039X disrupted IKs and IKr functions, respectively. The synergistic lesion, caused by KCNQ1-R174C and hERG-E1039X in IKs, is very likely why patients showed more severe phenotypes in the compound mutation case.

(AP), and play a critical role in controlling the ventricular AP duration (APD) 5,6 . The SCN5A gene encodes the α-subunit of the predominant cardiac sodium channel (Na V 1.5) that conducts the depolarizing sodium inward current and is mainly responsible for the initial depolarization in cardiomyocytes. Mutations in KCNQ1, hERG and SCN5A can cause LQTS through either a loss-of-function of potassium channels (I Ks and I Kr ) or a gain-of-function of sodium channel leading to an increase in the late I Na , lengthening the cardiac APD and manifesting as a prolonged QT interval 7 .
About 4-11% of LQTS patients host multiple mutations and typically present at a younger age with a more severe cardiac phenotype compared with individuals carrying a single mutation [8][9][10] . Patients with compound mutations were found to be associated with longer QTc, more frequent cardiac events, and earlier onset of cardiac events. However, the underlying mechanisms remain unclear.
We identified a LQTS family harboring three compound mutations in different genes: one missense mutation in KCNQ1 (R174C), one nonsense mutation in hERG (E1039X) and another missense mutation in SCN5A (E428K). To the best of our knowledge, this is the first report of LQTS associated with three different rare variants. We characterized the functional consequences of the I Ks , I Kr and I Na channels reconstituted with these three mutations in mammalian cells and provide important insight into molecular mechanisms underlying the LQTS associated with compound mutations. Specifically, we found that recombinant channel 'I Ks ' and 'I Kr ' interact and mutations in the two α subunits might produce a synergistic lesion in cardiac channel function. These findings may explain why patients with compound mutations show a more severe phenotype than those carrying a single mutation and suggest that the management of such patients should be tailored to their increased risk for arrhythmias [8][9][10][11] .

Results
Case description. The index patient was a 9-year-old boy (indicated by arrow in family pedigree of Fig. 1a), who experienced repetitive syncope while playing at school. He was identified as carrying three heterozygous mutations in three different genes: p.R174C (c.520 C > T) in KCNQ1, p.E1039X (c.3115 G > T) in hERG, and p.E428K (c.1282 G > A) in SCN5A. Figure 1b shows locations of three mutations in the relevant ion channel protein. The proband was admitted to a nearby hospital and diagnosed with LQTS. The Schwartz score was 4.5 points (T-wave alternans, notched T wave in three leads, low HR for age and syncope with stress). His basal ECG showed a negative T wave in lead III, aVF and V 1 -V 3 , and treadmill stress test uncovered a greater QT prolongation and the appearance of biphasic T wave on exercise (Fig. 1c).
The proband had a family history of syncope and QT prolongation (Fig. 1a). Clinical and genetic analysis of his familial members revealed that his 2-year-old sister also carried all three mutations and another younger sister had both KCNQ1-R174C and hERG-E1039X mutations. Both sisters showed QT prolongation, but were asymptomatic. All members with one (KCNQ1-R174C) or two compound mutations (KCNQ1-R174C + SCN5A-E428K) in his mother's family were asymptomatic and showed no QT prolongation. In LQT1 patients, QT prolongation can be sometimes detected along with the increase of heart rate 12 , however, they did not agreed for the exercise stress test. In his father's family (hERG-E1039X carriers), the grandmother (17-year-old at onset time) suffered syncope by a telephone ringing while she was sleeping, which is typical for LQT2. One of the proband's uncles experienced syncope, and 3/6 mutation carriers showed QT prolongation. Compared with QTc intervals in single hERG-E1039X carriers (440.2 ± 12.2 ms, n = 6), those in hERG-E1039X carriers with additional mutations (500.3 ± 16.7 ms, n = 3) were significantly longer (P < 0.05).

Electrophysiological study
The KCNQ1-R174C mutation produced a mild inhibitory effect on 'I Ks '. representative whole-cell current traces recorded from CHO cells expressing KCNE1 with KCNQ1-WT, KCNQ1-WT + KCNQ1-R174C and KCNQ1-R174C, respectively. Both the steady state and tail I Ks amplitude in WT + R174C conditions were mildly decreased compared to WT alone, whilst R174C KCNQ1 alone produced no currents. Figure 2d shows the current-voltage (I -V) relations for I Ks tails elicited after the voltage-step to −50 mV from various test potentials. Figure 2e summarizes I Ks tail densities measured at +30 mV. Compared with WT, WT +R174C KCNQ1 significantly decreased I Ks densities for voltages between −30 mV and +30 mV (Fig. 2d,e and Table 1). Voltage-dependent activation was quantified by fitting a Boltzmann equation to the I-V relations, and resultant data show that WT + R174C KCNQ1 significantly increased the V h value (Table 1).
Deactivation rates for I Ks were measured by depolarizing cells to +30 mV for 2 s, followed by repolarizing steps from −60 mV to −30 mV in 10-mV increments. Figure 2f shows the time constant for deactivation plotted as a function of repolarization potential. Compared with WT, WT + R174C significantly accelerated the deactivation rates between −60 mV and −50 mV (Table 1). Overall, the R174C mutation exerted a mild inhibitory effect on KCNQ1-WT channel 13 .
As the proband experienced syncopal episodes while playing with his classmates and his QTc interval was prolonged by exercise (Fig. 1c), we further tested whether KCNQ1-R174C might impair the response of I Ks to adrenergic stimulation in HEK293 cells co-expressing WT + R174C KCNQ1 with KCNE1 and Yotiao 12 . As typically shown in Fig. 2g, 100 nM isoproterenol increased I Ks by 93.5 ± 15.8% (n = 18) in cells expressing WT alone, but only mildly increased I Ks by 50 .4 ± 7.6%, (n = 13, P < 0.05 vs WT) in cells expressing WT + R174C KCNQ1. The result suggests that WT + R174C KCNQ1 partially blunted the activation of I Ks in response to isoproterenol, which is consistent with a previous report on the response of this mutant channel to forskolin in Xenopus laevis oocytes 13 . Figure 3a-c show representative whole-cell current traces recorded from CHO cells expressing hERG-WT, hERG-WT + hERG-E1039X and hERG-E1039X, respectively. I-V relations in Fig. 3d indicates that, although the steady state and tail I Kr amplitudes in E1039X hERG alone were notably decreased, the I Kr amplitudes in WT + E1039X hERG were not significantly changed (Table 2).

HERG-E1039X mutation caused an incomplete loss-of-function in 'I Kr '.
Figure 3e and f show normalized voltage dependence of activation/inactivation curves and time constants for deactivation under three different conditions, respectively. Numerical data pertaining to the biophysical properties therein are summarized in Table 2. Compared with those of WT hERG (V h : −52.50 ± 3.04 mV; τ of recovery from inactivation at −40 mV: 2.90 ± 0.20 ms; n = 18), however, the V h for the steady state inactivation of both E1039X (−64.63 ± 4.07 mV, n = 14) and WT + E1039X (−64.62 ± 3.75 mV, n = 14) showed a marked (P < 0.05 vs WT) negative shift, and the time course (τ) of recovery from inactivation at −40 mV was significantly (P < 0.05 vs WT) slower for both E1039X (3.85 ± 0.39 ms, n = 14) and WT + E1039X (3.70 ± 0.35 ms, n = 14). The pronounced hyperpolarizing shift of inactivation and slowed recovery from inactivation are likely to decrease the I Kr channel availability during excitation and to cause an incomplete loss-of-function in I Kr . The parameters for activation and time constants for deactivation were not significantly different between WT, E1039X and WT + E1039X hERG (Table 2). SCN5A-E428K increased the peak 'I Na ' currents but produced no late 'I Na '. Figure 4a shows representative whole-cell current traces recorded from CHO cells expressing hβ 1 with SCN5A-WT or SCN5A-E428K. Figure 4b shows I-V relations for peak I Na elicited by the protocol shown in the inset, and Fig. 4c summarizes peak I Na densities. Compared with WT, E428K SCN5A significantly increased peak I Na densities between −70 mV and −10 mV. The peak I Na density of E428K was 943.9 ± 93.8 pA/pF (n = 21) at −55 mV, which is significantly (P < 0.05) larger than that of WT (625.4 ± 124.4 pA/pF, n = 19,) at −50 mV (Fig. 4c). Figure 4d and f show conductance-voltage and steady state inactivation curves, representative late I Na traces recorded in the presence of 30 μM tetrodotoxin (TTX), and the properties of I Na recovery from inactivation for WT and E428K SCN5A. The mutation caused no significant changes to these parameters.
Co-expression of KCNQ1/KCNE1 with hERG. Based on the above electrophysiological findings on the three mutations, two loss-of-function mutations of potassium channels appeared to cause the clinical phenotype in this relatively large LQTS family. The proband's two sisters carrying the same combination of heterozygous compound KCNQ1 and hERG mutations showed long QT features very similar to those of the proband, suggesting that the pathogenesis of triple mutation carriers was mainly associated with the KCNQ1-R174C and hERG-E1039X. We therefore examined the interaction between KCNQ1-R174C and hERG-E1039X by co-expressing the two mutations into CHO cells. Figure 5a shows representative current traces recorded from a cell co-expressing KCNQ1-WT/KCNE1 and hERG-WT. In the presence of 1 μM E4031 (Kv11.1/I Kr inhibitor) and 2 μM HMR1556 (Kv7.1/I Ks inhibitor), the current was totally blocked, which confirms that the current was exclusively composed of KCNQ1 + KCNE1 and hERG channel currents. Figure 5b 1 Figure 5d shows I-V relations for tail currents elicited using the protocol in inset of Fig. 5b 1 in the presence of 1 μM E4031. Open circles indicate those measured for KCNQ1-WT alone. Compared with KCNQ1-WT, both KCNQ1-WT + hERG-WT (solid squares) and KCNQ1-WT + hERG-E1039X (solid triangles) significantly increased the tail current amplitudes (Fig. 5d and Table 1). In addition, KCNQ1-WT + hERG-WT caused a significant negative shift of both V h and K values of activation curve, and deactivation rates of KCNQ1-WT + hERG-WT were significantly slower than those of KCNQ1-WT between −60 mV and −50 mV (Table 1). These data suggest that hERG exerted a gain-of-function effect on I Ks when co-expressed with KCNQ1.    On the other hand, Fig. 5e shows that KCNQ1-WT + hERG-E1039X (solid triangles) significantly accelerated deactivation times between −60 mV and −50 mV (Table 1) compared with KCNQ1-WT + hERG-WT (solid squares), which suggests that hERG-E1039X led to altered I Ks kinetics when co-expressed with KCNQ1, although tail current densities were not significantly affected (Table 1). Figure 5f shows I-V relations for tail currents elicited using the protocol in inset of Fig. 5c 1 in the presence of 2 μM HMR1556. Compared with those of hERG-WT, tail currents and the V h values for the steady state inactivation of both hERG-WT + KCNQ1-WT and hERG-WT + KCNQ1-R174C were significantly lower, but the V h values for voltage-dependent activation and the recovery time from inactivation of both hERG-WT + KCNQ1-WT and hERG-WT + KCNQ1-R174C were significantly higher at −40 mV (Table 2). Taken together, KCNQ1 (including KCNQ1-R174C mutant channels) attenuated I Kr when co-expressed with hERG. In the meantime, Fig. 5f and Table 2 show that the tail current of hERG-E1039X + KCNQ1-WT (solid squares) was lower than that of hERG-WT + KCNQ1-WT (open squares) and the tail current of hERG-E1039X + KCNQ1-R174C (solid diamonds) was lower than that of hERG-WT + KCNQ1-R174C (open diamonds), which supports the above data that hERG-E1039X caused a loss-of-function in I Kr even in the condition of co-expression with KCNQ1. However, we failed to detect any significant changes in parameters between hERG-WT + KCNQ1-WT and hERG-WT + KCNQ1-R174C or between hERG-E1039X + KCNQ1-WT and hERG-E1039X + KCNQ1-R174C, which implicates that KCNQ1-R174C did not affect the function of I Kr when co-expressed with hERG.
Expression of channel tetramers on cell membrane was disrupted by either KCNQ1-R174C or hERG-E1039X but rescued by co-expression of WT. Figure 6A shows confocal images of CHO cells expressing KCNQ1/KCNE1 (upper panels) and hERG (lower panels). Both of KCNQ1-WT and hERG-WT proteins were amply transported to the cell membrane. In cells expressing KCNQ1-R174C alone, mutant proteins were mostly distributed in the cytosol but scarcely on the cell membrane, suggesting the trafficking defect of channel protein. While the hERG-E1039X mutant proteins were less presented both in cytosol and on cell membrane, implicating the inhibited protein synthesis and/or its potentiated degradation by hERG-E1039X. However, in cells expressing KCNQ1-WT + KCNQ1-R174C or hERG-WT + hERG-E1039X, channel proteins were expressed both on the cell membrane and in the cytosol, suggesting that the cell membrane expression of channel proteins was increased by the co-expression of WT subunits.  Fig. 6A, and suggest that KCNQ1 and hERG did not affect the expression pattern of proteins one-another, which is also confirmed by the merged figures in lower panels.

Discussion
Although LQTS caused by two compound mutations is relatively common, the arrhythmia associated with three different compound mutations is a rare case, which accounts for 0.2% of our LQTS cohort. The present study on the three-compound mutation case indicated that KCNQ1-R174C produced a mild inhibitory effect on I Ks and hERG-E1039X caused an incomplete loss-of-function in I Kr . In addition, the present study showed that I Ks and I Kr  interact: it is striking that hERG-E1039X caused a loss-of-function in I Ks when co-expressed with KCNQ1, which is likely to exacerbate the dysfunction of I Ks caused by KCNQ1-R174C. This result might reveal why compound mutations are associated with increased arrhythmic risk.
Of the three pathogenic mutations in the present study, KCNQ1-R174C was previously reported to be associated with both heterozygous LQT1 14 and homozygous autosomal-recessive LQT1 15 , in which the homozygous KCNQ1-R174C carrier displayed extreme QT prolongation and suffered multiple breakthrough cardiac events before succumbing to his malignant LQTS phenotype. Our clinical data show that patients with heterozygous KCNQ1-R174C or KCNQ1-R174C/SCN5A-E428K compound mutation were asymptomatic and their QTc intervals were not prolonged (Fig. 1a), suggesting that the phenotype caused by heterozygous KCNQ1-R174C is not severe and individuals carrying the same mutation exhibit diverse cardiac phenotypes clinically [13][14][15] . These results are well explained with our electrophysiological data: KCNQ1-WT + KCNQ1-R174C produced a mild inhibitory effect on I Ks channel and KCNQ1-R174 alone produced no I Ks current (Fig. 2b-d and Table 1). Immunocytochemical study show that, same as the cell surface expression pattern of another KCNQ1 mutation G269S 12 , the trafficking-deficiency in the homologous KCNQ1-R174C channel was greatly rescued by co-expression with the WT subunit (upper panels in Fig. 6A), resulting in the increased expression of channel proteins on the cell membrane. This result further explains clinical phenotypes of KCNQ1-R174C mutation carriers. Consistent with a previous finding that KCNQ1-R174C blunted the increase in I Ks with forskolin in Xenopus laevis oocytes 13 , we found that KCNQ1-R174C blunted the increase in I Ks with isoproterenol, which further confirms our previous speculation: a patient with KCNQ1 mutation showing an excessive prolongation of QT intervals on exercise is likely due to an adrenergic up-regulation of I Ca,L without concomitant up-regulation of I Ks 12 .
SCN5A-E428K was reported to be linked to atrial fibrillation (AF) 16,17 . The present clinical data show that two mutation carriers (harboring KCNQ1-R174C simultaneously) in proband's mother family were asymptomatic and did not exhibit QTc interval prolongation (Fig. 1a). Electrophysiological study revealed that SCN5A-E428K increased peak I Na but did not affect the late I Na , which indicates that this mutation might be associated with such genetic disorders as AF rather than LQTS because the increase of the late I Na is a characteristic indicator for LQT3 18,19 . In addition to above data, we predicted the pathogenicity of substitutions in SCN5A-E428K and KCNQ1-R174C mutations through the PolyPhen-2 system 20 . The results show that SCN5A-E428K is relatively benign, whereas the KCNQ1-R174C is strongly considered to be damaging.
HERG-E1039X is a novel nonsense mutation in distal C-terminus. Clinical data show that two of hERG-E1039X mutation carriers experienced syncope and half of the mutation carriers showed QT prolongation in proband's father family (Fig. 1a). Functional analysis indicates that hERG-E1039X mutation shifted the inactivation curve of I Kr in the hyperpolarizing direction and decelerated the time of recovery from inactivation (Table 2), while other gating kinetics and the current density were not significantly affected. Previous studies postulated that nonsense mutations in hERG cause abnormal transcription/translation of I Kr 21 . The present immunocytochemical data show that, in cells expressing hERG-WT + hERG-E1039X, channel proteins were amply expressed both on the cell membrane and in the cytosol (lower panels in Fig. 6A), suggesting that channel protein expression in heterozygous channels was very similar to that in WT channels although E1039X mutant alone disrupted the biosynthesis and/or degradation of hERG channel protein. The distinguishing features of I Kr kinetics are the rapid voltage-dependent inactivation and recovery from inactivation, subsequently coupling with a slow deactivation 21,22 . Most LQT2-causing mutations associated with abnormal channel gating/kinetics are involved in the accelerated deactivation 21,23,24 . Only a few studies reported that the loss-of-function in I Kr caused by hERG channel pore missense mutations (V644L 24 , G584S 25 , V603L and A614V 26 ) was associated with channel inactivation. The present study provides evidence that E1039X, a nonsense mutation located in hERG's distal C-terminus, caused LQT2 through inactivation mechanism, which gives us two notions: (1) in addition to affecting gene transcription/translation, a nonsense mutation in hERG can lead to LQT2 through disrupting inactivation gating of I Kr ; (2) the distal C-terminus is also involved in the inactivation in I Kr . In the present study, we cannot rule out the possibility that a nonsense-mediated mRNA decay (NMD) is involved in the phenotype of patients carrying hERG-E1039X. The position of E1039X is close to the other two hERG nonsense mutations (W1001 × and R1014X) which were reported to degrade mutant mRNA by NMD and to be associated with LQT2 21 .
The functional interaction between I Kr and I Ks is still in dispute [27][28][29] . Ren et al. showed that transiently expressed WT or mutant KCNQ1 downregulated Kv11.1 in both CHO and HEK 293 cells stably expressing hERG and the interactions of two channels occurred via the C terminus 30 , whose findings are consistent with the present study: both KCNQ1-WT and KCNQ1-R174C mutant decreased hERG channel currents when co-expressed with hERG in CHO cells. When co-expressed with KCNQ1, C-terminal mutation hERG-E1039X disrupted KCNQ1/ KCNE1 currents, whist hERG-WT produced a different effect in KCNQ1/KCNE1 currents. These data also support the above study that hERG C-terminus is involved in the interaction between I Kr and I Ks
It is well known that compound mutation carriers exhibit a more severe phenotype than those with a single mutation [8][9][10][11] . Westenskow at al found that a KCNQ1 mutant and a KCNE1 mutant could lead to cumulative lesion in I Ks 8 . Biliczki et al. reported that I Ks blocker chromanol 293B alone did not markedly lengthen dog ventricular APD, however, when repolarization had already been prolonged by I Kr blocker dofetilide, inhibition of I Ks with same concentration of chromanol 293B substantially delayed repolarization 31 . Their data suggest a synergistic prolongation of repolarization produced by I Kr and I Ks blockade. In the present experiment, hERG-E1039X caused a loss-of-function in KCNQ1/KCNE1 channels. This result indicates that a mutation in hERG not only can disrupt I Kr but can worsen I Ks function and superimpose to cause a synergistic lesion to the defective I Ks encoded with a mutant KCNQ1, leading to further prolongation of APD and the QT interval. Our in silico study also confirms that the synergistic effects of KCNQ1-R174C and hERG-E1039X in I Ks could prolong APD markedly (see Supplementary Material). Therefore, although phenotype of heterozygous KCNQ1-R174C carriers are mild, patients harboring additive pathogenic mutation hERG-E1039X showed more severe QT interval prolongations because of superimposed I Ks lesion caused by hERG-E1039X mutation and the proband even experienced a syncope evoked by exercise 32  above functional consequence, we suggest that compound pathogenic mutation carriers should be tailored to their increased risks for arrhythmias because these patients are more readily to be predispose to fatal arrhythmias. For example, a hERG-KCNQ1 compound mutation carrier should avoid QT-prolonging medications and swimming.

Conclusion
We identified a LQTS family harboring three pathogenic mutations in different genes and characterized the functional consequences of related three mutant channels. The synergistic lesion caused by different pathogenic mutation is very likely why patients with compound mutations showed a relatively more severe phenotype.

Methods
Clinical investigation and genetic testing. The study population consisted of 1,015 consecutive LQTS probands whose diagnosis was referred to the criteria of Schwartz et al. 1 . The protocol for genetic analysis was approved by the Institutional Ethics Committee of Shiga University of Medical Science and performed under its guidelines. Written informed consent was obtained from every subject before analysis, in accordance with the last version of the Declaration of Helsinki and with recommendations by the local ethics committee. Genomic deoxyribonucleic acid (DNA) used for genetic evaluation was isolated from venous blood lymphocytes. Genetic screening for mutations in LQTS-related genes including KCNQ1, hERG, SCN5A, KCNE1, KCNE2, KCNJ2 and CACNA1C was conducted by denaturing high-performance liquid chromatography (WAVE system, Transgenomic Inc., Omaha, Nebraska). For abnormal screening patterns, sequencing was performed with an automated sequencer (ABI PRISM 3100×, Applied Biosystems, Foster City, California).

Electrophysiological recordings and data analysis. Forty eight hours after transfection, cells attached
to a glass coverslip were transferred to a 0.5-ml bath chamber perfused with extracellular solution and maintained at 25 °C (for I Ks and I Kr ) or at 22-23 °C (for I Na ). Patch-clamp experiments were conducted on GFP positive cells. Whole-cell membrane currents were recorded with an EPC-8 patch-clamp amplifier (HEKA, Lambrecht, Germany) and a resistance of 2.5 to 3.5 MΩ (I Ks and I Kr ) or 1.5 to 2.0 MΩ (I Na ) electrodes.
Currents were evoked by depolarizing voltage-clamp steps given from a holding potential of −80 mV (I Ks and I Kr ) or −120 mV (I Na ) to various test potentials. Amplitudes of both I Ks and I Kr were determined by measuring the amplitude of tail current. All currents were normalized to the cell membrane capacitance to obtain current densities (pA/pF). The protocols used for the assessment of the voltage dependence of activation/inactivation and recovery from inactivation are provided as insets in the relevant figures. Voltage-dependence of activation/ inactivation were evaluated by fitting the I-V relation of currents to a Boltzmann as previously described 12 . The deactivation kinetics of I Ks after depolarization and the recovery from inactivation data of I Kr were determined by a single exponential equation: Y(t) = A 0 + A exp(−t/τ). The deactivation kinetics of I Kr after depolarization, the recovery from inactivation of I Na and decay phase of I Na data were fitted with a bi-exponential function of the form: Y(t) = A 0 + A f exp(−t/τ f ) + A s exp(−t/τ s ), where A f and A s are the fractions of fast and slow relevant components, respectively. The persistent inward (late) I Na , a hallmark of biophysical abnormality in LQT3, was determined as the tetrodotoxin (TTX, 30 µM)-sensitive current measured after 800 ms of depolarization at −20 mV.