Blockade of Permeation by Potassium but Normal Gating of the G628S Nonconducting hERG Channel Mutant

Abstract

G628S is a mutation in the signature sequence that forms the selectivity filter of the human ether-a-go-go-related gene (hERG) channel (GFG) and is associated with long-QT2 syndrome. G628S channels are known to have a dominant-negative effect on hERG currents, and the mutant is therefore thought to be nonfunctional. This study aims to assess the physiological mechanism that prevents the surface-expressing G628S channels from conducting ions. We used voltage-clamp fluorimetry along with two-microelectrode voltage clamping in Xenopus oocytes to confirm that the channels express well at the surface, and to show that they are actually functional, with activation kinetics comparable to that of wild-type, and that the mutation leads to a reduced selectivity to potassium. Although ionic currents are not detected in physiological solutions, removing extracellular K⁺ results in the appearance of an inward Na⁺-dependent current. Using whole-cell patch clamp in mammalian transfected cells, we demonstrate that the G628S channels conduct Na⁺, but that this can be blocked by both intracellular and higher-than-physiological extracellular K⁺. Using solutions devoid of K⁺ allows the appearance of nA-sized Na⁺ currents with activation and inactivation gating analogous to wild-type channels. The G628S channels are functionally conducting but are normally blocked by intracellular K⁺.

Introduction

The human cardiac delayed-rectifier K⁺ current (I_Kr) is carried by a channel encoded by the human-ether-a-go-go-related gene (hERG) (1–3), and this current contributes to repolarization of the cardiac action potential. Defects in channel surface expression or biophysical properties (4) can lead to the long-QT2 syndrome (5), which is caused by mutations (mostly missense) in the hERG gene. Although long-QT2 mutations in the pore are known to have more severe symptoms (6,7), and despite the fact that several long-QT2 mutations are located in or around the selectivity filter of hERG channels (8–12), only one long-QT2 mutation in this area, N629D, has been studied electrophysiologically. The mutation has been shown to lead to a change in the selectivity to K⁺ (13), but its effect is mostly due to decreased trafficking (7,14). Information is thus limited, probably because the study of the selectivity filter (SF) of hERG channels has been restricted due to the lack of expression of several mutations along the SF (15).

The G628S mutation was first detected in 1995 in a sporadic case of long-QT syndrome (16), and was found again later in patients with prolonged QT intervals (8,11). It is interesting that hERG G628S channels expressed in Xenopus oocytes or mammalian cells do not show ionic currents, which leads to the conclusion that they are nonfunctional (17,18), despite the fact that they process to the cell surface in a number comparable to that found for wild-type (WT) (14,18–20). When coexpressing G628S and WT proteins, a single G628S subunit suffices to render heterotetramers nonfunctional (17), so that the G628S causes a dominant-negative effect on normal hERG subunits (17,19–21). This property of the channel makes it an attractive tool for therapeutic research on the treatment of arrhythmias (22–25). The G628S mutant has also been shown to downregulate I_Ks in transgenic rabbits and in vitro (20,26).

Nonconduction of the G628S mutant makes the hERG channels relatively different from other Kv channels, where the same G→S mutation does not prevent conduction but the selectivity for potassium versus sodium is dramatically reduced (27,28). The study of the SF area is essential, as it is thought to be directly related, to some extent, to fast and unstable P-type inactivation (29,30) and/or slow C-type inactivation of hERG channels (31).

In this study, we used voltage-clamp fluorimetry (VCF) to follow time- and voltage-dependent fluorescence signals that report on voltage-sensor gating and pore opening/closing of hERG channels (32), to determine whether changes in the kinetics of activation of the G628S channel are responsible for its nonfunction. Once the proper function of the channel was confirmed, we determined the reason why the channels do not conduct detectable ionic currents in physiological conditions, using both Xenopus oocytes and HEK-293-derived TSA201 cells. We thus demonstrate that the major effect of the mutation is a block of conduction by intracellular K⁺. The data also reveal that selectivity of the G628S channel for sodium is increased.

Materials and Methods

Ethics statement

All animal protocols were performed in accordance with University of British Columbia animal care guidelines, which conform to regulations set out by the Canadian Council of Animal Care.

Cell preparation

Molecular biology and RNA preparation

Synthesis of oligonucleotide primers, generation of mutations, and preparation of RNA were performed as previously reported (32). For VCF experiments, a cysteine was introduced by site-directed mutagenesis in the S3-S4 linker at E519 in a WT or a G628S mutant background. The two endogenous extracellular cysteines (C445 and C449) in the S1-S2 linker reported in hERG channels to attach tetramethyl rhodamine methyl ester fluorescent probe (32) were replaced by valines. The introduction of the mutations C445V/C449V/E519C (E519C/C-less) did not affect the properties of the G628S channels (Fig. S1 in the Supporting Material).

Oocyte preparation and injection

Oocytes were prepared as previously reported (32). VCF recordings were performed three to seven days after injection.

TSA201 cells preparation and transfection

TSA201 cells were cultured in a solution of modified Eagle's medium + 10% fetal bovine serum (Gibco, Carlsbad, CA). Cells were transfected with 3 μg of channel DNA (WT or G628S in pGW1H vector) and 1.2 μg green fluorescent protein using Lipofectamine 2000 transfection reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. Patch-clamp recordings were performed 24–48 h after transfection.

Electrophysiology

Two-electrode voltage clamp and VCF

Two-electrode voltage clamping and VCF were performed as previously reported (32). To reduce endogenous currents from oocytes, the CaCl₂ concentration was reduced from 2 mmol/L to 0.18 mmol/L in the K⁺-free solution (17) (modified-ND99 solution).

Patch clamp

The standard/physiological bath solution contained (in mmol/L) 135 NaCl, 5 KCl, 1 MgCl₂, 2.8 sodium acetate, 10 HEPES, and 1 CaCl₂, adjusted to pH 7.4 using NaOH. The standard/physiological pipette filling solution contained (in mmol/L) 130 KCl, 5 EGTA, 1 MgCl₂, 10 HEPES, 4 Na₂ATP, and 0.1 GTP, adjusted to pH 7.2 with KOH. For the equimolar Na⁺ experiments, the intracellular solution contained (in mmol/L) 135 NaCl, 0.1 KCl, 1 MgCl₂, 10 dextrose, and 10 HEPES, adjusted to pH 7.4 with NaOH. The composition of the extracellular solution was the same, except that 1 mmol/L CaCl₂ was added. To measure the inhibitory effect of K⁺, extracellular or intracellular KCl was replaced by NaCl in a reciprocal manner. Unless otherwise stated, all chemicals were purchased from Sigma Aldrich (Mississauga, ON, Canada). Whole-cell current recording and data analysis were done using an Axopatch 200B amplifier and pClamp 9 software (Axon Instruments, Foster City, CA). Patch electrodes were prepared with resistances in the range 3–5 MΩ.

Data analysis

The inhibition-response curve was fit with a sigmoid relationship using nonlinear regression analysis with the program prism (GraphPad, San Diego, CA).

Conductance-voltage (G-V) relationships were derived by plotting peak tail currents as a function of the preceding depolarizing pulse. Fluorescence-voltage (F-V) curves were derived as stated in the figure legends. G-V and F-V curves were fit with a single Boltzmann function:

$$y=1/\left(1+\exp \left(\left(V_{1/2}-V\right)/k\right)\right),$$

where y is the Δ fluorescence or the conductance normalized with respect to the maximal Δ fluorescence or conductance, V_1/2 is the half-activation potential, V is the test voltage, and k is the slope factor. Unless otherwise indicated, data reported throughout the text and figures are presented as mean ± SE.

Results

G628S channels express well at the surface and show activation gating kinetics comparable to WT

Fig. 1 shows an alignment of the region of the SF from hERG, Kv1.2, and Shaker channels. This area is very well conserved among voltage-gated potassium channels and contains the GF/YG signature of the SF. The G628 residue forms part of the extracellular S0 ion binding site, which serves as an interface between the SF and the extracellular medium. The equivalent mutation to G628S in Shaker or Kv1.4 channels leads to a change in the selectivity to K⁺ (27,28). The same effect might be expected in hERG, as the channel is processed to the surface the same way as WT (14,18–20). It is then surprising that the channel is apparently nonfunctional according to those studies.

Figure 1.

Location of the G628 residue. (A) Sequence alignment of residues in the region of the SF of hERG, Kv1.2, and Shaker channels. The GF/YG signature is highlighted in bold and the G628 residue of hERG appears on a black background. (B) Side view of the Kv1.2 channel model in the open state (modified from Pathak et al. (61)). Only the S5 and S6 segments from two subunits are shown for clarity. Segments are represented as ribbons, the SF is in black, and G378 (which corresponds to G628 in hERG) is highlighted.

The VCF technique has been shown to provide reliable information on voltage-sensor rearrangements and on other conformational changes associated with pore opening/closing of hERG channels (32). To better understand the lack of function of the G628S mutant, the same technique was used to determine whether those gating and pore-opening properties were altered in the mutant. A cysteine was introduced at position 519 at the top of the S4 segment of hERG WT and G628S proteins expressed in Xenopus oocytes and labeled with tetramethyl rhodamine methyl ester probe. Significant fluorescent signals could be elicited from the mutant channel by depolarizations up to 60 mV from −80 mV, and repolarization to −110 mV. It came as a surprise that those records present overall characteristics very similar to those of WT (Fig. 2 A), suggesting functional activation gating and functional pore opening/closing. Two important features can be extracted from these data: the F-V_On (obtained from the inward signal upon depolarization), related to voltage-sensor rearrangements, and the F-V_Off (inward signal upon repolarization) associated with conformational changes that occur during closing of the channels (32). Comparing those curves between WT and G628S (Fig. 2, B and C), a significant 27-mV left shift in the F-V_On (p < 0.0001) was paralleled by a significant 20-mV left shift in the F-V_Off (p < 0.0001), reflecting the energetic coupling between the two events. It is important to note that the average size of the signals did not differ significantly between WT and G628S (p = 0.9), suggesting that the relative expression at the surface of both types of channels is comparable (Fig. 2, D and E).

Figure 2.

Comparison of fluorescence signals from WT (E519C/C-less)- and G628S (G628S/E519C/C-less)-expressing oocytes (gray and black traces, respectively). (A) Representative traces obtained from cells held at −80 mV, depolarized to potentials from −100 to +60 mV for 2 s, then repolarized to −110 mV. (B) ΔF, measured at the maximum amplitude of the inward signal upon depolarization, was plotted against voltage (squares, F-V_On). A Boltzmann function was used to fit the data. WT F-V_On: V_1/2 = −38.2 ± 1.8 mV, k = 14.7 ± 0.9 mV, n = 13; G628S F-V_On: V_1/2 = −65.8 ± 3.4 mV, k = 16.6 ± 2.4 mV, n = 5. (C) ΔF measured at the maximum amplitude of the inward signal upon repolarization (diamonds, F-V_Off) and plotted against voltage. Data were fit with a Boltzmann function. WT F-V_Off: V_1/2 = −20.6 ± 1.2 mV, k = 11.4 ± 0.6 mV, n = 16; G628S F-V_Off: V_1/2 = −39.3 ± 3.7 mV, k = 11.7 ± 1.5 mV, n = 5. (D) Representative fluorescence traces showing similar size of signals for G628S (black) and WT (gray). Cells were held at −140, depolarized to +30 (G628S) or +50 (WT) mV for 300 ms and repolarized to −140 mV. (E) Average F-V_off sizes for G628S (−2.5 ± 0.9% ΔF/F, n = 3) and WT (−2.6 ± 1% ΔF/F, n = 3) using the same protocol as in D. The fluorescence sizes are not significantly different (unpaired t-test).

G628S channels show altered selectivity for sodium versus potassium in oocytes

Since G628S channels expressed well at the oocyte surface, and their pore gating seemed to be similar to that of WT, the presence of conducting channels was assayed more than a week after RNA injection. A modified ND99 solution (extracellular K⁺-free solution) was used in the bath, due to the reported change in selectivity of the same mutant in other Kv channels (27,28). The predictable outward tail currents recorded from WT channels in those conditions (data not shown) were small due to the fact that extracellular sodium potently inhibits hERG channels when extracellular K⁺ is removed (33,34). Small currents were obtained from the G628S mutant, with inwardly directed tail currents (Fig. 3, A and B), suggesting that they are sodium-dependent. There was almost no contamination by endogenous currents as shown in uninjected oocytes (Fig. 3 B).

Figure 3.

G628S channels expressed in Xenopus oocytes show reduced selectivity to K⁺. (A) Comparison of ionic currents from WT and G628S channels expressed in oocytes. G628S currents were obtained in a K⁺_o-free, low-calcium extracellular solution (modified ND99), whereas WT currents were obtained in ND96. Membranes were depolarized from −80 mV to potentials ranging from −100 to +60 mV for 2 s, then repolarized to −110 mV. (B) Currents from uninjected- and G628S RNA-injected oocytes, recorded in modified ND99 solution. Oocytes were depolarized from −110 mV to potentials ranging from −110 to +90 mV for 200 ms, then repolarized to −110 mV. Currents were filtered offline (6 kHz). (C) The amplitude of G628S tail currents were normalized to the maximum and plotted against voltage. G-V_1/2 = −38.6 ± 3.4 mV, k = 11.6 ± 0.6 mV, n = 5. (D) Determination of the reversal potential of G628S current using the same solution as in B. Representative currents, evoked by a 1.2-s depolarization to +60 mV from a holding potential of −80 mV, followed by repolarization to potentials ranging from 20 to −110 mV for 4 s. (E) The maximum amplitude upon repolarization was plotted against voltage. E_rev = −33.7 ± 7.7, n = 3. The short lines to the left of the current traces denote zero current.

The G628S inward tail currents had a G-V_1/2 of ∼−39 mV (Fig. 3 C), close to that obtained from the F-V_off (−39.3 mV) in the fluorescence experiment (Fig. 2 C). Considering that [Na⁺]_i is estimated to be ∼15 mmol/L in Xenopus oocytes (35–38), E_Na should be ∼+47 mV in the modified ND99, and so outward currents upon depolarization obtained from G628S channels are expected to be mainly carried by K⁺. A two-pulse tail current protocol allowed measurement of a mean reversal potential of −33.7 mV (Fig. 3, D and E), suggesting that G628S channels have an altered selectivity to potassium but are able to conduct both potassium and sodium ions.

The G628S channels conduct in equimolar sodium solutions but not in physiological K⁺ conditions in TSA201 cells

To further characterize the properties of G628S channels while more easily manipulating the intracellular/extracellular solutions, the mutant was expressed in HEK293-derived TSA201 cells. Again, although WT channels were functional in standard K⁺/Na⁺ solutions, only an endogenous delayed-rectifier K⁺ current could be detected from G628S-expressing cells, as expected (n = 9) (Fig. 4 A). Considering the ability of the mutant to conduct sodium, shown in the oocyte experiments, recording in equimolar Na⁺ solutions was performed. No currents could be observed through WT channels in the absence of K⁺, as they are blocked by extracellular sodium (31,34). However, robust currents were recorded from G628S channels (Fig. 4 B). The voltage dependence of the conductance was again significantly left-shifted in the G628S mutant compared to WT (Fig. 4 C), with a G-V_1/2 of −30 mV, compared with −15 mV in WT (p = 0.008). The reversal potential was 0.96 mV, close to the Nernst equilibrium potential for sodium (Fig. 4 D).

Figure 4.

G628S channels conduct in equimolar Na⁺ solutions. (A) Representative traces of ionic currents obtained from WT and G628S hERG channels expressed in TSA201 cells, in standard solutions. Cells were held at −80 and stepped to voltages between −100 and 80 mV, then repolarized to −110 mV. (B) Representative ionic currents from WT channels (left; n = 4) and G628S channels (right) in equimolar Na⁺ solutions. Protocol was the same as in A. The short lines to the left of the current traces denote zero current. (C) Amplitude of tail currents (solid circles in A and B) were normalized to the maximum and plotted against voltage. WT G-V_1/2 = −15.0 ± 2.5 mV, k = 8.2 ± 0.2 mV, n = 9; G628S G-V_1/2 = −30.1 ± 4.3 mV, k = 11.2 ± 1.5 mV, n = 9. (D) A double-pulse protocol was used to obtain the reversal potential of G628S channels in equimolar Na⁺ conditions (representative traces are shown in the inset). Channels were activated by a voltage step to 60 mV, and repolarized to voltage steps from 30 to −120 mV. Peak currents (open circles) upon repolarization were plotted against voltage. The reversal potential was 0.96 ± 3.8 mV, n = 11.

The activation kinetics of G628S channels was unaltered compared to WT at potentials where the Po approached 1.0 (Fig. 5). Activation was measured during a tail envelope for WT channels after pulses to +40 mV in standard solutions, and in 135 mM Na⁺_o for G628S to maximize current size. Activation time constants were ∼70 ms at +40 mV for both channels (Fig. 5 B). Deactivation kinetics was measured on repolarization and was significantly slower in G628S channels than in WT channels. This effect was much greater positive to −80 mV and almost certainly reflects the more negative G-V_1/2 of G628S compared with WT channels. These data fully support the fluorescence findings in showing that G628S activation kinetics are not markedly altered from WT, apart from the hyperpolarizing shift in the S4 gating.

Figure 5.

Kinetics of activation and deactivation in WT and G628S channels. (A) Representative current traces recorded using an envelope-of-tails protocol, obtained from WT in standard solutions (gray) and from G628S channels in 140NMDG⁺_i/135Na⁺_o solutions (black). Membrane potential was stepped to 40 mV from −80 mV for varying durations and repolarized to −120 mV. (B) The peak tail currents were plotted against test-pulse duration, and fitted using a single exponential. (C) Representative WT and G628S ionic currents recorded in TSA cells, using a double-pulse protocol (shown). (D) Comparison of time constants of deactivation in WT and G628S channels. Decay of tail current upon repolarization (triangles in C) was fitted with a double exponential for WT or a mono- or double exponential, depending on the voltage, for the mutant. Fast time constants were plotted against voltage.

G628S channels are not permanently P-type-inactivated

The ability to record G628S currents in the presence of Na⁺ but not in the presence of physiological K⁺ concentrations is immediately reminiscent of the permanently P-type-inactivated W434F Shaker (39), W472F Kv1.5 (40), or F627Y permanently P-type-inactivated hERG channels (31,41). However, the F627Y mutant shows large K⁺-dependent tail currents when 140 mM extracellular K⁺ is used, due to the ability of high K⁺_o to inhibit P/C-type inactivation in both Kv (42–44) and hERG channels (45–47) through a foot-in-the-door mechanism (43).

To determine whether the G628S channels are nonconducting due to constitutive inactivation, ionic currents were recorded in symmetrical 140 mM K⁺ solutions. However, only small endogenous currents could be observed in these conditions in both untransfected and G628S-transfected cells (Fig. 6 A, n = 6). In addition, it appears that G628S channels undergo rapid inactivation upon depolarization, just like WT channels, which is relieved by hyperpolarization. A triple-pulse protocol to determine the voltage dependence of fast P-type inactivation (48) shows that, like WT, the amplitude of current after a brief hyperpolarization is bigger than that during the initial depolarization. This suggests that the channels are recovering from fast inactivation during the hyperpolarization, and thus that the initial outward currents are through open channels. The voltage dependence of fast P-type inactivation obtained from this protocol is shown in Fig. 6 C and, like the activation parameters, is shifted in a hyperpolarized direction compared with WT.

Figure 6.

Voltage dependence of P-type inactivation of G628S channels. (A) Currents from untransfected and G628S-transfected TSA cells exposed to 140 mM K⁺_o. Cells were pulsed to between −100 and +80 mV and repolarized to −120 mV. (B) To determine the voltage dependence of inactivation, cells held at −80 mV were depolarized to +40 or +80 mV to fully open-inactivate the channels, repolarized to potentials ranging from −140 to +10 mV for 20 ms to allow recovery from inactivation, depolarized again to +40 or +80 mV and repolarized back to −80 mV. Representative currents from WT and G628S channels are shown. (Insets) Brief period during the recovery and subsequent reinactivation of channels. (C) The maximum amplitude upon the second depolarization (B, arrow) was normalized and plotted against voltage. WT V_1/2 = −62.1 ± 4.3 mV, k = −22.5 ± 1.6 mV, n = 7; G628S V_1/2 = −88.0 ± 3.8 mV, k = −25.7 ± 1.5 mV, n = 3 (p < 0.01).

Further evidence for the idea that G628S channels are not permanently P-type-inactivated is shown in Fig. 7. Using high extracellular Na⁺ and intracellular NMG to maximize the size of currents, inward G628S currents carried by Na⁺ are apparent during voltage steps to increasingly positive potentials. At negative potentials, inward currents activate relatively slowly during depolarizations and show little inactivation. Tail currents show an immediate spike due to the change in driving force but then immediately relax without a hook. In contrast, at more positive step potentials, inward currents activate more rapidly but then decay back to the baseline. Also, G628S current tails on repolarization show a very prominent hook as channels recover from inactivation. This is clearly seen in Fig. 7 A (lower), and enlarged in the inset of the current hooks, where the +60-mV current trace crosses over the −50-mV trace during the depolarization. The relation of peak to current voltage obtained from the current at the end of the depolarizing pulse shows the effect of inactivation on peak outward current (Fig. 7 B).

Figure 7.

G628S channels conduct through the open state and recover normally from inactivation. (A) Voltage dependence of activation of G628S channels in 140NMDG⁺_i/135Na⁺_o solutions. Cells were held at −80 and stepped to voltages between −100 and 60 mV, then repolarized to −120 mV. Current tracings at +60 and −50 mV are shown below. (B) The amplitudes of ionic current at the end of the depolarizing pulse (triangles) and the peak tail currents upon repolarization (diamonds) were plotted against voltage. (C). Representative WT and G628S ionic currents recorded in TSA cells using a double-pulse protocol (shown). (D) Comparison of time constants of recovery from inactivation in WT and G628S channels. The rising phase upon repolarization (C, squares) was fitted with a single exponential, and the time constants were plotted against voltage.

The kinetics of recovery from inactivation was measured using a double-pulse protocol (Fig. 7 C) as the rising phase of current during repolarization, and as shown in Fig. 7 D, recovery was slowed in G628S channels compared with WT.

G628S channel permeation is inhibited by both intracellular and high extracellular potassium concentrations

In TSA201 cells, G628S channel currents are robust in the presence of physiological K⁺_o concentrations (5 mM) (Fig. 8 A and Fig. S2) when only Na⁺ is present intracellularly. However, using elevated K⁺_o concentrations of 140 mM (n = 7), 120 (n = 10), 100 mM (n = 5), or even as low as 35 mM (n = 7), inhibited the recording of inward or outward ionic current from G628S channels. This inhibition was progressively relieved by replacing K⁺_o with Na⁺_o (Fig. 8 B).

Figure 8.

G628S channels are blocked by intracellular K⁺. Cells expressing G628S channels were held at −80 mV, depolarized to 40 mV, and repolarized to −120 mV. (A) Representative currents recorded in 135Na⁺_i/135Na⁺_o//5K⁺_o solutions. (B) Progressive replacement of K⁺_o with Na⁺_o permits conduction of the G628S channels (dark gray to black). (C) Densities of tail currents obtained in extracellular standard (135Na⁺_o//5K⁺_o) solution plotted against the concentration of K⁺ in intracellular solution (n = 3-8). Intracellular concentrations of Na⁺ and K⁺ were varied in a reciprocal manner. The curve was fit with a sigmoidal inhibition relationship (IC₅₀ = 10.3 mM). (D) Evolution with time of the ionic current obtained from G628S-expressing cells, recorded every 7 s starting after whole-cell in equimolar Na⁺. Every other sweep is shown for clarity. Cells were held at −80 mV, depolarized to 40 mV, and repolarized to −120 mV. Note the increasing inward Na⁺ tail current as intracellular K⁺ is dialyzed out. The short lines to the left of the current traces denote zero current.

In another series of experiments, we varied the intracellular K⁺ concentration while using a standard 135 Na⁺_o//5K⁺_o extracellular solution (Fig. 8 C). It is interesting that only when intracellular potassium was reduced (IC₅₀ = 10.3 mM) could we record ionic currents from G628S channels, supporting the idea that in physiological conditions, those channels are functional but are blocked by intracellular K⁺. This was confirmed when recording the evolution of ionic current with time after whole-cell rupture (Fig. 8 D). Immediately after getting whole-cell access, no tail current could be recorded, suggesting that the channels were still blocked. A gradually increasing inward current was then observed as the intracellular potassium was diluted with the pipette solution, and this current reached its maximum ∼1.5 min later.

Discussion

The primary objective of this study was to understand the physiological mechanisms that prevent normally synthesized and expressed hERG LQT2 mutant channels (G628S) from conducting ions at the cell surface. Data from the VCF technique demonstrate that the mutant channel shows activation gating comparable to that of WT. The results from patch-clamp experiments in mammalian cells demonstrate the ability of the G628S channels to be conductive only in conditions in which the intracellular K⁺ concentration is dramatically reduced.

hERG channels can enter a fast-inactivated state (P-type) that is essential for their normal contribution to cardiac repolarization and allows permeation of Na⁺ ions when extracellular K⁺ is removed. The turret region (outer pore) mainly contributes to this state, as mutations along the structure lead to altered fast-inactivation (49) and to changes in the voltage dependence of this inactivation process (13,30,48,50). A mutant in the SF (F627Y) has been previously shown to conduct K⁺ only transiently upon depolarization in physiological solutions, as the channels are then subjected to an immediate entry into a P-type inactivated state. This mutant allows sodium permeation in symmetric sodium solutions through P-type inactivated channels, and allows potassium permeation in equimolar K⁺ solutions through the open state (31). The studies of F627Y hERG channels were modeled on W434F mutant Shaker channels, which are also nonconducting in physiological conditions but conducting in symmetrical Na⁺ solutions, because they are permanently P/C-type inactivated (39). A greatly reduced P_K/P_Na, combined with a high intracellular K⁺ normally prevents permeation of Na⁺ through the channel (44). The W434F mutation also shows voltage-sensitive activation and deactivation, albeit slower than that seen in WT. However, it is important to note that the different locations of the mutations, as well as major differences between hERG and Shaker SF structures (29), limit further comparison between the two mutants.

These mutant channels have some similarities with G628S, which is nonconducting in physiological K⁺ conditions but does conduct Na⁺ in equimolar Na⁺ solutions (Fig. 8). However, no K⁺ current can be recorded in equimolar K⁺ solutions in G628S (Fig. 6). The lack of effect of high K⁺_o suggests that ultrarapid inactivation is not responsible for the inability of G628S channels to conduct in physiological solutions. This is supported by the inactivation behavior of G628S in Na⁺, which is quite like that observed for WT channels in K⁺ (Fig. 6 B) but unlike the case for F627Y channels. The fast P-type inactivation relationship appears very similar to that of WT (Fig. 6 C), and a hook, representative of recovery from fast inactivation in hERG channels, can be seen in G628S (Figs. 4, 5, A and C, 6 B, 7, and 8). Also, recovery from inactivation occurs on the millisecond timescale, which is characteristic of P-type inactivation (Fig. 6 B and 7 D). A slower inactivation process can be seen at more depolarized potentials in Figs. 3 and 4. This is probably due to the fact that constriction of the SF is expected in situations where extracellular K⁺ concentration is reduced (31,51). These data all suggest that P-type inactivation occurs in this mutant rather like it does in WT channels.

The experiments discussed here suggest that G628S channels conduct sodium through the open state but fail to conduct K⁺ due to a change in selectivity. The fluorescence experiments revealed fundamentally normal activation-gating behavior related to both S4 and pore motions (Fig. 2). The activation and deactivation kinetics of Na⁺ currents are almost identical to that of WT K⁺ currents when the hyperpolarizing shift of voltage-dependent activation in G628S is taken into account (Figs. 4 and 5), again supporting the idea that G628S channels show an alteration of open-state permeability rather than a change in gating state. This suggests that, as in the case of the W434F channels (52), the hERG G628S mutant might prove useful for the recording of gating currents in physiological solutions.

The SF is known to possess a structure prone to accommodate dehydrated K⁺ ions, whereas Na⁺ ions do not fit as well (53). In contrast to other Kv channels, which possess a GYG signature that allows specific interactions with the remainder of the pore, hERG channels have a GFG motif, which might lead to a less rigid structure, facilitating a change to its K⁺/Na⁺ selectivity (54). This may explain why mutations in or close to the channel pore lead to an increased permeability to sodium (13,49). hERG channels have a high selectivity for K⁺ (29), with an approximate P_Na/P_K between 0.002 and 0.005 (49,55). In this article, we show that this high selectivity is lost when G628 is mutated into a serine. This change of selectivity can be seen in Fig. 3 E, where the reversal potential suggests a conduction of both Na⁺ and K⁺ (see also Fig. S2). The tendency of this outer pore location to result in channels with less selectivity to K⁺ is also observed in the noninactivating double mutant G628C/S631C (48).

The experiments here demonstrate that the nonconducting state of the G628S channel is a result of block by intracellular K⁺ (Fig. 8). Although an in-depth explanation of this phenomenon is not possible at this time, it may be related to a mechanical obstruction by the ions or an altered structural conformation in the presence of K⁺_i. The hERG channel's SF S0 and S1 sites are not selective for K⁺ (56), whereas the S2 site is. In parallel, it has been suggested from studies on Kv1.2 channels that the binding pocket at S4 is deeper than that at S0, which results in a favored outward direction for permeation (57). A potential explanation would thus be that intracellular K⁺ ions have access to the S2 site, where, due to their higher affinity for the site, they are able to block the ingress of sodium ions. Another explanation might reside in a comparison of hERG channels with MthK channels, and in the anomalous mole-fraction effect (58). In low K⁺/high Na⁺ conditions, a single K⁺ ion binds in site S1 or S3 of MthK, instead of two ions binding in the S1/S3 or S2/S4 configuration. This prevents Na⁺ permeation, which would otherwise occur in K⁺_o-free conditions (59). Paralleling this model, one can imagine that in physiological conditions, both a K⁺ and a Na⁺ ion could reside in the SF of G628S channels, and this could be possible because of the change in selectivity. The overall effect would be a lack of conductance. Alternatively, a serine residue in position 628 may change the structure of the outer part of the SF into a cage for this ion so that it becomes trapped there and does not allow sodium ions to bind and displace it.

Long-QT2 patients with mutations in the pore region of the hERG gene have been shown to be at greater risk for arrhythmia-related cardiac events. Most long-QT2 mutations are known to result in impaired synthesis or trafficking of channels (classes 1 and 2), whereas others result mostly in altered biophysical properties (class 3) (7,60). The N629D long-QT2 mutation (class 2) brought useful information about the SF, rarely studied, since most channels mutated in this area do not allow recording of ionic current. This substitution was first thought to result in a noninactivating channel with altered selectivity to potassium (13), although later investigations found that the actual consequence is a failure of trafficking (14). We confirm here that the G628S mutation is the only mutation found so far that is related to class 4 (altered or no permeability), and we would like to modify the classification proposed by Delisle et al. (60) and suggest that class 4 should then be defined as altered or no permeability under physiological conditions. The VCF technique has proved useful in investigating the activation kinetics of hERG channels (32). Here, it has allowed measurement of the gating kinetics of the nonconducting G628S mutant channel in K⁺-containing solutions. These data support the full functionality of a channel considered until now to be nonfunctional. This to our knowledge novel approach will certainly allow greater future insight into how mutations affect hERG channels.